UBC Faculty Research and Publications

Proceedings of the Second Trilateral Physician Workforce Conference, November 14-16, 1997, Vancouver,… Barer, Morris Lionel, 1951-; Wood, Laura 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-Barer_M_et_al_Physician_conference.pdf [ 8.14MB ]
Metadata
JSON: 52383-1.0075903.json
JSON-LD: 52383-1.0075903-ld.json
RDF/XML (Pretty): 52383-1.0075903-rdf.xml
RDF/JSON: 52383-1.0075903-rdf.json
Turtle: 52383-1.0075903-turtle.txt
N-Triples: 52383-1.0075903-rdf-ntriples.txt
Original Record: 52383-1.0075903-source.json
Full Text
52383-1.0075903-fulltext.txt
Citation
52383-1.0075903.ris

Full Text

--=-----Centre for Health Servicesand Policy ResearchTHE UNIVERSITY OF BRITISH COL UMBIAHealth Human Resources UnitResearch ReportsProceedings of the Second TrilateralPhysician Workforce ConferenceNovember 14-16, 1997VancouverDecember 1998HHRU98:4Proceedings of the Second Trilateral PhysicianWorkforce ConferenceNovember 14-16, 1997Vancouver, B.C.Editors: Morris L. Barer, MBA, PhDDirectorCentre for Health Services & Policy ResearchLaura Wood, MA, MHScResearch ManagerCentre for Health Services & Policy ResearchCentre for Health Services & Policy Research429 - 2194 Health Sciences MallVancouver, B.C.V6T 1Z3Health Human Resources UnitThe Health Human Resources Unit (HHRU) was established as a demonstration project by theBritish Columbia Ministry of Health in 1973. Since that time, the Unit has continued to be fundedon an ongoing basis (subject to annual review) as part of the Centre for Health Services and PolicyResearch. The Unit undertakes a series of research studies that are relevant to health humanresources management and to public policy decisions.The HHRU's research agenda is determined through extensive discussions of key currentissues and available resources with the senior staff of the Ministry of Health. Various health careprovider groups participate indirectly, through on-going formal and informal communications withMinistry of Health officials and with HHRU researchers.Three types of research are included in the Unit's research agenda. In conjunction withprofessional licensing bodies or associations, the HHRU maintains the Co-operative Health HumanResources Database. The Unit uses these data to produce regular status reports that provide abasis for in-depth studies and for health human resources planning. The Unit undertakes moredetailed analyses bearing on particular health human resources policy issues and assesses theimpact of specific policy measures, using secondary analyses of data from the Co-operativeDatabase, data from the administrative databases maintained under the HIDU, or primary datacollected through surveys. The HHRU also conducts specific projects pertaining to themanagement of health human resources at local, regional and provincial levels.Copies of studies and reports produced by the HHRU are available at no charge. (See final pagesof this document for a listing of HHRU studies and reports.)Health Human Resources UnitCentre for Health Services & Policy Research#429 - 2194 Health Sciences MallVancouver, B.C.V6T lZ3Tel: (604) 822 - 4810Fax: (604) 822 - 5690e-mail: hhru@chspr.ubc.caweb: www.chspr.ubc.caACKNOWLEDGEMENTSSpecial appreciation is extended to those who were involved in the planning andorganisation of the Second Invitational Trilateral Physician Workforce Conference, especially thePlanning Committee, Karen Cardiff, Mary Brunold, and Juliet Ho.Elements of these Proceedings are derived from notes taken by Jennifer Gait and Dr.William Webber. Kathy Campbell worked from less-than-adequate tapes to produce transcripts ofthe presentations by Bob Evans, Noralou Roos, Tony Mathie and Phil Lee. Juliet Ho and DougJameson patiently edited documents and retrieved tables and figures. Karen Cardiff co-ordinatedthe process of document preparation.Thanks are also extended to those who supported the Conference and the production of theseProceedings:B.C. Ministry of Health and Ministry Responsible for SeniorsCanadian Medical AssociationCollege of Family Physicians of CanadaWilliam A. WebberHealth CanadaJosiah Macy, Jr. FoundationRoyal College of Physicians and Surgeons of CanadaU.S. Bureau of Health ProfessionsU.K. Department of HealthTable of ContentsIntroductionM. L. Barer, L. Wood iDay OneOpening Plenary - Physician Resources Policy Challenges: Deja Vu All Over Again.R.G. Evans................................................................................................................... 1General Discussion 13Morning Panel - OverservicingUS If Doctors were Free, Would We Always Want More?David Goodman............................................................................................ 17UK Overservicing.Klim McPherson 38Canada What can administrative databases tell us about overservicing in Canada?Ben Chan 57DiscussantNoralou Roos 84General Discussion 95Afternoon Panel - Primary Care ReformUS Primary Care Reform in the US: The Best of Times and the Worst of TimesKevin Grumbacli 101UK Primary Care Reform in the UKAngela Coulter 121Canada Primary Health Care and Canadian Physicians c. 1997Michael Rachlis 143DiscussantTony Mathie 163General Discussion 172Evening Plenary - Medical Workforce Policy and Planning in AustraliaJohn Horvath 175General Discussion 193Day TwoMorning Panel - SubstitutionUS An Analysis of the Non-Physician Clinicians Who Provide "Physician Services".Richard Cooper, Craig L. Dietrich 197UK Labour Substitution in the UKSue Jenkins-Clarke and Roy Carr-Hill.: 221Canada "Substitution" For and Among Physicians: a Canadian ViewPeter Newbery 247General Discussion 258Closing Plenary - Perspectives on Overservicing, Substitution, and Primary Care Reform inCanada, the United Kingdom, and the United StatesDr. Phil Lee 265Conference Program 273List of Participants 277IntroductionThe First Trilateral Physician Workforce Conference involving the United States (US), UnitedKingdom (UK) and Canada was held in Washington DC in November 1996. The Conferenceobjective was to share experiences of the three countries with respect to physician workforceplanning and management, to identify degrees of commonality and differences in issues, solutions,and policies adopted by the three countries. The conference focused on physician supply andrequirements determination, physician immigration and emigration, geographic and socialdistribution, postgraduate medical education and residency training issues. Finally a panel ofexperts discussed drawing the link from data analysis to policy change. The Conference wasfunded by the Macy Foundation, US Bureau of Health Professions, Health Affairs and AMA.Canada hosted the Second Trilateral Physician Workforce Conference in Vancouver in November1997. The themes for the conference included issues related to 'overservicing', 'primary carereform' and 'substitution'. The themes were considered to reflect the common priority issues andconcerns confronting the three countries in the area of physician workforce planning. Physicianworkforce planning had again become a major issue, as some countries had begun to forecastphysician shortages in the next few decades. Canada is at the centre of this debate, as its physiciansupply-per-capita figures have begun to decline recently after over 40 continuous years of (at timesrapid) increase. Physician workforce planning and management requires the co-ordinatedinvolvement of a number of stakeholders, and the participants were chosen to represent thoseorganizations and interests that have to be part of the discussion to ensure continued dialogue andconcerted action in Canada.As with the previous meeting, the overarching objectives of this symposium were to:a) enhance understanding of the political and social forces bearing on physician resource policyin the participating countries;b) compare current policy approaches (and research methods and findings on which they maydraw, or which are indicative of their relative effectiveness), that show promise in each countryin addressing selected thematic issues;c) consider the potential for (and barriers to) international policy learning and cross-fertilisationin the physician resource policy area;d) continue the development of a network of academic and policy leaders in the three countrieswho can maintain channels of communication and policy learning, and promote same in theirrespective jurisdictions;e) produce a series of papers and discussions that could be disseminated through a number ofavenues, post-conference.This document has been put together in recognition of the last objective. In it, we present editedtranscriptions of the plenary speakers, the discussants, and summaries of the discussions, as wellas edited versions of the papers presented at the conference.The conference opened with a plenary on the relationship between cost constraints and workforceplanning which invited participants to ground their speculations in a consideration of theaccounting identity by which the inputs and outputs of the system were necessarily in balance, soiithat to change one aspect of the system was to have an impact throughout the system. Thediscussion which followed looked at the available strategies for workforce and budget management,the role of technology in affecting system costs and the relationship between political will and theimplementation of change based on research.The first session examined claims of, and evidence on, the provision by physicians of inappropriateor unnecessary care. The papers produced for this session explored the evidence on over-servicing,and the policy implications and approaches, in each of the three countries. The discussant thendemonstrated another approach to the problem of identifying overservicing, while the discussionwhich followed ranged around issues having to do with over- and under-servicing, the difficulty ofchanging physicians' patterns of practice, and the relationship between specialisation and over­servicmg.The second session explored developments in primary care organisation and financing in the threecountries, with the objective of sharing experiences regarding what was happening in each, whatseemed to work (and in meeting what objectives), what seemed not to be working, with what effectson costs, quality, patient satisfaction, etc., and what stood between intent and application injurisdictions (such as Canada) where very little has yet happened. The discussant gave his views onthe changes in primary care practice and service provision that he had seen during a long career asa primary care provider, and participants in the discussion that followed debated internationaldifferences in planning for and delivering primary care as well as the roles of medical educationand substitution in primary care.The evening plenary provided the participants with an overview of medical workforce planning inAustralia, and the next morning the conference went on to examine the evidence on, and political,legal, and other impediments to, substitution for and among physicians in the provision of clinicalcare. Nurse practitioners, physician assistants and midwives are much more widespread in theU.S. than in Canada, but in both countries, their scope of practice seems, at least in part, to be afunction of the number of physicians prepared to practice in the local area. In response to theU.K.'s new-found 'physician shortage', the issue of substitution is beginning to be explored. Butwhile this type of substitution (of non-physician personnel for physicians) is the most commonlystudied, there may be ample scope for a more generalized phenomenon of 'pushing down' therequisite skill-set in response to any given mix of clinical conditions. The extent of substitution, inthe end, seemed to have much to do with the available mix of personnel. The discussion whichfollowed touched on different professions' definitions of substitution and scope of practice, on therole of the organisational context of care in determining provider mix, as well as the impact ofbudget limitations and rising income expectations.The final plenary provided some commentary on the material presented, and sounded a warningnote about changes in medical service provision such as substitution, which might reduce thequality of care. The importance of evidence-based medicine was emphasised, as well as thechanging dominance of the medical profession in decision-making about access to care and theseemingly universal way in which costs seem to drive health policy.1Opening PlenaryPhysician Resources Policy Challenges: Deja Vu All Over Ag-iin,R.C. Evans23PHYSICIAN RESOURCES POLICY CHALLENGES: DEJA VU ALL OVER AGAINBOB EVANSLest people feel at all discriminated against because of the lighting in this room I want topoint out that this podium does not have a light socket, so that the only place I've been able to findin this building so far where I can actually read my notes is in the bathroom, and I thought thatmight get a little crowded. I brought my notes along, but I have no great confidence in being ableto see them. Whether this makes a difference to the presentation I' II leave you to judge. You'llsee. But enough with these vision jokes. The other thing about coming to a conference that isfundamentally about matters related to health, is that I've done a fairly poor job of looking aftermy own, so this talk will probably last about as long as my voice does, rather than necessarily aslong as is listed in the program.Morris mentioned some of the earlier publications in my official list. But in preparing forthis talk and this conference, and reading over the papers prepared for it, I was reminded of a stillearlier publication which I think is relevant to what I want to say. My first publication in healthcare, which is not on the vitae, was actually in the Toronto General Hospital nurses' yearbook of1963. I was just finishing up as an undergraduate at the time. Why was I publishing in theToronto General Hospital nurses' yearbook? Well, my fiancee was the editor, and so I learnedearly on, publish anywhere you can. The paper was a commentary on the so-called "CrispoReport". John Crispo was and is a distinguished labour economist in Toronto and he was juststarting his career then. His report was addressing, among other things, the perceived problems ofthe day in Ontario (this was in the early to mid-1960's) in nursing, and one of the principalproblems was of course, the desperate shortage of nurses. The Canadian hospital system wasgearing up and expanding at that point and the baby boom still had not quite collapsed (it collapsedbetween 1964 and 1966), but it took a while for people to get the message. And it was prettyobvious that a very large pool of trained nurses existed out in the general community. They hadgone through the training and worked for a year or two, perhaps, and then dropped out. This is along time ago. This is back when nursing was still the sort of thing that a respectable young ladymight get the training for, work a year or two, and then get out and get on with the more seriousbusiness of getting married and starting a family, and so on. Things change. The question at thetime was, could you use that pool of trained talent, could you draw that back into the field to dealwith the shortage?The answer that was being offered by the Registered Nurses Association of Ontario was,yes, and the best way to do that would be to raise nursing salaries. Now, even to an economicsundergraduate the thought rapidly occurred that if you raise nursing salaries, you will confersubstantial benefits on the nurses who are already in the workforce and are planning to remain inthe workforce. Whether this would have any impact at all on drawing people back in was whollyunclear, because no one really knew why they had left, or not even started practicing in the firstplace. So it is a policy recommendation which is at least questionable in its effect on the namedobjective, but will certainly have significant side benefits for people who are not supposed to be thetargets of the policy Of course, those people are the ones who recommended it.So that wasn't terribly mysterious, and it seemed like something that was worth writing alittle piece on, and it has its deserved obscurity, or so I thought, until I read the paper by Jenkins­Clark and Carr-Hill. Now you can see the Brits are really economizing all through the systemhere. You have four authors compressed into two. Lo and behold, I find that we've got a shortageof nurses in the UK. Now, the nature of this shortage, curiously enough, is simply that the number4of people employed in nursing seems to be falling. I would have thought that if the hospital sectorwere shrinking, that was probably a natural thing to happen- it would not necessarily represent ashortage. Anyway, the numbers are going down and something ought to be done about it. Whatdo we find? We find the professional organizations saying, yes, it's a serious problem here, peopleare dropping out of nursing, there's a pool of talent out there that we could draw on that's out inthe general community, if only we raised all our salaries. Apparently this happens.Then a little later on, the Ministry of Health suddenly tumbles to the fact that maybe this isnot the most efficient way of drawing people back into nursing, but maybe it is the most efficientway of rewarding the people who are already in nursing, and maybe that's why they reeommendedit. You will no doubt have noticed that this is the mid-1990's. Thirty years have gone by andwe're hearing about exactly the same thing. It's enough to make you feel young again. It'scertainly a comfort to know that the issues you thought you had something to say about thirty yearsago are still here and you can still say the same things. You don't actually have to learn anythingnew, and you still get invited to talk about it. One of my colleagues, formerly at McMaster, now inAlberta, David Feeny, said that health economics is a strange field, because you can benefit moreand derive more professional recognition from saying the obvious and repeating it than in any otherfield of economics. And the funny thing is, he's right.Now, apart from reminiscing, what is the point here? Well, there are a couple of points.One is the extraordinary persistence of issues in this field. Morris was kind enough to suggest thatmy publications have stood the test of time. I think all that says is that we've still the sameproblems out there, that to a very large extent we're dealing with exactly the same problems thatpeople were talking about in the 1960's, the 1970's, the 1980's. And we're not just talkingCanada here. There's an international process where very little changes in terms of the basicpolicy and planning issues facing health care managers in all developed countries. I'll try toexplore that a little more later. So persistence is, I think, one of the biggest things that you have towrestle with in this field. What makes us think that this conference is going to make muchdifference?The second point that I want to draw from this, and I think it tends to explain thepersistence, is that virtually everything you do, or try to do, or think about, or talk about, in thefield of whatever you want to call it, manpower, personnel, human resources (pick whatever nameyou prefer), is going to be tangled up in a conflict of objectives. It is going to involve multiple andoften directly conflicting objectives, and those will be strongly promoted right across the policyspectrum. We are never going to have the luxury of what Richard Nelson calls the steersmanmetaphor, where we have a well-defined problem, we all get together, we figure out what theproblem is, we figure out what the appropriate solution is, and we do it. The steersman metaphorsays we're all in the same boat, and we're just trying to figure out where best to direct the boat forour mutual benefit. The reality of the health care situation is that we're not anything like in thesame boat, there are highly conflicting objectives, and they are the same conflicting objectives theyalways have been. That is why it is so difficult to make much in the way of progress.Now, let me make that concrete as a pair of questions that it seems to me you have to askabout any proposal in the health human resources field. The first one is, whose jobs are you goingto "downsize"? In the case of physicians, you can either try to get rid of a job (although if they'reindependent fee-for-service practitioners, it's not immediately clear how) or you can simply reducethe rate of pay for that job, so that when we think about it, some of the proposals that are currentlyon the table represent ways not necessarily of putting doctors out of the field, but of just payingthem less. But in the case of nurses, and hospital workers generally, of course, it's more typicalthat we just get rid of their jobs. So, if you're going to do something new and different, it comesdown to whose job are you going to zap, or if the answer to that is nobody's, then we're left with5the second question - where are you going to get the new money? One or the other, maybe both,but you can't get away from that pair of questions.You can get away from them rhetorically, you can keep people confused, and I think that'spart of what was going on with the example I took from Jenkins-Clark and Carr-Hill's paper. (Bythe way, it's not a major part of their paper, it's just one that I jumped on because it sort ofrang anold bell for me.) The argument that was being made was that turnover is so expensive that if youpaid nurses more you could reduce the rate of turnover and the savings from lower turnover wouldmore than pay for the additional salaries for the nurses. And if you believe that, we have some verynice bridges here in Vancouver in which we'd like to interest you. They need a bit of repair, butonce you repair them, they'll be worth so much more. If you take it seriously, it immediately says,or should immediately lead you to ask, "Suppose you increase the pay for the working nurses,whose income falls?" Unless we're talking about putting more money in total into health care,somebody's income has got to fall because every dollar of income is a dollar of expenditure.Somebody's income has got to fall in this system if you're going to stabilize or reduce overallcosts. Well, maybe it's the incomes of the administration, the human resources department. Lowerturnover, you don't need to have as many people in administration, and you can fire some of them.That's always a good idea. Everybody in health care is in favour of getting rid of administratorsand substituting them with people who actually provide health care. Maybe you could do that.But of course those jobs weren't specifically identified by the advocates of the pay increase fornurses. I don't think there was any suggestion that you knew whose jobs were going to disappear,that you could say with conviction, these jobs will be gone.Alternatively, one might argue that because we don't have to be constantly training newpeople and having them learn the ropes and so on, we can get by with a lower overall nursingestablishment. In other words, by having people in their jobs longer, we will actually need fewernursing jobs. Now I'm pretty sure that's not what the professional organizations had in mind,fewer but more highly paid jobs for nurses, and enough fewer jobs that you actually save money intotal. But that would be another way of doing it. I suspect what they really had in mind, to theextent that they had anything specific in mind, was, if we don't have a lot of turnover, people willbe more effective in their jobs, we won't have to have new nurses constantly learning the ropes,and we will get more and better output from the same staff. OK, so you're telling us aboutimproved productivity and so on, along with perhaps higher costs? But originally you said thiswas all going to save money. Yeah, but we lied. You know we lied; you shouldn't have believed usin the first place.So what I want to emphasize is that the inevitable relationship between the number ofpersonnel that you employ and the level of income at which you employ them overrides virtuallyeverything that you tty to do in the field of health manpower. Where you sit on this tends to setyour objective, and it's impossible to aligu the objectives of all the interests affected by policychange.I don't want particularly to pick on nurses, because we'll see this coming up in a numberof other areas as well. Yesterday our Centre held a conference that some of you were at, righthere in this hotel, sub-titled "New Opportunities For Old Mistakes". That conference was focusingon Canadian physician supply, and observing that the forty year increase in doctors per capita inCanada has leveled out recently and may in fact now be turning down. And the immediate responseis, we've got to start training a whole lot more new doctors, rather than saying, maybe this is anopportunity to do other kinds of things, to take up opportunities we let slip away 30 years ago.In the discussion around the conference papers, one of the themes that came out toward theend was, yes, but we owe the children or the parents of this country the opportunity to have theirkids get into medical school. This was put with great passion from the floor, that it was grosslyinequitable and improper that people in some provinces of Canada have a lower chance of getting6into medical school because there was a lower ratio of places per capita than in other provinces.There was even a suggestion that this might violate the Canadian Charter of Rights and Freedoms,that there is this fundamental human right to go to medical school and that this had to be equalizedacross the country.Well, at this point what we have again is the proverbial tail wagging the dog. Rather thanthinking about manpower policy or manpower strategies as a way of ensuring services for apopulation, we've got them presented as ajob creation program for the upper middle class. Nowthat's always popular; those sorts of policies derive a great deal of political support. The point Iwant to emphasize is that we have as a significant consideration (because it has been a significantconsideration over the years on decisions about medical school sizes in Canada, and it's comingright back again from the medical schools themselves), this notion that we need to expandthroughput in order to provide employment opportunities, not only for the people who want to betrained as physicians, but also to maintain the employment opportunities for people who work inhealth sciences centres, which is also extremely important. The whole question of whether youreally needed the extra doctors drifts off the radar screen, kind of gets ignored, or suppressed - or,let's see if we can avoid thinking about it entirely.Now, I can formalize this set of questions with the construction of an accounting identity,and I think it's extremely important to recognize that it is an identity. This is not an economictheory. It is not subject to refutation. It can't be wrong. It can be irrelevant, it can be unhelpful,but it can't be wrong. That's sort of like accountants, you know. You can make mechanical errors,yes, obviously you can add things up wrong, make a mathematical mistake. But assuming that youhaven't done that, it is the case that in any country's health care system there will be an equalitybetween the total quantity of revenues that are raised to pay for that system, the total quantity ofexpenditures on the products of that system (the products are the goods and services), and the totalquantities of incomes that are earned within that system. Those things all have to be equal.Figure 1REVENUES ASSEMBLEDFrom taxes, user charges and private insurance premiums[T + C + R]must equalEXPENDITURE ON HEALTH CAREQuantities provided multiplied by prices[P x Q]must equalINCOMES EARNED FROM HEALTH CAREFactor inputs multiplied by input prices[W xZ]7We collect the revenues in three different ways. The predominant form of revenuecollection in all modern health care systems, including, interestingly enough, the United States, istaxation. Now, the United States data are not presented that way, because it tums out in fact thatsome of the public subsidy goes through the back door as tax expenditures, but the reality is thatin all modern systems the major share of revenue comes through taxation, with secondary andtertiary sources being private insurance premia and user charges. Those funds then get distributedand become the total expenditures on health care, which will be a function both of the quantities ofservices provided and the prices that are charged (implicitly or explicitly) for those services. Andin tum, that money will flow into the pockets of the people who make their livings from this sector.That will be a product of the total numbers of people and levels of effort that are provided, the "Z",and the wage rate or income, "W", so it, too, involves both 'prices' and quantities. Crudely, it'sthe number of jobs and the rates of pay, although it takes a lot more words to 'get it pinned downprecisely.That leads us to the balancing equation, which has to hold. So again, to reiterate,anything you do at some point in the 'manpower' system has to show up somewhere in thatequation, and indeed, it has to show up in at least two places in the equation. The accountants arekind of like the ecologists who say that it is impossible to do only one thing - if you are going to doanything to one part of a balancing relationship, it has to tnrn up somewhere else as well. Most ofthe proposals that we look at for changes in health care systems generally, not merely those relatedto personnel, can be traced through that equation. When you start talking about manpowersubstitution in an equation like this, what you are saying is that we could reduce the number ofhigh priced people in this system, particularly "Z"s, get rid of them, send them to the United States,export them somewhere, give them one way tickets, and replace them with less costly "Z"s, nursepractitioners let's say, people who could do the same jobs for lower wages. In this example, wemight have the same number of people in the system, but the overall average rate of pay for thosepeople will go down, because we will use less expensive substitutes. Now, if the "w" goes downand the "Z" stays constant, and if the new "Z"s are equally effective, then the "Q", the quantity ofservices will not change, but the average price will go down.Needless to say, that doesn't happen, or hasn't happened. While you are steadilyincreasing the supply of physicians, as we have done in Canada for forty years, until about1989/90, as you keep those "Z"s rising, unless you can figure out a way to keep cuttingphysicians' incomes at the same time, the increased costs have to be pushed forward into the rest ofthe system, and the way they are pushed forward has been through increasing rates of "T". Thelevel of taxation required to support this part of the system has continued to climb, and therevenues have had to increase in order to balance the numbers of income recipients "Z" coming in.One way you could deal with that, perhaps, is to cut the system somewhere else, or cutback elsewhere in the public sector, because this is a global relationship. So, for example, ifyou're going to increase your expenditures on physicians and on prescription drugs (they've alsobeen increasing quite rapidly, in which case the money goes to shareholders and researchers andpeople like that; it still fits into this equation, just a different group of people), maybe you couldpull the money out of the hospital system. And in fact we've been doing that. Acute care hospitaluse in Canada has been falling steadily since the 1960's. It has been masked or cushioned byconcurrent increases in capacity in long term care. So what that amounts to is eliminating peopleand jobs in the hospital sector to finance an increasing physician supply. You can do that.Needless to say, the nurses are not going to be particularly happy about that. Michael Rachlismade the point very strongly at yesterday's conference, and may again today, that what's beengoing on in recent years in Ontario has been a collapse of efforts to contain physician incomes,meaning that for physicians the wages have gone up, the "w" has risen, and that too has beenoffset, or is being offset, by firing nurses. So, in aggregate the "W"s for doctors go up, the "Z" for8doctors has gone up, the "2" for nurses goes down, you try to contain overall costs, but youredistribute within the health care sector, which brings us back to the initial question-whose jobare you going to downsize? If you want to add something new, who are you going to zap? And ifthe answer is "We're not going to zap anybody," then the next question is, "Where are you going toget the new money?". In the Ontario case, they eliminated a bunch of jobs and used the moneysaved from that to increase the rates of pay of another group of people in the same sector.Generally speaking worldwide, certainly in the Canadian setting, what seems to behappening is that the pharmaceutical sector is doing extremely well and the physicians seem to beholding their own (although there is a lot of unhappiness out there), and the hospital sector is beingvery radically downsized. The questions that are addressed at this conference about primary carereform, about over-servicing and about substitution, then have to fit into that reality, and into thisequation. You cannot avoid it; if you try to, it will find you. If you're concerned about over­servicing, what you're really saying is that the level and mix of quantities provided are too high;too many of the wrong things are being provided. The policy objective is to reduce the level of "Q".If you do that, and you keep the "P" constant, you will indeed save some money, but again,someone's income has to suffer. Either someone has to make less money or there have to be fewer"Z'ts as a byproduct. And if the response is, we'll do less of those things, but more of somethingelse, all you've done is change the mix of the "Q", you haven't changed the total, and you haven'tactually dealt with your cost issue (if cost is also an issue), but you have solved the threat to theincomes issue.You can take any policy and run it through this equation and expose similar trade-offs,The issues that we are addressing in this conference actually have a very long history, and theyarise, I believe, [rom the conflict between two different perceptions of the causal sequenceunderlying this identity. Accounting identities can be fascinating, because they tell you about theconstraints, they tell you about what must happen. If you do this, something else must adjust overthere, and you can interpret particular policies within that framework. They don't tell you anythingabout causality, and they don't tell you anything about the overall objectives or desirability ofparticular changes. They provide a framework within which different objectives can be reflected,but do not get us around the fundamental conflict among interpretations of what the causalsequence is that connects up the variables in the equation.Let me suggest what I think is the traditional viewpoint of how health care systems orhealth care management is supposed to work. This is pretty obvious stuff. Ideally, in principle,we'd like to believe that there is some kind of epidemiological relationship out of which emergesthe health needs of the population. We then have a clinical system that interprets those needs,figures out what the appropriate interventions are, what should be done in response to those healthneeds, and then chooses the appropriate mix of personnel and other technologies to meet them.Fignre 2POPULATIONHEALTH NEEDSAPPROPRIATE SERVICESHEALTH NEEDSAPPROPRIATE SERVICESREQUIRED PERSONNELWithin that framework, you can think of the first arrow as representing epidemiology, thesecond as having to do with clinical effectiveness, and the third as having to do with technicalefficiency. When we raise issues of over-servicing we're really asking questions about the second9arrow: are the services actually being generated by the needs or are they inappropriate for theneeds? When we raise questions of substitution, we're asking about the third arrow: are thepersonnel employed the most cost-effective, are they the most appropriate given the kinds ofservices that need to be provided to meet the needs? But this sequence essentially says, left alone,this is how a health care system operates, or should operate, or the logical way in which we wouldlike it to operate.In fact, the topics of and the papers for this conference suggest a strong counter-current, asit were, a belief that the system actually functions rather differently, that the dynamics are not at alllike this. Rather, if you find out what personnel and technologies are around (one of the papersmakes specific reference to the fact that the generation of new specialties in health care has to someextent been dictated by the generation of new technologies and new equipment, so that once youhave an ophthalmoscope you can have ophthalmologists, but prior to that time you can't), youravailable personnel and technology define the kinds of services that will get provided. If you havea hammer, everything looks like a nail. The services provided will then be used to infer needs, sothat if the services are being provided they must have been needed, and since all the services thatare now being provided meet needs, and since we can always identify some unmet needs, theanswer is that we always need more personnel.Figure 3PERSONNEL AVAILABLE ...SERVICE USE ..."UNMET NEEDS" ...SERVICE PATTERNSINFERRED NEEDSPERSONNEL SHORTAGEIn fact, this sequence was used to guide physician resources policy in Canada for most ofthe period up to about 1990. There is a classic review of the succession of exercises in physicianresources planning in Canada (Lomas, Barer and Stoddart's review of Ontario's McDonaldReport), which concluded that no more scarce research dollars should be wasted on manpowerplanning exercises of the type that begins from the assumption that every service provided isneeded, adds to that the observation that there is always some unmet needs and concludes that wetherefore need more personnel, irrespective of how many we have! So while this may seem like acynical alternative view of system dynamics, in fact there is a good empirical base for this sort ofrevised sequence of causality.Now, 1guess what is problematic about this, coming back to my first point, is that none ofthis is terribly new. This revised sequence, in John Galbraith's terms, has been around for a verylong time. It's been around for most of the time that I've been in this field, but knowing it's therehas not had much impact. We have until very recently continued to base our actual policy on thefirst causal view, while our research keeps coming back and telling us that the second is operative.As we saw at yesterday's conference, Charles Wright posed a question about some pretty powerfulindications of inappropriate patterns of care (and if you had quite strong evidence that patterns ofcare were inappropriate and that over-servicing was taking place, surely that ought to give pause tothose starting to talk about trying to train more physicians); the immediate response from a numberof members of the audience was completely to ignore that question and go on to talk about thedesperate need to provide more spaces so that people who want to become doctors can becomedoctors. It was quite a bizarre performance.10But what it did was it illustrated the conflict between the research that tends to support thisframework and the kind of continuing power of the other viewpoint in guiding policy. Coming backto the papers for this conference, we have some very interesting papers on over-servicing that tellus in more detail, and with more sophistication and power, things that we've been hearing for thirtyyears about practice variations. Nothing much new there. We have some very interesting newstuff on substitution of personnel, particularly nurse practitioners for doctors. Again, thatliterature goes back at least to the early 1970's. Nothing much new there. Peter Newbery's paperis saying that it is time to stop wasting our time talking about whether nurse practitioners can do anumber of services at the same quality or level of general practitioners. This has been conclusivelydemonstrated. Let's stop having that conversation. It's time to move on from there. But of coursethat conversation has continued for twenty-five years, because maintaining doubt is part of theprocess of blocking those changes from taking place.That, 1 think, then leads into where I want to finish up. If what we have here is a conflict ofobjectives and a conflict between policy and evidence, and I think we do (and that it's gone on fora long time), then the critical question is: what kinds of strategies do seem to be effective inbreaking that deadlock? If you go back to the three-part identity (Figure 1), what's really beendriving change in this sector is fiscal pressure. I don't think there's any doubt about it. But if youlook at that identity and say, why are things changing, why is there pressure to change, the answeris, because we're having trouble passing everything that we do back into increases in "T". "T" hasstopped increasing, and in some countries it has even fallen, and that is then washing across thewhole system.The immediate reaction by a number of people in health care has been to say, in that casewe need to raise "R" and "C". The obvious answer to government's failure to fund the system atthe level that we want is to see if we can't get the money from the patients. So we see theresurgence in Canada of a demand for various forms of private payment, being driven pretty muchentirely by the pressure to try to keep this whole game of expansion going. On the other side of thepolicy field, people are starting to look more seriously at more effective mechanisms for trying tochange servicing patterns and the distribution and use of personnel, and at the same time, or as aresult, mechanisms for controlling costs.I think we can categorize under three heads the strategies that one can employ. The firstis, just plain squeeze. Just keep cutting the budgets, or holding the budgets tight, restrict the supplyof different kinds of personnel, don't permit as many people as want to become specialists, aswe've done in Canada, as they do in Britain, and as they don't do in the US. Interestingly, youdon't do it in the US, they don't do it in Sweden, they don't do it in Greece - the "oversupply ofspecialists" problem is not specific to the payment system. It is a general problem, I think, ofhealth care systems if you don't impose a specific restraint on supply. Okay, so you can use bluntinstruments. You just say there just isn't going to he any more budget. There aren't going to beany more resident slots. This is how we handle it in Canada. That raises a number of problems,with which you will already be familiar.A second approach is reflected in the long-standing view that somehow if you change thepattern of organization and incentives within the system, things will be better, that there's a kind ofself-regulating approach whereby the system will somehow automatically move towards betterallocation of personnel and better choice of services, if we could only change the~tructure to makethis happen. The economists, I think, have a lot to answer for here, as do the 1St -centuryphilosophers. The view is that it would just be a whole lot hetter if the system were structured toorganize itself, rather than our having to keep beating it over the head with large bluntinstruments. I'rn not sure that's true. One of the very real challenges for a conference like this isto confront and interpret the evidence.IILet me offer an example. Back in the 1960's, everybody got excited about pre-paid grouppractices. Folks got excited in both directions. Some people thought they were the invention of thedevil. Once Paul Ellwood relabeled them Health Maintenance Organizations in the United States,that is, recast them as organizations competing in private markets, everybody said they werewonderful. The big evidence was that people who were enrolled in capitated, group practice-typeorganizations didn't use as much hospital care, and there was way too much hospital care beingused, everybody agreed on that. So, the obvious answer was to try to expand pre-paid grouppracticelHMO's/managed care, reduce hospital use, which would lead to more effective, moreappropriate care and lower costs. Good, wonderful stuff.Now in Canada, we did none of that. We talked about it a lot, and here, too, the talk goesback into the 1960's. The difference is that nothing happened, and our hospital use has droppedlike a stone anyway. Acute care hospital use in BC is down now to about one-third of where it wasin the 1960's. So we've had all the effects that were to be attributable to the U.S.-style systemrestructuring, but it's emerged from plain old "cut, freeze and squeeze". And it worked.Taking the major restructuring route in the U.S., you get the same reductions in hospitaluse, but you get a whole lot of micro-management laid down as well. The U.S. took a differentroute and wound up paying for a whole lot more managers. So, in terms of our equation, in theU.S. the numbers of people working in hospitals are down, but the numbers of people working ininsurance companies and managed care companies, or as consultants and so on, keep going up.The alternative to the blunt instrument, that of trying to bring in external management to controlthe system, seems finally to be having some results, but the net result is a lot more externalmanagers to pay, and there is no real sign yet of success on the cost control front. And it's not atall clear that you're farther ahead paying management consultants than paying nurses, certainly notif you're a patient.In the UK, you've gone the other route with GP fundholding, not only bringing in someexternal managers, but also trying to change the incentives so that the practitioners themselves willchange what they do. That seems to have had some effect, all right. But again, the net effectwithin the UK seems to be perceived shortages of clinical capacity, while costs are actually goingup. Since the Thatcher reforms, the UK system has become more expensive and it seems to havedone so because of all this extra management - different mechanisms, similar result to the U.S.So, what I am left with is this sort of broad brush impression that the history of the last tenyears or so is telling us that blunt instruments seem to get the results that we say we want, at leastin terms of cost control, and in terms of crude measures of appropriateness, such as reducinghospital utilization. Whether we actually get more effective patterns of care is a much morecomplex question, to which we don't really know the answer. We have certainly not achieved thepersonnel substitution that we wanted, but at least Canada has managed some changes in the rightdirection. Where we have either tried to overlay external management, or change the incentives toget more internal management, we have been much less successful.I think the difficulty is with what we might call the 'gatekeeper' concept, because itembodies two different concepts. If I could use the analogy of a concert hall, one notion is that wewant the general practitioner to serve as an issuer of tickets to the hall. Right now in Canada andthe UK s/he does serve as an issuer of tickets to the hall, but s/he's just giving them away. In theUS, you don't need a ticket to go into the hall. In the UK system, the change has been to say thatthe general practitioner can still give the tickets away, but the general practitioner will have to payfor them themselves, and maybe that will discourage giving away tickets. Inside the hall, a bunchof specialists, each organ, not just organs, sometimes trombones, stuff like that, each virtuosospecialist is doing his or her own thing. Nobody is conducting the orchestra. Sometimes groups ofspecialists get together and have trios and quartets and coordinate their services in various ways,but there isn't any orchestra and there isn't any score. It seems that we have moved away from12wanting the general practitioner to stand outside and restrict access to those services, toward anotion of having the gatekeeper also be the orchestra conductor.Now, I don't think there's any chance of that happening. I don't think the GP is ever goingto be the orchestra conductor, because the dynamics of information within the medical sector arenot going to let that happen. The whole principle of medical care is asymmetry of information. Iknow more than you do, and specialists know more than GP's about the things that specialists do,or at least that's the official doctrine. So, I don't see that that orchestra conductor role is actuallybeing taken on in any system, and I don't think it's realistic to expect it. The best you can hope foris that the GP is going to be a little more restrictive in handing out tickets.The problem you have there is that if at the same time you keep increasing the number ofspecialists who set up in the hall, they are going to find, as they are doing in the States, political orother mechanisms for preventing the GP's from playing that role. Once again, we are trying toimprove the quality of management of the system within the context of an environment in whichthere is conflict between the interests of the specialists, the interests of the GP's, and the interestsof the payers themselves. If you create conflict in an attempt to control costs, you're not likely atthe same time to get coordinated care. It's just not going to happen.And so we come full circle, back to saying what you really need are the blunt earlycontrols (e.g. saying we're just not going to train as many of a particular kind of health careprofessional). You've just got to control the numbers, and that's the key factor. After that, theremayor may not be ways of improving coordination, but it will require a dynamite set of tools.Again, this is pretty speculative, but it seems to me to be something that people ought to beworrying about in this conference. Primary care reform is not going to lead to GP's as orchestraconductors.A good part of primary care reform in Canada was in fact historically motivated by thedesire, not so much to improve the quality of mainstream health care, but to improve access toservices for underserved populations and to provide a wider array of alternative styles of practice,and some of that has happened. The interest here was not to change the mainstream structure ofmedicine. So it's not surprising that not much has happened in Canada. Now we're finding thatthe objectives of primary care reform have shifted. Michael Rachlis has documented this verynicely; the people who were, to a large extent, interested in doing something different, for differentpopulations, are now asking whether these same techniques could not be used to change the waymainstream medicine is structured.There was one line in Michael's paper that was very important, where he said there reallyisn't much of a political constituency for primary care reform. That's right, because the pressurefor doing it on a large scale is coming from people who think this would be a really neat way ofcontrolling overall costs, and maybe secondarily improving coordination of care and so on. Wekeep hearing those claims from the commissions, as Michael pointed out. It's not obvious to methat that is in fact the mechanism for doing it. I guess I used to believe it was, and I still want tobelieve it is, but that is something this group might want to think about, whether changing the finestructure of primary care is actually going to do very much to shift the patterns of service at thenext level up. Yes, we've seen the hospitals downsize, and we've seen all that change take place,but it really hasn't resulted from fundamental restructuring, and I'm not sure that we've reallyfound out or worked out the really good alternatives to plain old "slug the budget". That is kind ofdepressing, particularly for economists who believe in incentives, and I think it's also depressingfor folks who believe in rationality generally, but it seems to me that this is the challenge to befaced at the moment.13General Discussion following Bob Evans' plenary:It was stated that Bob had only talked about one strategy, the squeeze strategy. Bobresponded that he actually mentioned three. With the first, the squeeze strategy, he had talkedabout restricting the supply of personnel. An example would be to have no further residencyslots for physician training. Another example would be to say, "Do what you want" tophysicians, but to decrease the number of beds available at hospitals. The second would be theinternal management strategy, which would be to change the incentives for the delivery ofphysician services. The third was the external management strategy which would entailcontracting with a managed care organisation or a consultant or an HMO to come and managecare in the province. Internal management is philosophically attractive. External management isimitative of the Ameli cans and expensive. The squeeze method is at least as effective as theothers.It was noted that in Bob's equation, Z represented a number of personnel and not justphysicians. It didn't seem to make sense to reduce the number of physicians entering theprofession, as this was likely to be a strategy which entailed fighting more battles. Bob'sresponse was that it is essential to do this with physicians, as nurses are easier to get lid of oncetrained. Overall, it is better to start on the supply side. The problem with physicians is that it isdifficult to reduce the supply through training fewer because the trainers are powerful interests.For example, dental schools have downsized, but although medical associations have long beenuneasy over the entry size to medical schools, they were not downsized until around 1992because of the opposition of the medical faculties. That there was to be a problem with anemerging surplus of physicians was seen as early as 1975, when the immigration laws werechanged to reduce the inflow of foreign physicians, but educators fought decreased class size for20 years with questionable research.A question was asked about health human resources planning and the appearance ofconflict of interest for nurses when claims were made that supply should be increased. It wascommented that outcomes research should help to decrease the conflict of interest by sheddinglight on whether the issues were about job protection or the health of the population (health carequality assurance). Bob's response was that outcomes research wouldn't help because it's not aquestion of either job protection or population health. There always is job protection, but it mayhave a population health perspective. Nurses are involved in job protection but have alsoserious questions about outcomes that need to be answered. He is in favour of outcomesresearch, but it is true that good evidence of interventions with a positive impact will get lost orgo unnoticed, if it is politically not palatable. If the outcome is positive and those who do it arepowerful, the idea will be implemented. If the outcome is neutral and the research waspromoted by the powerful, the idea will still be implemented. If the outcome is damaging, theresearch may even stop, except where drug companies are concerned. Good evidence of nogain with a particular intervention will not reduce the intervention if it is a politically potentintervention. Outcomes research tends not to affect policy. The problem has to do with who isthe client for the outcomes research? Who will apply it?14It was commented that Bob had not discussed the impact of technology as a driver ofcosts. This could be important as Canada has had policies in place to restrict technology todecrease cost, and the UK had gone even further. How does technology affect P and Q in theequation - it would come in at P? The response was that technology is not a driver. We aregetting more technology but it is not driving the system per se, it is the range of applicationwhich drives the system. The fact that technology increases world wide but the costs aredifferent in different jurisdictions shows that it does not drive costs - a logical mistake by Fuchsand Newhouse. What does drive costs are the characteristics of systems that lead to differentuptake of technologies. The key problems with the technology is that it's front-end-loaded; itcan cost millions to bring to the market, so therefore companies need to sell it so they force it onus. Some countries are more successful than others in resisting but the intemational grouping ofcompanies means this is going to get more difficult. For example, the National Institute ofHealth budget supports technology advancement. This is the American disease - to convert apolitical question into a technological/scientific question so that it is believed that the value ofhuman life can be technically determined. The key issue is the structural response.15Morning PanelOverservicing16Ijjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjj17IfDoctors Were Free, Would We Always Want More?David Goodman, MD MS*Elliott Fisher, MD MPH**The Center for the Evaluative Clinical SciencesThe Department of Community and Family MedicineThe Department of PediatricsThe Department of MedicineDartmouth Medical SchoolHanover, NH 03755The Department of Veteran Affairs Medical CenterWhite River Junction, VT 05001Supported by a grant from The Robert Wood Johnson FoundationPresented, in part, at the Second Invitational Trilateral Physician Workforce Conference,Vancouver, British Columbia, November 14,1997.18It is hard to resist more of an apparently good thing, whatever the cost. The physicianworkforce in the United States is a case in point. Compared to other large developedcountries, the U.S. is notable for it's high number of physicians per capita, particularlyspecialists. 1 At first glance, this abundance of doctors should be no more of a societalproblem that a bumper crop of corn. Even if a high physician supply is partly responsiblefor the equally high per capita medical care costs in the U.S., what better way for anaffluent nation to spend its loose change? The public has a thirst for more medical care toaddress an ever greater range of perceived human ills. The underlying assumption is thatmore physicians are always better for a population - greater access to preventive services,a quicker response to acute illness, and more consistent care of chronic illness. But thisassertion remains largely unstudied. Consider two fictional metropolises in the midwest:Moredoc and Fewerdoc are nearly identical twin cities divided by a liverand a twist of fate. Their health care economies are predominately fee-for­service with one important difference: Ten years ago, Moredoc wasbequeathed a fantastically large legacy with the stipulation that it only beused to encourage the recruitment and retention of more physicians. Newphysicians were provided with relocation bonuses and then guaranteed anincome greater than the national median of their specialty. Physicianscould not charge for their services. Physician eligibility for the programcontinued as long as they were fully engaged in the usual activities of theirspecialty. Today the citizens of Moredoc consider themselves truly blessedwith an abundant supply of physicians.On the other side of the river are the unhappy residents of Fewerdoc withhalf as many physicians per capita. Although it once considered itself aperfectly good place to live and work, the Fewerdoc population now seesitself as medically disenfranchised.Stranger than fiction is the actual variation in physician supply across cities andregions in the U.S. The per capita number of clinically active physicians ranges,not two-fold, but almost four-fold.2 Without obvious benefits, San Francisco hasalmost twice as many physicians as Wichita KS (282 vs. 148 per 100,000). Threeother regions have even more physicians per capita: White Plains, NY (334 per100,000), Hackensack, NJ (300 per 100,000), and Royal Oaks, MI (289 per100,000). Nor is Wichita unique for its modest supply. Other regions withsimilarly sized workforces include Las Vegas NV (149 per 100,000), Dayton OH(148 per 100,000), and Winston-Salem NC (147 per 100,000). Putting costsaside, do we know which region is better off?19Understanding the consequences of increasing physician supply for population outcomes iscritical if we are to know how many physicians we need or want. Many fear that recentefforts to constrain the growth in physician supply in an effort to curb costs will lead torationing medical care and will harm patients.L 4What if a higher per capita workforce, however, causes more harm than good? Then,constraining the supply of the specialists is not rationing but improving medical care.This paper will tackle the subject of physician supply and population effects in three parts.First, we will briefly discuss a theoretical framework for approaching the question, "Ifdoctors were free, would we always want more?" Then, we will present the sparseliterature that has tried to directly or indirectly answer this question. Finally and mostimportantly, we will discuss the implications of these studies that will guide futureresearch directed toward understanding the marginal benefits of a greater physiciansupply.THE QUESTION'S THEORETICAL FRAMEWORKThe relationship between physician supply and net patient benefits is almost certainlydescribed by the law of diminishing returns. (Figure 1) The curve is initially steep as thepopulation receives basic preventive measures and as those with untreated disease receivecare. In this region of the curve, a greater physician supply should lead to a population'smarkedly improved functional status and greater satisfaction with health and medical care.As the physician supply increases, however, the population benefits may be less apparent.In the relatively flat portion of the curve, increased supply leads to minimal or noimprovements in health. Instead of helping those still in need of basic medical care,utilization may increase for those already receiving ample services. If utilization continuesto increase with physician supply, then benefits may eventually be outweighed by theharms of needless medical care - too many screening tests of dubious benefit.' andexcessive surgery for minor ills.6While the curve in Figure 1 represents the general case, it should be noted that the shapeand slope of the curve will differ for different specialties and populations. The change innet benefits (or the slope of the curve) should be greater for indigent populations, as wellas for the elderly, for pregnant women, or for the chronically ill. Across similar20populations, the incremental benefits of a greater dermatologist supply are likely to differfrom a greater supply of cardiovascular surgeons.Characterizing the physician supply-benefit curve for specific specialties and populationswill be difficult because of the other variables that can alter the effectiveness of thephysician supply or the health outcomes of the population under study. (Figure 2)Potential confounders of measures of physician supply include the organization of care,the productivity of the physicians and the use of mid-level providers. Potentialconfounders of patient utilization and outcomes include both organizational and patientcharacteristics.WHAT DO PAST STUDIES TELL US ABOUT THE POPULATION EFFECTS OF VARYINGPHYSICIAN SUPPLY?In the past twenty years, only three studies have specifically examined populationoutcomes associated with varying physician supply. (Table I) Nesbitt? and AllenS lookedat neonatal outcomes of pregnant women and found that a greater supply of obstetricalservices was associated with lower rates of prematurity and neonatal deaths. Althoughthese findings may represent a causal relationship, differences in population characteristicsmay have confounded the results of the study. Krakauer, for example, did not detectdifferences in Medicare mortality by physician supply across Health Care Service Areas.l'The limited scope of these studies is primarily useful in understanding the magnitude offuture research challenges.Other investigators have examined the relationship between physician supply andutilization and have found conflicting results. Welch and his colleagues, examiningMedicare utilization in MSAs, found that areas with a higher proportion of primary carephysicians had lower admission and expenditures rates for outpatient services. 10 Thenumber of "patient care physicians" per capita appeared to have no influence. Markobserved lower Part B reimbursements in urban counties with high supplies of familypractitioners and general internists. I I Escarce concluded that greater surgery rates in highsupply areas reflected increased rates of initial contact with surgeons. 12 Krakauer usedHealth Care Service Areas as devised by the National Center for Health Statistics (Countyaggregates based upon Medicare utilization) and found lower Medicare admission ratesfor Ambulatory Care Sensitive Conditions in areas with a high physician supply, but onlyfor those areas already falling within the lowest supply decile.V In the most recent study,21Grumbach and colleagues studied office visit rates and use of preventative services inurban California Primary Care Service Areas. No association was noted between primaryphysician supply and utilization by those age 18-64 years.!3Although definitive studies linking physician supply and outcomes are lacking, a growingbody of literature suggests that more medical care is not necessarily better for patients. Ina recent VA Cooperative Study,14 for example, patients with diabetes, chronic obstructivepulmonary disease and congestive heart failure were randomized to either usual care orearly and frequent primary care. Yet the rate of readmission, the study's primary outcome,was significantly higher for recipients of shorter revisit intervals. Although not reachingstatistical significance, mortality was higher in the group who received the primary careintervention. Several studies have also failed to find better outcomes with higher rates ofinvasive cardiovascular interventions.!5-18 One recent study, however, found that thehigher rates of intervention in Texas were associated with both higher mortality and worsefunctional outcomes than in New York. 19 These findings could plausibly fit the law ofdiminishing marginal returns: some populations are on the flat of the curve, but others areexperiencing net harm from rates of intervention where the risks exceed the benefits.LESSONS FROM THE PAST THAT WILL ENHANCE FUTURE RESEARCHStudies that have examined the relationship between physician supply and outcomes pointto the difficulties in measuring either. What can we learn from these previous efforts?Attention should be focused on five areas -- research design, the units of observation,measurement of physician supply, covariates that could confound analyses, and expandedmeasures of outcomes. (Table 2)Research designRandomized clinical trials of differing levels of physician supply will provide the mostconvincing evidence of the effectiveness of the physician workforce. When these trials arenot feasible, longitudinal cohort studies still offer substantial advantages in inference ofcausation over the cross-sectional design used in all of the studies that we reviewed.Although a relatively efficient approach to hypothesis generation, cross-sectional studiesrarely lead to certainty about causation.22The units ofobservationIf physician supply is our policy interest, then the attribution of group "exposure" toindividuals must follow. By definition, a "physician supply" is a measure of physicians inaggregate and is logically expressed as a ratio to the population served. Though ecologicstudies are frequently maligned, in this instance there is a clear "logic to the ecologic" ifthe supply with minimal heterogeneity is matched to the population20, 21 With somecare, this can occur within a geographic area or a practice. (Table 3) With the exception ofGrumbach's work, all of the research previously discussed used large and internallyheterogeneous geographic unitsAt one end of the geographic hierarchy is the entire U.S. Though this unit of observationis the one most commonly used for workforce studies, it has important weaknesses.Measures of the overall physician supply in a country of 260 million people have littlemeaning for any particular patient or health care provider. Primary care is localized towithin fifteen minutes for most patients22, 23 and tertiary care centers serve regional, butrarely national, populations.24Smaller geopolitical units have their own limitations. States and counties are attractive tohealth care policy makers because many health care programs in the U.S. are administeredat these levels2 5 A closer look, however, shows that they suffer from the same extremeinternal heterogeneity as the overall U.S. Since their boundaries are ignored by patientsseeking medical care, physician numbers are mismatched to the population; the resultingratios have limited meaning.An alternate approach that we have used in the Dartmouth Atlas ofHealth Care project isto define utilization-based health care market areas using Medicare inpatient (Part A)data 2 4, 26, 27 ZIP Codes are the fundamental geographic units and are assigned tohospitals on the basis of patient travel for hospital care. Since physicians tend to belocated close to hospitals, these hospital service areas (HSAs; N=3,436) are useful formeasuring medical care resources at a community level including primary care and somespecialty care physicians. However, physician specialties with regionalized practicesrequire tertiary health care markets. HSAs are further aggregated into Hospital ReferralRegions (HRRs; N=306) on the basis of travel for major cardiac and neurosurgicalprocedures. Some patients, of course, still choose to travel beyond their own service area.We adjust the per capita physician measures for this travel, in proportion to the out-of23area inpatient utilization. Further adjustments are made in the measures of specialtysupply for population differences in age-sex across areas using data from the AmbulatoryMedical Care Survey.A further advance in measuring physician supply would be physician service areasconstructed using Medicare physician claims (Part B) data. Defining and characterizingthese areas is one of the objectives of the Dartmouth Atlas group during the next year.The validity of areas defined with utilization patterns of the elderly to younger populationswill be tested using Medicaid data in selected states. Beyond our current horizon isphysician service areas defined using all payor data. That would require a data set, atpresent imaginary, that sampled utilization for all specialties, ages, and areas. Currently,health care markets defined using Medicare utilization are the best available for studiesinterested in regional supply for the entire U.S.An entirely different approach would be to define populations based on health careorganizations. This is difficult in fee-for-service-f or IPA environments, but conceptuallysimple in I-IMOs where both the population and number of physicians may be readilyascertained. An observational study would require settings where the panel sized varied;this might only be achieved by observing across a number of HMOs. More interestingwould be to randomize physician panels to different patient population sizes.Analyses will also have greater validity when there is congruence in the supply numerator,the population denominator, and population utilization. A study of neonatologists, forexample, would logically use newborns as its supply denominator. A meaningful generalistsupply for children must delete the internists and discount the family practitioners.Measures ofsupplyIt almost goes without saying that workforce investigators desire data that moreaccurately identifies the practice location, specialty, and activities of providers. For smallareas or individual specialties, accurate information data may be obtained from statelicensing boards, specialty organizations, or even the phone book. The AMA and AOAMasterfiles remain the sole sources of data for studies of large geographic areas. Releaseof data with sufficient details for research has not always been easy in the past and is neverinexpensive. These organizations could facilitate workforce research by inviting24investigators to participate in the design of data collection and the guidelines for datarelease.ConfoundersThe importance of understanding factors that could modify the association between thesupply of physicians and utilization is well exemplified with physician productivity. Simplephysician-to-population ratios are crude measures. The productivity of physicians isheterogeneous, even within each specialty. Surgeons who are self-employed, for example,perform 30% more surgical procedures per week than those who are employees orindependent contractors.29 Women physicians, on average, work II % fewer hours perweek than their male counterparts.30 The physician supply may itself influenceproductivity as the number of patients available to each physician decreases.29, 31 Theleft-hand chart in Figure 3 shows that there is a negative association (R2=0.39) betweenthe supply of office-based physicians and the mean weekly number of office visits perphysician at a Census Region level. But this decline in productivity is not great enough tooutweigh the activity associated with more physicians delivering care. The chart on theright-side shows that the office visits per capita still increases with a higher per capitaworkforce (R2=0.66).The adjustment of clinicians per capita should also account for mid-level providers. 32This reservoir of clinician supply might be called the "fourth compartment" (the otherthree are federal, non-federal, and HMO physicians; See Tm'lov33) and rarely receives itsdue because it is presently difficult to measure. The dimensions of this fourth compartmentshould be systematically measured; then the relation between this clinician supply and thethree physician compartments can be analyzed.What about the organization and financing of health care? Underservice will still occur,even in areas of high physician supply, because of structural barriers to care. 13 Medicalcare is less available to those with inadequate insurance, cultural differences, and forvulnerable or segregated populations - migrant farm workers and those who are homelessor incarcerated. Managed care, as previously discussed, can influence utilization not onlythrough low physician staffing levels,27, 34, 35 but also through a combination of limitingother resources and rewarding physicians who contain costs. Providers of health care withconstrained employment of physicians should be closely examined to understand how theorganization of health care alters physician productivity and patient outcomes. Staff and25group-model HMOs remain valuable natural experiments for studying the provision ofmedical care with relatively few physicians.Non-experimental designs must also account for possible confounding by differences inthe characteristics of the patient populations across different levels of physician supply.Although age, sex, and race are often available from administrative databases, futurestudies should build on the model of the Medical Outcomes Study and incorporatebaseline measures of patient characteristics such as severity of illness, level of comorbidityand functional status.36Measures ofoutcomesAlthough mortality remains an important and incompletely studied outcome, it is usefulonly for certain specialties. We must look beyond this single measure to other outcomesthat are valued by patients. Since this data is usually not found in national data sets, somestudies will need to be restricted to specialties or practice settings where more detailedclinical information can be collected.Two EXAMPLES USING SUGGESTED METHODSManaged care offers favorable settings for future studies of the effectiveness of physiciansupply. The physician-to-population ratio is clear and could be adjusted for patient-mix,determined through patient surveys. Cohorts may be followed over time withmeasurement of utilization and outcomes including functional health status and patientsatisfaction. While it may be difficult to find a large enough range of panel sizes that aresimilar in other respects, an altemative is to randomize panels within a health care plan tovarying sizes. Managed care plans should be enthusiastic participants in such a study sincethe results would assist them in planning clinician staffing levels.In our current work, we have chosen to focus upon a specific specialty, neonatologists,where the intervening mechanisms of the relationship between neonatologist supply andnewborn outcomes can be more completely understood. This study, conducted withfunding from the Robert Wood Johnson Foundation, was stimulated by the four-foldgrowth in neonatologists, from 1.4 in 1981 to 5.9 per ten thousand live births in 1996, orone neonatologist for every 11 extremely low birthweight infants « 1,000 gms)37.26Although the efficacy of neonatal intensive care in reducing mortality and morbidity invery premature or ill infants is well established,38 it is not known if broadening theseservices to infants with less severe problems is helpful or harmful3 9-43 We will studytwo cohorts -- all newborns in the United States (1995) and in New York State (1996)The NY State dataset includes particularly detailed information about the infant's medicaltreatment. Both datasets include extensive individual information on the newborn's healthstatus (e.g. birthweight, gestational age, complications) and subsequent outcomes (deathwithin one year of birth, readmissions).The study has adequate power to detect even a small increase in NICU use rates andinfant mortality rates across regions of differing capacities. We will use traditional smallarea analytic methods to determine the local per-capita supply of neonatal resources(NICU bed supply, neonatal care physicians) and general pediatric clinicians. The detailedinformation on the birth records will allow excellent adjustment for differences acrossregions in population risks. In addition to the direct tests of the specific hypotheses, thesmall area analysis will provide the first comprehensive national information on regionalvariations on resources for the care of newborns.Bur DOCTORS ARE NOT FREE!We have chosen to emphasize research concerned with the effectiveness, not theefficiency, of the physician supply. Having made the point that there may be limits tophysician benefits aside from costs, we should note that the resources associated with1hcse benefits are likely to be linked by a simple relationship: As the supply of physiciansincreases the cost per marginal net benefit will increase more and more. At the peak ofthe physician supply-patient benefit curve, costs become infinite. Beyond the peak existsan irrational domain where costs per benefit cannot be defined -- there are no net benefits.Presumably, this is a region of physician supply where society would benefit if physicianswere paid not to practice!CLOSING THOUGHTSThe conviction that more physicians for a population is always helpful and never harmfulis ingrained in patients and physicians, alike. The lament of two physicians is a reminderof the medical profession's belief in its unequivocal effectiveness:27"...troubling to me ... as a neurologist ... is the whole issue of the primarycare physician gatekeeper. I cringe when somebody says we have anoversupply of specialists... How can you have an oversupply of a goodthing,?,,44"...for many people it is difficult to understand why too many doctors isbad...How can we have too many professionals dedicated to relieving painand sUffering?"45To our question, "If doctors were free, would we always want more?," the residents ofour two fictitious cities, Moredoc and Fewerdoc, would join these physicians inresponding with a resounding "yes!" But, the effectiveness of a physician supply should beexamined no less thoroughly than that of a new treatment. Effectiveness, or the limitsthereof of medical care, can be understood only through systematic study, not throughassertions of "common sense" or anecdote. Our first step is to recognize that the answerto the question is not presently known. The greater challenge is to generate the evidencenecessary to understand when a constrained physician supply leads to harm and when itleads to improved patient outcomes.Table 1. Recent studies (1) examining relation of pbysician supply to utilization or outcomes.Author Units of(study date) Design Population observation Measure of supply Key findingsNesbitt Cross- Pregnant women in Hospital service Physicians providing Lower supply associated with higher proportion of(1986) sectional rural WA State areas OB care complicated and premature births.Escarce Cross- Medicare MSAs Surgical specialties Increased supply correlated with higher rates of initial(1986) sectional contact.Allen Cross- Newborns in rural Counties MD delivery capacity Infant mortality rates higher in counties with lower OB(1986-90) sectional Iowa minus live births physician availability.Welch Cross- Medicare MSAs Patient care .MDs; % Overall supply not correlated with admissions or(1989) sectional primary care outpatient payments; proportion of primary care :MDsnegatively correlated with same.tvKrakauer Cross- Medicare Health Care Multiple measures Reduction of admission rates for ACS (3) conditions 00(1992) sectional Service Areas observed with a greater supply only within lowest(2) decile of supply. Slight effects were noted in the higherdeciles of supply. No effect on mortality.Mark Cross- Medicare Urban counties FP, IM, non-primary Higher supply of family practitioners and generalist(1992) sectional care per capita internists associated with lower Part B costs perbeneficiaries; higher supply of non-primary care MDsassociated with higher reimbursements.Grumbach(1993,94)Cross­sectionalAge 18-64 yrs inurban CAPrimary CareService Areas(4)Primary care :MDs percapitaSupply not correlated with office visits or %preventative services received.Notes:(1) Restricted to past 20 years.(2) Couoty aggregates based upcn Medicare hospitalizations.(3) Ambulatory care sensitive conditions(4) Neighborhoods reflecting residential trade patterns, topography, and demographic characteristics.Table 2. Suggested methodologic improvements in research analyzing physician supply - patient benefits.Study SettingMethodologic issueHealth care market-based(geographic areas) Practice-basedResearch designUnits of observationCross-sectional designs rarelyestablish causation.Supply should accuratelymatch the population understudy.Analyses biased by mis­matching of supply,utilization, and outcomes.Use cohort designs wheneverfeasible. Take advantage ofnatural experiments.Develop physician serviceareas for U.S. using Part BMedicare data. Use all payordata for regions whereIncorporate all payorutilization data.Use cohort or experimentaldesigns if feasible (e.g.experimentally manipulatepractice panel size).Identify managed care settingswith discrete physician­population groups (e.g. staffor arouo model HMOs).Measure out-of-plan use.t6Measnres of supply Current physician data is notalways accurate.Further develop physician Masterfiles with input from researchcommunities. Validate non-AMA, AOA data sources forregional studies.Confounders Physician clinical activity isinfluenced by non-patientfactors.Improved data sets of mid­level providers. Analyzevariation in physicianproductivity.Investigate the role of otherproviders. Study settings withlow physician staffing.Measures of outcomesOutcomes andutilization areinfluenced by patientcharacteristicsMortality is an insensitiveoutcomemeasure.Measure patientcharacteristics through surveys. Use widerrangeof data from administrative data sets (e.g. outpatientutilization, pharmacy).Expand use of administrative datato measuremorbidity. Surveypopulations to assess functional healthstatusand satisfaction.Table 3. Possible units of observation for measuring supply in workforce analyses.Units of observationGeopolitical areasUSStatesCountiesHealth care market-basedHospital Service Areas·- defined by Medicarepatient travel forinpatient care. (l)Physician Service Areas(2)Practice-basedPhysician or PracticePanelsAdvantagesSimple; easy to compare with othercountries.Important geopolitical units for healthpolicy development and implementation.Population and medical care data is highlyavailable.Functioning health markets for secondarycare. Better matching of physician supplyand population.Physician health markets defined bypatient utilization of physicians.Discrete units of physician - patients.DisadvantagesMedical care is localized and physicians areunevenly distributed.Physicians unevenly distributed. Not health caremarkets: physicians serve populations in otherstates.Not health care markets: physicians servepopulations in other counties.Patient travel may vary for different specialties.Travel for care by non-Medicare patients may bedifferent than for Medicare.Doesn't exist, yet; Medicare Part B is the onlyphysician utilization data covering the entire US.May be biased by high turnover rates. Mostphysicians do not have panels of patients.woNotes:(1) Developed for The Dartmouth Atlas of Health Care (Ref 24). Aggregations of ZIP Codes based upon the plurality of travelfor Medicare hospitalizations.(2) ProposedTable 4. Possible clinical settings to test the relationship between physician supply and utilization and outcomes.MeasuresUnits ofHypothesis Design Observation Supply Utilization Selected Covariates OutcomesHealth care market-basedGreater supply of Retrospective Neonatal Neonatologists, Admission to Birth weight; Neonatal mortality;neonatologists and cohorts service areas pediatricians, and family intensive care race/ethnicity; parental infant mortality;intensive care nurseries is practitioners per births; nurseries; leN days; educational attainment; readmission ratesassociated with higher presence of intensive care diagnostic and presence ofutilization in less ill nursenes procedure rates perinatologistnewborns without betteroutcomes.w~Practice-basedGreater panel size is Randomized Primary care Patients per primary care Office visits and Random assignment of Functional statusassociated with lower trial or physician physician panel (in admission rates per panels will minimize and satisfaction ofutilization and unaltered retrospective panels observational or capita; office revisit confounders survey sample.outcomes. cohorts experimental study) rates; utilization ofpreventative services.32Figure 1. Hypothesized physician supply - patient benefit curve. The linerepresents a curve of diminishing returns. Beyond the relatively flat part ofthe curve is a region where a further increase in physician supply leads topatient harm.Physician Supply33Figure 2. Theoretical relationship between physician supplyand health outcomes.Price of medicalservicesPotential Confounders:Physiciansper capitaOf supply measuresOrganization of care 1ProductivityMid-level providersAdjustedcliniciansper capitaOf utilization and //ljJutcome measures /Organization of care UtilizationPopulation of servicescharacteristics<,per capita<:AlteredpopulationoutcomesFigure 3. Physician supply, productivity, and clinical activity for office-based physiciansby Census region, 1994.140 I 7 IESC"90 , , Iw..I200NE ..ATL"PAC""MNTENCSATLR2=0.66140 160 180Office-based physiciansper 100,000 population" WSCWNC"ESC"90 I I I I I120c.~.S 130CfJ-I-''M os>"5<lJ 0...S:: 0 120~~0 0o>,0~ 6 110<lJ0<lJ'"""';s: ~0.. 100200NE"MATL"R2=0.39PAC"SATL'"ENC"WNC"'WSC" MNT"65 I \ \ I1207085CfJ-I-''MCfJ'M><lJ C.~ co'-'-< 'MO.~ 80>,>'~..c..:.;p..,8:! ~ 75;S:o..Cos<lJ::;sNotes: Data adapted from AMA data (see references 29 and 31). Only office-based non-federalphysicians included. Lines represent linear regression. Symbol key: NE-New England; MAIL-MidAtlantic; S AIL-South Atlantic; ENC-East North Central; WNC-West North Central; ESC-East SouthCentral; WSC-West South Central; PAC-Pacific35References1. Organization for Economic Development. OECD Health Data 97- FrequentlyAsked Data -- Available at: http://www.oecd.org/:AccessedOctober 1,1997.2. Wennberg J, Cooper M (series editors). The Dartmouth Atlas ofHealth Care -1998. 2nd ed. Chicago, IL: American Hospital Association; 1997.3. Schwartz W, Sloan F, Mendelson D. Why there will be little or no physiciansurplus between now and the year 2000. N Engl 1 Med. 1988;318:892-897.4. Schwartz WB, Mendelson D. No evidence of an emerging physician surplus.lAMA. 1990;263:557-560.5. Black W, Welch H. Advances in diagnostic imaging and overestimations of diseaseprevalence and the benefits of therapy. N Engl 1 Med. 1993;328:1237-1243.6. Wennberg JE, Blowers L, Parker R, Gittelsohn AM. Changes in tonsillectomyrates associated with feedback and review. Pediatrics. 1977;59:821-826.7. Nesbitt TS, Connell FA, Hart LG, Rosenblatt RA. Access to obstetric care in ruralareas: Effect on birth outcomes. Am 1 Publ Health. 1990;80:814-818.8. Allen D, Kamradt J. Relationship of infant mortality to the availability ofobstetrical care in Indiana. 1 Fam Pract. 1991;33:609-613.9. Krakauer H, Jacoby I, Millman M, Lukomnik JE. Physician impact on hospitaladmission and on mortality rates in the Medicare population. Health Serv Res.1996;31: 191-211.10. Welch WP, Miller MK, Welch G, Fisher E, Wennberg J. Geographic variation inexpenditures for physicians' services in the United States. N Engl 1 Med. 1993;328:621­627.11. Mark D, Gottlieb M, Zellner B, Chetty V, Midtling J. Medicare costs in urbanareas and the supply of primary care physicians. 1 Fam Pract. 1996;43:33-39.12. Escarce Jl. Explaining the association between surgeon supply and utilization.Inquiry. 1992;29:403-415.13. Grumbach K, Vranizan K, Bindman AB. Physician supply and access to care inurban communities. Health Affairs. 1997;16:71-86.14. Weinberger M, Oddone E, Henderson W. Does increased access to primary carereduce hospital readmissions? N Engl 1 Med. 1996;334:1441-7.15. McClellan M, McNeil BJ, Newhouse JP. Does more intensive treatment of acutemyocardial infarction in the elderly reduce mortality? lAMA. 1994;272:859-866.3616. Whittle J, Conigliaro J, Good C, Lofgren R. Racial differences in the use ofinvasive cardiovascular procedures in the Department of Veterans Affairs. N Engl 1 Med.1993;329:621-627.17. Ayanian J, Epstein A. Differences in the use of procedures between women andmen hospitalized for coronary artery disease. N EnglJ Med. 1991;325:221-225.18. Ayanian J, Udvarhelyi I, Catsonis C, Phasos C, Epstein A. Racial differences in theuse of revasculazation procedures after coronary angiography. lAMA. 1993;269:2642­2646.19. Guadagnoli E, Hauptman P, Ayanian J, Pashos C, McNeil B, Cleary P. Variationin the use of cardiac procedures after acute myocardial infarction. N Engl 1 Med.1995;333:573-578.20. Susser M. The logic in ecological: 1. The logic of analysis. Am 1 Public Health.1994;84:825-828.21. Susser M. The logic in ecological: II. The logic of design. Am 1 Public Health.1994;84:830-834.22. Goodman D, Barff R, Fisher E. Geographic barriers to child health services in ruralNorthem New England: 1980 to 1989. 1 Rural Health. 1992;8:106-113.23. Williams AP, Schwartz WB, Newhouse JP, Bennett BW. How many miles to thedoctor? N Engl 1 Med. 1983;309:958-963.24. Wennberg J, Cooper M (series editors). The Dartmouth Atlas ofHealth Care inthe United States. . Chicago, IL: American Hospital Association; 1996.25. Rivo ML, Henderson TM, Jackson DM. State legislative strategies to improve thesupply and distribution of generalist physicians, 1985 to 1992. Am 1 Publ Health.1995;85:405-407.26. Wennberg J, Gittelsohn A. Small area variations in health care delivery. Science.1973;182: 1102-1108.27. Goodman D, Fisher E, Bubolz T, Mohr J, Poage J, Wennberg J. Benchmarking theUS physician workforce: An altemative to needs-based or demand-based planning. lAMA.1996;276: 1811-1817.28. Roos N. Linking patients to hospitals -- Defining urban hospital service areas. MedCare. 1993;31:YS6-YS15.29. Socioeconomic Characteristics ofMedical Practice 1995.. Chicago, IL:American Medical Association; 1995.30. Kletke PR, Marder WD, Silberger AB. The growing proportion of femalephysicians: Implications for US physician supply. Am 1 Public Health. 1990;80:300-304.3731. Randolph L, Seidman B, Pasko T. Physician Characteristics and Distribution inthe U.S. 1995-96 ed. Chicago, IL: American Medical Association; 1996.32. Mundinger MO. Sounding Board. Advanced-practice nursing--Good medicine forphysicians? N Engl J Med. 1994;330:2Il-214.33. Tarlov AR. HMO enrollment growth and physicians: The third compartment.Health Affairs. 1996:24-35.34. Mulhausen R, McGee J. An alternative projection from a study of large, prepaidgroup practices. JAMA. 1989;261:1930-1934.35. Weiner JP. Forecasting the effects of health reform on US physician workforcerequirements. JAMA. 1994;272:222-230.36. Tarlov A, Ware Jr J, Greenfield S, Nelson E, Perrin E, Zubkoff M. The MedicalOutcome Study: an application of methods for monitoring the results of medical care.JAMA. 1989;262:925-930.37. Goodman D, Little G. General pediatrics, neonatology, and the law of diminishingreturns. Pediatrics., in press.38. Williams R, Chen P. Identifying the sources of the recent decline in perinatalmortality rates in California. N Engl J Med. 1982;306:207-214.39. Silverman W. Is neonatal medicine in the United States out of step? Pediatrics.1993;92:612-613.40. Silverman W. Overtreatment of neonates? a personal retrospective. Pediatrics.1993;91:971-976.41. Kliegman R. Neonatal technology, perinatal survival, social consequences, and theperinatal paradox. Am J Public Health. 1995;85:909-913.42. Phibbs C, Bronstein J, Buxton E, Phibbs R. The effects of patient volume and levelof care at the hospital of birth on neonatal mortality. JAMA. 1996;276:1054-1059.43. American Academy of Pediatrics. The role of the primary care pediatrician in themanagement of high-risk newborn infants. Pediatrics. 1996;98:786-788.44. Iglehart J. Listening in on the Duke University Private Sector Conference. NewEnglJ Med. 1997;336:1827-1831.45. Letter to the Editor. New York Times.; Dec 16, 1997.38Overservlcing,Klim McPhersonProfessor of Public Health EpidemiologyLondon School of Hygiene and Tropical MedicineVancouver November 14th.l997Introduction.Overprovision of health care is a crucial issue in all health care systems. Since they are expensi ve, andinexorably seem to consume greater proportions of GNP, it is natural to ask for increasinglydemanding justifications for what is provided. The designation of overprovision is complex, however;while it clearly must include care for which no benefit (for the patient) can be expected, it might alsoinclude a wide spectrum of care which is merely inefficient relative to other use of scarce resources.Thus overprovision includes care which should not be provided under some strategic rationing policy.A problem is that each clinical decision is taken in a context of an assessment of individual balancesof benefit and risk, This is generally done in a situation of a chronic asymmetry of information',constrained by resources and often with real or implied urgency. These decisions have had to be madefor decades under a model of sophisticated biological understanding, strong professional hegemonyand a dominant belief in a medical model of disease. What we observe is the evolution of verydifferent medical practice styles", Since measuring (and comparing) outcomes for many treatments isgenerally difficult and often unreliable, and anyway often disparaged and considered unimportant,practice styles have evolved", under this asymmetry, in circumstances of crucial and real informationdeficit about the actual benefits and risks associated with particular decisions'. (See Figure 1 below.)Thus knowing about the attributable benefits and risks of treatment choices (including no treatmentoptions where relevant) for individual patients is generally much more difficult than is commonlyrealised. It is far too easy to be glib and reassuring when that is what ill patients and their advisors39Figure 1GIOLECYSTECTO'1Y41\ Q41\ N,B,300•• NEWF.NC •£It ....• W 41\2CD ,. ..'6'. • W ..It§,8.---;<,ccu.Jo.• •lOJ • B '1/ ..1 of •0 fItIt •... • ~ ..• It iE &Vi CPN75 75USA WEST OXFORD MAINE NORWAYMIDLANDS GPs75 75-77 75-77 73 7740generally prefer. But the true uncertainties may actually be profound, even if their extent is notsuspected. Certainly many well established treatments have been found to be wanting when assessedrigorously.Practice variationsThus, as would be expected, practice in differing organisational structures, even if supply and need areconstant, can be shown to be very different in highly predictable ways. Fee-for-service providers withunrestricted third party insurers provide more care while constrained or capped levels of supply leadsto lower levels of provision and managed care - with second opinions and greater resource toguidelines - has the explicit objective of containing costs via controlling provision. It is much moreimportant to control hospital admission rates than expenditures per admission in achieving such anobjective. Since per capita expenditures are the product of average cost per case and admissions percapita, it is not difficult to demonstrate that where there is discretion on admission itself it is essentiallythis that determines costs. (See Figure 2 below.)Best estimates suggest that around 10% of hospital admissions constitute care which is determinedby unambiguous medical need. Some interventions do in fact only vary with morbidity rates,because there is very little discretion or uncertainty associated with treatment choices. There areother instances where elements of medical care may vary enormously, but the cost and outcomevariations are minimal, which nonetheless might represent over-provision. The general rule is thatmost hospital admissions are discretionary, to the extent that systematic two to three fold variationsin standardised admission rates are seen between neighbouring hospital market areas. These haveimplicationson costs which are largely obvious and on health gain which are largely poorlyunderstood, because the micro level effectiveness has not been established. (See Figures 3 and 4below.)241Figure 2HysterectomyStandardised Bed Days Ratio.~----~---~-.__.__.._-0.5 1.5 2 2.5Standardised Incidence Ratio._---_._~~~-_._--_..~-~..~-_._-HysterectomyStandardised Bed Days Ratio2 c-------------------===_=-----------------,15() ~j05Standardised Length of Stay RatiolS42Figure 3Variation in per Capita Incidence of Hosmtanzation forCases Classified by Diagnosis-Related Group (DRG) among 30Hospital Market Areas In Maine, 1980-1982.'RA no OF RATESPERCENTAGEv .\-RIA TlON :--;0. OF COEffiCIENT OF OF CASESSCALE DRGs SCV VARIATION EXTREME QUARTILE (ACCUM.ILow 3 17 0.16 2.3 l.J 1.1 (1.1)Moderate 2-l 39 0.2-l 2.8 1.-1 8900.1)High 135 90 0.3-l -l.6 1.6 -l2.3 (52.-1)Very high 1 1-l9 190 0.-19 7.-1 1.9 31.7 (84.1)Verv high II 134 506 0.70 12. 1 2.8 15.9000.0)"Vanauon staus ties are wcrghted averages. Number of cases = 428.056. See text for expla-nanon or vanauon scale and S(V tsystemauc component of variation>. Accurn. denotes accu-mutated.Figure 4Relationship between systemic variation, an estimate of the ratio of rates between extremedistricts (fold), and typical surgical operation in the eight bands.rr=_...__...I Amount of SCV Range Approximate Fold Typical Surgical ProcedureI Variation RangeIil Very Low <1.8 <1.5 Hernia Repair:1----'---'il__Lo':"_____.. 1.8 - 3.2 2 Hysterectomy..-..._---Ii Medium 3.2 - 5.4 2.5 Hip Replacement.,".".~~-~--_._~.._--~---_. •.il High 5.4 10 3 Tonsillectomyii -.. .. -,n Very High I 10 · 20 4 Haemorrhoidectomy,il --_ ...,---_.._~"--_._------- •.-;1 Very High II 20 30 7 Adenoidectomy, -'I -- -~~------------ •... _-_. -_._- f------.---....ii Very High III 30 · 50 10 Removal of toothIir.-------_._~- .._--~-,_..._ .. ._...-,. Very High IV 50+ 20 Carotid Endarterectomy'I: ,...-.~c;~:,,=_=~.;;;;:;~=:::...___=:..=~::..__ · -43Figure Sa'Percentage of the 32 selected surgical procedures, categorised by grouped SCVALL FOUR RHAsSCV No. 01 Percent of all Cumulative %Cases Surgical admissions<1.8 78281 2.57 2.571.8103.2 136450 4.48 7.053.2 to 5.4 93831 3.08 10.135.4 10 10 475510 15.62 25.7510 to 20 90218 2.96 28.71201030 47112 1.55 30.2630 to 50 94076 3.09 33.3550+ 27805 0.91 34.27Tolal 1043283 34.27The total number 01 surqical procedures on which the percents are based is 3,043,926.Figure 5bPercentage of admissions for the 45 selected hospital admissions, categorised bygrouped SCV.All RHAsSCV No. 01 Percent 01 all Cumulative %Cases Hospital admissions<1.8 0 0.00 0.001.8103.2 25935 0.44 0.443.2 to 5.4 1078133 18.37 18.815.4 to 10 1626176 27.70 46.5110t020 1564083 26.99 73.50201030 166764 2.84 76.34301050 235413 4.01 80.3550+ 0 0.00 80.35Tolal 4716504 80.3544However, frank overprovision is rare, because, in the end, if the appropriate consensus ever existed todefine any provision as such, then it would tend not to be provided. Usually there is sufficient noise inthe decision-making process to disguise things which might be considered unnecessary as appropriate,What we have instead is indirect evidence for overprovision coming from systematic variation, givinglise to massive differences in provision, which clearly have cost implications but rarely any measurableattributable health gain consequences.Figure 6160Cro;:: 150~ <n"'-;: c'" '",,-'0sz .V)- '".- 0:5: 0<no0:0o:~:c ~o '"_ 0.'" <n>'0'';::; <lJroro"iU",0: .'"0>'"ro0:1401301201 101000900.80- II Hlp FraoureII SurgeryIB ~-'o DeadII HVMCFewer 2.5 3.5 4.5than 10 10 or2.5 3.5 4.5 moreHospital Beds per 1,000Residents in HRRsAll research so far, which seeks to compare mortality rates (for example) with hospital use rates,uncovers positive associations rather than negative ones. Clearly since provision itself might be inresponse to need, and it certainly determines thresholds of use rates, the determination of causalpathways is complex 13. However the evidence for relatively high levels of hospital provision having abeneficial effect on health is restricted to a few specific procedures, mostly very rare. Of course this45might be a function of insensitive measures, but prima facie the empirical evidence supports the notionof decreasing aggregate benefit associated with increasing expenditures above the norm".Figure 7Net ettect of an Increasing rate ofnospttaflzatlonB' ............." ,,,uzNew HavenIncreasing faie of hospitalizationNOTE. As hospitalizations 3fC increased. inevitably J point willbe reached where the net effect is harmful to patients This figurerepresents alternative interpretations of where tn-at point may ~.The first curve shows thai the marginal benefits plateau at POintA and decline into a zone where the iatrogenic effects of hospitalcare exceed the benefits. The second curve shows a much broaderLone of decreasing but still positive utility extending past poi,nl A10 point B. The present tate of clinical knowledge as exemplifiedb the practice patterns of New Haven and Boston does nOIdistinguiSh between 'besc two possibilities.46Intensive CareA very interesting, and seemingly paradoxical, example comes from the provision of (adult in this case)intensive care. Intensive care is clearly highly technical and labour intensive care for the acutely sick 15 .Hardly any of it has been subjected to randomised trials to examine the attributable effect of this extraexpenditure and provision and yet there are, in common with most other kinds of hospital admission,apparently large variations in use rates.Figure 830Severity distribution of AP'ACHE II scoreson patients admitted to the ICU.Gil USA - 5,030 patients, 13 hospitalsII1II NZ - 1,005 patients, 2 hospitals[] UK - 5,327 patients, 18 hospitalsV) 20 .z0V;'""0<u,0....zwUc;wc:100'I' S''"0''"0''"+~ ~ 0J 0J ('C) U')0 en 0 cf, 0 , , ('C)en 0~ ~ 0J 0J ('C)APACHE II SCOI\E47It epitomises much of health care as the need for such care is deemed to be self evident. Some keyworkers in the field have particularly singled out intensive care as being so self evidently beneficial 16that randomised trials might be unethical. However herein lies the central problem with health care,that assumptions are made on grounds of intrinsic plausibility without the necessary empiricalvalidation of efficacy. As a case study in overservicing, intensive care is illuminating, since in the UKthe average cost per day is around 800 pounds.These data show that in the US the expenditure on intensive care (IC) is more than 1% of GDP whilein the UK the figure is more like 0.1%. Since our GDP is lower anyway the per capita expenditure iseven lower relatively. In the UK there are serious questions of under-provision of intensive careaugmented sometimes by questions of the quality of life associated with the outcome. However, muchpublicity in the UK is routinely given to the fairly frequent occasions when persons are refusedadmission to intensive care units because there are no beds or no staff to operate the provision. Theclear assumption is that the refusals will give lise to many unnecessary deaths and that the solution is toprovide more beds or staff.Figure 9IC costs per person in some countries (1990;HC health care, CAP capita, GOP gross domestic product._..COUNTRY HC % GOP£ HC £ IC % IC £GOP ICAP' ICAP HC ICAP.US 12.4 11634 1442.6 10 144.3FRANCE 8.9 11805 1050.1 ?5 (52.5)"N.Z. 7.2 7211 519.2 "5 (26.0)"JAPAN 6.5 13164 855.7 "5 (42.8)"UK 6.1 9540 581.9 I 58• gross domestic productJperson, market prices 1990 (Chew 1992)" estimated IC costs as 5% health care costs48The first question of interest is what is the relationship between Ie bed provision and refusal rate in'the England, as each extra bed would cost more than a thousand pounds per day in extra staff andequipment. We did a survey of intensive care provision in England in 1993 for the Department ofHealth of all intensive care unirs'". We surveyed 234 units and received replies from around 75%,We asked for estimates of the number of patients who had been refused admission because of ashortage of staff or beds during 1992. When we compared these estimates with the bed provisionrelative to the catchment population the relationship was very strong. The strongest determinant ofrefusals by far' was the bed provision and even after adjusting for many other aspects of supply, therefusal rate had an odds ratio of around a third when the bed provision was relatively high, Thus, atleast as far as refusals were concerned, the strongest determinant does seem to be number of beds perhead of catchment population. Very roughly, 1.5 beds per 100,000 population is associated withrefusal rates of around 10% while an increase to around 5.25 beds per 100,000 is associated with 4%.Thus, very roughly, a fourfold increase in beds might result in a halving of the refusal rate.49Figure 10Odds ratios (ORs) + 95% confidence intervals [or (actors affecting leU rcfw;,11 ratesADJUSTED ODDS RATIOS FOR leu REFUSAL RATES10ly' -: .'I' , ,o "" ..h::: : ' ...'J!::: ~of-::: : ~ ., ,• ~';, .. ,.,~~.J*:f-:' , ,J!:::.,;" ,'"~: 4--',{f-: ~' , ,71-. " .. .yenlowmediumhighhighmediumlowmed7Lg~lowDGHDGHlnmad scrc onmixehigmediu10shareexctusivexclusiveshared0,1DIRUNITTYPEHI·TECBEDSEXCLCONSHOSPTYPENURSEWTECONSSESS~~~TREBEDPROVODDS RATIOS (log scale)baseline: OR=1Confounding vanabtcsUNIT TYPE rc. or tc +/- HD -t, CCHI·TEe BEDS other high-technology beds wuhm the hospuatEXCL CONS exclusivity of consultant coverHOSP TYPE DGll/t· district general hospital. postgraduate reacfung rcsponsrbiliucsDGH/Ol . DGll. no postgraduate tcnclung rcsponsibilurcs111<:0 sell . hospital wHh attached medical schoolREr: C:NTHE hoxpunl a spcciaiisr referral centreSupply van abies!lED PROYNURSE WTECONS SESS1)lf{number of beds per 100,000 populationnurxc whole uruc equivalents per bedweekday consuunm sessionsleu has director cxctusrvc 10 unl( (or sh,!fed)A halving of the refusal rate to 4%, if achieved, would still be regarded as unacceptable, since anyrefusal to intensive care, if appropriately referred, is still prima facie evidence for under-provision.Hence pressure for more beds still would not cease. However, what is the relationship of bedprovision with admission rates? Here we seek information on one of the most crucial relationships in50health care provision, that between the supply and the use of services. The broad implicit assumptionis that supply is in response to need and use is determined by medical need -- essentially. Nobodybelieves that if asked, but it is nonetheless assumed at a macro level by policy makers and politicians.The observed relationship in this study between bed supply and admissions is shown below.Figure 11Intensive care unit admission & refusal rates (1992)in areas with different bed provisionflOO,OOO populationleu REFUSAL & ADMISSION RATES (1992)IN AREAS WITH DIFFEREN, BED rr10VISION PER UNIT POPULATIONrefusal rate %12ro refusal .ato101-8-64f-2L Mbed provisionHL low 0.81-1.85. M medium \.86-2.81, H high 2.82-13.83 beds/IOa,OOO populationAnd here we can see a really very serious relationship where high bed provision is associated withmore than three times as many patients per capita than low bed provision. So to increase beds byfourfold might decrease refusals but would also increase admissions. Of course this is not a verysurprising finding, but nonetheless it is indicative of the kind of problem faced in health care. Imaginetwo communities of 500,000 people, one with 7.35 beds and another with 26.2 beds. The first would,on the above basis, admit 500 patients each year to its intensive care and the second 1700. Assumingagain the above refusal rates the number of patients refused admission per annum would be 56 and 7151precisely how we know there is overservicing, because the pressure to increase bed provision is notassuaged at all by such a dramatic increase in bed provision, in fact it is intensified.Figure 12Intensive care unit admission & refusal rates (1992)in areas with different bed provision/iOO,OOO populationleu REFUSAL & ADMISSION RATES (1992)IN AREAS WITH DIFFERENT BED PROVISION PER UNIT POPULATIONrefusal rate %12108ac.Jmissions/100,000 POPr--------~ 350.. admission rete 0 rofusal rate30025020064215010050L Mbed provisionHL low 0.81.1.85. M medium 1.86-2.81, H high 2.82-13.83 bedS/100,000 populationFigure 13Two communities in England of half a million populationleu beds /l00,000Expected admissions per annumRefusals per annum1.5500565.25170071, More beds need more beds'52Figure 14•; III :e : 1•: I--..-<>-- •I •AdmItted appropriateLow APACHE IIMedium APACHE IIHigh APACHE IIRefused appropriateAge(years)16-4950-69~70Number of interventions<11-2;,3Surgical/EmergencyNon-surgical/ emergencySurgical! emergencySurgical/ non-emergencyNon-surgic a1/ non-err-ergenc ySex.Fe-narcMaleleU bed provisionHighMediumLow, , I0"./ 0:,,) 0.,.;0"" 06'0(9""0 ~o <10Relative risks (log scale)Odds ratios and 95% Cis for factors affecting mortality forselected admission and refusal categoriesData from the national part of our study indicated that approximately 73,000 cases werereferred to intensive care units in England in one year. We estimate from our data that 80% ofthese cases were appropriately referred, and that 90-day mortality was 37%. If the relativerisks for mortality for refusals compared with admissions were between 1.5 and 2.0, theattributable risks would be 9.9% and 11.6% respectively. Based on the results of our study,the numbers of deaths in 1 year attributable to refusal could therefore lie somewhere between2,100 and 2,500 per annum for appropriately referred cases. This number of potentiallyavoidable deaths is in the same order of magnitude as those due to road traffic accidents" (butfar less than the 138,000 avoidable deaths due to smoking").53Notwithstanding the extremely low level of provision of IC in the UK, which is widely understood,and the observational evidence described above, these data do not indicate any degree of under­provision. What they indicate is a much lower apparent attributable mortality associated with refusedadmission to ICU, and serious confusion about who should be appropriately referred and seriousuncertainty about who should be admitted. Obviously such uncertainty will give lise to overprovisionboth at the individual patient level and also at the macro provision level. But increasingly, the rationalresponse is to find out more about the role of high dependency beds and about the attributable role ofaspects of intensive care in increasing health gain and preventing mortality. Broadly, the health caresystems are still set against that, in spite of the fact that, increasingly, they will have to do that byproving the need for what they offer.As it is this work demonstrates yet again the dominance of seemingly plausible pressures to provideservices which match crude measures of unmet need. But need is generated by professionals whoundeniably have concerns which are outside the needs of the community they serve. In a state ofmanifest uncertainty it is seems sensible to cover the most extreme possibility - which is precisely howoverservicing gets to exist without its being obvious. Yet it is probably dominant in even the mostfinancially constrained systems. Practice styles have a momentum or their own, much of which hassurprisingly little to do with the health of populations.254REFERENCESWennberg JE, BarnesBA,Zubkoff M. Professional uncertainty and theproblem of supplierdemand. SocSciMed 1982; 16: 811-24.McPherson K. Whydo variations occnr? In: The Challenges of Medical PracticeVariations. Eds MooneyG & Anderson TF. Macmillan 1989.McPherson K, Downing A & Buirski D. Systematic Variation in Surgical Procednres and HospitalAdmission Rates. London School of Hygiene and Tropical Medicine Report to DoH April 19664 Wennberg J (1986) Whichrate is right?NewEng J Med 314 310-3115.6.8910II.12J.P. Bunker (1970), Surgical Manpower: A comparison ofoperations and surgeonsin the United States and in Englandand Wales(1970) NewEng J Med282135-144J A Glover (1938). The incidence ofTonsillectomy in School Children.Proceedings of the Royal Society of Medicine, xxxi, 1219-36.Anderson TV & Mooney G. The Challenges of Medical Practice Variations. Macmillan Press 1990M R Chassinet al (1987a) Does inappropriate nseexplain geographicin the useof healthcare services. JAMA, 258,2533-37.McPherson K. The bestand theenemyof the good: Randomised controlled trials,uncertainty, and assessing the roleof patient choicein medical decision makingThe CochraneLecture. J Epidemiol Community Health 1994; 48: 6-15.Evan RGThe dog in the night time: medical practicevariations andhealth policy. In The Challenges of Medical Practice Variations Anderson & Mooney op citWennberg JE, McPherson K,CaperP. Will payment based on diagnostic relatedgroups control hospital costs. N Engl J Med 1984; 311: 295-300.McPHERSONK. How should health policy be modified by the evidence of medical practicevariations?; Controversies in Health Care Policies 1994: Challenges to Practice; edited by Marshall Marinker;Chapter 4;56-74.13 Centre for the Evalnative Clinical Sciences. The Dartmouth Atlas of health Care. AHA 19961415Wennberg JE. What is outcomes research? In: Gelijns AC ed. Medical innovations at the crossroads.Vall Modern methods ofclinical investigation. Washington DC: National Academy Press, 1990:33-46Rowan KM, Kerr JH, McPHERSON K, Short A, VesseyMP. Intensive Care Society's AcutePhysiologyand Chronic Health Evaluation (APACHE II) study in Britain and Ireland; A multicenter, cohort1655study comparing two methods for predicting outcome for adult intensive care patients. Critical Care Medicine1994. Volume 22; No.9, 1392-1401.Nick Black Why we need observational studies to evaluate the effectiveness of health care BMJ 19963121215-121817 Metcalfe M, McPherson K. Study of Intensive Care in England 1993. Department of Health, London, 199518 Metcalfe AM, Sloggett A, McPHERSON K. Mortality among appropriately referred patients refusedadmission to intensive-care units. The Lancet 1997; 350:7-1219 Mortality Statistics. Office of Population Censuses and Surveys. Series DHZ no. 20. HMSO London 199520 Peto R, Lopez AD, Boreham J, Thun M, Heath C. Mortality from Smoking in Developed Countries 1950-2000. Oxford University Press, Oxford 19945657What can administrative databasestell us about overservicing in Canada?Dr. Ben Chan, MD, MPH, MPAScientistInstitute for Clinical Evaluative Sciences in OntarioOctober 1, 1997Prepared for the Second Invitational Trilateral Physician Workforce ConferenceDo not quote or circulate without the permission ofthe primary author58IntroductionHow much overservicing is there in Canada? Policy-makers and practitioners across the nationhave a great interest in knowing whether care is being provided which is ineffective or inefficient,or inapproptiate or unnecessary. Recent fiscal pressures and deficit reduction battles have placedthe entire health care system under intense scrutiny. Health expenditures typically account for athird of provincial budgets; yet, while the size of health care budgets make them prime targets forfiscal restraint, the growing demand for high quality health care services makes it difficult forpolicy-makers to reduce budgets without enduring the wrath of the population. Hence, anyopportunity to extract savings from the system by reducing overservicing is met with greatinterest from all stakeholders.Allegations of inapproptiate use abound. Large, persistent regional variations in the use of a widevariety of in-patient and out-patient services have been well documented in numerous practicesettings, both in Canada and internationally [Wennberg, 1982; McPherson, 1982; Naylor, 1994;Black, 1995], and such variations suggest either overuse or underuse of services in particularregions. Some conjecture that overservicing occurs because of inadequate training of physicians,or growing physician supply, which in turn leads physicians to increase their provision of servicesinappropriately to protect against lower market share and income. Physician organizations arequick to point out that patient-driven demand is a major cause of inappropriate care [Stewart,1995]. Unnecessary visits to physicians for minor complaints, abuse of emergency departments,defensive medicine generated by ftivolous lawsuits from patients are all believed to put pressureon physicians to perform more services than are necessary.The traditional approach used in the United States to measure overservicing is through thedevelopment and application of appropriateness criteria to physician practices [Brook, 1994].Such studies typically start with an extensive literature search and meta-analysis to establish theDo not quote or circulate without the permission of the primary author59effectiveness of different treatment options for a given clinical condition. Consensus panels thenconsider different clinical scenarios in which such treatment options could be used, and rate theirappropriateness of options on a scale of 1 to 9 for each clinical scenario. The extensive criteriaare then applied retrospectively on medical records to examine the extent of inappropriate orappropriate care.Such studies have faced several criticisms. First, they are time-consuming and expensive to use,and may be cumbersome to apply in day-to-day practice. Second, it may be difficult to anticipateall clinical conditions in which such criteria could be applied. Third, the validity and reliability ofthe appropriatness instruments have been questioned [Phelps, 1994]. There is no gold standardby which to evaluate these criteria, and the criteria have been shown on many occasions to differdepending on which body of experts developed them. For example, there are importantdifferences between consensus guidelines on cholesterol testing between Canada, the US andother jurisdictions [Naylor, 1996].In this paper, we examine alternative approaches to measuring overservicing. In particular, whatinferences can we make on overservicing based on admistrative datasets? Such datasets arereadily available, and in many cases provide comprehensive data on an entire population. Hence,they provide an opportunity to observe what is occurring on a system-wide basis at lower costcompared to in-depth reviews. The disadvantage of administrative data is that they are oftengenerated for other purposes (e.g. for remunerating physicians), and hence may not have all of thedesired datafields, or may be lacking in data quality or detailed clinical information.Given these caveats, what can we conclude about overservicing? We present, below, four caseDo not quote or circulate without the permission ofthe primary author60studies of examples of what we can learn from administrative data, the limitations in their use, andstrategies to address these limitations.Example 1: Use of benchmarks for efficient use of hospital resources in OntarioThe Institute for Clinical Evaluative Sciences first began using administrative data in the early1990's to examine variations in use of hospital services [Basinski, 1994]. Measures of interestincluded length of stay for different diagnoses, or proportion of surgical cases performed on asame-day basis, for specific types of surgery. For each major diagnostic group, institutionsoperating at the 75th percentile for low length of stay or high same-day surgery rate wereidentified as benchmarks. The assumption was that such institutions were likely operating with ahigh degree of efficiency. While higher percentiles might be even more efficient, they could bemore likely to represent outliers, local situations which were not generalizable, or underuse ofresources. The estimated savings in hospital bed days were then calculated as the number of bedswhich could be saved if all hospitals operated at the 75th percentile. The conclusion of theanalysis was that some 143,000 hospital bed days in 1991/92 could be saved if all hospitalsoperated at the benchmark.At the same time, the Joint Policy and Planning Commission (JPPC), a partnership between theOntario Hospital Association and the Ministry of Health, conducted reviews of benchmarkhospitals to identify the processes of care which allowed these institutions to operate efficiently,and to verify that patient outcomes at discharge were acceptable. Detailed case studies of leadinginstitutions were compiled and distributed across the province, with descriptions of howefficiencies were achieved (e.g. early discharge planning, close cooperation with homecare,specialized teams, patient education). These case studies were designed to give health careproviders practical suggestions on how to improve their lengths of stay [JPPC, 1994].Do not quote or circulate without the permission ofthe primary author61More recently, there has been interest in Ontario in selectively applying explicit criteria to measurehospital appropriateness. The InterQual ISD-A tool has already been used in other Canadianjurisdictions and institutions, including Saskatchewan [HSURC, 1994], Manitoba [MCHPE,1996] and British Columbia [Blackstein-Hirsch, 1993; BC Provincial Advisory Committee onClinical Resource Management, 1997]. Recently, the Joint Policy and Planning Commission(JPPC) of Ontario commissioned a study on non-acute hospitalizations [JPPC, 1997]. Samplingabout one half of the hospitals in the province on six major medical and surgical diagnoses, theJPPC found that 18% of admissions and 45% of subsequent bed days were not acute. Furtherresearch is planned to examine how consistent the results of the InterQual analyses are with theadministrative data analysis.This case study illustrates how administrative data can lead investigators to best practices in theprovince. In this example, the focus was on improving hospital efficiency. The fact that someinstitutions consistently discharged certain types of patients early suggested that these wereleading institutions; however, a subsequent detailed examination of those institutions wasnecessary to confirm that the quality in these institutions was adequate. This approach hasobvious limitations. For example, institutions may appeal' to have low lengths of stay if they havea large number of inappropriate one-day admissions. These issues, however, can be addressedthrough the selective use of more refined tools such as explicit appropriateness criteria.Example 2: Prenatal Ultrasound in OntarioPrenatal ultrasound is used in the detection of fetal abnormalities, multiple pregnancies,intrauterine growth retardation, as well as to estimate gestational age and locate the placenta. Itmay also be used for such medically unnecessary reasons as sex determination or to obtain aDo not quote or circulate without the permission of the primary author62souvenir, in-utero photograph. Several elinical trials suggest that in uncomplieated pregnancies,the benefits of routine prenatal ultrasound are questionable [Thacker, 1985]. Nonetheless, theSociety of Obstetricians and Gyneeologists of Canada (SOGC) has published guidelines [SOGC,1997] suggesting that in general, pregnant women should receive one routine ultrasound. Theseguidelines gave the benefit of the doubt to performing somewhat more ultrasounds than has beenestablished by the scientific literature. The Health Services Utilization and Research Commission(HSURC) in Saskatchewan has issued more restrictive guidelines, reeommending prenatalultrasound screening only for women at risk for abnormalities [HSURC, 1996].The number of ultrasounds per pregnancy has been rising steadily in Canada. In British Columbia,age-sex adjusted rates of ultrasounds per pregnancy rose from 0.90 in 1981/82 to 1.75 in 1989/90[Anderson, 1993]. In Ontario, similar rates of growth were observed in the 1980's [Anderson,1993]. A more reeent analysis we prepared for the Ontario Ministry of Health demonstrates thatultrasounds per pregnaney have risen from 1.7 in 1990/91 to 2.3 in 1994/95 (Figure 1).Furthermore, 62% of women reeeive two or more ultrasounds, in excess of the SOGCrecommendations (Figure 2).Was there any apparent justification for this increase? During this time period, there has been nosignificant ehange in the rate of complications in pregnancy. One factor which may have drivenincreased use of ultrasounds was the introduction of triple marker screening for neural tubedefects in July, 1993. This blood test is heavily dependent on accurate estimates of gestationalage, and an ultrasound is required to confirm both the dates and the diagnosis. However, thispolicy change took place before much of the utilization increase occurred.Do not quote or circulate without the permission of the primary authorFigure 1: Prenatal Ultrasounds per Pregnancyi"--:jg.§W;~EiW~.¥ ~'*~iiB;·'.-)'·',.."'''"'0)_ .~'"c""" "";:;;;S "1.'<';''''·'J','''''3''' -Ultrasounds perPregnancy2.5 T'---------------------~.i->:1.5 +1-------------------- _'"o­w1 +1--------------------__0.5 +1--------------------__1994/951993/941992/93Fiscal Year1991/92o -I , , ,1990/91Source: Ontario Health Insurance Planphysician billing dataDo not quote or circulate without permission of the authorFigure 2: Distribution ofNumber of Ultrasounds per Pregnancy;'E~;@';'i2\.~iW§il&. iZ :z: ?' L n d01",'1::Mii'Ci,,;;,;.:.gt}';f;'<'·,~-,."" ... "'~~" ,"""us, =;;~'"Y;';V::-:;C7;'<''P'''''''3 to 4(24% )\2(31% )Source: Ontario Health Insurance Planphysician billing data5+;(7%) 0)5%)1(33% )0,..,.Do not quote or circulate without permission of the author65The amount of clinical information in the OHIP database is negligible. Hence, it is impossible todetermine the indications for ultrasound on each patient. Many of these utrasounds may havebeen performed for complications of pregnancy, such as loss of fetal movement or threatenedabortion. Women with such cases may have had multiple ultrasounds. However, some researchsuggests that only 30% of women develop complications during their pregnancy requiringevaluation by ultrasound [Waldenstrom, 1988]. Yet, the proportion of women received two ormore ultrasounds was double this complication rate.It appears that the increase in use of prenatal ultrasounds has occurred without any apparentjustification. These findings are suggestive of overuse of the technology. Patterns of practice donot conform with established clinical practice guidelines, which themselves are considered to erron the side of caution. Clearly, the lack of clinical data limits our ability to determine withcertainty whether overuse occurred. The complication rates quoted in the literature maythemselves be inaccurate or not generalizable to the Ontario context.These results were presented to a working group of the Ministry of Health on reform of thephysician fee schedule in March, 1997. The working group recommended that the preamble tothe fee schedule be revised such that only one routine prenatal ultrasound in pregnancy would beremunerated by OHIP, and that if maternal serum screening was ordered, one limited prenatalultrasound could also be ordered [Ontario Ministry of Health, 1997]. However, the newguidelines also state that additional ultrasounds may be ordered in the case of pregnancycomplications. The frequency of and indications for ultrasound in these circumstances were notclearly stated. It was anticipated that more specificity to this clause could be added once newguidelines from McMaster University on appropriateness indications for ultrasound are released.Do not quote or circulate without the permission ofthe primary author66Example 3: Use of Doppler Echocardiography in OntarioEchocardiography refers to a family of tests that use ultrasound to image the heart. Two­dimensional echocardiography creates 2-D images of the anatomic structures of the heart, and isuseful in evaluating changes to heart structures from hypertension, myocardial infarction andother cardiac conditions. Doppler echocardiography is an add-on test to conventional 2-Dechocardiography, and assesses blood flow by detecting frequency shifts of ultrasound wavesreflected back from moving blood cells. Doppler studies are particularly useful in assessingvalvular disease and septal defects [Chan, 1995].Echocardiography is one of the fastest growing physician expenditures in Ontario. Totalexpenditures on all types of echocardiography rose from $28.3 million in 1989/90 to $47.9 millionin 1994/95 [Anderson, 1996]. While part of this increase was related to an increase in the numberof 2-D echo studies done, an important driver of expenditure growth was also the growing use ofDoppler as an adjunct study. The percent of 2-D echo studies in which a Doppler was performedrose from 53% in 1989/90 to 77% in 1992/93 (Figure 3) [Chan, 1996]. Secondary data analysisfurther revealed that there were wide variations among physicians in how often they used Dopplerstudies. 48% of physicians used Doppler in at least 95% of the conventional echocardiogramtests they performed, 13% did not perform Doppler at all, and the remaining 39% used Doppleron a selective basis (Figure 4).Many of the observed trends were certainly suspicious of overservicing. The rapid rise inexpenditures in the absence of any major change in disease patterns was of concern, and the rise inDoppler expenditures was also perplexing. When Doppler was first introduced in the early1980's, the capital costs of adding Doppler to a standard 2-D echocardiography machine wereDo not quote or circulate without the permission ofthe primary authorFigure 3: Doppler / Echo Ratio over Time;:::;-i-2;'};:(.:,'t¥~~j t t &~~~T! ;LZ'"'",. ';~""';. 'h....I.'" & .,._,;l'"I:):,Y~'"7:·:'}",",:.~:N:"',··'C-0.80 0.7.-~ 0.6~0 0.5~o~ 0.4~C)........ 0.30..~ 0.2Q0.1--<>- ~------- ~ ...-0-- ~~ ~ ~~.:-+- Doppler-~ -a- Colour Doppler/0­-..Jo89-90Source: Ontario Health Insurance Planphysician billing data90-91 91-92Fiscal Year92-93Do not quote or circulate without permission of the authorFigure 4: Distribution ofDoppler/Echo Ratios Among PhysiciansYij,';@; ':",{·:t'1f(;"!M"i1!!$.iRWW},iGM WI i i <&X1&J;';;'<h'§7A;%).1;~;S1Bf:·; ,.wDoppler-Echo Ratio;;..,"QQ.eeJ)e.....'"0;..<:':~o"Q~~ee=.....E.s;..Q.l~'"~~....e~eo"Tor<)o01o-oo 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9o.00Source: Ontario Health Insurance Planphysician billing dataDo not quote or circulate without permission of the author69high, and remuneration rates were set accordingly. In the late 1980's, however, Doppler becamerecognized as a standard feature of all echocardiography equipment. Yet, the remuneration feefor Doppler was not readjusted downwards. Hence, the financial incentives for performingDoppler were greater. The tendency of some physicians to use Doppler routinely as an add-onprocedure was also in question; while Doppler is clearly useful in assessing valvular disorders, itwas not clear what purpose it served where the echocardiography test was requested primarily toassess structural changes in the heart (e.g. left ventricular wall thickening from hypertension).At the same time, there were arguments, on clinical grounds, to justify the growth ofechocardiography. A Canadian consensus statement on management of congestive heart failurehighlighted echocardiography as a key tool for assessing and managing the disease [Johnstone,1994]. Secondly, some cardiologists have argued for use of Doppler to evaluate diastolicdysfunction, believed to be a form of left ventricular failure [Nishimura, 1997; Gaasch, 1994].However, this claim is controversial, as there is no clear evidence yet to suggest that attempts totreat diastolic dysfunction lead to meaningful improvements in quality of life [Pagel, 1993].To investigate this issue further, we examined records from Sunnybrook Health Sciences Centre,an academic institution which has routinely performed Doppler echocardiography on almost all itspatients over the last several years. The Sunnybrook database records a variety of patientdemographic data and the reason for referral, as well as all of the major clinical parametersrecorded from an echocardiography exam. Using this database, we tested the hypothesis thatinformation obtained from reason for referral and 2-D echocardiography can identify cases whereno new clinically significant information is obtained on the Doppler study. Logistic regressionmodels were developed for finding significant valvular regurgitation on Doppler examination.For ten common referral diagnoses, a clinically significant flow abnormality warranting assessmentby Doppler could be confidently ruled out, with a negative predictive value of 95%, through theapplication of simple scoring algorithms that make use of readily available demographic and two-Do not quote or circulate without the permission of the primary author70dimensional echocardiographic variables [Cox, 1997]. If applied on a province-wide basis, thisalgorithm could reduce by 40% the number of Doppler studies and result in savings of about $1million, under current remuneration schedules.Example 4: Use of spirometry in OntarioSpirometry is a diagnostic test used in the management of respiratory diseases. Its most commonuse is for the diagnosis and follow-up of patients with asthma. To perform the test, the patientblows into a spirometer, a machine which measures the rates of airflow from the lungs over time.[Crapo, 1994]. There are two types of spirometer. The closed circuit spirometer is used bylaboratories and is considered the gold standard. In the late 1980's, a new, hand-held portablespirometer was introduced into the market, at substantially lower cost. These units, however,may yield inaccurate results if not calibrated properly [Nelson, 1990].Spirometry use has risen sharply over the past several years, from 353,000 tests in 1989/90 to463,000 in 1994/95 [Chan, 1997]. Even after controlling for changes in prices, populationgrowth, ageing of the population, and substitution from spirograms to flow-volume tests (i.e.substitution from a less expensive to a more expensive variant of spirometry), there was a 19.6%increase in utilization over this time period. Among GPIPP's who perform the test, the per capitarates of testing by GPIFP's was seven times higher in Metropolitan Toronto than in NorthernOntario. The distribution of individual physician billings is heavily skewed to the right. A smallproportion (7%) of physicians account for 40% of the GPIPP billings on spirometry (figure 5).These high volume physicians are concentrated in the Metro Toronto region, and account formuch of the regional variation in use (see figure 6).This case study illustrates the inherent difficulty in making inferences about overservicing in manyDo not quote or circulate without the permission ofthe primary authorFigure 5: GPIFP Spirometry BillingsRanked from Lowest to Highest Billings'~~igW,,*-W:';:~§( L&&Jiilllli W 2 ..&~lP'w'J0}>"81}!E~t';Th&k~;-';";-)2iibf.f"/<'".."''''''~'''''~"'''.==~= :::::;;;; :><:=;",,,,,<·,,,c.,,~;·· ,::;.,:."OHIP 1993 First Quarter Billings$20,000$15,000~~&'$10,000"'Eiii$5,000$0 ' " ,," ,"!; '!Hl!,"!!! '+1 li,li'ilIT....wlil !II, I, ,jll, ,1ill1Jj, ' ,-.l~o 100 200 300 400 500 600 700 800GP's Ranked by Billings1300 1000 1100 1200Source: Ontario Health Insurance Planphysician billing data Do not quote or circulate without permission of the authorFigure 6: GP/FP Spirometry Billings per Capita byDistrict Health CouncilComparison of High & Low Billing GPIFPs, 1994/95'~f;k;;, i~:;,:i.t':H[!;%1lf:m:iW:f ;;;:;;z& TFflf:';il@it.~~'£~:K$9k1fi&t ,?B·:'c',:""""""''',o~ • "'Q;ff"'it',;ti-.Jtv800700600500400300200100o~-~<., -,-- --•..,.~ .. . .,-"~.--.,."" " .. "..-,I!IIi Hig h Bille rsIjj Low Billers~AlgomaEasternOntarioRideau ValleyBrantWaterlao RegionHaldimand-NorfolkRenfrewWeflingt o n-DufferinLambtonPeelEssexCochraneHam iltu n- WentworthMetropolitan TorontoOnawa- Carleton RegionalEast Muskoka-Parry SoundM an ito ulin - SudburySource: Ontario HealthInsurance Planphysician billing dataGP/FP Spirometry Billings per 1,000 populationDo not quote or circulate without permission of the author73areas of clinical practice. Clearly, the seven-fold difference in utilization rates between MetroToronto and rural districts does not have any justification. However, there is not enough clinicaldata to ascertain whether this is due to overservicing in the Toronto region or underservicing inNorthern regions. More significantly, the greatest difficulty in making the case for overservicingis the lack of specificity of practice guidelines. Guidelines developed in both the US and inCanada state that spirometry is important for diagnosis and may be useful for follow-up [NHLBI,1992; Ernst, 1996; Hargreave, 1990], but do not describe the indications for ordering the test fordiagnosis nor the frequency with which it should be used for follow-up. Furthermore, theguidelines do not specify clearly under what circumstances peak expiratory flow metermeasurements, a much cheaper but unremunerated test, may be substituted for spirometry.DiscussionAnalysts of administrative datasets frequently encounter patterns of utilization which do not haveany apparent justification. One common pattern is small area regional variation (SARVs); anotherpattern is growth in utilization of health services in the absence of any obvious change inprevalence of disease or population health status. Yet, most analysts are reluctant to claim thatoverservicing exists, based on these trends alone. SARVs may represent persistent underservicingin low rate regions, and several studies have shown that inappropriate care occurs in both highand low utilization regions [Leape, 1990; Chassin, 1987]. Similarly, rampant growth alone mayrepresent a response to unmet demand rather than overservicing.Despite these limitations, administrative data clearly have an important role to play in theidentification of overservicing. In all of these examples, administrative data analysis may beconsidered as a screening test, identifying patterns suspicious of, but not definitive enough toprove,overservicing. The administrative data analysis then pointed researchers to specific areasDo not quote or circulate without the permission ofthe primary author74for further study, from developing case studies in scenario 1, to assessing deviation from practicepattems expected from practice guidelines and clinical evidence in example 2, to further analysisof detailed clinical databases in example 3.Each of the first three case examples illustrates different approaches to addressing overservicing.Example 1 identified best practices and gave health care providers practical solutions on how toimprove the delivery of care. Example 2 led to an attempt to change reimbursement policy todiscourage overuse. The findings from example 3, in theory, could be used to reformreimbursement policy, or could be programmed into existing software supportingechocardiography machines to inform the technician performing the study when a Doppler studyis unnecessary.Trying to reconcile utilization pattems with clinical practice guidelines is an important tool forresearch. Data on prevalence of disease at different stages or different levels of complication cangive some hints as to whether overservicing is occurring. The example of prenatal ultrasound isone example of this approach. Other examples from our research include analyses of drugprescribing behaviour. Ranitidine ranks first in expenditures for individual drugs in the OntarioDrug Benefit Plan, and, along with other expensive H2 antagonists, are prescribed nine timesmore frequently than cimetidine, which costs two-thirds less. Current guidelines, however,suggest that for most patients cimetidine can be safely substituted [Centre for Evaluation ofMedicines, 1993]. Another example is the analysis of blood pressure medications. Highestexpenditures are for calcium channel blockers and ACE inhibitors [Anderson, 1996], both ofwhich are considered second or third line treatment for hypertension, according to Canadianconsensus guidelines [Reeves, 1993]. Data from these analyses have been used to supportcampaigns to encourage greater use of less expensive medications [ORTAP, 1997].Do not quote or circulate without the permission ofthe primary author75The greatest barrier to identifying overservicing, however, may not be the data themselves but thelack of specificity of many of the existing practice guidelines. This case is illustrated in example 4on spirometry. Should an asthma patient have a spirometry test once a year, once a month or atevery visit to the physician? Should smokers or patients in certain occupations at risk ofdeveloping respiratory diseases be screened routinely? Should peak flows be the mainstay offollow-up, with spirometry reserved for specific complications? The reluctance to add more detailto guidelines may be reflective of the general desire not to add too much complexity or rigidity tothe guidelines.One major area of our research is examining the behaviour of physicians in areas of ill-definedparameters on appropriateness. In such areas, a small number of physicians appear to positionthemselves at the high utilization end of the appropriateness spectrum. This is manifest through apersistent and marked right-skewing of utilization distributions for many different types ofservices. In example 4, a small group of physicians accounted for much of the seven-folddifference between high and low regions as well as 40% of expenditures. A recently submittedstudy on general practitioners and family physicians in Ontario billing over $400,000 annually hasnoted similar patterns and concerns about possible overservicing [Chan, 1997 (2)].The Berwick theory of quality management stipulates that efforts to improve quality or physicianperformance should be directed towards all physicians and not just outliers, or "bad apples"[Berwick, 1989]. The underlying assumption of this theory is that majority of the care is providedby the average typical physician, and focussing on the behaviour of outliers does not capture thebulk of clinical practice. Our research, however, shows that the right skewing phenomenon is sopronounced that outlier physicians typically account for a disproportionately large percentage ofphysician expenditures. Provincial governments have attempted to address this issue by imposingthreshold reductions, or decreases in fees paid for services once a physician's billings enter a highDo not quote or circulate without the permission ofthe primary author76billing zone [Barer, 1996]. These policies, however, are blunt instruments which do notdistinguish between appropriate and inappropriate service provision. One may speculate whethergovernments, operating in a context where difficult, immediate decisions on cost containment hadto be made, had enacted these policies on the imperfect information that such patterns weresuspicious, but by no means conclusive, of overservicing.The amount of clinical data available on databases varies widely. In Ontario, there are reasonablygood diagnostic data for hospitalizations, and an increasing amount of data documenting severityof illness. Physician billing data have some diagnostic information, but the quality of thisinformation is suspect. However, there are almost no data on clinical outcomes related toservices provided, collected on any system-wide basis. As such, inferences on overservicing fromadministrative data tend to be focussed on identifying care that is inefficient, rather than care thatis ineffective. Some attempts have been made to construct proxy measures for outcomes fromadministrative data; readmission rates, for example, have been used as a measure of unstablemedical condition at the time of discharge. Such measures, while useful, are clearly limitedbecause they are a function of other utilization patterns and not just the patient's condition.Collecting post-intervention outcomes data would be the ideal for health services researchers, butits great cost would make such efforts difficult to justify to cash-strapped governments. In theinterim, however, such data could be collected on a selective basis to investigate patterns ofconcern identified through administrative data analysis.Do not quote or circulate without the permission ofthe primary author77ConclusionAdministrative database analysis provides important insights to policy-makers on potentialoverservicing. Patterns suggestive of overservicing include regional variations in utilization,utilization growth in the absence of any change in patterns of disease or clinical indications for aservice, and utilization patterns which do not appear congruent from what would be expected ifclinical practice guidelines were followed. This type of analysis, however, should be considered ascreening test for overservicing; the definitive proof for overservicing, in most cases, should beobtained through more formal and detailed evaluations. An emerging area of research is thebehaviour of physicians in areas of clinical practice where there is little or no consensus onappropriateness, Our preliminary findings suggest that in these areas, some physicians appear toposition themselves at the high end of the spectrum of utilization patterns, and in the processaccount for a grossly disproportionate share of expenditures. Policy-makers need to arrive at aconsensus on how to deal with such situations, through options such as more detailed guidelinesand accountability mechanisms.Do not quote or circulate without the permission ofthe primary author78ReferencesAnderson GM. An analysis of temporal and regional trends in the use of prenatalultrasonography. In Royal Commission on New Reproductive Technologies. Current practice ofprenatal diagnosis in Canada. ISBN 0-662-21387-4, Ministry of Supply and Services Canada,Ottawa 1993. 509-534.Anderson GM, Chan B, Carter JA, Axcell T. An overview of trends in the use of acute carehospitals, physician and diagnostic services, and prescription drugs. In Goel V, Williams IT,Anderson GM, Blackstien-Hirsch P, Fooks C, Naylor CD (eds). Pattems of Health Care inOntario. The ICES Practice Atas. 2nd edition (Ottawa: Canadian Medical Association, 1996).Barer ML, Lomas J, Sanmartin C. Reminding our Ps and Qs: Medical cost controls in Canada.Hlth Affairs 1996;15(2):216-34.Basinski A. Use of hospital resources. In Naylor CD, Goel V, Anderson GM (eds). Patterns ofHealth Care in Ontario. The ICES Practice Atas. First edition (Ottawa: Canadian MedicalAssociation, 1994).Basinski ASH, Theriault ME. Patterns of hospitalization. In Goel V, Williams Jl, Anderson GM,Blackstien-Hirsch P, Fooks C, Naylor CD (eds). Patterns of Health Care in Ontario. The ICESPractice Atas. 2nd edition (Ottawa: Canadian Medical Association, 1996).Berwick D. Continuous improvement as an ideal in health care. N Engl J Med 1989; 320:53-56.Do not quote or circulate without the permission ofthe primary author79Black CD, Roos NP, Burchill CA. Utilization of hospital resources. Medical Care 1995: 33(12);DS55-DS72.Blackstein-Hirsch P, Basinski ASH, Naylor CD. Management of hospital bed utilization inCanada - current activities and potential strategies. Institute for Clinical Evaluative SciencesWorking Paper # 10 (June 1993), Toronto.British Columbia Provincial Advisory Committee on Clinical Resource Management. Acutemedical beds: how are they used in British Columbia? Victoria, 1997.Centre for the Evaluation of Medicines. The Optimal Prescribing Bulletin, 1993: 1(1). McMasterUniversity.Chan B, Cox JL, Anderson G. Trends in the utilization of non-invasive cardiac diagnostic tests inOntario during fiscal years 1989190 to 1992/93. Institute for Clinical Evaluative SciencesWorking Paper # 40, Toronto, 1995.Chan B, Cox JL, Anderson GM. Trends in the utilization of non-invasive cardiac diagnostic testsin Ontario from fiscal year 1989190 to 1992/93. Can J Cardiol1996; 12(3): 237-248.Chan B, Anderson G, Dales RE. Spirometry utilization in Ontario: practice patterns and policyimplications. Can Med Assoc J 1997; 156(2): 169-176.Chan B, Anderson GM. GP/FP's billing over $400,000 in Ontario: How do they do it? Can MedAssoc J 1997 (accepted for publication).Do not quote or circulate without the permission ofthe primary author80Chassin MR, Kosecoff J, Park RE. Does inappropriate use explain geographic variations in theuse of health care services? JAMA 1987: 258(18); 2533-7.Cox JL, Chan B, Anderson OM et al. Is colour flow imaging needed to exclude clinicallysignificant valvular regurgitation in adult patients undergoing trnasthoracic echocardiography?Can J Cardiol1997; 13(3): 261-269.Crapo RO. Pulmonary function testing. N Eng1 J Med 1994; 331: 25-30.Ernst P, Fitzgerald JM, Spier S. Canadian asthma consensus conference: Summary ofrecommendations. Canadian Respiratory Journal 1996: 3(2); 89-100.Gaasch WHo Diagnosis and treatment of heart failure based on left ventricular systolic or diastolicdysfunction. JAMA 1994: 271(16): 1276-80.Hargreave FE, Dolovich J, and Newhouse MT. The assessment and treatment of asthma: Aconference report. Journal of Allergy and Clinical Immunology. 1990; 85:1098-1111.Health Services Utilization and Research Commission (HSURC) of Saskatchewan. Barriers toCommunity Care. Saskatoon, 1994.Health Services Utilization and Research Commission (HSURC) of Saskatchewan. Routineprenatal ultrasound guidelines. Saskatoon, 1996.Joint Policy and Planning Committee. Non-acute hospitalization project (adult): Final report.Publication # RD6-3A, April 1997, Toronto.Do not quote or circulate without the permission ofthe primary author81Joint Policy and Planning Committee. Reducing length of stay: how do you compare? Aresource manual for Ontario hospitals. Case studies. Publication # 254, June 1994, Toronto.Johnstone DE, Abdulla A, Amold JMO et al. Diagnosis and management of heart failure.[Canadian Cardiovascular Society Consensus Conference]. Can J Cardiol1994; 10: 613-31.Leape LL, Park RE, Solomon DH et al. Does inappropriate use explain small-area variations inthe use of health care services? JAMA 1990: 263(5); 669-672.Manitoba Centre for Health Policy and Evaluation (MCHPE). Altematives to acute care.Winnipeg, 1996.National Heart, Lung, and Blood Institute. Intemational consensus report on diagnosis andtreatment of asthma. Clinical and Experimental Allergy. 1992; 22 (Suppl.1):l-72.Naylor CD, Paterson JM. Cholesterol policy and the primary prevention of coronary disease:reflections on clinical and population strategies. Ann Rev Nutr 1996:16;349-82.Naylor CD et al. Variations in selected surgical procedures and medical diagnoses by year andregion. In Goel V, Williams Jl, Anderson GM, Blackstien-Hirsch P, Fooks C, Naylor CD (eds).Pattems of Health Care in Ontario. The ICES Practice Atas. 2nd edition (Ottawa: CanadianMedical Association, 1996).Nelson SB, Gardner RM, Crapo RO, Jensen RL. Performance evaluation of contemporaryspirometers. Chest 1990: 92; 288-97.Do not quote or circulate without the permission ofthe primary author82Nishimura RA, Tajik Al. Evaluation of diastolic filling of left ventrical in health and disease:doppler echocardiography is the clinician's Rosetta Stone. 1 Am Coli Cardiol 1997: 30(1); 8-18.Ontario Round Table on Appropriate Prescribing (ORTAP). Information brochure, Institute forClinical Evaluative Sciences, North York, Ont., May 1997.Ontario Ministry of Health. Report of the expert panel on the modernization and tightening of thephysicians' schedule of benefits. Toronto.Tune 1997.Pagel PS, Grosman W, Haering 1M, Warltier DC. Left ventricular diastolic function in the normaland diseased heart. Perspectives for the anesthesiologist (2). Anesthesiology 1993: 79(5); 1104­20.Phelps CEo The methodologic foundations of studies of the appropriateness of medical care. NEngl 1 Med 1993: 329(17); 1241-5.Reeves RA, Fodor lG, Gryfe CI et al. Report of the Canadian Hypertension Society ConsensusConference: 4. Ypertension in the elderly. Can Med Assoc 11993; 149: 815-820.Society of Obstetricians and Gynecologists of Canada (SOGC). Guidelines for routine ultrasoundin obstetrics. Website http://sogc.medical.org/sogc_docs/public/guidelines/ultra4.htm, October1997.Stewart 1. Manitoba targets patient overuse as it tackles abuse of health care system. Can MedAssoc 1 1995: 152(9); 1483-4.Do not quote or circulate without the permission ofthe primary author83Thacker SB. Quality of controlled clinical trials. The case of imaging ultrasound in Obstetrics: areview. Brit JObs Gyn 1985:92; 437-44.Waldenstrom U et al. Effects of routine one-stage ultrasound screening in pregnance: arandomised controlled trial. Lancet 1988: 585-588.Do not quote or circulate without the permission of the primary author84NORALOU ROOSDiscussant PresentationWhen I met Klim McPherson coming into the meeting today he said that he is anunreconstructed small area variationist. I think that's how he introduced his first slide ongall bladder rates across the world. In thinking about it, and listening to the presentationsthis moming, I realise that I have become a reconstructed, small area variationist. What Ihope to show you today is how to look at the kinds of data that were presented thismoming, reorganise them a bit, make some sense of them, and in fact decide which is "theduck"; i.e., which rate is too high.I don't think anybody in the room could have listened to those presentations, seenthose graphs, and walked away saying there is no overservicing: Let's simply take it offthe table and go on to the next discussion topic. The papers, and certainly the discussions,have covered all the points which are useful in illustrating the degree to whichoverservicing exists. I'm not going to repeat that. However, it's very important toappreciate that despite the fact that overservicing exists, it's not easy to tell who are theover-servicers and who are the over-servicees. In fact, the Federal/Provincial committeewhich looked at issues relating to the control of health care costs asked me, along withDavid Naylor, to develop rules for identifying the overserviced, and we had to say it's notthat simple. That was our bottom line and we ended up walking away from that request,(although if I'd had that graph showing the forty-fold variation in spirometry, I'm not sureI'd have been so reluctant to identify overservicing). But even in that area, you'll noticethat Ben was saying there are clinicians who are going to make a very strong case thattheir practice style or the promise of the new technology will deliver benefits. These highservicing clinicians will argue that there will be patients who will suffer, who will not getthe best care possible, if they don't receive spirometry testing. What's going on?It occurred to me in reading these papers for the second, third, I'm not quite surewhich time, that there's a problem here with levels of analysis. All of the analyses whichwe typically use to demonstrate overservicing take place at the population level. So we'relooking at rates which describe populations. We then buy into the over-arching ideology85of the health care system, that more health care must be better. No matter what theamount of care delivered, there will be individual patients who benefit, or who mightpotentially benefit. Klim gave you the example of intensive care use. I think in his paper heestimates that 2,000 more patients might die if intensive care in Britain is not provided atthe highest level. So those are 2,000 lives that might be saved. We frequently get Lettersto the Editor saying, "What is the value of the human life? Are you saying you won't allowthis kind of practice if it has the potential for saving only one life?" Here is thefundamental dilemma. We have no problem demonstrating overservicing at the populationlevel. Our problem is identifying the individual who has been "overserviced",Life becomes much easier if one stops trying to identify the overservicedindividual, and stays focused at the population level, distinguishing between what isoverservicing, and what represents appropriate high levels of service to high need,unhealthy populations. Let me illustrate this alternative approach. In Manitoba we havenew regional health authorities and we've been trying to put together data which would beuseful to them in managing the system and helping them to decide when too many servicesare being provided, when too few; data which identify areas where they need to applymanagement skills to get that use down, and where they have high use which is in factappropriate. Here we focus on hospital use across physician service areas within oneregion of our province. What is the total number of days of hospital care provided percapita to residents in each of these areas') (See Figure I.)As you can see, there are two areas which are, by any measure, very high users ofhospitals, and we've got one area where use looks low. Now, whore's "the duck?" Well, Itoo used to wring my hands and say, "How can we say that these areas are high or thatthat area is low?" We have to look for something that's going to tell us that. I wouldargue that what you want to find is a measure of the health status of the populations forwhich you are comparing service rates. I'm using premature mortality here. In otherwords, we're looking at the population aged 0-74 years in these areas, and we're askingat what rate are these individuals dying (from all causes of death), over a recent five-yearperiod. This is a measure which is widely used as a measure of population health status,particularly in Britain and other countries. In Canada we are now using it to adjust needs-86Days of Hospital Care(Sub-Areas of Parkland Region)p{)eJsex adjusted days per 1,000 residentsGilbert PlainsDauphinAloneaSwan RiverRoblinPine CreeKMb ~ ;::',' '::( :II::' ' ,)./: '" ,'. ~'/';' ~. :',r, ".'. "~. .:o 500 1000 1500 2000 2500 3000••••11118111I'111••••• 11111I11 1111I IlIIi•• •• ••IImlll. • •••III... ..tlllI•• lUll •••• Ill. II.11l'lnJIlIIi••• IIU •••••87based funding formulae in several provinces, Indeed, when we look at those same areasand ask what are the health characteristics of the populations of these areas, we find thisvery interesting pattern, (See Figure 2 - Premature Mortality Rate.)These are exactly the same areas that you were looking at before, arranged inexactly the same order. Pine Creek, which as you recall had very high hospital use, has oneof the highest premature mortality rates in the province, so it has one of the least healthypopulations in the province, whereas Gilbert Plains, which had a very low hospitalisationrate, turns out to have one of the healthiest populations in the province.Alonsa was the other area that had a high hospital use rate. Now, looking at healthstatus of these residents, where's "the duck?" As a first cut (and I hope that you begin tobuy into this as a final cut), I would suggest that where you have high use amongpopulations of similar health status, the area with the very high use rate is a duck. This isan outlier, where usage patterns do not relate to the needs of the population. On the otherhand, we don't want to start trying to reduce the rate at which Pine Creek residents arehospitalized. Residents of this area have a high need for health care. Now, if you want toget a little more sophisticated, you might say, "Wait a minute. You're calling this a littleearly." What if Alonsa residents are also a high risk population? Perhaps by providinglarge amounts of hospital care to these residents relative to other populations, the healthcare system is reducing the health risks in a poor health population and this should havebeen considered a success story; with high health care use this population has achievedreasonably good health status. An interesting, important, alternative hypothesis.So, next we look at the socio-economic characteristics of area residents using asocio-economic risk index developed by researchers at the Manitoba Centre for HealthPolicy and Evaluation. We use a variety of indicators for identifying a high-riskpopulation, including things like unemployment, the value of an average house, andeducation rates. We've standardised and identified each area according to its deviationfrom a provincial mean, which is zero. So, across several of these areas, there's essentiallyno difference in the socio-economic characteristics of their populations: Alonsa residentsare no more at high socio-economic risk than are residents of the healthiest, low hospitaluse area - Gilbert Plains. However, residents of Pine Creek, the unhealthy, high users of88Premature Mortality Rates(Sub-Areas of Parkland Region, 1991-1995)~eJsex adjusted rajas ofprematura deathper 1,000residents age 0-74Gilbert PlelnsDauphinAlonsaSwan RiverRoblinPine CreekMbo 2 3 4 5 6 700•••• Ill•.......•• •• ••_ Bill a_IlUlIilli ••••IIllIa. • ••••• 1111 II.lUll •• ••IIS.WIllIS8S••1118••••MANITOBA CEN11ffi1: rOB F1EAler H ROUCY AND EVAWATION89hospitals, score more than two standard deviations higher than the provincial mean interms of socio-economic risk characteristics. (See Figure 3.)Let me state a controversial position: I am unaware of any study in a developedcountry (with high or low levels of health care) which has demonstrated a clearrelationship between health care supply and health status. You can find high levels of usein healthy populations, you can find high levels of use in unhealthy populations. I wouldargue that we need to change the burden of proof as to what underlies our assumptions asto which rate is right. After we've taken account of health status, or population needscharacteristics, the burden of proof has to move to those who are delivering high rates ofcare; those areas which are using high rates of specialist services, or high rates ofprocedures, need to prove that this is in fact beneficial to health. We should not beaccepting that more is necessarily, or plausibly, better.There was a physician in the audience at yesterday's CHSPR policy conferencewhose comment relates back to why it's so difficult to identify who the overserviced are.There was discomfort in the room when he said this, but he said, ask those 30% of theindividuals who are being overserviced, who are being over-hospitalised, to come into myoffice before I admit them, and identify themselves as overusers. I'll be happy not to admitthem to hospital. In fact, you researchers are telling me after the fact that this wasunnecessary, after they've been in hospital, after they've had their spirometry test. That'snot very useful to me as a physician.What I'm suggesting to you is to move this discussion of overservicing and relative"need" for health care to the population level, and to identify the population which hashigh socio-economic risk, low education, high unemployment. I can guarantee you eightout of ten times, if not nine out of ten times, that populations at high risk socio­economically are going to score poorly on measures of health status. They won't justscore poorly on mortality measures; they will also score poorly on self-reported health, interms of functional disability, in terms of acuity of condition, for which they're admitted tohospital. That high risk population is going to need high levels of health care andconversely, low risk populations are going to need low levels of health care. That's notnecessarily going to solve the dilemma facing the individual physician dealing with the90Socioeconomic Risk Index (SERI) Scores(sub-Areae of Parkland Region)Standard deviations from provincial mean (zero)Gilbert PlainsDauphinA10naaSwanRiverRoblinPino CrookMb (lero)0.0 0.5 1.0 1.5 2.0 2.5 3.091individual patient's problem. But from a planning perspective that is indeed going to helpyou sort out these issues.At last year's trilateral conference, I was beginning to make this kind of anargument with a fellow delegate and he said well, wait a minute. You're telling me thatyou can't demonstrate that health care makes a difference. At the same time you're tellingme we should be providing more health care to high need populations. Explain to me,please, how that makes sense. A not unrelated question emerged when I presented asimilar set of overheads focusing on differences across Winnipeg neighborhoods to a localaudience: the high hospital use high premature mortality area was that of the inner core.The audience member remarked that we were talking about a largely aboriginalcommunity which was influencing this very high premature mortality rate. He went on tosay, "We know the kinds of things which those people are dying from. They haveaccidents, homicides, suicides, and deaths which come from needle sharing and HIVinfection. Tell me what I'm supposed to do about this as a physician?" So we looked atthe causes of death among residents in the poorest Winnipeg neighbourhoods, andresidents of the more affluent communities. (See Figure 4.)When this is done in Manitoba, we find that in the lowest income neighbourhoods,8% are dying of accidents and infectious disease and HIV deaths, compared to 3% in thehigher income neighbourhoods. There is a reduced proportion of deaths from cancer inlow income neighbourhoods, but when you look at the actual number of deaths in theseneighbourhoods from cancer they are higher than those in the high incomeneighbourhoods. The size of the pie wedge indicates the proportion of deaths from thatcause. There are equal population numbers in each of these circles (that is, 20% of thepopulation in each). Still, we have four times the rate of homicides, accidents, and HIVdeaths. The total is 10% of the deaths in this group. In the other group it's about 3%. Sothere is a variation, but that's not what we're talking about in terms of the ill health burdenwhich individuals in these high-risk groups bear.If we look at neoplasms, it looks like we've got substantially fewer cancer deathsin the lowest income group than in the higher income group. Certainly 44% of the deathsin the highest income group are from cancer, compared with only 27% in the lowestType of Deaths Among Residents of Low and High Income NeighborhoodsAccidents4.2% ~Infectious Diseases>2.3%Homicides I.U /U '>1.4%Neoplasms27.4%Other34.9'%Lowest IncomeCirculatory27.9%Accidents1.4%InfectiousDiseas~0.9% ~HIV0.5% -----7 'iz~~~/'Homicides0.5%Other25.5%Neoplasms43.8%Highest IncomeCirculatory27.4%'Dtv93income group. However, if we look at the numbers of deaths which occur in Winnipegevery year, we had 161 deaths from this 20% of the population which were caused byneoplasms, and only 133 deaths from neoplasms in the 20% of the population which wasthe highest income group in Winnipeg. So we can't ignore the health problems of thesehigh-risk groups on the grounds that they are not relevant to the health care system.That's simply not true. These deaths, these diseases, look very much like those ofeverybody else in the system.To conclude my observations about what should one do with these kinds ofphysician practice variations data, the kind of small area variation data which we've seentoday, please note I am not talking about cross-country comparisons. What we're talkingabout is within a health care system where you have a given level of expenditure. Howdoes it make sense to allocate those expenditures if you're interested in doing your best,given the existing health of the population? It has nothing to do with where the next extradollar should be spent, whether it should be spent on CAT scans for low income groups orspent on early childhood education. That's not what we're talking about. What we'retalking about here is moving towards vertical equity in the delivery of care, given what wecan learn from exploring population based data. The quote which I like comes fromMaynard and Bloor in 1995 saying that health care funding should always be modest. Itsbenefits are unproved and clinical practice exhibits large, unexplained variations. That Ithink is what we've heard about in spades today. I think we've also heard aboutalternatives being developed in Ontario that may be helpful in dealing with overservicing.However, whenever we get into this overservicing debate, we must be conscious ofdistinguishing what is overservicing from what is legitimately high use driven by highhealth care needs. Usc this to find "the ducks."9495General Discussion after Morning Panel:Discussion began with the comment that expenditure planning for resource allocationrequires sophisticated measures. The mortality rate does not mean the same thing in developedcountries as it does in the undeveloped world. You need to look at the size of the small areas,for example, many range from the size of 8,000 to over 50-100,000. We need to try and getsome consistency in size. After controlling the number of beds in an area, the number of doctorsin the area, community care, and after recognising interaction between supply and need, itseems that 40% of the variability in utilisation is at the district health authority level in GreatBritain. Within these authorities, access to hospital beds makes no difference in utilisation butaccess to physicians does influence utilisation.Another participant remarked that approximately ten years ago Kaiser Permanente hadlooked at what could change physician behaviour. They decided to examine practice patternsfor physician outliers to determine the difference in their type of practice, and then they went onto ask these physicians why their type of practice was different. In many cases the physiciansrelated this to their past experience, either of litigation or of their own ill health. KaiserPermanente also sent utilisation data back to physicians with names of physicians attached andthis resulted in a reduction of utilisation of specific interventions. This group is now looking atuse of health status indicators in reducing utilisation.It was then remarked that overservicing is a major concem in Australia because theyhave had ten years of uncapped fee-for-service and have an estimated 4,000 surplus GP's. InAustralia, GP's can charge a user fee over and above the amount paid to them by the MedicalService Plan. Researchers find that in capital cities where GP's have an average of 982 patientsand earn $173,000, 7.5% of the GP's charged user fees, whereas in rural areas, where GP's havean average of 1362 patients and make and average of $176,000, 20% of GP's charge user fees.It was clear that where there was an oversupply of physicians, the practitioners did more workon smaller populations. He said it's very easy to get convictions of fraud, but no one's been yetconvicted for overservicing, even when taken to the Supreme Court. They are trying to linkadministration to peer review.One conference member remarked, in relation to Ben Chan's paper, how Saskatchewantook two years to make recommendations on ultrasound which run counter to the Society ofObstetrics and Gynaecologists of Canada guidelines. When published, the physicians said thatthe public pressure for ultrasound was driving the increasing utilisation. The Ministry of Healththen put out a communication to the public on ultrasound and the need for it. This upset theprofession and also the group that made the recommendation in the first place and they wantedto withdraw the recommendation. How do you deal with the main core of physicians whogenuinely believe that what they are doing is appropriate even though often it is not?Another conference member complained that there was a frustrating issue that no onehas been addressing, which is the underservicing issue. He stated that he knows there isoverservicing but the problem of underservicing is ignored by health economists who shouldgive it at least equal attention. For example, for pap smears, the shift has been from96overservicing to underservicing. He felt there has been a tendency to restrict the focus onvariables of interest to health economists, rather than variables that affect patient outcomes.With regard to the paper addressing the ICU study, the study addressed variables of interest tohealth economists rather then variables which may affect utilisation. For example, whatalternative care was provided in the absence of an ICU, and what are the costs of looking afterpeople in wards versus care in ICU's? The biggest factor affecting those going into hospital isthe availability of another place to care for that patient. In reference to Noralou's work, in area'YC' where large numbers are hospitalised, it may be because there are no other facilitiesavailable. The basic problem with this type of research is that it doesn't consider all the variablesthat could potentially affect the results.Another speaker had a question for Noralou regarding her presentation and Bob Evans'earlier discussion of the use of blunt instruments. What are the plans for areas where there areincreased hospitalisations despite decreased need? It was then noted that over-servicing is not ameaningful term because it implies a point where there is a need for a certain level of service.There is no correct level, it depends on how we choose to spend the money. Therefore, whathealth services should do is determine the budget, don't allow this budget to expand and thendistribute it equitably through weighted capitation. Unless this is considered, you end up facingBob Evans' warning about political versus technical solutions. He also asked about changes inthe health care system or health care services provision resulting from increased doctor training.For example, a change in the mix of available capacity can lead to a change in the whole healthcare system or shifts from effective health promotion to less effective interventions.Another speaker noted that we don't just operate within a medical care box, in whichwe've decided what to allocate to it. It's tough enough to operate within the box but we need toinclude expenditures on the environment and other health determinants in our allocationdecisions.Then followed responses from the presenters and the discussant. David Goodmanreplied that in the US, an increase in physician supply rapidly leads to further specialization ofthe workforce. Arguments about how this trend should be altered and the efficacy of alternativeinterventions are largely untested. There is an argument about how to halt the trend towardsspecialisation with a conflict of the stakeholders but there is no conflict about the efficacy ofworkforce. He wonders that if there is no further benefit from increasing supply, maybe weshould not worry about it. There are also cost issues. We have to know the benefits before wecan calculate the cost to benefit. If we don't know that, we don't know what we're buying.Ben Chan replied that he agreed with the statement that underservicing is as much aproblem as overservicing. For example, in his study, some of the spirometry rates in some areasof the province wore extremely small; it looks as though there is underservicing in those areas.Clinical practice guidelines do not give enough guidance on when procedures should be used,how they should be used, and who they should be applied to, and need to be much morespecific. Though the overservicing issue tends to focus on the supply side and incentives, wealso need to look at the demand side. For example, in Ontario there are a certain number ofGP's who have been earning over $400,000 a year, which would lead one to assume that there97was overservicing, but the College is unable to act because there are no clear guidelines on manyservices.Klim McPherson responded that we need to differentiate between macro and microoverservicing issues. Also, there's a need to educate the public regarding the potential downsideto more medical care. More does not always mean better. Research does not necessarilyanswer questions on the utility and the value of medical care. It is possible that in the UK theyare near the flat of the curve for intensive care unit beds, but they don't know whether they areactually on the flat of the curve yet. Researchers need to look at the social aspects of care aswell as the medical issues; for example, ultrasound may be important for women to see theirbabies although not medically indicated.Noralou Roos commented that small area variation data may be used to bring to theattention of policy-makers that some areas may be relatively underserviced. She will beincluding data about the availability of home care and nursing homes in coming studies. Inrelation to the question about whether to worry about over-supply, the answer is to look at BobEvans' equation.9899Afternoon PanelPrimary Care Reform100101Primary Care Reform in the US:The Best of Timesand the Worst of TimesKevin Grumbach, MDPrepared for the Second Invitational Ttilateral Workforce ConferenceVancouver, British ColumbiaNovember, 1997Primary Care Research CenterUniversity of California, San FranciscoBox 1364San Francisco, CA 94143USA415-476-8260415-206-5586 (fax)keving@itsa.ucsLedu102For most of the latter half of the 20th Century, primary care has been largely aneglected facet of the US health care system, obscured by the shadows cast by the citadelsof specialty medicine. These lofty icons of American culture were richly furnished withall of the latest technology that US affluence could afford, that a biomedically-entrancedpublic could desire, and that imaginative, well-intentioned physicians could devise. Thehumble OP in his fraying tweeds became a second class citizen. The cardiologist, slippingfrom Armani suit into surgical scrubs to enter the catheterization suite to inject, dilate,stent, and laser the coronaries of an increasingly sedentary, atheroslcerosis-ladenpopulation, became the new American medical hero.Has a new era arrived for primary care in the US in the 1990's? Is a convergenceof public opinion, policy consensus, and market prerogatives occurring that is reinstatingthe primary care practitioner to a commanding role in the health care enterprise? Areinnovative models of primary care being developed in the US that might prove instructivefor planners and policy makers in other nations?In this paper, I will briefly review the historical role of primary care in the US, anddiscuss its resurgence in the 1990' s. The key dynamic in the shifting stock of primary carepractitioners has been their assignment to the role of "gatekeepers" in managed care healthplans. I will discuss various features of this gatekeeper role, including organizationalstructures, patient flow, and payment and financial incentives, and how these changes maybe affecting both practitioners and patients. I will also highlight a few primary care grouppractices that may be considered innovative in the US context. (A note of caution here forthose who must skip to the end of a book and read the final passages first: what a USaudience considers innovative in primary care tends to be old hat in the UK and someother European nations.) I will conclude by offering opinions about the lessons that reformof US primary care may hold for observers from other nations.103The History of Primary Care in the US in 500 Words or LessIn the beginning, there were general practitioners. (There were also a fewsurgeons, but they evolved from barbers.) Then god created specialists. This occurred inthe early 1900's. The first specialist in the US was the ophthalmologist, the result of theinvention of the ophthalmoscope--establishing the well-respected pattern of subspecialtiesbeing defined by innovation of a new device or technology that became the signatureelement of that specialty's work. Specialties developed board certification processes andtraining programs, leaving OP's by 1970 as the only group of physicians in the US thathad no board certification credentials available and that required only a single year oftraining following graduation from medical school.By 1990, the US had 7 specialists for every OP-family physician. Oraduatingstudents faced virtually no limitations on choice of specialty. For residency trainingprograms, the "Field of Dreams" principle was in full operation: "If you build it, they willcome" (and the federal government will foot most of the bill).Chaos ensued. Primary care lacked definition as a particular scope of practice andexpertise. OP's emulated specialists by entering the hospital setting and attending tohospital inpatients as well as providing ambulatory care. Specialists proliferated to such anextent that there were insufficient numbers of patients with special problems to occupytheir practices. Many specialists began devoting a substantial proportion of their practiceto providing general primary care services outside of their particular specialty domain. Afamous study in the 1970's found that nearly 20% of patients reported relying on aspecialist as their principal source of care. Patients, insured by a fee-for-service system,directly routed their medical problems to the specialist of their choice, often entirelybypassing a OP in the process. A patient visited a dermatologist for a rash, agastroenterologist for a stomach ache, an orthopedist for a back ache, and a gynecologistfor an annual pap test. Coordination of care suffered, expenses soared, and rumblings ofdiscontent about the sorry state of primary care occurred among some of the public andpolicy leaders.By the early 1970' s, the rekindling of interest in primary care had led to someimportant developments in the US. In 1971, family practice became a board certifiedmedical specialty, transforming 1 year OP internships into 3 year, comprehensive familymedicine residency training programs. 1978 marked the issuance of the WHO's Alma-Altadeclaration and an influential report on primary care by the US Institute of Medicine.Experiments were occurring in developing innovative models of primary care practice,such as collaborative team models and community-oriented primary care. Usually, these104experiments took place in community health centers serving indigent patients or inteaching programs, and did not enter mainstream medical culture.The 1970's also marked the beginning of the modern HMO era, or at least therechristening of this type of insurance and delivery model. Plans such as KaiserPermanente had developed in the 1930's-1940's under the appellation "prepaid grouppractices." These plans were vilified by organized medicine as socialist conspiracies duetheir promotion of salaried group practice--an organizational model viewed as a threat tothe independent, entrepreneurial practice of private medicine. (The socialist label mayactually have been earned by some prepaid group practices that were founded asconsumer-physician cooperatives by left-leaning social activists. It is not certain whetherHenry Kaiser, the US industrial tycoon who formed the Kaiser-Permanente plan,appreeiated the irony of this characterization for the Kaiser health plan.) In 1972, prepaidgroup practices were "reborn" as "Health Maintenance Organizations"--a term coined byPresident Richard Nixon's health policy advisor, Paul Ellwood, to escape the politicallegacy conferred by the term prepaid group practice. Nixon's promotion of privatelyfinanced HMO coverage as a conservative alternative to government financed nationalhealth insurance led to the enactment of the HMO Act, which spurred expansion ofHMO's. (Nixon's overall health care reform proposal actually had many of its centralelements, such as mandatory employer-purchase private health insurance, eventuallyincorporated two decades later into President Clinton's health care reform program, givingcredence to the opinion of many political analysts that the US has a 1.5 party system thatspans the spectrum from right to right of center.)The HMO's of this early era were not, however, uniformly innovative in theirapproach to primary care. The flagship Kaiser Permanente plan of Northem California, forexample, eschewed hiring OP's and, later, family physicians, preferring instead to featureintemists, pediatricians, and internal medicine subspecialists as primary care providers.This decision to hire board certified specialists was in part driven by the desire to competewith office based physicians and avoid the criticism of second class medicine.By the mid-to late 1980's, primary care had reasserted its presence on the UShealth care scene, but was still struggling to gain full acceptance. Family physicians werereplacing the aging stock of OP's, but together FP's and OP's constituted only 13% of thetotal pool of practicing physicians. Many general internists and general pediatricians alsooccupied the primary care ecological niche, representing another 20% of physicians. Two­thirds of practicing physicians were non-primary care specialists. The average specialistphysician earned twice as much or more in net income as the average primary carephysician. Nurse practitioner and physician assistant programs had come into existence,105and were training modest numbers of primary care practitioners who faced widespreadrestrictions on their ability to practice independently from physicians. Many primary carepractitioners strived to provide continuity and coordination of care for their patients, butthis was not enforced by any administrative limitations on self-referral to specialists orrequirements to formally register with a particular primary care practitioner.When I graduated from medical school in 1985 and decided to train in familymedicine, several medical school faculty admonished me for wasting a promising career inmedicine and sentencing myself to a future of drudgery in primary cm·e. No familyphysicians were included in the UCSF medical center's clinical practice group at that time.And the Winner is, Primary Care (...or is it?)There is a popular saying in the US: 4 times out of 5, the right thing happens forthe wrong reason. This seems to be the case for the sudden ascendance of primary care inthe US in the 1990's.Leading policy analysts, blue-ribbon government commissions, prestigious privatefoundations, and groups of professionals and citizens have all in the past decadepronounced the wisdom of reorienting the US health care system toward more of aprimary care foundation. Yet the most influential factor promoting a stronger role forprimary care has been the actions of managed care health plans operating in a competitivemarket for health insurance. Although some enlightened managed care firms may, perhaps,manifest a deep appreciation of the full virtues of primary care, the most alluring feature ofprimary care for managed care firms has been its perceived utility as a means of containingcosts.Managed Care Structures in the USUnderstanding how managed care has changed the dynamics of primary care in theUS requires a basic understanding of the nosology of HMO's. The first generation ofHMO's in the US were the non-profit plans such as Kaiser-Permanente, Group HealthCooperative of Puget Sound, and the Health Insurance Plan of New York. These were thetraditional "prepaid group practice" models in which a single, large physician groupprovided care on a prepaid (i.e., capitated) basis to subscribers of the health plan. Theseplans became known as staff-model HMO's or group-model HMO's. In staff-modelHMO's, the plan directly employs physicians (e.g., Group Health Cooperative) (Figure 1).In the group-model, the plan contracts with a single, semi-autonomous physician group(e.g., the Permanente Group, which serves patients in the Kaiser plans) (Figure 2). As106mentioned above, these traditional I-IMO's varied in the extent to which they promoted aprimary care model of care. Although some embraced family physicians and encouragedcontinuity of care with a personal, primary care provider, others functioned more as largeclinic operations (or "poly-clinics" in the European mode) and did little to emphasizeprimary care and continuity of care.The traditional staff- and group-model I-IMO's exemplify "vertical integration" ofcare. Under one organizational roof and common ownership, these plans consolidated alllevels of care, from primary to tertiary care, and the facilities and staff necessary toprovide this full spectrum of care.An altemative to the traditional group- and staff-model I-IMO's developed as asecond generation of managed care plans in the 1970's and 1980's. These I-IMO's, whichhave come to be known as "network" or "IPA (Independent Practice Association)"I-IMO's, differ in several ways from the first generation I-IMO's. The second generationI-IMO's have not built vertically integrated systems under common ownership, but rely oncontractual relationships to piece together the various elements of a full-service plan (whatsome observers have called "virtual integration" rather than vertical integration). TheseI-IMO's contract with a variety of physicians in office practice, rather than employing theirphysicians directly or contracting with a single large group (Figure 3). Unlike themonogamous arrangement between the Kaiser plan and the Permanente Medical Group,physicians in second generation I-IMO's may establish promiscuous contractualrelationships with numerous HMO's. The result of this more open I-IMO-physicianrelationship is a series of physician panels in the same community that overlap partially,but not completely, for patients covered by different HMO's.Although many second generation HMO's were initially created by associations ofphysicians and hospitals as locally-based, non-profit organizations, most of these I-IMO'shave been purchased by large, for-profit commercial insurance plans. As these large,commercial insurers realized that they needed managed care products to compete in anincreasingly price- sensitive insurance market, they bought up fledgling organizations andconsolidated them into large corporations. Whereas the first generation I-IMO's grewslowly over decades and built a cohesive practice culture among physicians who soughtout these alternative systems, the second generation I-IMO's have been referred to as"Instant Soup" I-IMO's: just add water (or capital, as the case may be) to an assortment ofphysicians and hospitals, stir, and voila, an HMO! Most of the growth of enrollment inI-IMO's in the past 10years has been in these network/IPA style plans, and most of theseplans have become for-profit, investor-owned entities during that time.107Staff Model HMOVertically IntegratedSystemFigure 1Health Plan_PionHospitals& StaffFacilitiesGroup Model HMOHospitalsPhyelclans­TeamsPharmaciesFigure 2Figure 3IPAM~Health Plan108Adding to the confusion of comprehending the multiplicity of practiceorganizations developing in the US, physicians have responded to these networklIPAHMO's by creating some new practice structures at the provider level. One of the mostcommon of these structures, especially in a "vanguard" managed care state such asCalifornia, is the Independent Practice Association. IPA in this case refers not to the typeof managed care plan, but to a particular physician organization. (My apologies for thehopelessly convoluted language of US managed care!) An IPA in this sense of the termconsists of a network of physicians who agree to participate in an association for purposesof contracting with I-IMO's and other managed care plans. Physicians maintain ownershipof their practices and administer their own offices. The IPA serves as a vehicle fornegotiating and administering HMO contracts. The IPA also accepts the capitationpayment from I-IMO's and distributes these revenues to the physicians participating in theIPA.)The traditional group- and staff-model I-IMO's were not originally developed tocontrol costs. Instead, they were designed to be more responsive to patient needs andinterests and to provide physicians an alternative to the dominant entrepreneurial, fee-for­service private practice system. As cost containment became a more pressing policy issue,however, the traditional HMO's recognized that they had many assets in this regard. Theypaid their physicians on a salaried basis, had prospective budgets, regionalized facilities,and tended to operate their hospitals at near-maximal occupancy. In essence, they had ahandle on managing supply-side capacity. In fact, these I-IMO's were the closest thing inthe US to a national health service, with similar cost-control tools at their disposal.Because of this situation, innovation in primary care became less of a central issue for costcontrol in these organizations. Most consistently, these I-IMO's simply kept their staff ofspecialists fairly lean.109The second generation HMO's faced a different situation. They had less controlover capacity in their loosely assembled provider networks, had little means for enforcingglobal budgets, and had not steeped their physicians in a practice culture that respectedparsimony. Their cost containment tools became cumbersome, case-by-case utilizationmanagement, price discounting, and most germane to primary care, use of primary care"gatekeepers" paid by capitation.The Gatekeeper StructureA typical HMO capitation arrangement in California is shown in Figure 4. AnHMO pays an IPA a capitation sum for each of the plan's subscribers who enrolls with aprimary care physician in that IPA. (In US managed care parlance, the HMO pays a"pmpm" (per member, per month) fee for every "covered life" (enrolled patient).)Typically included in this capitation payment are funds to pay for specialty consultantvisits and procedures as well as for laboratory, radiology, and related ancillary services.Surpluses in these referral "risk pools" are typically returned (at least in part) to primarycare physicians, augmenting the net income of these practitioners. In many managed careplans, these incentives and bonuses may comprise a substantial portion of the physician'sultimate income, creating a potentially powerful incentive to restrain referrals,CapitationIndependentPracticeAsaociation;~~;:; : ··'II>.:;}/: ".' ... ,/::: ......, r', .•.. ~, I ".,'I : ........ ~.' Referral: ~ services, '~BonUs, " ~'", "'.~'",,,• 46:511PrimaryphysiciancapitationHealth planFigure 4110This shift to capitation linked to an individual primary care physician marked aradical departure from business as usual for patients and physicians in the US. All of asudden, several things happened: patients in these plans had to register with a primarycare physician. Moreover, patients now had to initiate all of their care with this primarycare physician. No more would patients be able to exercise unlimited choice and directlyshop their organ-specific symptom to the specialist of their choice.Undeniably, this organizational shift brought tremendous potential to empowerprimary care physicians and elevate the role of primary care in the US. In theory, primarycare physicians would be better able to coordinate referral services for their patients,avoiding inappropriate self-referrals; no longer would patients disappear into the blackhole of specialty care, only to return several months later with a variety of scans, incisionsand missing organs. Primary care physicians would be liberated from prejudicial fee-for­service pay schedules that failed to reward talking and management services. Primary carephysicians could exercise greater flexibility and control over services by managing acapitated budget. Primary care physicians would have an explicit panel of patients, andcould begin to practice population-based medicine. The Golden Age of Primary Caremight finally arrive!Unfortunately, the peculiarities of the US system have left many primary carephysicians worrying whether it's an era of Fools Gold. Although some managed careorganizations may be aspiring to implement a visionary model of primary care, researchhas failed to document a systematic improvement in primary care practices in HMO'scompared with the more traditional fee-for-service sector. The liabilities of a fragmentedand non-universal system of financing health care also undermine potential progress on theprimary care front. Primary care physicians, rather than relishing a revival of their calling,often appear beleaguered by the pressures of an unstable competitive market, questionablefinancial incentives, and threats to their autonomy.In the following section, I will highlight recent research on the current status ofprimary care in the US, including presenting some recent findings from the work of ourteam at the UCSF Primary Care Research Center.IIIEvidence on Primary Care Performance in the Changing US SystemPrimary Care-nessSeveral research groups have developed instruments relying on patient self-reportto evaluate the degree to which patients receive care that satisfies the essential elements ofprimary care, such as continuity, comprehensiveness, coordination, accessibility, andeffective communication or "partnership." These instruments have been dubbed "primarycare-ness" measures.Dana Safran and colleagues have used these measures to compare the primarycare-ness of HMO and fee-for-service care in the US. Theil' first study, conducted in theearly 1990's, found that HMO's did not achieve uniformly higher marks on primary carescores. A more recent study, conducted within the past 2 years and as yet unpublished, hasfound similar patterns.One of the most basic elements of good primary care-continuity with a personalpractitioner-vis frequently undermined in the US by involuntary disruptions in insurancecoverage. One disruption is complete loss of health insurance. Patients may lose theiremployment-based private insurance if they become unemployed, self-employed, orchange jobs. Patients also frequently move in and out of coverage by the public Medicaidinsurance program. A forthcoming study shows that two-thirds of new Medicaid enrolleeslose their Medicaid coverage within 12 months.Even without completely losing insurance, patients may face severance of theirprimary care relationships. Many employers offer their employees only a single managedcare plan. Employers may change the managed care plan they offer in response to changesin premium prices. A new plan may contract with a different array of primary carephysicians in a region, resulting in a patient being forced to select a new physician fromthis different provider panel. One recent study found that 25% of patients in managed careplans said that they had been forced to change primary care physicians within the past 2years because of "changes in their insurance plan." (Flocke et aI., 1997). The patients whohad experienced forced discontinuity rated their current physician significantly lower onseveral primary care-ness measures than their counterparts who had not been forced tochange their primary care physician.The Primary Care Physician's Experience ofManaged CareStudies have suggested that physicians overall in the US are experiencingdiminishing satisfaction with the practice of medicine. A 1995 survey of US physicians112(including both primary care physicians and specialists) found that 55% of physiciansbelieved that the health care system became worse in the past year (Donelan et al., 1997).Physicians practicing in states with high managed care market penetration were muchmore likely to rate the system as getting worse than physicians in low managed carepenetration states. In this same study, two-thirds of physicians reported problems with"movement of patients in and out of your practice because of changes in insurancecoverage," and more than half reported problems with limitations or administrative hasslesin referring patients to specialists. Another recent study conducted among physicians inMinnesota found similar patterns (Borowsky et aI., 1997). Many physicians expresseddissatisfaction with utilization authorization procedures and other polices to restrict care inmanaged care plans.A 1995 survey of primary care physicians also found that physicians tended to beless satisfied with the care they were able to provide patients covered by capitatedmanaged care contracts than with the care they could provide to non-capitated patients(Kerr et aI., 1997). The primary care physicians were less satisfied with the quality ofcare, their ability to treat patients according to their own best judgment, and their ability toobtain specialty referrals for capitated patients.Some of our recent research at UCSF suggests that part of this dissatisfaction maybe related to the new financial incentives primary care physicians are experiencing inmanaged care plans.In 1996, we surveyed nearly 900 primary care physicians in practice in California.Over 90% of these physicians either worked at a group model HMO such as Kaiser or hadat least 1 contract with an IPA or network-model HMO. We questioned physicians aboutthe types of financial incentives they were experiencing under managed care. Over one­third of physicians reported that they had some sort of bonus or incentive arrangementover and above their base salary, capitation, or fee-for-service payment. These incentiveswere linked to different performance measures. 14% of all physicians had an incentivebased on restraining use of referral and ancillary services, and nearly 20% had an incentiverelated to their overall productivity. 23% also had an incentive targeted to measures ofquality of care or patient satisfaction. For three-quarters of physicians experiencing abonus incentive, their individual performance-sand not just their overall physician group'sperformance-factored into calculation of their bonus.113Table1Percent of Primary Care PhysiciansExperiencing Bonus IncentivesType of IncentiveReferral useHospitaluseProductivityQuality/PatientSatisfactionAny type of bonusincentive% Experiencing1419182038source: California Primary Care PhY5idull Survey, 1996Includes PCPs Involved in managed cereWe also asked physicians about whether they felt pressured about certain aspectsof their practice: scheduling many visits per day, not referring patients to specialists, andnot fully informing patients about treatment options. For each of these types of pressureson their practice, we asked physicians to indicate one of the following responses: "don'texperience," "experience it but it doesn't compromise care," or "experience it and itcompromises care." Figure 5 shows the distribution of responses to each of thesequestions. Overall, a substantial proportion of primary care physicians reported pressuresthat they believed compromised care in these areas.We then examined theassociation between bonusincentives and the806040% of PrimaryCare Physicians20l1li-. ..Experience w/lhaul compromiling care... .. ExperlenC<l willi eompromlllnll c....oFigure 4Practice Pressures Experienced byPrimary Care PhysiciansL1mJl # ofreferralsSee large # ofpatientssource: California Primary Care Physician Survey, 1996Includes PCPs Involved in managed careLim It wh at I tellpatientsexperience of pressures inpractice. In multivariatemodels that controlled for avariety of practice andphysician characteristics,the presence of a bonuswas one of the factors moststrongly associated withreported experiences of pressures that compromised care. Presence of a referral-relatedincentive predicted the experience of high pressure to limit patient referrals, and thepresence of a productivity incentive predicted the experience of high pressure about seeing114a large number of patients. Some of these bonus incentives were also strongly andnegatively associated with primary care physicians' sense of satisfaction in their work.Table 2Association Between Incentives & PracticePressures (Adjusted Aualyses)Type of Bcnus-IucentiveType of Pressure Referral Bonus Productivity BonusTo Limit Referrals 0.24** 0.15To See More Patients RO.I0 0.19**p<.05**p<.01Figuresarc regressioncoefficientspredictingpressure, on a scaleof 1 (no pressure)to 3 (pressure comprisingcare).Another recent, unpublished study has also revealed the flip side of the coin for thenew, expanded role for primary care physicians. In this physician survey, many primarycare physicians reported that they felt pressured by managed care policies that discouragereferrals to expand their scope of practice beyond a level that was comfortable for them(St. Peter et al., 1997). Nearly one-third of primary care physicians reported that theirscope of practice had increased in the past 2 years, and 41% of these physicians believedthat their scope was now "greater than it should be."Our own team's survey of primary care physicians also documented anotherdynamic that may be interfering with physicians' ability to fully appreciate their new role.Although concerns about excessive supply and potential unemployment among physicianshave been most acute for specialists, primary care physicians have not been completelyimmune to the unsettling dynamics of the competitive market. We found that almost onequarter of primary care physicians in office-based practice in California had experienced amanaged care plan either denying a contract application or terminating an existingcontract. Concerns about job stability and control over one's practice in the managed caremarket have not left primary care practitioners untouched.The Empire Strikes BackGatekeeping has become a central issue in a mounting public backlash againstmanaged care in the US. In my opinion, managed care in the US deserves sound criticismlISfor its fixation with the corporate bottom-line, its lack of public accountability, and itsscandalously excessive rates of administrative overhead and profits (many plans operatewith only 75 cents of every premium dollar being spent on medical care). However,gatekeeping has become a lightning rod for public dissatisfaction with managed care.Americans believe that they have a constitutional right to life, liberty, and the pursuit ofthe specialist of their choice. They resent new restrictions on their access to care,especially changes that compromise privately insured patients' previous ability to directlyrefer themselves to a specialist.Specialist physicians have not, to say the least, discouraged public resentmentabout these managed care arrangements. They have worked both through political actionand public policy (e.g., laws that require women to be able to have "direct access" toobstetrician-gynecologists without needing a referral from a primary care physician) andthrough research. US medical journals now frequently carry research articlesdemonstrating that cardiologists are better than generalists at treating patients withmyocardial infarction, or that patients with rheumatoid arthritis fare better if a specialist isinvolved in their care. This "Empire Strikes Back" literature tends to focus on singledisease entities, rather than comparing specialist and generalist care for the whole patientand the patient's full complement of medical and psychosocial needs.A new development that is posing a major threat to primary care in the US is thephenomenon of "disease management." Disease management initially was the brainchild ofthe pharmaceutical industry. Pharmaceutical firms marketed medical care packages to takeon management of specific chronic diseases, such as asthma and diabetes. These programsdid not catch on terribly well. Many providers and organizations suspected thatpharmaceutical companies were more interested in promoting their own drug products forthese conditions than in pursuing a non-denominational approach to enhancing the qualityof care.However, organizations other than pharmaceutical manufacturers have nowdeveloped disease management programs, and these new programs are beginning to catchon with the managed care industry. Typically, these programs are organized around a teamof practitioners headed by a physician specialist and include nurse case managers,nutritionists, health educators, and other personnel--but not primary care physicians. Theprograms focus on single chronic disease entities, such as congestive heart failure, andremove care of these conditions from the primary care setting. Often, payment for theseprograms is "carved out" of the primary care capitation fee to the primary care physicianand transferred to the disease management team. Many of these programs tout their116success at reducing hospitalization rates and costs and improving functional status,although most documentation consists of internal evaluations performed by the diseasemanagement firms themselves rather than rigorously performed, peer-reviewed publishedevaluations.These programs represent a centrifugal model of care in which patients withchronic disease are pulled away from the primary care practitioner and into the specialist'sdomain--at least for the particular identified chronic condition. The fact that many of thesetypes of patients forget to limit their chronic ailments to just one condition, and tend tohave not just heart failure but also diabetes and depression and chronic lung disease andosteoarthritis, raises obvious questions about the success of these disease managementprograms in caring for the whole patient. However, there is a perception in some quartersthat primary care providers have not performed adequately in the routine care of patientswith chronic disease, and these specialty-focused models are marketed as superiorproducts in the managed care environment. Few organizations seem to have thoughtcreatively about more centripetal models of collaborative care that keep care based in theprimary care setting and harness specialty and interdisciplinary resources to supportprimary care practitioners in their management of patients with chronic disease.Diamonds in the Rough:Some positive models of innovative primary care in the USDespite these dispiriting trends in the US, there are some instances of good thingshappening in primary care. (I must confess, however, that I solicited the opinions ofseveral experts in the area of health policy and primary care in the US to help me identifysome shining examples that would impress an international audience. I needed to take anextra Prozac in response to the difficulty everyone had in identifying such examples. Iftruth be told, I learned about the first of the following models while attending a meeting inCanada sponsored by Michael Rachlis.)117The Community Medical AllianceThe Community Medical Alliance (CMA) of Boston was developed in 1989 as analternative model for caring for patients with severe disability, particularly patients withquadriplegia. Dr. Bob Masters, the director of this program, observed the extremefragmentation of care provided these patients under the conventional Medicaid program inMassachusetts, with vast amounts of money being spent on things like hospitalization forsevere urinary tract infections that could have been managed at home if they were detectedsufficiently early in their course. The CMA's organized a primary care team that relies onnurse practitioners as front-line providers in the patients' homes, with back up fromprimary care physicians. Nurse practitioners serve as case-managers for panels of 25-50patients, and function as "gatekeepers" to facilitate access to early interventions ratherthan to restrict services. Various other personnel are included, such as social workers andphysical therapists. Masters negotiated with the Medicaid program to have care paid foron a capitated basis, allowing the primary care team to control financial resources andreallocate them in a more efficient manner. The contracts include limited risk sharing bythe program and reasonably plentiful payment levels. The program proved successful atimproving quality and value, and has expanded to include patients with advanced AIDS.Certain features of CMA are important to note: The program has remained non­profit and small, with a case load of approximately 400 patients. Because of its relativelysmall size, it has not become a significant threat to specialists and hospitals because it hasnot diverted major financial sums away from these sectors relative to the overall medicalcare expenditures in the region. Enrollment by patients is voluntary. CMA was created byindividuals committed to a clear vision of a better model of primary care, and has remainedunder their leadership.Group Health Cooperative ofPuget SoundOne of the oldest and most respected of the traditional staff-model HMO's in theUS is Group Health Cooperative (GHC) of Puget Sound. Located in the Seattle area andserving a population of almost one-half million enrollees, GHC was founded as a non­profit, consumer-directed cooperative. It recently merged with the Kaiser-Permanentesystem.GHC has become a model of a "centripetal" approach to chronic diseasemanagement (Wagner et aI., 1996). GHC has attempted to take advantage of the capitatednature of enrollment to define registries of patients with chronic diseases and tailor118interdisciplinary, primary care based models to these patients' needs. These models,termed "Population-Based Management," begin by identifying patients and then workwith practitioners to develop and implement evidence-based care guidelines, buildinterdisciplinary teams to augment care from an individual primary care physician or nursepractitioner, and organize group visits and educational sessions. Many of the elements ofthis program are derived from European innovations such as the UK "chronic care mini­clinic" and the German primary care based group education sessions. One of GHC's initialpopulation-based models had focused on diabetes care.Dr. Ed Wagner of GHC's Center for Health Studies has commented, "[There is]genuine uncertainty among managed care leaders as to whether to enhance their primarycare system's ability to provide chronic illness care, or to delegate care to specialized careproviders such as medical specialists or case managers for various patientgroups....Environmental pressures for the [specialist] approach come from theoverabundance of specialists, increasingly intense marketing of 'disease management' orcarve-out programs, and cultural beliefs in the superiority of specialists over generalists."GHC is attempting to demonstrate the advantages of resisting the specialized model andintegrating chronic disease management into the primary care team.Lessons from the US ExperienceWhat are the lessons in all of this? The first lesson is that a culture must exist or bedeveloped that values and supports primary care, particularly among the public and amongpractice organizations. It is not easy to wean patients and practice organizations from along-standing over-dependence on specialty care, especially when it threatens the self­interests of specialists.Second, patients need to be convinced that they will stand to gain something undera system that insists that they register or enroll with a primary care physician and thatinsists that the primary care physician help to coordinate their care. Otherwise, the primarycare "gatekeeper" (such an unhappy choice of words) runs the risk of being perceived assimply the gateshutter, deterring patients from desired specialty care.Third, a sure way to undermine public confidence in, and physician enthusiasm for,a capitated, primary care model is to saddle it with ethically suspect financial incentivesand to place it under the auspices of unabashedly profiteering organizations.119A corollary of this lesson is that true innovation in primary care is unlikely to occurin a context of organizations fixated on immediate cost savings rather than on long-terminvestment in enhanced models of care. Hard work, long term vision, and the buy-in ofpractitioners are required to shape better models of care.The final lesson is that empowerment of primary care by giving primary careproviders control of a capitated budget may not always have the desired effect. Moneymay be power, but it also may mean having to act an awful lot like an insurance company,with the attendant burdens of administrative and budgetary-driven decisions for individualpatients.Dickens best sums up the peculiarly ambivalent state of managed care and primarycare in the US:"It was the best of times, it was the worst of times, it was the age of wisdom, itwas the age of foolishness, it was the epoch of belief, it was the epoch ofincredulity, it was the season of Light, it was the season of Darkness, it was thespring of hope, it was the winter of despair, we had everything before us, we hadnothing before us, we were all going direct to Heaven, we were all going direct theother way."For a primary care physician in the US, the experience of managed care is just sucha lesson in the "superlative degree of comparison only." Primary care physicians are loved,and they are defiled. They are the answer to the problems of too easy access to specialtycare, and they are the cause of the problem of too difficult access to specialty care. Theyare empowered by controlling capitated budgets, and they are damned for controllingbudgets. They have been liberated from the evil temptations of fee-for-service, and havebeen seduced by the wickedness of financial incentives to limit care. They are expected tocompensate for a seriously delinquent system of financing health insurance, and they arevictims of the grave limitations and discontinuities of the fragmented financing system.They are the type of physician everyone really needs, unless of course the person isactually ill, at which point a more qualified specialist should take over.We await the unambiguously golden age of primary care in the US.120ReferencesBodenheimer TS, Grumbach K. Understanding Health Policy: A Clinical Approach. 1sted. Stamford, CT: Appleton & Lange; 1995.Bodenheimer TS, Grumbach K. Capitation or decapitation: keeping your head in changingtimes. JAMA. 1996;276: 1025-1031.Borowsky SJ, Davis MK, Goertz C, Lurie N. Are all health plans created equal? JAMA.1997;278:917-921.Donelan K, Blendon RJ, Lundberg GD, et al. The new medical marketplace: physicians'views. Health Affairs. 1997;16 (5):139-148.Flocke SA, Stange KC, Zyzanski SJ. The impact of insurance type and forceddiscontinuity on the delivery of primary care. The Journal ofFamily Practice.1997;45: 129-135.Kerr EA, Hays RD, Mittman BS, et al. Primary care physician's satisfaction with qualityof care in California capitated medical groups. JAMA. 1997;278:308-312.Master R, Dreyfus T, Connors S, et al. The community medical alliance: an integratedsystem of care in greater Boston for people with severe disability and AIDS. ManagedCare Quarterly. 1996;4(2):26-37.Safran DG, Tarlov AR, Rogers WHo Primary care performance in fee-far-service andprepaid health care systems. JAMA. 1994;271:1579-1586.St. Peter R, Reed M, Blumenthal D, Kemper P. The scope of care provided by primarycare physicians: physician assessments of change and appropriateness. (Presented atAnnual Meeting of Association for Heath Services Research, Chicago, June 1997).Wagner EH, Austin BT, Von Korff M. Improving outcomes in chronic illneses. ManagedCare Quarterly. 1996:4(2): 12-25.121PRIMARY CARE REFORM IN THE UKAngela Coulter, PhDKing's FundPaper for Second Invitational Trilateral Physician Workforce Conference, Vancouver, 15-16November, 1997The primary care-led NHSThe universal problems in health policy - how to cope with increasing demands resulting fromdemographic change, technological advance and the gap between public expectations andwillingness to pay - are just as intractable in a cash-limited, taxation-funded system such as theBritish NHS as they are elsewhere. The recognition that countries with more highlydeveloped systems of primary care tend to have lower health care costs (Starfield 1994) hascontributed to the widespread search for ways of shifting the balance of power and resourcesfrom secondary to primary care. The 1990' s have seen a number of policy initiatives in theUK aimed at controlling demand for expensive secondary care services by investing in primaryand community care in the hope that this will deliver greater efficiency. In recent yearsDepartment of Health rhetoric has been stressing the need to develop 'a primary care-ledNHS'. This paper examines what is meant by a primary care-led health service in the Britishcontext, traces its evolution and achievements and speculates on future developments.General practice in the UKThe relatively well-developed primary care system, with general practice as its cornerstone,has always been considered one of the great strengths of the British National Health Service.General practitioners (GP's) are independent contractors and patients can choose to registerwith any GP in their locality. The registered list system facilitates co-ordination of servicesand continuity of care and provides population registers for use in screening and otherpreventive services. The GP is responsible for general medical care and for acting as agatekeeper to the rest of the service, referring patients to hospital-based specialists orcommunity health services (nurses, midwives, physiotherapists, occupational therapists,speech therapists, counsellors, dieticians, podiatrists, social workers, etc.) when necessary.Virtually the whole population is registered with GP's who retain a monopoly of medical careoutside the hospital. In contrast to hospital doctors who work as salaried employees, GP'sare self-employed independent contractors. The majority work in group practices on apartnership basis, owning their own premises and employing administrative and some nursingstaff. Of the 29,000 GP's in England, about 10% work as single-handed practitioners, 47%are in partnerships of two to four doctors, and 43% work in groups of five or more. Theaverage list size per GP is 1,900 registered patients. On the whole patients do not have directaccess to non-emergency hospital or community health services, except via referral from their122OP. The OP's role as gatekeeper to hospital care is one of the main reasons why the NHS iscost-effective relative to other systems of health care delivery.However, during the 1980' s it became clear that demand for health care was outstripping thesupply of services. The most obvious manifestation of the mismatch between demand andprovision was lengthening waiting lists, but there was also growing awareness of unexplainedvariations in practice patterns. Studies of OP's' prescribing patterns and referral rates drewattention to rising trends and unexplained variations between general practices, suggestinginefficiencies and a lack of consensus among doctors about appropriate treatment strategies(Acheson 1985, Forster and Frost 1991). Policy analysts pointed to perverse incentives in thesystem which tended to fuel demand and inefficiency (Enthoven 1985, Maynard et aI1986).OP's did not have to consider the cost implications of their decisions to initiate treatment orrefer patients to hospital specialists and the system of funding hospitals according to bedprovision rather than throughput did little to reward efficiency.The internal marketFollowing a review of alternative methods of funding and organizing health services, theThatcher government opted to preserve the existing taxation-based funding system and toconcentrate on various methods to increase efficiency. This was to be achieved by devolvingresponsibility, promoting competition and encouraging consumerism. The main mechanismwas the introduction of an internal market in which health care purchasing was to be separatedfrom the provision of services (Department of Health 1989). The internal market was to beestablished by encouraging hospitals and community units to become self-governing providertrusts. In this way they were to be separated from direct management by the district healthauthority (DHA), allowing them to control their own finances, staffing resources and capitalinvestment. Meanwhile, not one but two different types of purchaser were introduced: DHApurchasers and OP fundholders. These represented two different models of purchasing.Whereas DHA purchasers were exhorted to carry out formal needs assessments as the basisfor their purchasing plans and to balance priorities for the complete range of health care needsin a large (approximately 500,000) population, OP fundholders were expected to respond totheir patients' demands by purchasing a selected range of services for the relatively smallpractice population (approximately 10,000) patients.GP fundholdingPurchasers were supposed to choose between competing provider trusts, agreeing servicecontracts specifying price, volume and quality of care. DHA purchasing was similar to thetraditional top-down planning model, but with the addition of an explicit expectation thatOP's and local people would be consulted about priorities. OP fundholding, on the otherhand, was bottom-up and demand-led, with responsiveness to patients as its key characteristic.The scheme meant that fundholders' prescribing and outpatient referral costs were brought123within a cash-limited budget for the first time. The budget also covered admissions forelective surgery, diagnostic tests and investigations, community health services andparamedical services, but accident and emergency services, medical and psychiatricadmissions, maternity care and patients with expensive health care needs were excluded.The OP fundholding scheme was introduced in April 1991 on a voluntary basis with only 300practices, whose registered patients comprised 7% of the population, entering in the firstwave. Despite initial opposition from the British Medical Association, the opposition partiesand the majority of OP's, fundholding rapidly gained favour. By April 1996 13,400 OP's in3,700 practices were fundholding. Between them these practices catered for 52% of thepopulation of England. Initially the scheme was restricted to practices with registered lists inexcess of 11,000 patients but this lower limit was gradually decreased in response to pressurefrom OP's in small practices who felt excluded. By 1994 smaller practices and single-handedOP's were allowed to group together to hold ajoint budget or multi-fund. Although thescheme represented a substantial transfer of resources to OP's, fundholders had purchasingpower over only approximately 20% of the total hospital and community health care costs fortheir patients, the remainder being in the control of the DHA. Nevertheless the emphasis wason competition between purchasers rather than collaboration. The Conservative government,which introduced the scheme, and fundholders themselves, liked to stress the fact that itstrengthened OPs' autonomy and released them from bureaucratic constraints imposed byDHA's. Critics argued that it led to fragmentation of services and increased inequity, sincefundholders' were able to negotiate reduced waiting times and improved access arrangementsfor their patients.In announcing their new policy drive 'towards a primary care-led NHS' (NHS Executive1994), the UK government made it clear that they were primarily interested in strengtheningOPs' role as purchasers of secondary care services. The fundholding scheme was to beextended and the 'benefits' of direct involvement in purchasing health cm'e for their patientswere to be made available to all OP's. The belief that OP's were well placed to define andarticulate the health needs of their patients had some force, given their responsibility forproviding comprehensive and continuous services to defined populations, but three years intothe scheme the available evidence suggested that OP fundholding had had only a marginalimpact on the pattern of services (Coulter 1995, Petchey 1995, Audit Commission 1996).Despite incentives to restrict prescribing and referral costs, fundholding OP's had notmodified their behaviour very significantly. Prescribing costs rose less steeply amongfundholding practices in the first two years of budget-holding, but this advantage was notsustained in the longer term (Stewart-Brown et al 1995, Harris and Scrivener 1996, Raffertyet a11997, Gosden and Torgerson 1997) and referrals continued to rise in fundholdingpractices although the increase was slightly less steep than in non-fundholding practices(Surender et al 1995). Fundholders did manage to secure shorter elective surgery waitingtimes for their patients, an advantage which was seen as unfair by doctors and patients outsidethe scheme (Dowling 1997). Critics argued that this arrangement increased inequity and ledto a two-tier service. Despite this, the 'primary care-led NHS' strategy announced in 1994underlined the govemment's confidence in fundholding and signalled their intention to place agreater proportion of health service funds under the direct control of OP' s.124Following this announcement, a number of 'total purchasing' experiments (TPP's) werelaunched in which OP's at selected sites were encouraged to take over the entire budget forhospital and community health services. Since existing legislation did not allow for acomplete hand-over of budgets, the DHA retained ultimate responsibility requiring the OP' sto work collaboratively with the health authority. In theory this collaborative arrangementcould combine the best of top down, strategic purchasing for the needs of a population, with abottom up approach responding to the demands of individual patients. Currently 92 TPP's areunder evaluation, the majority being multi-practice groupings with a mean population size of35,000. Early evidence from the pilot schemes suggests that OP's are enthusiastically tacklingsome previously intractable problems, for example trying to stem the rise in emergencyadmissions, or securing improvements in co-ordination of care for patients with long-termneeds, but they have been reluctant to take on responsibility for the full range of services andco-ordination between multi practice groups and the DHA has not always been easy (Mays etaI1997).GP commissioningWhile fundholding was absorbing the energies of a growing number of OP's, many remainedimplacably opposed to the scheme. Some of these began to work together with local healthauthorities to influence DHA purchasing plans. A group of non-fundholding or'commissioning' OP's was formed to represent their interests. The National Association ofCommissioning OP's has defined commissioning as 'the process of gathering and analysingthe wants and needs of a population for which services are to be purchased, and ofmonitoring those services as they are delivered' (NACOP 1996). This is distinct frompurchasing which they define as 'the interpretation of commissioning plans and theconstruction and implementation of time-related purchasing plans'. Thus commissioning isintended to be strategic, population-focussed, and collaborative with the health authority,whereas purchasing is primarily concerned with contracting and resource allocation forspecific patient groups. According to these definitions those involved in purchasing are alsocommissioners, but it is possible to be involved in commissioning without having responsibilityfor purchasing and without having a delegated budget.By 1996 a number of different models of purchasing or commissioning were in existence, fromhealth authority-led schemes in which OP's were merely consulted but had no budgetaryresponsibility, through formal arrangements for collaboration involving indicative budgets, tofundholding where budgets were delegated to OP's. The Labour Governrnent, which came topower following the May 1997 general election, had committed themselves while inopposition to the abolition of the internal market. They were keen, however, to retain OPinvolvement in commissioning, but they wanted to avoid the perceived inequalities of thefundholding scheme. The new Secretary of State announced his intention to begin a debate oncommissioning models for the future. While remaining committed to the development of aform of collaborative commissioning involving health authorities and OP's, the Governmenthad no organisational blueprint. The intention was to evaluate the existing models and to pilotsome new variants, prior to deciding which forms of commissioning should be encouraged.125Teasing out the relative risks and benefits of the different models will not be an easy task, butthe opportunity to learn from these natural experiments poses an exciting challenge to policyanalysts and health services researchers.Features of commissioning modelsThe different purchasing or commissioning models vary according to a number of key featureswhich can be classified on three main dimensions:a) degree of budgetary control, where the main difference is between those in which budgetsare delegated to GP's and those where the health authority retains budgetary control, butpractices are encouraged to participate in decision-making;b) population coverage, which can vary from a single practice through multi-practice groups,to the division of the health authority into geographical patches which do not necessarilycoincide with practice boundaries;c) range of services included in devolved budgets, which can be limited to prescribing andcommunity nursing only, or selected hospital and community services as in standardfundholding, or the complete range of hospital and community health services as in totalpurchasing.Any model of GP-Ied commissioning carries potential risks as well as potential benefits.These are summarised in Table 1:Table 1Potential benefits and risks of GP-Ied commissioningBenefits RisksGP leadership and enthusiasm doubts about sustainabilitypatient centred GP as rationerresponsive increased inequalitieslocal knowledge lack of accountabilitycontinuity of care fragmentation of servicesunderstanding of individual needs lack of population focusdevelop new services stimulate new demandsfewer unnecessary interventions reduced access to specialist servicesincentives to control costs reduced cost-effectiveness126Degree of budgetary controlThe enthusiasm with which many OP's had embraced fundholding confounded the sceptics. Itwas clear that some OP's relished the opportunity to get involved in non-clinical activitiessuch as contract negotiations and budget monitoring despite the substantial increase in theirworkload that this represented. For some it seems to have provided an answer to the mid-lifemalaise that afflicts OP's faced with up to 35 years in the same job and few opportunities forcareer development, but a substantial number did not want to accept direct responsibility fordelegated budgets. Many had ideological objections to the scheme (Robinson and Hayter1995). Some argued that time spent on budget management had a detrimental effect on OP's'ability to provide good clinical care. Others felt they did not have the requisite skills infinancial management. Experience of non-fundholding commissioning groups demonstratedthat it was possible to involve primary care staff in setting priorities without delegatingbudgetary control (Black et aI1994), but the incentives for involvement in these non­fundholding schemes were weaker, raising doubts about long-term sustainability.Budgetary control implies a need to balance population needs against individual needs. Thiscan cause role conflict for the OP. Direct involvement in decisions about resource allocationplaces the OP in the role of rationer, a task with which many OP's feel uncomfortable becauseit conflicts with their preferred role as patient's advocate (Robinson and Hayter 1995, Ayres1996). Patients may be less willing to accept advice that they do not need treatment orreferral if they believe the OP's decision is influenced by budgetary considerations. SomeOP's feared this would undermine the doctor-patient relationship with possible adverseconsequences on the effectiveness of clinical care (Cornell 1996).Transaction costs are high when budgets are devolved to numerous small-scale purchasers.OP fundholding has led to a dramatic increase in practice administrative costs, as well asincreased costs to the providers and to the DHA's responsible for overseeing the scheme.These can be balanced against increased incentives for practices to control demand byrestricting referrals or investing in practice-based facilities as a substitute for more expensivesecondary services. So far, while there is no doubt that it has increased the level of practice­based investment, there is little evidence that the fundholding scheme has succeeded incontrolling demand. Indeed by raising expectations and providing for needs that werepreviously unmet, it may have increased it.Fundholders have used their financial leverage to achieve a number of beneficial changes,including investing in practice-based services such as physiotherapy, counselling anddiagnostic equipment, but the relative cost-effectiveness of these on-site services isquestionable (Coulter 1996). It is also clear that fundholders have secured beneficial accessfor their patients, at least in terms of waiting times and new practice-based services. The factthat they hold a budget has given them some leverage over secondary care and for the mostpart they have used this to improve access rather than seriously to challenge the pattem ofprovision. Some efficiency savings have been achieved, but at a cost. In 1996 the AuditCommission attempted to calculate a balance sheet. They estimated that fundholders hadmade savings worth £206 million, but practice management costs had increased by £232million (Audit Commission 1996).127These costs would be justified if they had led to substantial improvements in the quality ofcare, but it is hard to demonstrate that this is the case and few studies have addressed theissue. There is little evidence of an impact on clinical effectiveness. Few fundholders havedared to challenge the clinical decisions of their consultant colleagues (Douglas et al 1997).The Audit Commission found that most fundholders were not making full use of theincreasing body of knowledge about clinical effectiveness to change the way they commission.Of the fundholders surveyed, only a third said that their purchasing plans had been influencedby literature on evidence-based medicine (Audit Commission 1996).Financial risk resulting from unpredicted demand is greater when the population pool is small.Perceived wisdom, based on the experience of American HMO's, suggests that purchasing forpopulations of less than 50,000 involves punitive transaction costs and unmanageable risk(Sheffler 1989), although this view has been challenged (Weiner and Ferris 1990). Modelshave been devised for spreading the risk over 3-5 years, but these have not yet been tested(Bachmann and Bevan 1996). Attempts to devise a resource allocation formula forfundholding which would be robust at the practice level have so far proved unsuccessful(Sheldon et al 1994). In the meantime fundholders' budgets have been based on a mixture ofcapitation allowances and historic costs and there have been allegations of over-fundingrelative to DHA allocations for the patients of non-fundholding practices (Dixon et aI1994).If devolution of budgets is to continue, strategies will have to be devised for transparent andfair resource allocation and for managing variability in utilization and unpredictable demandson the budget (Martin et al 1997).Population coverageA commonly stated advantage of involving GP's in the commissioning process is that they arecloser to patients and therefore can help to ensure that services take account of patients'needs and preferences. This assumption that GP's' views and priorities are congruent withthose of their patients is still an act of faith - few studies have investigated concordancebetween patients' views and those of GP' sand fundholders have been slow to establishformal mechanisms for consulting their patients (Audit Commission 1996). The new localitycommissioning pilots which the Government intends to set up from April 1998 will covermuch larger populations than the average fundholder. These new commissioning groups,which will cater for populations of more than 50,000, will have pooled prescribing budgetsand will be expected to work in co-operation with health authorities to commission hospitaland community health services. Larger scale commissioning groups may turn out to be morecost-effective than purchasing or commissioning by individual practices, but it may be moredifficult for larger groupings to secure consumer involvement. While patients might find iteasy to identify with their own general practice, aggregates of practices or localities may beless meaningful to them. Much will depend on the extent to which the localities coincide with'natural communities', for example a small town or a defined part of a city or larger urbanarea.128A population approach to health commissioning requires some knowledge of epidemiologyand access to data on the distribution of disease. OP-Ied commissioning or fundholding isalmost always based on single practices or clusters of practices, but much health data, forexample those used to assess health needs, are based on electoral wards, i.e. geographicalboundaries rather than practice boundaries. Because patients can choose which OP to registerwith, practice boundaries do not necessarily fit into 'natural' communities, nor are theycoterminous with local authority boundaries used by social services and other agencies. Co­ordination of information sources can be especially difficult in urban areas, where practiceselection effects operate more powerfully. Few OP's are trained in epidemiology orpopulation sciences. Most have had little experience in manipulating data. A demand-ledsystem increases the likelihood that the focus will be on individual needs at the expense ofpopulation concerns and equity.In view of strong popular attachment to the ideal of a universal health service accessible to allaccording to need, the suggestion that fundholding introduces greater inequity into the systemhas been politically damaging and has caused considerable public antipathy to the scheme.Fundholding has tended to attract well organised practices from better-off parts of thecountry, with inner city practices particularly under-represented (Audit Commission 1995).The government's intention is that the new locality commissioning pilots should be establishedin places where fundholding or OP commissioning has not yet taken off, but there is a real riskthat OP's and practice staff working in 'difficult' areas will not have the time or the inclinationto get involved and that their patients' needs will be less well catered for than those whoseOP's are involved in commissioning or fundholding.There are concerns that fundholders' patients benefit at the expense of patients of non­fundholders, for example in shorter waiting times as a result of their ability to jump the queue,or practice-based consultant clinics which mean that the specialist is less often available in thehospital clinic. It is also true that fundholding can have a destabilizing effect on local servicesby contracting with private providers, for example, for pathology services, thus removingresources from local hospitals and causing a leakage of NHS funds into the private sector.However, getting 'a better deal' for their patients is a main motivator for OPs' involvement inpurchasing. Increased inequity in access to care may be an inevitable consequence ofdevolving budgets to the practice level. Locality commissioning involving larger practicegroups may avoid some of this perceived unfairness, but at the expense of reduced incentivesfor OP' s to get involved.Range of servicesIn absolute terms the appropriate population size for commissioning depends on what servicesare to be commissioned. One argument, based on theories of 'epidemiological stability', isthat community nursing and other community health services, elective surgery and outpatientfacilities are logically purchased at practice level, while most other secondary care services,including accident and emergency and maternity services, may better be purchased underblock contracts by organisations covering populations of around 300,000, leaving tertiary and129more highly specialised services to be covered at a 'regional' level (roughly 1 million pop.)(Murray 1993).Effective purchasing requires a wide range of skills, including needs assessment and planning,contracting, monitoring and performance management, accounting and budget management.Beyond an understanding of the processes of commissioning, some specialist know ledge isrequired to make strategically coherent purchasing decisions. This knowledge may not alwaysbe vested in general practice. For example, people with learning disabilities have special needsof which their GP's may not always be aware (Howells 1996). The arguments for restrictingthe range of services purchased at a practice or locality level are based on assumptions aboutthe level of expertise required to purchase effectively. Patients with complex needs, such asthose with long-term mental illness, may not be best served by a decentralised system wherethe risk of service fragmentation may be greater. On the other hand, if the GP or primary careprofessional is to act as care manager helping these patients through the maze of services, itmay help if he or she has sufficient leverage to ensure that the patient gets appropriate care.Balancing the needs of the individual and the population is never going to be easy.Future developmentsThe current policy concern to shift the balance of care from secondary to primary settings isled by a desire to contain rising health care costs. To date there are few signs that givingsecondary care budgets to GP's achieves the desired shift. There is a real risk that GP-Iedcommissioning will encourage responsiveness at the expense of appropriateness and cost­effectiveness. Some analysts argue that it is too early to draw conclusions from thefundholding experiment (Gosden et al 1997). The new emphasis on evaluation is welcome ­the previous Government refused to fund formal evaluations of the 1991 reforms - but someare sceptical about the genuineness of the Labour Government's espousal of evidence-basedpolicy-making.Meanwhile even more radical developments are on the horizon with the introduction oflegislation allowing for deregulation and merged budgets for primary and secondary care. The1997 NHS (Primary Care) Act offers scope for experimentation with unified budgets,although piloting of these arrangements has been deferred until 1999. In theory unifiedbudgets could provide more powerful incentives to shift the balance of care and to ensure thatinvestment in primary care is matched by reductions in expenditure on specialist services.However the risks are considerable (Coulter and Mays 1997). The creative tension ofcontestability introduced by the purchaser provider split could be diluted if GP's were able toincrease the range of services they could provide and there would be greater potential forserious conflicts of interest for GP' s. The role of the DHA as monitor and regulator wouldneed to be strengthened and it would become even more essential to develop robust systemsfor monitoring the quality and outcome of care. Vertical integration of budgets couldencourage a shift in the opposite direction to that desired if secondary care providers were tobecome the budget holders, a possibility allowed for under the 1997 Act. They could use their130controlling influence to generate demand for their specialist services and to squeeze resourcesout of primary care.ConclusionCurrent developments in the British NHS offer an excellent opportunity to learn more aboutthe merits of centralised versus devolved systems for commissioning health services. Theevidence is thin as yet, but it seems likely that devolving budgetary control to GP's increasesresponsiveness and sensitivity to local needs and encourages innovation at practice level, butat the expense of equity. In the absence of evidence about impact on the quality of care andhealth outcomes and in the face of insufficient data on costs, it is impossible to draw hardconclusions about the relative cost-effectiveness of the different models.AcknowledgementsI am grateful to my colleagues at the King's Fund, Nicholas Mays, Jennifer Dixon andStephen Gillam, for helpful ideas.Pros and Cons of Offering Patients ChoicePROincreased satisfactionincreased awareness of treatmentreduced distress and anxietyimproved quality of lifebetter health outcomesreduced litigationless inappropriate treatmentCONgreater confusionloss of confidence in doctor~increased distress and anxiety ~burden of responsibilityincreased morbidityincreased time pressuresmore inappropriate demandsInformation and Participation PreferencesAmong 256 American Cancer PatientsAge (years) 20-39 40-59 60+% % okParticipation preferencesprefer participating in decisions 87 62 51prefer leaving aecisons to doctor 13 38 49Type of information desired -v.>'"want all information - good and bad 96 79 80want only minimal or good information 4 21 20Preferences for detailed informationprefer minimum 15 40 31prefer maximum 83 60 69B . Cassileth et al, Ann Int Med 1980Information Needs of 250 Cancer Patients inScotlandQUESTION Do not want Would like Absolute needto know to know to know0/0 .0/0 0/0Whether illness is cancer 4 24 72Medical name of illness 25 46 30 ~wwWeek by week progress 10 42 48Chances of cure 9 31 60AU possible treatments 14 32 54AU possible side effects 6 21 73How treatment works 20 36 43C Meredith et al. BMJ 1996Proportion of Women Choosing BreastConservation Treatment in Preference toMastectomyFirst author/dateFallowtleld 1990Cotton 1991Wilson 1988Morris 1987Wolberg 1987Sauer 1992Pozo 1992Proportion choosingbreast conservation0,/069453565497538CountryUKUKUKUKUSAGermanyUSA-v.>..,..M Richards et al, Eur J Cancer 1995Preferences for Participation in Decision...makingActive Collaborative Passive0,/0 0/0 0/0Canadian cancer patients 12 29 59Canadian householders 64 27 9Canadian breast cancer patients 22 40 38 ~wU>British breast cancer patients 20British benign breast disease patients 2428455231M Richards et al, Eur J Cancer 1995Evaluating New Developmentsin Primary Care• Why is change necessary?• How can the changes be evaluated?• What problems have to be overcome?~w0,The Secondary Care Paradox• length of stay reduced• day case surgery increased• faster throughput• longest waits reducedBUT .• emergency admissions increased• outpatient referrals increased• more people on waiting list• beds still blocked~W-..JThe Primary Care Paradox• greater power and control over resources• increased staffing• better vocational training• better prevention and chronic diseasemanagementBUT .... w ••• rising demand• increased time pressures• variations in quality• low morale­w00Learning From New DevelopmentsInvolves:• systematic monitoring• critical evaluation• weighing up benefits, harms and costs­W'0140ReferencesAcheson D (1985) Variations in hospital referrals. In: Health Education and generalpractice. London: Office of Health Economics, 21-23Audit Commission (1995) Briefing on GP fundholding. London: HMSaAudit Commission (1996) What the doctor ordered: a study of GP fundholders in Englandand Wales. London: HMSaAyres P (1996) Rationing health care: views from general practice. Social Science andMedicine 42; 7: 1021-1025Bachmann M a, Bevan G (1996) Determining the size of a total purchasing site to managethe finance from risks of rare costly referrals: computer simulation model. British MedicalJournal 313: 1054-6Black D G, Birchall A D, Trimble I M G (1994) Non-fundholding in Nottingham: a vision ofthe future. British Medical Journal 309; 930-2Cornell J (1996) Has general practice fundholding been good for patients? Public Health;110: 5-6Coulter A (1995) Evaluating general practice fundholding in the United Kingdom. EuropeanJournal of Public Health 5: 233-9Coulter A (1996) Why should health services be primary care-led? Journal of HealthServices Research and Policy 1: 122-124Coulter A, Mays N (1997) Deregulating primary care. British Medical Journal 314: 510-3Department of Health (1989) Working for patients. London: HMSaDixon J, Dinwoodie M, Hodson D, Dodd S, Poltorak T, Garrett C, Rice P, Doncaster I,Williams M (1994) Distribution of NHS funds between fundholding and non-fundholdingpractices. British Medical Journal 309: 30-4Douglas H R, Humphrey C, Lloyd M, Prescott K, Haines A, Rosenthal J (1997) Promotingclinically effective practice: attitudes of fundholding general practitioners to the role ofcommissioning. Journal of Management in Medicine 11: 26-34Dowling B (1997) Effect of fundholding on waiting times: database study. British MedicalJournal 315: 290-2141Enthoven A C (1985) Reflections on the management of the National Health Service.London: Nuffield Provincial Hospitals TrustForster D P, Frost C E (1991) Use of regression analysis to explain the variation inprescribing rates and costs between family practitioner committees. British Journal of GeneralPractice 41: 67-71Gosden T, Torgerson D (1997) The effect offundholding on prescribing and referral costs: areview of the evidence. Health Policy 40: 103-114Gosden T, Torgerson D J, Maynard A (1997) What is to be done about fundholding? BritishMedical Journal 315: 170-1Harris C M, Scrivener G (1996) Fundholders' prescribing costs: the first five years. BritishMedical Journal 313: 1531-4Howells G (1996) Situations vacant: doctors required to provide care for people withlearning disability. British Journal of General Practice 46: 59-60Martin S, Rice N, Smith P C (1997) Risk and the GP budget holder. Discussion Paper 153.Centre for Health Economics: University of YorkMaynard A, Marinker M, Pereira Gray D (1986) The doctor, the patient and their contract.British Medical Journal 292: 1438-40Mays N, Goodwin N, Bevan G, Wyke S (1997) What is total purchasing? British MedicalJournal 315: 652-5Murray D (1993) Patterns in NBS commissioning? A review and discussion of alternativemodels of integrated locally sensitive commissioning. London School of Hygiene andTropical MedicineNational Association of Commissioning GPs (1996) Report to the Secretary of State forHealth on national standards for GP commissioning (unpublished)NHS Executive (1994) Developing NHS purchasing and GP fundholding: towards a primarycare-led NHS. EL(94)79. Leeds: Department of HealthPetchey R (1995) General practitioner fundholding: weighing the evidence. Lancet 346:1139-42Rafferty T, Wilson-Davis K, McGavock H (1997) How has fundholding in Northern Irelandaffected prescribing patterns? A longitudinal study. British Medical Journal 315: 166-70142Robinson R, Hayter P (1995) Why do OPs choose not to apply for fundholding?Southampton: Institute for Health Policy StudiesSheffler R (1989) Adverse selection: the Achilles heel of the NHS reforms. Lancet i: 950­952Sheldon T, Seith P, Borowirz M, Martin S, Carr Hill R (1994) Attempt at deriving a formulafor setting general practitioner fundholding budgets. British Medical Journal 309: 1059-64Starfield B (1994) Is primary care essential? Lancet 344: 1129-33Stewart-Brown S, Surender R, Bradlow J, Coulter A, Doll H (1995) The effects offundholding in general practice on prescribing habits three years after introduction of thescheme. British Medical Joumal311: 1543-7Surender R, Bradlow J, Coulter A, Doll H, Stewart Brown S (1995) Prospective study oftrends in referral pattems in fundholding and non-fundholding practices in the Oxford region1990-4. British Medical Journal 311: 1205-8Weiner J, Ferris D (1990) OP budget holding in the UK: lessons from America. London:King's Fund143Primary HealthCare and CanadianPhysicians c.1997Michael RadIUs MD, MSc, FRCPCThe Second Trilateral Physician Workforce ConferenceVancouver November 15, 1997144IntroductionIn 1972, Dr. John Hastings of the University of Toronto issued a report to the Conference ofDeputy Ministers of Health on the Community Health Centre (CHC).! Dr. Hastings' commissionrecommended that, as the provinces introduced Medicare,' they also reorganize medical practiceand the deliveIl. of primary health care services. Quebec did develop a network of 160 communityhealth centres.• (These are discussed later in this paper.) However, other provinces developedfew, if any, new models of primary health care. Most primary medical care continued to beprovided by family doctors in private practice who were reimbursed on a fee-for-service basis.However, provincial commissions and inquiries on health in the 1980's and early 1990's againhighlighted the importance of primary health care to make the rest of the health care system runmore efficiently and to improve population health. The British Columbia Royal Commission onHealth Care and Costs concluded:"We believe there is enough evidence and track record that community healthcentres can reduce institutional costs and maintain the quality of care. This aloneshould be of interest to the govemment."The Federal/Provincialtrerritorial Advisory Committee on Health Human Resources report onCommunity Health Services' concluded:"The literature on community-based health services models reviewed for thiscomponent indicated that, in general, integrated, multi-service, multi-disciplinarymodels are less costly, and more cost-effective, than comparable services provided bysingle-service providers and institutional providers. This is particularly evident whencomparing the community health centre organizational model with solo fee-for-servicephysician practice. The major cost saving appears to occur through the reduction inthe use of hospital out-patient and inpatient services by populations receiving servicesfrom community health centres."Later provincial ministry of health planning documents usually recommended the communityhealth centre model for primary health care. For example, the BC ministry introduced their policyfor health reform with the release of New Directions for a Healthy British Columbia in Februaryof 1993. New Directions envisaged CHe's playing a major role in the restructuring of the healthcare system:"In the decentralized system, community health centres will serve as a key mechanismto integrate and coordinate services at the community level."But, as in 1972, it has proven harder to actually implement new models of primary health carethan it has been to recommend their adoption. The 'health care reform' in the 1990's has, actually,Public health insurance is widely known in Canada as "Medicare" which is also usuallyused to mean the entire health care system.En francais, centre local services communautaire.145been more like hospital restructuring. However, in 1997 there are signs that primary health carereform is making some progress in some provinces.This paper summarizes the situation in Canada's ten provinces' for physicians and new models ofprimary health care, analyzes why new models have and have not been implemented, and thenmakes some predictions about the next ten years.MethodologyStructured interviews were conducted with ministry representatives in all provinces between Julyand October 1997.ResultsThe current situation in Canada for physicians and new models of primary medical careIn this paper new models of primary medical care include:1. Community health centres where family doctors are salaried or contract employeesand they work in interdisciplinary practice with other professionals such as nurses and socialworkers. Most CHe's have global or program-based budgets although some in Ontario havecapitation funding.2.capitation.Family physicians' private practices where the prime mode of payment is3. Private practice family physicians who hold contracts with ministries of health andare primarily not paid on a fee-for-service basis.There are a number of doctors on salary through university departments of family practice.Unfortunately, there is no central collating of these data and this survey does not include thesedoctors. The survey results are shown in summary in table one.The reorganization of health services in the 1990'sIn the 1970's, as Quebec implemented Medicare it developed 18 regional health authorities whichgradually were given full executive authority over all services except physicians andpharmaceuticals. Duling the 1990's all the other provinces except Ontario have also devolved theday-to-day management of health services to regional health authorities. There are manydifferences in the details of this devolution. Typically, the new regional health authorities(variously referred to as regional health boards, regional health corporations, or district healthboards) have been given budget authority for all health services except for physicians andpharmaceuticals. However, New Brunswick's regional health corporations only administer,Canada does have two territories (Northwest and Yukon) but their large geography (4million square kilometres), and sparse mainly aboriginal population (less than 100,000) means thatthese jurisdictions already have unique primary health care services.146hospitals and home care services. In Newfoundland there are six regions but only two have theresponsibility for institutional and community services. The four larger regions have separateboards for institutional and community services. In Manitoba, the regions outside of Winnipeghave control over all health services (except for physicians and drugs) while Winnipeg will have(at least initially) separate boards for institutional and community services. In Nova Scotia, thedevolution to regional authorities is occurring more slowly and so far the regional boards onlyhave jurisdiction over hospitals and public health although the province does plan to eventuallyinclude other community services.In most provinces the boards of individual hospitals and other facilities have been disestablishedwith their governance being assumed by the regional authorities. However, the Catholic facilitieshave been allowed to maintain their boards although they have greatly reduced autonomy. InQuebec, the hospitals, long-term care facilities, and CLSC's (en francais establishments) havebeen allowed to maintain their boards but receive their budgets and overall direction from theregional authorities.All provinces started with appointed boards although most have announced their intention toproceed with elections. However, only Saskatchewan has conducted elections (in the fall of1995). Eight out of the 12 board members of Saskatchewan's district health boards are electedand the other four are appointed by the Minister. Alberta had planned to hold elections in the fallof 1998 but these have been postponed for at least one year.British Columbia, Alberta, and Nova Scotia had announced their intentions to create smallerboards (to be called community health boards or community health councils) within their regionalboards to administer community based services. However, these plans have changed somewhat. InNova Scotia their role is still being deliberated although thus far the community boards are strictlyadvisory. In Alberta, each regional health authority is to create at least one community healthboard but their roles are also unclear. In BC, the government in 1994 created 23 regions andapproximately 70 community health councils. However, in 1996 government changed its courseand now has 18 regions. The nine in the most heavily populated areas have the responsibilities forall services but physicians and pharmaceuticals while the other nine were split into 34 communityhealth councils. The community health councils are responsible for primary health care servicesand small hospitals.Ontario, Canada's largest province with over II million people, considered the regional model inthe early 1990's but rejected the concept. In 1996, the provincial government created the HealthServices Restructuring Commission. The Commission is appointed by the Minister of Health butoperates at arm's length. The Commission has the authority to reorganize hospital servicesincluding the power to close hospitals or merge them. However, the Commission can only makerecommendations to the minister about investments in community services. Ontario initiated aseries of discussions in 1995 about Integrated Health Systems (IHS's). ICS's were planned tointegrate all health care services under member controlled boards. Although the planning wasnever completed the intention was to have competition between enrollment-based integratedsystems like US Health Maintenance Organizations (HMO's). It was very unclear as to howprimary health care would fit into these models. However, a cabinet paper on ICS's originally duein May/97 has now been indefinitely postponed. It appears that there will be little restructuringaside from hospital closures until after the next provincial election which is not due until 2000.As mentioned earlier there have been recommendations for primary health care reform and147changing family doctors style for remuneration in the past 25 years. In 1995, theFederal/Provincial/Territorial Advisory Committee on Health Services issued a discussion ~aperon the reorganization of primary care and new methods of remuneration for family doctors andthen conducted a national consultative process over the next year. Several medical organizationsalso issued discussion papers'v'" showing a readiness to consider change and several nursing.. I I d di h I h f 78910 II A ti Iorgamzations a so re ease reports recommen mg pnmary ea t care re orm. . ... na ronasurvey of physicians confirmed that more than half of all famil¥ physicians would consider movingto a salary even if the change entailed a slight loss of income. IDoctors in Community health centres and working on service contractsAll provinces have some community health centres (CHC's) but no province other than Quebechas a full network. There are approximately 160 CHC's (en francais, CLSC's) in Quebec. Thefirst CLSC's were initiated by community groups often with federal government grants in the late1960s and early 1970's. However, as Quebec implemented its Medicare program after 1970, theCLSC's gradually became the focal point for community based services. The CLSC's receiveapproximately 8% of the health budget and this proportion continues to increase. In the mid­1980's, the CLSC's took on the home care mandate and gradually have become "one-stopshopping" facilities for health and social services. The CLSC' s are the exclusive providers ofhome care, public health and certain specialized services for individuals (for example, programsfor children's mental health). During the hospital downsizing of the 1990's, Quebec has reliedupon the CLSC's as the cornerstone of its virage ambulatoire (move to the community). Budgetsand personnel have literally been moved from the hospital system to the CLSC's. The centreshave an average of 7-8 physicians and provide general medical services and support to particularprograms (e.g. home care).'During 1997-98, Quebec is merging the boards of about 50 CLSC's in rural areas with the boardsof small hospitals. There are concerns that this might threaten the mandate for health promotionand community development which is already under attack because of increased demands forhome care services.Ontario's first CHC's developed in a similar fashion to those in Quebec where short-term federalgrants allowed the start-up of "street clinics" and other special purpose facilities. In 1973 theOntario government established the program development and implementation group which wasto develop globally-budgeted CHC's and capitated funded health service organizations (HSO's).By 1975, there were 10 CHC's which had started or had been given promises of start up fundsand another 30 groups which were in various states of development. However, in April, 1975 theCHC program was frozen and the ministry attempted to force all CHC's onto capitation fundingwhich would have closed most of the centres.In 1982, a Conservative government began to gradually expand the program and this expansionwas accelerated under a Liberal government in 1988 when 6 new centres were to be funded peryear. However, this was scaled back under a New Democratic Party" administration to 3 per year,Some of the newer CLSC's just have physicians to support their specialized programsrather than to provide general primary medical care.The New Democratic Party or NDP is Canada's social democratic party.148as part of general government cutbacks in 1992. Finally, a new Conservative government frozethe development of new centres in 1996:" The 60 Ontario CHC's do not provide mandatedservices such as home care or public health. Rather they have been seen as facilitating access forgroups which otherwise would have difficulty accessing the health care system (e.g, the poor,elderly, multicultural groups, aboriginals, etc.)In Saskatchewan, there are 5 community clinics which were originally developed in 1962 whenthe social democratic govemment of Tommy Douglas introduced Medicare and 95% of theprovinces' doctors went out on strike. The clinics were established by party activists and governedas cooperatives, The later NDP governments of Alan Blakeney and Roy Romanow have not usedthe model as a prototype on which to restructure medical practice and primary health careservices. The district health boards were supposed to have taken over the govemance of theseclinics in 1995 but this decision has been postponed several times.In the summer of 1997, the Saskatchewan Department of Health appointed the province's chiefmedical officer, Dr. David Butler-Jones, to assist district health boards in developing a primaryhealth care strategy. Dr. Butler Jones and his group are working with the districts to integrate thepreviously separate community services as well as facilitating the involvement of family doctorswith these new structures. Saskatchewan is focussing on signing doctors to service contractswhere the doctors agree to give up their fee-for-service billings in retum for a fixed payment andcontract which clearly delineates their responsibilities including on call arrangements.A senior Manitoba official, Dr. Ted Tulchinsky, was a member of the Hastings commission intoCHC's and during the early 1970's, the Manitoba &,ovemment started to implement the Hastingsproposals. Six CHC's were developed in Winnipeg and seven were opened in rural areas. Theseven rural CHC's provide home care and public health nursing services and often incorporate along-term care facility but the six urban ones don't have mandatory services. However, with theregionalization begun in Manitoba in 1996, the rural CHC's are being integrated under theregional health authorities and in Winnipeg 22 neighbourhood resource networks have beenannounced. These networks will formally link the various community services within theirboundaries. The existing CHC's in Winnipeg will be so-called network "anchors" in theneighbourhoods in which they are located. Other anchor clinics will include some privatepractices.British Columbia's short-lived social democratic government of 1972-1975" established a numberof community health centres but many of these were cut after the government's defeat. Theremaining seven stayed marginalized from the rest of the system. There was great anticipationafter the 1991 election of the New Democratic Party. The Royal Commission on Health Care andCosts had just issued its report and it had recommended CHC's as the main approach for'" There are about a dozen CHC's being developed in southem and mid-northern urbancommunities to provide services to first nations members in Ontario.Winnipeg, Manitoba's capital city has a population of approximately 650,000, 60% of thetotal population of the province." The New Democratic Party government of David Barrett was sandwiched in betweennearly 40 years of rule by the conservative Social Credit Party.149integrating community services with salaried doctors. However, the province has had considerabledifficulty in enunciating a strategy for primary health care. The Vancouver-Richmond health boardis creating six community health centres in the inner city of Vancouver (population approximately400,000) to better coordinate the integration of mental health, public health, home care, andaddictions services. They plan to use some of the funding for public health physicians (19 full timeequivalent positions) to fund the salaried doctors in these clinics.Newfoundland's position is unique because historically it has had a large number of salaried familydoctors in its remote towns and villages. The doctors were salaried in part because there was notdeemed to be the population to support a fee-for-service doctor. These doctors correspond tothose in other provinces paid on service contracts. Recently the ministry of health has begun todevelop CHC's and there will be approximately 4 in operation by the end of 1998.Other provinces have had few CHC's in the past and are finding it somewhat difficult to movetowards this new model. Alberta is the only province where public health nurses haveresponsibility for all childhood immunizations and they are located in offices with approximatelyone per 20,000 - 40,000 people. There have been some recommendations that Alberta use theseoffices as a base for a new system of primary health care built upon public health. However, thishas not happened and there are pressures from some doctors to discontinue the public healthclinics and second the public health nurses to doctors' offices.The number of Alberta doctors with service contracts is imprecise because the provincialdepartment of health does not have complete data on the arrangements which involve regionalauthorities. There are over 230 family doctors with some sort of service contracts but almost 90%of them continue to bill on a fee-for-service basis for at least some of their work. Alberta is 'exploring expanding the option of service contracts as are Nova Scotia, New Brunswick andPrince Edward Island.Capitation arrangementsThe Ontario Health Service Organization ProgramOntario has had almost thirty years of experience with a capitation model for primary care -- theHealth Service Organization or HSO program. In 1963, the Sault Ste Marie Group HealthAssociation Clinic opened in this Northem Ontario city of 80,000. 13 The Clinic had beenorganized by local citizens and had a community board. Major funding and expertise had beenprovided by the United Steelworkers of America' and, to a great extent, it was patterned on theUS prepaid group practices like Group Health Cooperative of Puget Sound and the KaiserPermanente Clinics. After the implementation of Medicare in the early 1970's, negotiationsbetween the clinic and the Ontario government led to the HSO program which was subsequentlyopened up to private doctors.At present, there are approximately 80 centres involving 300 family doctors. There are twocentres (including the Sault clinic) which are govemed by communities and four which aregovemed by university family practice departments but, most HSO's are small private practices, The main employer in 'the Soo' has historically been the Algoma Steel Company and theSteelworkers union has always played a prominent role in the community.150with less than 5 doctors. The Sault clinic does have about 35,000 patients and the largest privateHSO, the Grandview Medical Centre in Guelph, has 25,000 patients. The program grew slowlyuntil the mid-1980's and then the provincial government began to encourage doctors to switchfrom fee-for-service to capitation highlighting that their incomes would improve in the process.Unfortunately, the Ontario HSO program had significant administrative problems. There were norules for creating or administering a patient roster. To complicate matters further, the HSO' s onlylost the payment for that months capitation if a rostered patient received care outside the practice.(This process is referred to as negation.) However, the HSO would receive the payment thefollowing month if the patient was not seen elsewhere.Some doctors registered patients who weren't really theirs but continued to be paid for them mostmonths. Some rostered patients were long dead but the province still paid for their care. Anotherproblem was that HSO doctors were allowed to bill fee-for-service for so-called 'transient'patients. But this supplied an incentive for doctors to roster healthy people who they never sawand bill fee-for-service for sick patients whom they labelled as transients. The HSO's were paidbased upon provincial average payments for various age and gender groups (e.g. females 0-4years of age, males 55-59, etc.) but the HSO patients appeared to be healthier than average.' TheHSO's were supposed to hire other professionals like nurses and provide better preventiveservices but most didn't. 14 By and large, the HSO doctors just kept the extra revenue as personalincome with some family doctors receiving payments of more than half a million dollars a year.And, finally the HSO's were given bonuses if they reduced hospital use by their patients (theambulatory care incentive plan or ACIP) but a study showed that there was no difference in theirpatients hospital utilization after they converted from fee-for-service."In the early 1990's as the Ontario Ministry of Health developed concerns about this program andbasically froze the program to new applicants. The government gradually negotiated newarrangements with the Ontario Medical Association which was given bargaininfi rights for theHSO doctors as part of the 1991 agreement with the provincial government. 16. ACIP paymentswere discontinued although some of the money was invested in specific program budgets. Thenegation for outside use by rostered patients was increased to 50% of the actual cost of outsideservices and mechanisms were put in place to better verify rosters.Other capitation plansBy the mid 1990' s many Ontario physicians wanted to change the way they were paid. TheOntario Medical Association (OMA) proposed a rostered system dubbed "reformed fee-for­service" in 1996.18The plan would involve rostering of patients but the doctors would be paid ona fee-for-service basis up to a maximum calculated on the basis of the estimated capitationpayments for their rostered population. In July of 1996, the Ministry of Health announced aprimary care task force which was to fund pilot programs based on the OMA proposal as well asa so-called "interdisciplinary model" which would have doctors and other professionals workingtogether. Interestingly, the government showed no interest in the existing capitation organizations(the HSO'S)19 or interdisciplinary practices (the CHe's). However, the announcements of pilotprojects has been delayed three times (the original announcement was to be made in January, Most HSO's were located in middle to upper income areas and there were few rosteredpatients in long tern care institutions.1511997) and it is widely anticipated that there will be little if any progress in the development ofcapitated models until after the next provincial election (not expected until 1999 or 2000).However, other provinces have begun to develop capitated models. Saskatchewan has brought onstream three capitated centres since 1994, one in Regina and the other two in rural areas. Thereare three types of funding -- one covering all family practitioner services, one covering only officevisits, and one covering all services except surgery, obstetrics, and anesthesia. The capitation feeincludes protected money for certain ancillary staff such as nurses. Patients are automaticallyenrolled if they receive more than 75% of their family practice services from a clinic doctor. Theclinics are charged the full value of all outside services but once they have lost 150% of the annualcapitation payment, the clinic may disenroll the patient. After an estimated value of $1000 ofmedical services has been provided (the clinics submit 'dummy' bills to the provincial medicareplan), the doctors can bill fee-for-service to these clients. The capitation model has not proven aspopular as was previously anticipated. However, Ministry staff feel that through publicity aboutthis model, doctors have become more interested in the service contract option.Manitoba is in the process of initiating its first capitated pilot project. A 14 family doctor clinic ina Winnipeg suburb will convert to capitation funding on January 1, 1998. During the 1997transition year, the clinic has been on a global budget based on 1996 billings. The programoperates in a very similar fashion to that in Saskatchewan. The clinic's patients are automaticallyrostered if they have received more than 75% of their family practitioner services from the clinic.The clinics are charged the full costs of any outside use up to 150% of the annual capitationpayment. The initial budget included a protected allotment for 4.5 FfE nurse practitioners. Therehas been a lot of interest displayed in this model but it will likely only be applied to the suburbanareas of Winnipeg and possibly some rural areas. Ministry staff note that the core area of the cityhas a very transient population and much greater needs for interdisciplinary care.Evidently, the Winnipeg doctors are fairly satisfied with the transition year. They have been ableto cut their outside use by more than 90% through the extension of office hours and the use of thenurse practitioners. In fact, as the roster is about 20% lower than planned, the doctors areconsidering hiring more nurses if they have to replace a doctor.The Alberta Medical Association released a discussion paper on capitation in 1995 and hassubsequently been negotiating with the Provincial Ministry of Health and the Regional HealthAuthorities for pilot programs: As of November, 1997, the details have not been officiallyannounced, but the first pilot will likely be in a rural community where the two doctors willreceive capitation payments for all the residents within a particular geographic area. Other pilotsmay include an innovative ambulatory long term care program.In most other provinces there is also interest in capitation from either ministries of health orphysicians or both. In Quebec, capitation is mainly being considered for innovative ambulatorycare programs for the frail elderly." The main exception to this rule is British Columbia where theprovincial medical association has been particularly resistant to non-fee-for-service payment. 21DiscussionBarriers to primary health care reform152The above review of developments in primary health care and physicians services displays a mixedpicture. Quebec has always been ahead of the rest of the country in developing salaried practice.Saskatchewan which led the country in the implementation of public health insurance in the1940's, 1950's, and 1960's has reasserted its leadership in English Canada in the 1990's. StevenLewis, the executive director of the province's Health Service Utilization and ResearchCommission has predicted that fee-for-service payment for Saskatchewan family doctors willdisappear within 5-7 years. However, Ontario which pioneered capitation payment and alsodeveloped the second largest network of CHe's has, if anything, moved backwards in the 1990's.The collective agreements between provincial medical associations and ministries of health whichwere signed in the 1990's have presented a significant barrier to alternative payment and practicearrangements. Starting in 1991 in Ontario, most provincial medical associations and provincialministries of health negotiated the creation of 'joint management committees' which were toprovide 'co-management' of the health care system." Typically, these agreements establish a fixedbudget for fee-for-service payments and if doctors wish to move to other methods of payment, theprovince is only allowed to remove that doctor's Medicare billings from the fee-for-service pool offunds. Because the doctors who wish to change payment methods tend to have lower thanaverage fee-for-service billings, the provinces have to come up with new money to changedoctors' payment methods.In some provinces, the ministry of health must negotiate the conversion of physician practices ona case-by-case basis. In all provinces which have such agreements, co-management is felt to be atleast somewhat of a barrier, However, there is still progress in some of these jurisdictions (e.g.Manitoba, Alberta) if there is political will and no active opposition by the provincial medicalassociation. However, Saskatchewan's lack of such a co-management agreement is widelyconsidered to have facilitated its lead in English Canada. The 1997 physicians' agreement inOntario has added even more barriers in that province. The new agreement stipulates that nomoney may be moved out of the fee-for-service pool to pay doctors in another manner. Ontariomust come up with totally new money to change the way any doctors are paid. It is widely feltthat this section of the new agreement will prevent the Ontario govemment from making anysignificant changes in physician payment until after it expires in 2000.There are other barriers as well. In provinces without comprehensive labour adjustment strategies,labour is threatened by a shift to community care and the reorganization of community services.Typically, institutional jobs are unionized while most of those in the community are not: As aresult many community jobs pay poorly compared to institutional ones and they are more likely tobe part-time. Less than five of the approximately 60 Ontario CHC's have unionized staff and,therefore unions have not supported their expansion. In some provinces there are problems withdifferent unions representing different workers in different settings. In British Columbia, two ofthe most powerful public sector unions are on opposite sides of this divide with the HospitalEmployees Union representing most of the institutional support workers and the BC GovernmentEmployees Union representing most of the support workers in community care. This conflict hasadded to the policy paralysis in Be, where the govemment is heavily influenced by organizedlabour. On the other hand, a clear labour agreement in Saskatchewan has facilitated thatprovince's move to new primary health care models.,In Quebec, BC, Saskatchewan, and Manitoba most workers in community health servicesare unionized.153Other non-physician health providers are also somewhat resistant to the CHC model. In mostprovinces the assorted community services (variously organized in the different provinces intopublic health, community mental health, home care, and addictions and alcohol services) had beenunder different administrations prior to regionalization. The workers in the field are somewhatreluctant to change their previous methods of working and many do not want to be integratedwith others. Administrators in regional authorities report that it has been somewhat difficult toconvince these workers even to move to one location much less to create new kinds of primaryhealth care teams.There is little political demand for primary health care reform. The public is much more concernedabout waiting lists for acute care. While a convincing case could be made that more Canadians aresuffering or dying because of lack of effective primary health care than a lack of high technologyservices, the public pressure is for more high technology services (in the vernacular 'hips andhearts'). This limits the ability of government to put more resources and policy attention into thisarea. However, there has been some advance in the understanding of the public and decision­makers of these issues in the past ten years.Confusion moving to clarity on the relevant information for primary health careIn general, the information about primary health care has been fairly confused for the public andpolicy-makers. There are many ways of reviewing the literature on primary health care. There is asmall body of controlled studies which has evaluated the performance of different whole modelsof primary health care in Canada. However, even when this literature is added to the larger bodyof research from other countries, researchers and interest groups differ on what policy conclusionsshould be drawn. Some researchers and policy makers feel the literature conclusively proves thatthe community health centre model is superior to traditional practice models.""We found that community participation health care centres (CHC) offersignificant economic and non-economic advantages. Their overall costs per patientare lower. They have consistently brought about significantly lower hospitalutilization rates. their emphasis on multi-disciplinary teams makes them better andmore cost-effective users of health care professionals..." (Angus and Manga 1990)On the other hand, some interest groups believe that there is a paucity of useful evidence toindicate a move away from the status quo:"Although it is claimed that community health centres constitute a cost-effectivemethod of providing health Cal'e, little information is available on this or on patienthealth outcomes compared with private office practice." (Canadian MedicalAssociation, 1994)24In fact, the literature comparing whole models of primary health care is very limited in its abilityto assist policy-makers. First, there will always be conflict about the generalizability of the results.For example, the studies from the early 1970' s of the Sault Ste Marie Group Health AssociationClinic ~the largest capitated clinic in Canada) are often dismissed because they are claimed to bedated." .26.27 Studies of the Saskatchewan CHC's are sometimes dismissed in Ontario because it isclaimed that the two provinces are too different." It is conceivable that an excellent study couldbe conducted of a CHC in east Toronto and it could be dismissed in Ottawa, (or even in west154Toronto).A second problem with the literature on whole models is that there are conflicts about whichcomponents of the models make a difference. Physicians' groups are quick to claim that betterprimary care performance could still be achieved by continuing doctors exclusive right of'ownership' and gatekeeping and simply tinkering with other components.'By the mid 1980's, provincial governments had come to believe that fee-for-service paymentincreased overall costs although this was not necessarily accepted by others in the health policycommunity. However, by 1997 many physicians' organizations have published their own critiquesof fee-for-service and there is a heightened awareness by other groups. For example, the Albertaprovincial auditor criticized fee-for-service payment in his 1997 annual reporr" noting that,"Some believe that a volume-driven payment system poses the risk of encouragingthe provision of unnecessary services."Another recent newspaper story reported that fee-for-service payment increased antibioticprescribing and linked this issue with the rise of drug-resistant bacteria in Canada.j"Policy makers increasingly accept the following argument for primary care reform.1. Current Canada arrangements for primary health care promote ineffective healthcare delivery and Canada's performance in this area could be improved.31.32.33.34.35.36.37..38. . 2. Changing the way doctors are paid but not changin~40~heraspects of primary careIS unlikely to make much difference In overall system performance. .1.3. Changing the ways doctors are paid and introducing elements of the communityhealth centre model could make a difference in patient outcomes and the utilization ofinsti tutions.39,40.41 ,42,43,44,45,46,47,48,49In particular, policy makers in other provinces have become fairly familiar with the evidence andother lessons from Ontario's HSO program, Even though the HSO funding eliminated manybarriers to better organized primary care and provided a major incentive to reduce hospitalizationthrough a bonus system, overall the physician-run capitation practices did not perform any betterthan their fee-for service counterparts. Also, the HSO's did not use non-physician providers muchbetter than fee-for-service practices. 14.50 This is consistent with other evidence that physicians areless likely to work collaboratively with nurses and other professionals if this decision is left totallyin their hands." As a result, nursing organizations and policy-makers are attending to the detailsof primary care contracts to ensure that these efficiencies are built into new funding models, Forexample, as mentioned above, the Manitoba government has insisted upon a protected budget forprimary care nurses in their capitation pilot in Winnipeg,This overall line of argument in favour of the CHC model is also consistent with major USThe subtitle of the Ontario Medical Association proposal on primary care is particularlyrevealing of this approach -- "A Strategy for Stability", (emphasis not in original)155literature, in particular the Rand Health Insurance Experiment which concluded that the costs forthe HMO patients were 25% less than those seeing fee-for-service doctors -- due almost entirelyto the fact that HMO patients spent 40 percent fewer days in hospital. 52.53Looking to the futureWhile the 1990's has not seen a wholesale move to change Canadian family physicians' paymentor toward new models of primary health care, there are signs of policy movement.After 25 years of regionalization in Quebec and 2-5 years of regionalization in eight otherprovinces, it appears that a new institution, the regional authority, has changed the balance ofinterests within the system. In 1975, Alford referred to the corporate rationalizers as thoseadministrators and planners within the system whom he saw increasingly challenging the medicalprofession's professional dominance." In the United States third party payers and theadministrators of integrated systems have taken control over the organization of health careservices from physicians and other professional interests." However, in Canada private third partypayers are still mainly potential interests and, up until regionalization, Canada had few integratedsystems. Particularly within medium and small-sized communities, physicians presented aformidable interest group to a series of administrators in institutional and community agencies.However, regionalization introduced a unified a management structure with increasedadministrative powers, more political influence, and a broad mandate to improve the health oftheir citizens.Originally physicians were able to stay out of regional budgets and maintain a direct politicalconnection with the province. But, over time, physicians are becoming more connected to theirregions. Increasingly, physicians are being credentialed at the regional versus facility level andprovincial medical associations are using regional boundaries for political purposes. InSaskatchewan the physicians who have service contracts sign them with the district health boardsand the district has a complementary contract with the province. In time, it is likely that physicianswill have even more direct connections with their regions which is likely to lead to more directaccountability to the administrators and governors of the system than has occurred in Canada upuntil now.Physicians are increasingly interested in being remunerated in a different fashion. Fewer doctorswant to run their own business as opposed to the security of a regular paycheck and the comfortof having someone else administer the practice. There is a growing public consensus that fee-for­service produces undesirable outcomes. There is also a growing consensus among policy-makersand health system stakeholders that simply changing the way doctors are paid will not achieveother policy goals.While it is still too early to predict with certainty, within 10 years it is overwhelmingly likely thatmost family physicians in most provinces will receive the majority of their income from non fee­for-service payments. It is also likely that primary health care will be more influenced by non­physician providers and administrators than it has been in the past. This may have profoundimplications for overall health care policy. Primary health care plays a pivotal role in humanservices system. A primary health care system which is dominated by physicians and concentrateson referrals to medical specialists tilts human services in the direction of individual treatment witha biomedical focus. On the other hand, a primary health care system with a concentration onhealth promotion where physicians are members of collaborative teams could re-direct humanservices in a more social direction.156157Table 1. NEW MODELS OF PRIMARY MEDICAL CARE IN CANADA - 1997(population c. 1996, general/family doctors* c. 1995)NEWFOUNDLAND (pop 550,000, general/family doctors 606)salary> 75NOVA SCOTIA (pop 900,000, general/family doctors 931)salary < 10service contracts - 20PRINCE EDWARD ISLAND (pop 130,000, general/family doctors 100)salary 0NEW BRUNSWICK (pop 750,000, general/family doctors 660)salary 43-4 CRC's7CRC's2CRC's2 CRe'sQUEBEC (pop 7,000,000, general/family doctors 7528) 160 CLSC'ssalary - 1200 capitated ambulatory long term care project?ONTARIO (pop 11,000,000, general/family doctors 10,230)salary 125capitation 300MANITOBA (pop 1,100,000, general/family doctors 1012)salary 60service contracts 35capitation 14SASKATCHEWAN (pop 1,000,000, general/family doctors 932)salary 95service contracts 70capitation 13ALBERTA (pop 2,700,000, general/family doctors 2453)salary < 5service contracts - 20-235BC (pop 3,700,000, general/family doctors 4080)salary < 3060CRC's77 capitated clinics14 CRe's1 capitated clinic5 CRe's3 capitated centres3 CRe's6 capitated centres in 1998?7 CRe's,The numbers are for active civilian general/family doctors. These doctors are non-militarydoctors who are licensed in the province and have a valid address.158Endnotes:1. The Community Health Centre Project (JF Hastings chair). The Community HealthCentre in Canada, report to the Conference of Deputy Ministers of Health. Health and WelfareCanada. Ottawa. 1972.2. Church WJB et al. Organizational Models in Community-Based Health Care: A Reviewof the Literature. A report prepared for the Fed/Prov/Terr Conference of Deputy Ministers ofHealth through the Advisory Committee on Health Human Resources. Health Canada. 1995.3. The Federal/Provincial/Territorial Advisory Committee on Health Services. The VictoriaReport on Physician Remuneration: A model for the reorganization of primary care and theintroduction of population-based funding. July, 1995.4. The Alberta Medical Association. Alternative remuneration for primary care: fee forcomprehensive care option. Edmonton. 1995.5. The College of Family Physicians of Canada. Managing change: the family medicinegroup practice model. September 19, 1995.6. The Ontario College of Family physicians. Bringing the pieces together: beginning theprocess. December 1, 1995.7. Manitoba Nurses' Union. Community Health Centres: The Better Way to Health Reform.Winnipeg. October 1995.8. Staff Nurses Associations of Alberta. Alberta Business Plan to Develop CommunityHealth Centres. Edmonton. April, 1995.9. The British Columbia Nurses' Union. Action Plan to Develop Community HealthCentres in British Columbia. Vancouver. September, 1995.10. New Brunswick Nurses' Union. Community Health Centres: The Better Way to HealthReform. Fredericton. March 1995.11. Newfoundland and Labrador Nurses' Union. Community Health Centres: The BetterWay to Health RefOlID. St. John's. October 1995.12. Medical Post National Survey. September, 1995. Pages 52-54.13. Lomas J. First and Foremost in Community Health Centres: The Centre in the Sault SteMarie and the CHC alternative. University of Toronto Press. Toronto. 1985.14. Abelson J, Lomas J. Do health service organizations and community health centres havehigher disease prevention and health promotion levels than fee-for-service practices. CanadianMedical Association Journal. 1990;142:575-581.15. Hutchison B, Birch S, Hurley J, Lomas J, Stratford-Devai F. Do physician-paymentmechanisms affect hospital utilization? A study of health service organizations in Ontario.159Canadian Medical Association Joumal. 1996;154:653-661.16. Ontario Ministry of Health. New Beginnings: Draft Discussion Paper on the Review ofthe HSO Program. February, 1991.17. Brooke, Jeff. HSOs -- the future of health care. Medical Post. March 9, 1993.18. Ontario Medical Association. Primary Care Reform: A Strategy for Stability. February 2,1996.19. Anon. Health service groups in the dark. The Globe and Mail. September 27, 1997.20. Bergman H, Beland F, Lebel P, et al. Care for Canada's frail elderly population:fragmentation or integration. Canadian Medical Association Joumal. 1997;157:1116-1121.21. British Columbia Medical Association. Capitation: a wolf in sheep's clothing. Vancouver.November, 1995.22. Barer ML, Lomas J, Sanmartin C. Re-minding our Ps and Qs: medical cost control inCanada. Health Affairs. (Summer 1996): 216-234.23. Angus DE, Manga P. Co-op/Consumer sponsored health care delivery effectiveness.Canadian Co-operative Association. Ottawa. 1990.24. Canadian Medical Association. Strengthening the Foundation: The role of the physicianin primary health care in Canada. CMA. Ottawa. 1994.25. Hastings JEF, Mott PD, Barclay A, Hewitt D. Prepaid group practice in Sault Ste.Marie, Ontario: Part I: Analysis of utilization records. Medical Care. 1973;11:91-103.26. Mott PD, Hastings JEF, Barclay A. Prepaid group practice in Sault Ste. Marie, Ontario.Part II: Evidence from the household survey. Medical Care. 1973;11:173-188.27. DeFriese GH. On paying the fiddler to change the tune: Further evidence from Ontarioregarding the impact of universal health insurance on the organization and patterns of medicalpractice. Millbank Memorial Fund Quarterly. 1975;53(2):117-148.28. Saskatchewan Department of Health. Community Clinic Study. Regina. 1983.29. Walker R, Chase S. Doctors' fees blamed for health-care costs. Calgary Herald.September 26, 1997.30. Papp L. Fee-for-service doctors give more antibiotics, study says. Toronto Star.September 29, 1997.31. Starfield B. Is primary care essential? Lancet. 1994;344:1129-1133.32. Asthma sufferers must take more care MDs warn. Toronto Star. March 20, 1996.16033. MacGowan J. Neurologist critical of FPs. Family Practice. November 6, 1995.34. FitzGerald JM, Swan D, Turner MO. The role of asthma education. Canadian MedicalAssociation Journal. 1992;147:855-856.35. Deutsch N. FPs can do more to prevent neural tube disorders: Not enough women arebeing told to take folic acid, study reveals. Family Practice. March 18, 1996.36. Pica LA, Boucher MM, Grignon R, Guibert R, Dery V. Hypertension follow-up surveyLaval, Quebec, 1988. Canadian Journal of Public Health. 1993;84:174-176.37. Abelson J, Lomas 1. Do health service organizations and community health centres havehigher disease prevention and health promotion levels than fee-for-service practices? CanadianMedical Association Journal. 1990;142:575-581.38. Vayda E, Williams AP, Stevenson HM Pierre KD, Burke M, Barnsley 1. Characteristicsof established groups practices in Ontario. Healthcare Management Forum. Winter/89: 17-23.39. The Diabetes Control and Complications Trial Research Group. The effect of intensivetreatment of diabetes on the development and progression of long-term complications ininsulin-dependent diabetes mellitus. New England Journal of Medicine. 1993;329:977-986.40. Bojestig M, Arnqvist HJ, Hermansson G, Karlberg BE, Ludvigsson J. Decliningincidence of nephropathy in insulin-dependent diabetes mellitus. New England Journal ofMedicine. 1994;330: 15-18.41. Vetter NJ, Jones DA, Victor CR. Effect of health visitors working with elderly patientsin general practice: a randomised controlled trial. British Medical Journal. 1984;288:369-372.42. John Pathy MS, Bayer A, Harding K, et al. Randomized trial of case finding andsurveillance of elderly people at home. 1992;340:890-893.43. Stuck AE, Aronow HU, Steiner A, et al. A trial of annual in-home comprehensivegeriatric assessments for elderly people living in the community. New England Journal ofMedicine. 1995;333: 1184-1189.44. Rich MW, Beckham V, Wittenberg C, Leven CL, Freedland KE, Carney RM. Amultidisciplinary intervention to prevent the readmission of elderly patients with congestiveheart failure. New England Journal of Medicine. 1995;333:1190-1195.45. Hall N, De Beck P, Johnson D, Mackinnon K, Gutman G, Glick N. Randomized trial ofa health promotion program for frail elders. 1992;11:72-91.46. Hendriksen C, Lund E, Stromgard E. Consequences of assessment and interventionamong elderly people: a three year randomized controlled trial. British Medical Journal.1984;289: 1522-1524.47. Cupples ME, McKnight A. Randomised controlled trial of health promotion in general161practice for patients at high cardiovascular risk. BMJ. 1994:993-996.48. Hart IT, Thomas C, Gibbons B, et al. Twenty five years of case finding and audit in asocially deprived community. British Medical Journal. 1991;302: 1509-1513.49. Renaud M, Beauchemin J, Lalonde C, Porier H, Berthiaume S. Practice settings andprescribing profiles: The simulation of tension headaches to general practitioners working indifferent practice settings in the Montreal area. American Journal of Public Health.1980;70: 1068-1073.50. Vayda E, Williams AP, Stevenson HM Pierre KD, Burke M, Barnsley J. Characteristicsof established groups practices in Ontario. Healthcare Management Forum. Winter/89:17-23.51. Johnson RE, Freeborn DK, McCally M. Delegation of office visits in primary care toPAs and NPs: the physicians' view. Physician Assistant. 1985;9:159-169.52. Manning WG, Leibowitz A, Goldberg GA, Rogers WH, Newhouse JP. A controlled trialof the effect of a prepaid group practice on the use of services. New England Journal ofMedicine. 1984;310:1505-1510.53. Ware JE, Rogers WH, Davies AR, et al. Comparison of health outcomes at a healthmaintenance organization with those of fee-for-service care. The Lancet. 1986;i:1017-1022.54. Alford RR. Health Care Politics: Ideological and Interest Group Barriers to Reform.University of Chicago Press. Chicago. 1975.55. Anders G. Health Against Wealth: HMOs and the Breakdown of Medical Trust.Houghton Mifflin Company. New York. 1996.162163TONY MATHIEDiscussant PresentationI'm delighted, if somewhat surprised, to be here. I have washed the straw out ofmy hair before coming, but as you've heard, my background is very firmly rooted in actualday to day, ordinary, common or garden general practice. I did chair a health authority fora while, but that was before the reforms and seems an awful long time ago. Before I gaveup practice I was visiting my patients one day in a rural area, and I went to a farm, andoutside the front door of the farm, much to my surprise, was a pig with a modern, hightechnology, fully articulating, artificial leg. After I'd seen the farmer I said, "Bert, Icouldn't help noticing the pig outside the front door." And he said, "That's the mostwonderful animal. A few months ago we had a fire in the farm at night, and we were allasleep in bed. The pig smelled the smoke. It got out of the sty, nuzzled up the latch of theback door, came upstairs, woke us up, led us to safety through the smoke." I said, "Isuppose it was injured in the fire." He said, "No, no, nothing happened to him in the fire atall." I said, "What's the story about the artificial leg?" And he said, "You see if you've gota pig that smart, you don't eat it all at once."Now, I tell that story for two reasons. The first reason of course is to get a cheaplaugh. The second reason is that it often seems to me that when things aren't going well,that pig is an allegory for primary care. People keep telling me how important it is, butthey're chopping it apart bit by bit by bit.I've enjoyed the papers that we've heard this afternoon tremendously. I have tosay, Michael, that from the other side of the pond, the Canadian system looks rather moreintegrated than you've painted it. But certainly community health centres do have a feelthat they might be something like we're used to in the United Kingdom. I suppose the badnews for you is that you've asked if there's a light at the end of the tunnel. I have to tellyou that it's a train coming.Angela, when I read your original paper, I found it very balanced, because I keptwriting at the side, oh my God, and yes, and oh my God and yes. At the end there were anequal number of both, so I assume that means that it's a balanced paper. But there were a164number of things, having been a fundholder myself, that I think it perhaps would be helpfulif I just added to the equation. You quite rightly said that health authorities performed anamazing task. When fundholding first started health authorities said, don't worry, don'tinvest money in this, it will wither on the vine, nothing's going to happen, it's just apassing phase. You join the system year on year, so that on the first of April each year youhave an opportunity to join. The doctors who joined in the first wave were not pilots forthe scheme, but they were certainly the Kamikaze pilots of British general practice. Theyinvented methodologies and ways of working which no one had ever thought about. Theyput together limited companies which allowed them to buy care from themselves at aprofit. They may not be allowed by fundholders to keep money directly from the scheme,but you'd be very surprised, I'm sure, to hear that a number of fundholders have boughtbigger and better premises with their funds. They've certainly had money to help them buycomputerised systems for their practices and the penetration of on-desk computers inBritish primary care is extremely high. And, although they may not get money in-year, as itwere, many fundholders will profit when they retire. As to the future of fundholding, wehear that the Secretary of State has already said that if you are a manager in a fundholdingpractice, start looking for another job. But the Prime Minister has told him that's not quitethe party line, so it's really quite difficult to know, and the white paper keeps on gettingput back. So we're not quite sure what's going to happen there.The big problem, I think, for fundholders was that, since 1948 when the NationalHealth Service started, general practitioners have had no voice. The hospitals simply didnot listen to a word we said. Fundholding did seem to be an opportunity to get thingschanged, to get things done, and many people joined from that point of view. Then whenit gathered momentum, in many parts of the country there was huge pressure to becomefundholders. Many practices were grouped together in really unholy alliances, and calledfundholding practices, managers were put in from health authorities to help them run, andin no sense were they a collaboration or in any way a grouping that would normally havegot together, but were rather forced into this situation. So, although penetration as you'veseen has reached just over 50% of the population, certainly in the later wave, this has beena rather contrived experience for many practices, and for us to see some of the practices,165who have gone in the sixth or seventh wave, some of them (fair enough) have beenstruggling with internal practice dynamics, and having got those sorted out they've comein. But many of them, having said no, no, no for five and six years, have then come in tofundholding in rather a strange way. There arc quite marked differences in the makeup ofthe waves. As I say, the people in the first wave really were a different sort of animal frompeople in later waves.I think the thing that finally persuaded the current government to get rid offundholding is what has been called two-tierism in the National Health Service. Angela hasreferred to this idea that fundholding patients get into hospital easier, see specialists easier,and so on. I think this is a fallacy. I mean, I'm not saying that it didn't happen, but I don'tthink it's got very much to do with fundholding. There's always been at least two tiers inthe National Health Service, leaving aside the private tier as a back way into the NationalHealth Service. There's always been those OP's who would, in a phrase that might befamiliar to some of you from over here, go the extra mile for their patients, and those whosimply would say, well I'm sorry, waiting lists are very long, you'll just have to wait thetwo years or whatever is okay. There's always been two tierism, and in general the peoplewho were prepared to go out for their patients more have been the doctors who havejoined the fundholding system.Kevin's paper I thought was wonderful, and I was delighted to read it. There werea number of things in it which rather struck me as, not bizarre, that's not fair, but theseven-to-one ratio of specialists to family doctors is something that I had to read severaltimes. In the UK, as you probably know, the ratio is 2:3 at the moment. It's approachingparity because specialists are growing at the rate of around 5% a year, whereas familydoctors are growing at the rate of less than 1% a year. But 7: 1 is a ratio I can't evencomprehend in my mind. The discontinuity caused by the movement of plans is extreme,where I think in your paper you said 25% of people over a two year period have forceddiscontinuity in their medical care due to decisions made by the people running the plans.This was something which I find very difficult, particularly coming from a leafy lanes, aswe call it, from a rural area in England where I've looked after people from generation togeneration, so that until you've been in the practice as a partner for about 15 years you're166still known as the young doctor. If you don't come from their part of the country you'rethe young, foreign doctor, and it takes a very long time. It's said in English generalpractice that until you deliver a baby from a mother who you delivered, you haven't reallyarrived, Now, of course the situation's changed quite a lot, and it's going to change evenmore in the future.But I have to tell you, I'd been a OP for 15-16 years at least before 1 realised I wasbeing the gatekeeper. For the first 15 years or so of my time as a OP, this is what Ithought I was doing. I was providing personal, primary and continuing care to a registeredpanel of, in our case, about 13,000 people. Tending them in their homes, in my practicewe would do, as recently as 1990, over 10,000 home visits a year between us. Weaccepted responsibility for making initial decisions on every problem that presented,consulting with specialists when we thought it appropriate to do so; neither more nor less.As you've heard, in our case in a rural area, probably around 5% of our contacts would bereferred for specialist advice. So, I think that gatekeeping was something a bit likespeaking prose, in that I didn't realise I was doing it for many years, But having looked atthe gatekeeping role, and accepting that there are good things about it, there are thingsabout patient support and advocacy, things about continuity of care and responsibility forpatients, there are the good things like the OP being the care manager for the patient, theregistered list concept, the unified medical record, the referral and feedback system whichwe have, which allows us all sorts of advantages that many other systems don't have.However, there are many considerations now, which I think are coming in, whichwill limit our ability to work in this way in the United Kingdom, and I suspect elsewhere.The first is the availability of doctors. We have managed on about half the undergraduateoutput in the United Kingdom. Until we got short of doctors, we'd always assumed thatmost of those people were volunteers who wished to be general practitioners. Now thatwe're short of doctors, the latest cohort study from Parkhouse and his colleagues suggeststhat we're down now to about 18% of final year students for whom primary care is theirfirst choice of career. This is half the number that was coming out 10 years ago. We arerecognising more and more that many people are becoming primary care doctors becauseother avenues have shut to them, and that in many cases, negative attitudes about primary167care from hospital specialists with whom they have worked and trained have forced theminto primary care. Fifty-five per cent of our trainees in general practice are women now.This is not something I've heard yet today, and I apologise if this is not politically correct,and I will try and avoid doing what I normally do, and referring to them as the contrarygender. But in England where 55% of our trainees in general practice are women, weknow that 30% of the principals, 30% of the fully trained general practitioners actuallyworking are women, and that number is rising year on year. We also know that, at anyonetime, only about half, less than half of these women work full time. But if you look at themacross a career, they deliver about 3/4s of full-time. That's not meant to be pejorative inany way, because we know that men in England are working much less than what wetraditionally regarded as full time as well. The effect of this is that there will be 100 peopleretiring from the top of the profession, who will mostly be men, who will mostly haveworked full time throughout their careers, who often will have been in one practice for 35years or more. Although it's a very crude calculation, and it's based on other things likethe fact we're also training doctors to go back to the European community under theserather curious regulations that we find ourselves saddled with, in order to replace themwith the equivalent amount of career productivity, we need to train 150 doctors, andapparently we're training about 80. So we only have a capacity to replace half our retiringfte general practitioners in England. This is going to cause us problems. You need to addto that the fact that in the older age groups of general practitioners, particularly in innercities, we have a very large number of doctors who trained outside the United Kingdom,often in the Asian subcontinent. Due to changes in immigration regulations in England,they will not be replaced. We're facing a very severe shortage of career generalpractitioners.In addition, of course, the people who are coming in have absolutely no intentionof working for the same length of time or with the same intensity that we worked and ourforebears worked. I recently had the unpleasant experience of advertising for a newpartner in the practice and we had 50 applicants. When we got down to the short list, wetook the two or three doctors aside and gave them all the details about projected income,about the financial situation with the building and equipment, about the legal tying into the168practice, the way we do it in England, and this doctor looked at me and he said, "I'mflattered to be shortlisted, but there's a bit of a problem." I said, "Well, I'm sure we'll beable to work it out, what's the problem?" He said, he liked to get off to the cottage on aFriday. I said, yeah, well that'll be fine, one Friday in five or something, I'm sure we can ..He said, "No, we like to go to the cottage every Friday, and we like to be away beforelunch because the traffic builds up." This guy turned down the opportunity of being in apartnership in a rural area of England, good quality of life, earning at that time £55,000Sterling a year, job for life, turned it down because he wanted to be away to the cottageon a Friday afternoon. When I'd revived my senior partner and explained to him that hewasn't having a holy emanation, we had to face the fact that it would no longer be likelythat we would find people who would come out of vocational training, enter our practiceand work full time for 35 years as most of us had done. When I left my practice on thefirst of January 1996 to become a full time director, I was the first doctor to leave mypractice since 1973. Now that's not a particularly common situation in the UnitedKingdom, but certainly in rural areas it's not unheard of.So, we have major problems. Now, skill mix is coming into the equation. We'regoing to talk about that, I know, tomorrow. But in British general practice, we've movedfrom a situation of having 1,500 practice nurses in 1982 (these are nurses employed bygeneral practitioners, working in their offices, working in their surgeries), to 18,000 in1995. This is a huge increase. So skill mix has been taken on board. It is regarded bygeneral practitioners as a very threatening thing. And GPs' views about what is needed todeliver primary care is, I think it would be charitable to say, extremely naive, When nursesand doctors were asked about what would be needed to deliver perfectly trained nurses totake over some of the role in primary care of doctors, while the nurses were very clearabout their educational needs, the doctors just thought they would need a bigger room towork in.The other big issue that we're facing and that's been alluded to is variable quality.I'm rather surprised that it took so long in the day for people to bring it to mind the factthat outliers are at both ends of the scale. Certainly, we have major problems in quality inthe United Kingdom, such as unexplained variations in performance, referral rates to169secondary care varying ten times between practices in the same location dealing withbroadly similar patients. We have just brought the United Kingdom regulations to dealwith what I call "doctors behaving badly", but is actually called "poorly performingdoctors." This is below the level of malpractice, but at the level of persistent poorperformance, where our General Medical Council, which is currently headed up by ageneral practitioner, is proposing that these doctors should be offered remedial help, and ifthat help doesn't deliver the goods they should cease to be able to practise. Many of thesedoctors, unfortunately, are single-handed, overseas-trained doctors in inner city practice.The current prediction is that primary care doctors will be hit by these regulations in aratio of 4: 1, compared to their specialist colleagues. I hold no brief for these doctors. Isee a lot of them as part of my role in postgraduate education, and I'm very sad to seethem. But to be honest, many of these have never had any guidance of any sort throughouttheir careers, and have been able to do almost what they like. The health authorities whoarc supposed to be managing them have turned their backs on them, simply because theyknow it's difficult to recruit replacements. So they've been allowed to get away witheverything. To reflect on what my colleagues have already said, there is almost nopressure from patients to remove these doctors from their posts, or to change the systemunder which they work. Indeed, one of the problems we're going to have with poorlyperforming doctors is persuading enough patients to make a fuss, or to make complaints,to allow us to go to the General Medical Council with carefully thought out cases tosupport taking the remedial route with these doctors. So that is a big problem that wehave.There are perverse incentives in our system, of course, as there are in all systems.The major one in our system is that the current funding situation means that the less youdo the more money you get, and we've heard this from elsewhere. The situation is thatthere is partial reimbursement for staff and for premises. It makes no sense for somedoctors to spend a lot of money on staff and premises if you can get away with less. Againin the inner cities, there are doctors who are single-handedly looking after up to 5,000patients on a registered list, in very high dependency areas, and are continued to be170allowed to do this, on the grounds that they are looking for partners. And when youactually look at what they're offering partners, it's not a surprise that they can't find them.I was interested in Kevin's description of centripetal as against centrifugal systems.Certainly in England, one of the things the OP fundholding system has done is to be veryovertly centripetal; we've brought specialists into the community and into outreach clinics.In addition to a vast increase in the numbers of administrative staff, we have all theseoutreach clinics with specialists coming to our surgery to see only our patients. But we'restill the gatekeeper in that these are all referred patients with referral letters and referralresponses from the specialist. We have a very multi-disciplinary-looking team. I say"looking team". I will be scarred for life by the experience of sitting in a room like this fullof doctors and primary care nurses, and asking the question, how many of you think youare working in a functional, multi-disciplinary team? All the doctors in the room put theirhands up, and all the nurses looked out of the windows. To this end, I have, ten days ago,appointed to my own director team in Liverpool (the first such appointment in thecountry), a nurse to be an associate director to demonstrate that we are actually not onlytalking multi-disciplinary teamwork, but we're actually going to try and deliver it anddemonstrate that we have it. So that's one of the issues also for us.We have, as you've heard, a monopoly on delivery of specialist care, in the sensethat people cannot go anywhere except to emergency care, although in some places somespecialists have tried to bypass this by introducing direct access for what are called at-riskpatients, people like severe asthmatics in childhood, chronic respiratory obstructivepatients and so on. But this has not proved to be very popular.I don't want to say any more about fundholding. I'm not sure that commissioningwithout money is going to get many OPs out of their beds early in the morning, but timewill tell. But what to me has been the central ethical dilemma of fundholding is that whenyou're in your consulting room with your patient, that's the person who you areadvocating for, that is the person who you fight for, that is the person who you want thebest possible care for. You can then step outside your consulting room and look to getgood care for the rest of your practice, Frankly, if you get better care for your patientsthan the guy down the road gets for his, then so be it. The problem then is when you step171back one step further and start talking to colleagues in other practices about how you getthe best care for the population in your district, you are in a dilemma. Because you dothen go back to your practice and fight tooth and nail for your own patients, and if theycan't fight tooth and nail for their patients, then that's bad luck. That has been a dilemmaand the rationalisation of that dilemma is something which has caused a lot of fundholdersa great deal of grief.Talking informally with colleagues, there is clearly a cry for some sort ofcentralised policy and planning of health, simply to work on an incremental basis fromwhere we are now. All I would tell you is that, in the United Kingdom, having started offwith a new govemment who promised that they would be policy-driven, their policy ofcourse was to beat the Conservatives, and now they've done that, they are stuck a littlebit, because they're not quite sure what Plan B is, and certainly we're not seeing muchsign of policy-making. I'm sure my colleagues from the UK would agree that neither arewe seeing much evidence that we can actually get to the table to help them make policy. Ifthat of course is the case, then we'll be stuck as many of us are with political dogma, andwe'll be stuck with just trying to make the best of the systems we've got.172General Discussion after Afternoon Panel:The discussion opened with the panel chair commenting on the difficulty of the issue ofprimary care, that it is a blend of disciplines, but we train in different ways as well and we tendto define primary care around workforce, not practice. There appear to be a couple of agendas:1) a practitioner or practice that is medically effective, and 2) social justice or equitable accessto service. Medical effectiveness and social justice sometimes overlap and sometimes do not.There followed a comment from the floor that the afternoon and morning speakers had movedtoo far from workforce policy and planning which should be concerned with training anddistribution and making changes in training and policy. When asked to clarify, he replied that hewas concerned that we not put more dollars into training but that we should concentrate on howto train the workforce better, e.g. to be more sensitive to things like ethnicity.The next speaker commented on the situation in Winnipeg, population one million with600 physician Fl'E's. They have one capitated clinic with 14 doctors serving 28,000 patientswhich is a ratio of 1 doctor per 2,000 people. What will happen to these physicians, that is theones that are not in the capitated clinics, who will be in surplus as more capitated clinics startup? Therefore, don't put more doctors into rural areas and don't gear up training programs buttake advantage of the change in the doctor-population ratio in metropolitan areas as newerdelivery models take hold. The training of nurse practitioners and midwives will only make thesituation for surplus physicians worse. Also, in Ontario there's at least one clinic caring for5,000 mental health patients, with only 0.25 F'I'E's in psychiatry. If this concept is extended toother clinics, there will be a decreased need for specialists. The panel chair then commented thatit seemed as though if we can constrain workforce growth, it would be a good idea.A speaker from the UK then commented that all three countries seemed to be indulgingin the same fashion. The issue in the UK, for example, was that the National Health Service issaid to be primary care led, but what does this actually mean? There tends to be a lack ofleadership in primary care, there tends to be little evaluation of practices, and there is nounbiased national database which shows what primary care physicians actually do. There havebeen attempts to tie money to the performance of GP' s, although the evidence base for this ispoor, partly because it is difficult to measure. There is an inability to get more physicians intoprimary care, but why do we need them, given that we have more help from other disciplines?When looking at policy initiatives we should ask, is this just the fashion? What's the evidence?Have we a golden opportunity here to change skill mix? He asked if we do know theeffectiveness of inputs in relation to the difficulties of outcomes measurement. He asked whyit is a problem to replace GP's when we can use nurse practitioners and could go for larger listSizes.The discussant replied that he didn't think that patients are getting better care - just morecare. For example, there are now counsellors for the worried well but the worried wellcontinue to come to the GP. John Fry always said a patient-doctor ratio of 1:4000 ismanageable. With regard to NP's - at this time there is no training program in the Ll.K.Primary care practices are only are able to recruit nurses. There is also no description of what173they should do - and at the same time, lots of primary care nurses are about to retire and can'tbe replaced with trained NP's. He expects an increase in medical school training. Anotherpanel member then replied that there's no strong case for an increase in the number of generalpractitioners. There's limited evidence on the efficacy of primary care. GP or nurse primarycare, for certain diseases, is not shown to be more cost-effective than secondary care, andrecruiting more people into primary care practice does not seem to enhance the quality of care.The debate then shifted as another panel member commented that there's good evidencethat primary care reduces hospitalisations and increases good outcomes and the use ofprevention measures. The challenge is to come up with good outcomes. How do we measurethis Gestalt of a whole care model versus the single organ model? A conference participantremarked that there's a difficulty in providing good primary care in the current system in Canadasince we need a different organisation of care and to readdress some of the problems of training.Another conference participant raised the issue of training for primary care, as itappeared the three countries were offering conflicting advice. The Americans were advisingprimary care training, while the Canadians were saying no more GP's and the UK were sayingnot only primary care, but more of it, if possible. A paper he had seen from three years ago,trying to predict the future, had stated that many primary care physicians would not be required.He wondered what the panellists had to say? The discussant replied that, as far as he wasconcemed, medical students should go into general practice. He is still practising in the careerthat he started in 1969. General practice means more hours, independence, working as ageneralist and general practitioners are generally more prepared to live with uncertainty than arespecialists. Another conference participant then added that the need for GP's will continue togrow cause the public demands primary care from doctors. Right now we have shortages inrural areas and we can't get primary care physicians out of the metro areas. It's difficultenvisioning areas in which the population knows which specialist to go to and the role of the GPin terms of an advocate or helper is not required.One of the panellists then commented on the lack of primary care policy levers in theU.S. What are the effective policy initiatives to promote primary care, for example, exhortationor bribery or blunt instrument? There is federal financing for primary care but no accountabilityaround how the money is spent. In the US the big player is the market. Planning does notappear to get it right. The US talks about underemployment but you still don't really see it. Inthe US, 13-14% of students now go into family practice, but it used to be a much lower figure,and now others go into the other generalist specialities as well. He thinks that the market onlyoperates at the margins and there will be no huge unemployment or loss of income forspecialists. He went on to say that we can not talk about primary care in isolation. It is tied tospeciality care. He asked if care in the UK would be different if they had more specialists? Thepanel chair commented that when he went into practice in 1969, GP practice was anathema atthe University of Chicago, but generalism in medicine and paediatrics is now respectable. Thereis an upward trend in primary care despite gate -care is not important. Why is it that we woulddiscuss reform of primary care without talking about specialist care? Maybe what we areactually talking about is a change in the whole system. A panellist replied that he, an American,174admires the UK system - but it is a system that is wedded to salaried remuneration and lownumbers, neither of which is likely to come to America any time soon.A participant commented that he was surprised at the direction of primary care in theUS. There are powerful forces including I-IMO's and the market in the US, which talk about theneed to provide primary care to reduce cost. In Canada, the OP's are the ones that USrecruiters are looking for. There are also increased patient expectations that someone will lookafter them as a total and not as a collection of organs or body parts. Medical schools work inthe abstract and are concerned about what students should learn. The Ministry of Health hasideas on funding and organisation, the profession has ideas about practice. We need moreintegrated planning between all three organizations which implies the risk of a loss of autonomy.Someone else then pointed out that health care reform in the US did not fail for lack of healthservices researchers. She went on to say that we must identify areas where policy change canmake a difference and identify what other questions we are trying to ask in each country andwhere is the common ground.It was then pointed out that the concept of primary care that emerged from the Alma­Alta Conference on Primary Care was aimed at developing countries and how to provide care tothe relatively underserviced. In the UK, OP's are paid more for having patients in deprivedareas. Directing labour to areas won't work. Squeezing those in the metropolitan areas won'twork either; they would just leave and go to another country. In the UK they tried parachutingphysicians into underserviced areas. They took them shortly after training and placed them indifficult-to-manage practices under supervision. There was weekly support (this was a keyissue), and also the ability to get academic input. Most have opted to work in these practicespart-time, in Manchester, Liverpool and other difficult areas, provided that they can keep upwith the weekly support. However, they don't want to be full time. OP's in the inner citiesobviously need more flexible alternatives.One of the panellists then commented that we need to talk about the efficient productionof health. For example, there is a huge difference between a primary care system which screenseverybody for cholesterol versus a system that only screens 2-3%. Primary care needs to playabroad role in allocating resources. For example, CLSe's in Quebec were involved with guncontrol legislation, accident prevention and similar programs. In terms of getting physicians intounderserviced areas, the Cuban model where doctors are assigned only seems to work in aautocracy. Paying more for those in rural practice does seem to have some effect, as doincentives for young physicians such as scholarships and loan repayments.Lastly, the panel chair raised the issue of distribution. How could physicians be allocatedto under-serviced areas? Some ideas have been a social security year or a return of servicerequirement. The Cuban model with assigned place of practice is very efficient. Twoincentives have been used in the U.S. - pay more (i.e. some benefit from the Medicare bonus) orreturn of service through NHSC. It is possible that there are other options more palatable I tostudents.175Evening PlenaryMedical Workforce Policy and Planning in AustraliaJohn Horvath176MEDICAL WORKFORCE POLICY AND PLANNING IN AUSTRALIAProfessor John HorvathChairman, Australian Medical Workforce Advisory CommitteeAddress to Second Trilateral Conference on Physician Workforce PolicyVancouver, Canada15-16 November 1997Thank you. I am pleased to be here and honoured to have been invited to address theconference.As the program says I have been asked to speak on what is occurring in medical workforcepolicy and planning in Australia. In doing so, I also intend to speculate on some of thechallenges for us in this area for the future, which I trust you will find relevant not only toAustralia but to all of the countries represented at the conference.I am afraid that as with all medical workforce issues, the Australian situation requiresconsideration within the context of the type of health system we have in Australia. I am suremost of you are familiar with the basics, although I will summarise the structure quicklybecause I think that will contribute to a better understanding of the Australian medicalworkforce situation.The Australian health care systemAustralia is a federation of six States and two Territories. The powers of the national, 01'Commonwealth government are defined in the Australian constitution; broadly speaking thosethat are not specifically vested in the Commonwealth, remain the responsibilities of the Statesand Territories.In the area of health care, a distinguishing feature of its organisation is the extent to whichresponsibilities are split between the three different levels of government. (And as an aside Ishould add the cost shifting between governments this split encourages). The Commonwealthgovernment is responsible for the funding of medical services, aged care services andpharmaceuticals. The States and Territories are responsible for the provision of public healthservices, principal among which is the running of public hospitals, although because of itswider taxation powers, the Commonwealth makes a contribution to the cost of running thehospitals. Local government responsibilities vary but are mainly in the area of environmentalcontrol and home care services. In practice, there is considerable overlap in some areas,particularly aged care.177Payments by the Commonwealth to the States and Territories are usually made via tied grantsthrough a system known as Medicare, which has operated since 1984. Medicare is fundedthrough an annual levy on individual taxable income, currently set at 1.7%.For the individual Australian citizen, Medicare entitles you to universal access to 'free' healthcare, either in a public hospital or through a practitioner. Medicare does this by reimbursingthe provider at what is known as the scheduled fee, which is the amount set by theCommonwealth government as fair payment for a particular service. Providers are at liberty tocharge above the scheduled fee, with consumers paying any difference.Agreements between the Commonwealth and State and Territory governments provide for allMedicare beneficiaries to obtain inpatient and non inpatient care at public hospitals withoutcharge. Medical care is provided for such public patients (at no cost to them) by doctorsappointed by the hospitals.Private health insurance is also available and it covers individuals or families who take out theinsurance for private services provided in either public or private hospitals. Currently,approximately 30% of Australians have private health insurance.Medicare is popular with consumers, but has obvious impacts on the medical workforcemarket place. It provides practitioners with a price floor in the market and to some extent aguaranteed income, although the level of income is of course related to throughput, givenAustralia's reliance on fee-for-service medicine.Characteristics of the current Australian medical workforce.As you can see, Australia is quite a large country in area, being a bit smaller than theUnited States or Canada, but roughly the size of Europe and the United Kingdom (SeeAttachment 1.)Australia is a country of some 18 million people. It is a highly urbanised country, with most ofthe population clinging to the coastal fringes. Despite its vastness, much of the inland is desertor semi-arid. Rural areas account for most of area, but just 29% of the population. Most ofthe population is located on the eastern seaboard, including the three largest States of NewSouth Wales, Victoria and Queensland, which account for approximately 60% of the totalpopulation. State populations range from six million in New South Wales, over half of whichlive in Sydney, to several hundred thousand in Tasmania and the Northern Territory.To serve this population, Australia currently has a medical workforce of 46, IIIpractising clinicians. Essentially the workforce comprises four main segments:1. 21,037 general practitioners, who provide primary care services; 22% of whom178Attachment 1179practise in rural areas. General practitioners currently comprise 45.6% of the workforce.Essentially they are private consultants working in either solo or, more usually, grouppractice. In rural areas, some general practitioners will also have hospital visiting rights.2. 15,318 specialists, who quite obviously provide specialist consultation and proceduralservices. Specialists make up 33.2% of the medical workforce. 12% of specialists are locatedin rural areas. Specialists practice in either private rooms or hospitals, or most commonly in amixture of both.3. 4,468 specialists in training, representing 9.7% of the workforce. These are medicalpractitioners who have been accepted by a specialist medical College into an accreditedtraining program.4. 5,288 hospital non-specialists; these are non-vocationally-trained specialists whopractice solely in public hospitals in salaried positions. Hospital non-specialists comprise11.5% of the workforce.Some other interesting characteristics of the workforce are:27.2% of the current workforce is female53.6% of female clinicians work in primary care (compared to 42.7% of maleclinicians)Of the 15,318 specialists, only 14% are female. and in the surgical specialtiesonly 3% are female20.7% of the Australian medical workforce is aged 55 years and over (28.7% of thespecialist workforce)74.8% of the female workforce is aged under 45 years (compared to 49.9% of the maleworkforce)the average age of the urban medical workforce is 44.7 years and of the rural medicalworkforce is 45.4 yearsThe Australian Medical Workforce Advisory Committee (AMWAC)You might be thinking, well, what is the Australian Medical Workforce Advisory Committeeanyway? Essentially, we have been around for nearly three years now; we are basically aresearch organisation funded by govemment, reporting to the Australian Health Minister'sAdvisory Council and through it to Australian Health Ministers. We were formed to assistwith the development of a more strategic focus to national medical workforce planning inAustralia; and in one sense we grew out of a period of conflict between government and the180medical profession following several government reports that touched on medical workforceissues but were very much hostile to the profession, particularly a report by Professor PeterBaume of which some of you may be aware. A more co-operative approach was necessary toprogress issues to resolution and at the same time, governments quite rightly recognised thatmany of the medical workforce issues actually required considerable research and better databefore conclusions could be conclusively made.The prime focus of AMWAC's work is Australian medical workforce research and dataanalysis, although AMWAC also aims to provide workable policy solutions where appropriate.AMWAC' s role as a workforce planning agency is centred around examining:the structure, balance and geographic distribution of the medical workforce;medical workforce supply and demand;present and future medical workforce training needs;models for projecting future medical workforce requirements and supply; anddevelopment of medical workforce data collections.The membership of the AMWAC comprises representatives from Commonwealth (Australian)government departments, State and Territory health authorities, the Australian Institute ofHealth and Welfare, the Australian Medical Council, peak organisations representing varioussections of the Australian medical workforce (including the Australian Medical Association,the specialist Medical Colleges and University Medical Schools), and consumers. TheCommittee is supported by a small Secretariat based in Sydney.I should hasten to add that the nature of the Australian system of government dictates that weare by no means the sole body; but we certainly operate at the peak national level and have abroad ranging terms of reference.The aim of Australian medical workforce planningIn its very simplest form, we like to consider medical workforce policy and planning to beabout ensuring the right people are in the right place, at the right time, with the right skills tocompetently and proficiently perform the tasks expected of them in accordance with world'sbest practice. And whilst a strict balance between supply and demand may always be difficultto achieve in medical workforce management, planning to minimise the divergence between thetwo should not be that difficult.From our perspective, to ensure that all Australians have access to quality medical services, thesize of the medical workforce must be appropriate to the needs of the community. In verysimple terms, too few medical practitioners will mean there is under-servicing, poor healthoutcomes, and overworked practitioners; too many practitioners and there is the likelihood ofover-servicing and underemployment of practitioners. Both scenarios involve the nation in anunnecessary financial burden.181Key Australian medical workforce issuesAt the present time the general view is that Australia has too many doctors and that many ofthese are in the wrong place. So, broadly speaking, we are currently dealing with issues ofoversupply and maldistribution - both geographic and structural. (As an aside I shouldprobably add that you can, however, come across views that consider Australia not to beoversupplied with doctors). This in turn, creates the paradox of medical workforce policyhaving to simultaneously develop ideas to deal with the somewhat mutually exclusive problemsof surplus and shortage.As a result, in Australia, planning for, and management of, the medical workforce is regardedby many as one of the most important challenges facing the health industry today. At the sametime, because of the complexity of the medical workforce and the number of groups involved,medical workforce issues can be some of the most complicated and difficult issues in healthpolicy, as I am sure you are all well aware.The range of significant medical workforce issues around at the present time in Australiaillustrates this point.As I indicated, there is a general feeling that Australia has too many doctors. Yet many ruraland remote communities are just as certain there aren't enough doctors or, if there are enoughdoctors in Australia, then some of them are certainly in the wrong place.Junior doctors are expressing concern about future career options and are unhappy becausethey believe they are overworked, and that they may be unable to access training in the areaswhere they want to train.Debate continues about the appropriate number of specialists in some disciplines, and in mostabout provision of services to rural communities.University medical schools have been told to reduce their student intake, which in tum raisesconcerns about the future viability of some of the smaller schools. Medical schools are alsoincreasingly focusing on how they select their students and the type of graduate they produce,which has led in recent years to a number of changes to the medical student selection process.There is an array of extremely complex issues surrounding foreign doctor entry into Australia,and examination of such entrants to ensure they meet our standards of proficiency and skill.Similarly, there is a range of potential effects associated with the increased participation ofwomen in the medical workforce, none of which I should add are considered negative, butsome of which may certainly change the dynamics of the workforce and the work place.On top of all this, consumers are often unhappy with a lack of timely access to services, andgovernments are constantly having to balance providing a quality service and better healthoutcomes with the available resources and the now ever present need to use the resources182more efficiently.This of course is a listing that relates to the present; there are other issues that would arise ifthe nature of the Australian health care system changed, for example by a move to managedcare. However, we have not got into modelling the effects of major policy changes, preferringto work within the structures of the here and now, rather than the what might be. That is notto dismiss the importance of maintaining an eye to possible future structural and fundingarrangements; it is more a statement of the practical reality of the need for good quality dataand research analysis on the present and the short term.I should probably also point out that focusing more on medical workforce policy and planning,and integrating it into planning for the health system as a whole, is a comparatively recentphenomenon in Australia. In the past, workforce planning was largely ignored for a range ofreasons, including the fact that the economy was always growing, as was the population. Thepost World War II demand for doctors was undoubtedly high and in any case, there were noreliable data or projection modelling around to inform the policy process. The lack ofenthusiasm for planning probably also reflected a belief that the market would solve anyproblems so long as sufficient doctors were produced and trained.The level of workforce analysis usually focussed on the simple need for more doctors asdictated by population growth, and the boosting of medical school intake to achieve this. Atthe same time, specialist practice was expanding, particularly as technological advance enabled,and indeed encouraged, greater sub-specialisation.Yet, while the actual growth in numbers was comparatively unchecked, standards and qualityhave always been key issues for the profession and indeed can be traced back to the mid­1800's and early 1900's, when government devolved responsibility for overseeing medicalpractitioner registration to the profession through the establishment of State medicalregistration boards. Similarly, specialist medical Colleges have always had responsibility foroverseeing the training and examination of specialist trainees.In the 1980's concerns about the basically unchecked growth in the medical workforce beganto surface. In part, this was clearly related to the slowdown in the economy and thecomparatively rapid increase in government health expenditure. It was probably also related toa growing realisation that oversupply does not lead to redistribution. In addition, when datawere called for, they were often found to be basic, non existent, inconsistent, patchy, or poorlycollected. Hence, these factors combined to move us quickly into a phase which requiredbetter workforce planning.Accordingly, over the past decade, we have moved to a situation where we now have some ofwhat I consider to be the key factors for successful medical workforce planning in place:recognition of its need and the role it can play in informing policy; a national quantitative datacollection on the medical workforce, which is derived from an annual survey which goes outwith each doctor's annual registration renewal; and a peak research body, involving all keyplayers, reporting to government, which aims to bring a national perspective and focus to183medical workforce planning.To help you further appreciate the current Australian situation I thought I should run throughthe results of some of our research and then finish up with some speculation about challengesfor the future.Australian medical workforce benchmarksUpon our establishment in early 1995, one of AMWAC's first tasks was to produce an overallstrategic framework within which its work would be conducted; which would, at the sametime, update a target of 200 doctors per 100,000 population that was adopted as a desirablemedical workforce supply target by Health Ministers in 1992.And, while benchmarks and the use of doctor:population ratios are sometimes criticised, it hasto be recognised that in a strategic sense it is essential that benchmark supply is defined.Without any benchmark, rational workforce supply planning and management is not possible.The need for a strategic goal is also important, given the long lead time before the decision toact and the actual reaction. Without a strategic framework, a system of largely ad hocplanning and abrupt crisis management when absolutely necessary will, by default, become thepreferred policy approach to medical workforce issues.The resultant report - Australian Medical Workforce Benchmarks - published in January 1996,recommended that the 1994 benchmark for supply of medical workforce in Australia be 205FIE clinicians per 100,000 of the population. The report was particularly significant because,for the first time in developing the benchmark, the quantum of work conducted by practitionerswas examined.It is estimated the benchmark will increase to 220 FIE clinicians per 1 00,000 population 'in2005,245 FIE clinicians per 100,000 population in 215, and 270 FIE clinicians per 100,000population in 2025. To arrive at these estimates, it was assumed that the growth in demanddue to population growth and ageing of the population will be 1.17% per yeat· and that growthdue to other demand factors will be 0.6% per year.On the supply side it was assumed that:the number of graduates of Australian medical schools qualified for permanentresidence in Australia will fall from the present 1200 per year to 1000 per year from theyear 2002;average net addition of overseas trained doctors to the Australian medical workforcewill be 200 per year;hours worked by male doctors will gradually fall by an average of five hours per weekby 2025 (female hours worked will remain steady); and184the proportion of total medical practitioners who are female will gradually increase andcomprise an estimated 38.9% of total medical practitioners in 2025.The 1994 benchmark of 205 FIE clinicians also assumes appropriate distribution; andit compares to an actual number of 209 FIE clinicians in 1994.You can see from the accompanying graph (Attachment 2), that if our supply and demandassumptions are correct, they translate into a situation where it is expected that Australia willsee continued growth in supply in excess of benchmark requirements for the next ten years,after which there will be a gradual convergence to around 2019, when supply and demand areprojected to move into balance, and from there into a situation of undersupply.The benchmark was adopted by Australian Health Ministers in July 1996.Perhaps more importantly, however, the report highlighted the degree of maldistribution in theAustralian medical workforce. This is clearly a significant issue for the Australian healthsystem, and Health Ministers agreed, in adopting the benchmark, that effective action to reducemaldistribution must be taken in the same time frame as any supply changes are made, that isover the next twenty years.In summary, the data indicated that there are some 4,356 (2,911 FIE) too many generalpractitioners, principally located in the suburbs of Sydney, Melbourne and Brisbane and some511 (445 FIE) too few general practitioners in rural areas, especially in New South Wales andQueensland (Attachment 3).The evidence also suggests that in areas where there is an oversupply, the average doctor seesfewer patients, receives a smaller average co-payment, and provides more services per patientthan the national average. In areas of undersupply the reverse applies.Other issues highlighted by the disaggregated benchmarking exercise included an absoluteshortage of specialists (estimated at 1,255 FIE), a shortage of non-specialist hospitalpractitioners (estimated at 360 FIE), and a shortage of locums (estimated at 150 FIE).'As well as highlighting oversupply and maldistribution as key issues, the benchmark work alsoshowed there was a need for more information on several key areas of the Australianworkforce, and a number of research projects were initiated. These projects are examining:the impact on the workforce of increasing female participation;the changing characteristics of medical students; andthe requirements for sustainable specialist practice (both urban and rural).Work on the later two projects is still underway so I am unable to report on any results, but we185have completed one of two planned reports on female participation. Some of the key issuesraised in the first AMW AC female report were:It is estimated that the proportion of female medical practitioners will increase from the currentlevel of 27% to 38% in the year 2014 and to 42% in 2025.At present, female clinicians are more likely to be working in major urban centres than maleclinicians: 83% of female primary care practitioners and 93.0% of female specialists wereworking in capital cities and major urban areas, in comparison with 77.5% of male primarycare practitioners and 87.3% of male specialists who were working in these areas.On average, female clinicians work fewer hours per week than males, reflecting the higherproportion of females who work part-time. For females, the proportion working part-time washighest in the 30 to 44 years age group, coinciding with ages when large numbers of womensubstitute family commitments for labour force activity (in 1994,46.8% of female cliniciansworked part-time (defined as less than 40 hours per week) compared with only 15.3% of maleclinicians. 65.5% of female primary care practitioners and 43.7% of female specialists workedpart-time).I should also add that the average hours worked per week by junior hospital doctors,specialists in training and full-time clinicians are roughly similar for both males and females.Lower participation in the labour force and lower average hours worked mean that the averagefemale medical practitioner has a significantly lower lifetime hours worked contribution thandoes the average male practitioner. For clinicians as a whole, the average female contributionis estimated to be 68% of the average male contribution. For general practitioners, thisproportion is estimated at 63%, and for specialists at 75%, although this varies amongspecialties.Despite efforts from the specialist Colleges in recent years, there remains, in the majority ofspecialties, a comparative absence of female practitioners and female trainees. Generally,young female clinicians are more likely to be training to become a general practitioner and lesslikely to be a specialist in training than young male clinicians, although some specialties aremore popular with female practitioners than others. If the current preference for generalpractice continues to predominate, it could be expected to contribute to a continued shortageof specialists. (Specialties with high levels of female participation are psychiatry (21.2%),anaesthesia (11.5%), paediatric medicine (7%) and radiation oncology (6.3%».There is some evidence of a revealed preference from women patients for female clinicians,which is particularly prominent in care for reproductive health. It is a trend that we feel can beexpected to continue, and as a result, there is some concern that those female clinicianspractising may become overwhelmed by the demand for their services. This could be aparticular problem in rural areas, where there is a shortage of female clinicians and sometimes ashortage of women's health services.186Attachment 2AUSTRALIAN MEDiCAL WORKFORCE BENCHMARKSFTEs Actual1. 1994 medical workforceTotal practitioners 40,317 44,337Practising clinicians 37,300 40,8592. Estimated oversupplyPrimary care (GPs & OMPs) 2,911 4,3563. Estimated undersupplyPrimary care (GPs & OMPs) 445 511Hospital non-specialists 360 400Specialists 1,255 1,838Locums 150 200Total undersupply 2,210 2,950Benchmark workforce (= 1 - 2 + 3)Total practitionersPractising clinicians39,61636,59942,93139,453Benchmark workforce per 100,000 population *Total practitionersPractising clinicians222.0205.1240.6221.1Actual 1994 workforce per 100,000 populationTotal practitionersPractising clinicians* calculated on aggregated data226.0209.0248.5229.0187The second stage of our female participation research project is some retrospective qualitativeresearch aimed at establishing the key influences on career choice and workforce participationdecisions; that is, what factors might be responsible for the workforce looking as it does andwhether there are any gender- and age-related similarities and/or differences. This workshould be completed by the end of the year.As well as the macro research projects, AMWAC is also progressively reviewing each of themain specialist workforces. These reviews were initially established because of concerns fromgovernment about shortages in several workforces and the impact these shortages were havingon service delivery. As a result we have developed a methodology for describing thecharacteristics of the workforce, assessing the adequacy of the workforce, and projectingsupply and requirements. If necessary, we then try to provide recommendations based onadjustments to training program intake that should move expected supply more into line withestimated future requirements.To date, AMWAC has completed nine specialist reviews and there are a further six in progress.Our reviews have highlighted specialities that:I. have current shortages (for example anaesthesia);2. are likely to experience imminent shortages if not promptly corrected (forexample urology and car, nose and throat surgery);3. are in a growth phase, because they are relatively new areas of specialisation,but will then taper off significantly (for example emergency medicine);4. are not experiencing much requirement growth but have a large cohort of olderpractitioners moving through to retirement who will need replacing (forexample general surgery and ear, nose and throat surgery, where 40% of theworkforce is aged over 55 years); and5. have CUITent shortages of both practitioners and trainees and appearcomparatively unattractive areas in which to work (for example rehabilitationmedicine and geriatric medicine).The overall trend to date has been a need for more specialists, which is consistent with theconclusions of the benchmark report. It is possible that some of the specialist workforcesunder review at present will not show much, or any, need for growth.188Attachment 3PROJECTED MEDICAL WORKFORCE, 1994-2025:CLINICIANS - FULL TIME EQUIVALENTS6500060000tilc.!!! 55000eco'050000 -....Ql..c~ 45000Z4000035000 --.---------..-..~...----..--.-....,.- .._-- ·,--------------~·-·---~~--·-·--~:-·-·'--'-I I i I ~1994 1996 199820002002200420062008201020122014201620182020 2022 2024YearSupply (recommended scenario 5)Demand (based on benchmark supply according to moderate demand assumption)189Policy responsesSo out of all this work, what has been the response of government?Well, at the macro level, the activity has continued to focus on the supply side, with theCommonwealth government introducing several changes to restrict growth in medical studentnumbers and the entry of foreign doctors into the workforce. Specifically these are:Medical school intake, which is regulated through Commonwealth government funding of. places, is currently restricted to approximately 1200 places per year; with the possibility of thisfailing to 1000 places by 2002;Foreign doctor entry has been tightened through immigration points penalties and changedeligibility arrangements for temporary resident overseas-trained doctors' ability to claimMedicare benefits; andThe continued examination of those foreign doctors who are allowed to enter Australia toensure maintenance of standards and quality.At the same time, the Commonwealth government has also introduced legislation requiring allnew doctors to be vocationally trained to be eligible to claim Medicare benefits.Similarly, at the more micro level, and particularly in response to the AMWAC specialistworkforce reviews, which. as I indicated. have to date generally highlighted the need for moretraining positions in particular specialties, the relevant specialist medical Colleges haveaccredited extra positions and State and Territory governments, who are largely responsible forthe funding of these positions, have ensured that the funding is available. This is of course amore long term solution; and it is the solution favoured by AMWAC.On the issue of geographic maldistribution, all governments, both Commonwealth andState/Territory, have introduced a range of initiatives aimed at encouraging clinicians to gointo rural practice. The focus has been very heavily on incentives, based around the provisionof extra funding, and includes initiatives such as funding for additional rural-based vocationaltraining places, financial incentives to work in rural areas, and support for continuing medicaleducation. Most of the policy activity has also been focused on primary care providers. Inaddition, university medical schools have generally increased the number of rural-basedmedical students they accept. Increasingly, local government has also become involved,offering additional local financial incentives to attract practitioners, such as providing a house,surgery and motor vehicle.190CHALLENGES FOR THE FUTUREI would now like to conclude with some thoughts on what I see as some of the possiblechallenges facing the Australian medical workforce over the next decade.medical workforce planning is not just a numbers gameIncreasingly we are finding that Australia now has comparatively well developed quantitativedata collections, even if they have not been collected for too many years yet; but that we havelittle qualitative data to assist us to make some judgements on why the workforce looks as itdoes and assessments of what factors may be influencing workforce participation decisions. Iknow longitudinal qualitative research has been underway in the United Kingdom for someyears now and I think that in Australia we will have to be looking to collect this type ofinformation also. It is of course more expensive and time consuming to collect, but potentiallyof greater value in understanding the workings of the workforce.the continued impact of increased female participationI have already touched on this subject, but its impact cannot be underestimated. There isnothing wrong with the trend, but we all need to be aware that it will have an impact, both interms of total numbers and in terms of geographic and structural distribution, and increasingly Ithink in terms of work practices. The particular problem for Australia in the short term is thatwhilst the number of female practitioners is increasing, compared to male practitioners, femalepractitioners have a preference for practice in the areas of the workforce that are oversupplied,that is urban general practice.the changing attitudes of junior doctors, especially to their treatment in the workplaceand training requirementsThis appears as another universal issue. As with the United Kingdom, we are noticing thatjunior doctors are increasingly voicing concerns about long work hours, the nature of thetraining process, access to the training program of their choice, and the need for more familyfriendly, and therefore flexible, work arrangements, It is unlikely that all concerns will beresolved to their complete satisfaction, but in Australia they are certainly expressing a desirefor reform in some key areas; and again any reform will have an impact on the workforce.the changing entry requirements to university medical schoolsUntil recently, entrance to medical school in Australia was based almost exclusively on astudent's score in final year school examinations, with the cut off for medicine being essentiallythe top half to one percent of any year. In the past four years, three of the ten Australianmedical schools have moved to a graduate entry program, where selection is based onsatisfactory completion of a first university degree and a medical school entrance examination.Most of the other seven medical schools, whilst maintaining an undergraduate program havealso moved to incorporate interviews into the selection process. Increasingly, medical schools191are also reserving places for students of rural background and Aboriginal background, in thebelief that this will ultimately be of some assistance in alleviating the shortages in these areas ofthe medical workforce.In the longer term it is expected that these changes in admission policies have potential toinfluence doctors' lifetime workforce participation and on the nature of the medical workforceas a whole.rural retentionUniversally this is a problem, although possibly not so much for the United Kingdom. InAustralia, whilst governments have introduced a range of incentives, many of these focus onobtaining rural practitioners and comparatively less on retaining them. In Australia the annualturnover remains high, particularly in the more remote areas where it is at about 30% of theworkforce (in the less remote rural areas annual turnover is around 15% of the workforce).Most of the reasons are well researched and well known to us all. However, it would seemthat the policy responses may still need to be better focused.Aboriginal workforceAssociated with rural retention is the issue of an expanded Aboriginal health workforce andbetter access for Aboriginal Australians to health services. In Australia, the health status ofAboriginals, particularly those living in remote areas, requires urgent and innovativeapproaches to service provision.constant vigil to expand and improve data collectionThis is self explanatory; but in any listing of future challenges the point needs to bemade. Without data you are left groping in the dark.importance of regular reviewingThe medical workforce is dynamic. Again this is a simple statement that is well known to all ofus; but what I think it means in the workforce policy and planning context is that you have tobe constantly updating your profiles and projections of the workforce as new data becomeavailable and new trends are seen to emerge. It is especially relevant if national healthstructures and funding arrangements change, as the United States shows with the recent movesto managed care.But even if change is not that dramatic, we have found there is value in committing yourself toregular review and update. In the Australian context, this has been especially useful in gainingthe profession's support for the notion of workforce planning. In a sense it means you candevelop a direction but no one feels too locked in, and the flexibility this approach affordsseems to provide comfort but at the same time actually assists with implementing reforms.192maintenance of political willIn Australia there is quite a degree of government involvement in the medical workforce. As aresult the success of any work in this area ultimately requires the maintenance of political will;both in recognising the role government has to play, in the importance of data and research tothe process, and in being prepared to act in a co-operative way with the profession.ConclusionsSo, in Australia, I can probably sum up the current situation on medical workforce planningand policy as being that, while we have some large challenges to overcome, we have severalimportant things in place:1 . recognition of the need for good data, and data collections in place that provide reliableand credible information, particularly on the characteristics of the workforce;2. a broad strategic framework in place to guide policy thinking through the developmentof national medical workforce benchmarks;3. a medical profession which is largely onside and part of the workforce planningprocess; and4. political will, with govemment prepared to fund research and implement suggestedactions.The challenge for both governments and the profession in the future is likely to be to continuethis process, to make sure policy action involves everyone in contributing to a solution, andthat the solution is always part of an accepted strategic goal. Ultimately, despite all the hardwork associated with providing some definition of the size of problems and what could becausing them, it may prove that getting some numbers together was the easy part. It is alwaysto be hoped that this is not the case; and I am sure that this sentiment would be viewed withuniversal approval by those of you from Canada, the United States and the United Kingdom.Thank you.193Question and Answer Session after Evening Plenary.What is the source of the graduates from foreign medical schools in Australia?They used to be mostly from the UK, but this source has now dried up. This wasfollowed by intake from India and South Africa but now they are coming in mostly frommainland China. Australia is a huge net gainer of physicians. They have a net gain of 500physicians annually from overseas and 22% of the workforce are graduates from foreignmedical schools, most of them from the UK.What vocational training is available for rural physicians?In the last budget the minister put out dollars for federally funded rural programs.Until the last budget, most rural incentives were local initiatives.Australia has a tight wing government and an oversupply of physicians, so why not simplyreduce fees?There has been no fee increase for general practice and a progressive decrease ofthe floor price over the last few years. Further fee decreases would not be accepted.Background practice costs increase a lot but fees have not increased.The federal minister of health is a doctor. In Australia, is this an advantage or not?He is neutral - and does not see himself as a doctor. Shortly after becoming aminister he fell out with the College and other medical organizations. He is not seen as acaptive of the profession.One of my moles told me Australia was adopting a national billing numbers scheme. Whathas been the effect of that in moving folks to rural areas?It's in place but it will take two years to assess the impact. Most new graduateshave two years in training, after which their only options are rural training posts.Are there a lot of FMOs in training?Most in training are Australian citizens. Most FMOs who are non-citizens don'tstay. The issue can be confused by the guest workers, but by the year 2000 they are notto be regarded as medical practitioners.How prevalent are user charges? Is this seen as creeping privatization?Among urban physicians, 90% charge no co-payments. Approximately 90% ofphysicians' income comes from the floor price and 10% from user charges. In rural areas,only 20% of doctors charge the floor price, the other 80% charge user charges. Thesefigures are for general practitioners and would be different for specialists .. We don't seethis as creeping privatization.194Are you thinking you may want to maintain the oversupply in order to keep outprivatization?Take ophthalmologists - if you take out the top 15% who charge lots in co­payment, the rest earn less than most other specialists because they do not charge abovefloor price.Is provider number limited to a community or does it go with the physician? Is there COUItaction in place regarding billing number restrictions?Provider numbers go with the physicians wherever they move, and they cannot berestricted by area. We can't do geographic provider numbers - it is against theconstitution.How many Aboriginal doctors are trained?Overall there are 200,000 Aboriginal people in Australia and 15 in med schoolnow. We have tried some extra education to bridge the gap between the normaleducation of these people and what they require for medical school. However, medicalschools are far from tribal lands and people are unwilling to go. For example, inQueensland, three places in medical school were saved for Aboriginal people but theynever managed to fill them all in one year and only 1/3'd made it into later years.Is rural training provided for OP's?Now all medical schools must provide rural exposure.195Morning PanelSubstitution196197AN ANALYSIS OF THENON-PHYSICIAN CLINICIANSWHO PROVIDE "PHYSICIAN SERVICES"Richard A. Cooper, M.D.Craig L. DietrichHealth Policy InstituteMedical College of WisconsinMilwaukee, WI 53226Presented at theSecond Trilateral Physician WorJiforce ConferenceVancouver, B. c., CanadaNovember 16,1997198The delivery of health care services in the United States is evolving rapidly, not only interms of its organization and financing but also in the ways that clinical responsibilities arebeing distributed among practitioners in the various health care disciplines. This lattertrend follows a period of almost a century during which physicians held a virtualmonopoly in the delivery of health care services, a monopoly that was created during theearly years of this century by state licensing and regulation, enhanced during the century'smiddle decades by third party reimbursement, and facilitated over the past 20 years by theproduction of physicians in numbers that eclipsed any others who might enter their domainof practice. However, two important dynamics are changing the balance. First, recentchanges in state laws and regulations are enhancing the ability of non-physician clinicians(NPC's) to practice independently within the domain of "physician services." Secondly,the number of NPC's now being trained is outpacing the production of new physicians.The accelerated entry of NPC's into the health care workforce is occurring at a time whenthere is increasing concern about an impending oversupply of physicians relative to thedemand for their services. Indeed, in some communities an oversupply exists already.Although the magnitude that this physician imbalance will attain over the coming years isdebated, it seems clear that, at a minimum, a super-abundance of physicians will exist for adecade or more. It is less clear just how the growing supply of NPC' s capable ofundertaking the independent responsibility for patient care will have further impact on thisphysician imbalance.ELEMENTS OF INDEPENDENCESix elements contribute to the ability of NPC's to practice within the realm of "physicianservices" (Figure 1):First is state licensure, which establishes the light of NPC' s to practice, although it neitherassures their autonomy nor, by itself, delimits the scope of their practices. This right isgranted not only through licensure but also through certification and registration or simply"approval to practice." Recognition in these latter forms acknowledges the education andtraining attained by nonphysicians and allows them to provide care within the bounds oftheir expertise, This generally is outside of the domain of physician services, unlessdelegated by a physician.199The ability to use titles such as "doctor" or "physician" represents the second elementcontributing to the independence of NPC' s. This right, which is defined in the licensingprocess, is seen as advantageous by practitioners in many of the nonphysician disciplines.The third and fourth elements are the regulations that define the scope-of-practice of eachdiscipline and those that determine the degree to which clinicians may exercise theirpractice prerogatives independent from physician supervision. Both are codified withinstate practice acts. Licensing boards are the fifth element. In many states these boardsoversee more than one discipline, but most disciplines seek to have a separate board fortheir discipline. The final element is reimbursement. This includes both the range ofservices that health plans cover and the ability of NPC's to bill health plans for theseservices. All six of these elements have been the object of an increasing degree of stateand federal legislative action over the past few years.THE REGULATED HEALTH CARE DISCIPLINESThe health care disciplines are regulated in 51 jurisdictions (the 50 states and the Districtof Columbia). Thirty-six distinct health care disciplines are regulated in one or more ofthese jurisdictions. Two are the disciplines of medicine (allopathic and osteopathicmedicine), and a third (homeopathy) is limited to other licensed professionals (physicians,naturopaths and chiropractors).Ten disciplines strongly overlap the scope of physician practices and, like physicians,include the ability to take the principal responsibility for the care of patients (Figure 2).Six fall within the category of primary or general care, including three traditionaldisciplines [nurse practitioners (NP's), certified nurse midwives (CNM's) and physicianassistants (PA's)] and three alternative disciplines (chiropractors, acupuncturists andnaturopaths). Four others fall within the realm of specialty care [optometrists, podiatrists,nurse anesthetists and clinical nurse specialists (CNS's)]. It is these ten disciplines that arecollected under the term "non-physician clinician" and that are the subject of furtheranalysis in this paper.In addition to these ten disciplines, a group of five disciplines overlaps, to varying degrees,the work-scope of psychiatrists (Figure 2). Although also non-physician clinicians, theyare not considered in this analysis. The remaining 15 licensed disciplines (Exhibit 3) haveless overlap with the work-scope of physicians, although clear exceptions do exist.200The purpose of this study is to assess the ten general and specialty NPC's disciplines. Itwill examine the current numbers of practitioners in each of these disciplines, analyze theirlicensure and scope-of-practice, measure the size of their training programs, and projecttheir numbers and the contributions that they are likely to make to the health careworkforce in the future.LICENSURE (Figure 4)Traditional NPC's: Separate licensure, certification or registration is afforded to NP'sand CNM's in all jurisdictions except Illinois (which recognizes advanced practice nursesas registered nurses) and to PA's in all but Mississippi. While licensure is most common,almost many of these jurisdictions offer only certification or registration for one or moreof the disciplines.Alternative NPC's: Chiropractors are licensed in all 51 jurisdictions, but acupuncturistsare licensed or otherwise recognized in only 33, and naturopaths are licensed in only 11.However, the number of jurisdictions recognizing these latter two disciplines has increasedfrom 24 and 10, respectively, since 1994. Moreover, legislation that would result in thelicensure of acupuncturists was considered in 10 additional states during the past year, andthe licensure of naturopaths was considered in nine.Specialty NPC's: Optometrists and podiatrists are licensed in all 51 jurisdictions.However, nurse anesthetists are separately recognized in only 44 states and clinical nursespecialists in only 36 states.SEPARATE BOARDS (Figure 4)All of the NPC disciplines have sought boards that are separate from the medical boardand, preferably, are specific for only one discipline (or, in the case of nursing, for thenursing disciplines collectively). The reasons are clear. In the analysis of practiceprerogatives described below, it was observed that the scope-of-practice and autonomy ofeach of the ten NPC disciplines was greater in states that had established separate boardsthan in states that regulated these disciplines through combined boards, and were least instates in which NPC's were regulated by the Board of Medicine.201Traditional NPC's: NP's are governed by either a separate board or by a board specificfor nursing in 48 jurisdictions. In the other three they are governed by a board that,although combined with other disciplines, is not the Board of Medicine. CNM's aregoverned by separate boards in 40 states and by the Board of Medicine in two states. Theconverse arrangement exists for PA's. The Board of Medicine governs PA's in 38 states,and separate boards for PA' s exist in only six states.Alternative NPC's: Chiropractors are governed by separate boards in 48 jurisdictions,but in two of the other three they are governed by the Board of Medicine. The other twoalternative disciplines are less frequently governed by separate boards .. Acupuncturistsare governed by the Board of Medicine in 13 of the 33 states that recognize theirdiscipline and by combined boards in 9 states, but by separate boards devoted toacupuncture in only 11 states. Naturopaths are govemed by boards that combinedisciplines other than medicine in 6 states and by separate boards for naturopathy in 5statesSpecialty NPC's: Optometrists have a long tradition of independence, and in all but twostates the boards governing then are specific to optometry. Although the majority ofboards that regulate podiatrists are also separate, the Board of Medicine regulatespodiatry in 25% of the states. Nurse anesthetists and clinical nurse specialists aregoverned principally by separate boards of nursing, but in 20% of states they are regulatedby combined nursing and medicine boards.TITLESThe titles "doctor" and "physician" connote positions of stature in health care. Althoughnurses and PA' s have not sought their use, practitioners in many other disciplines havecoveted these titles. Most states permit chiropractors to use titles such as Doctor ofChiropractic (DC) or Chiropractic Physician (CP), but five states restrict their title to"chiropractor." Similarly, all states that license naturopaths consider them to be physiciansand permit them to use the titles "Doctor of Naturopathic Medicine" (ND) or"Naturopathic Physician" (NP). However, among the 33 states licensing practitioners ofacupuncture and herbal medicine, only three permit practitioners to use the title of"doctor," and most of the others require the title of "acupuncturist." Moreover, a numberof states specifically prohibit the use of "doctor" by acupuncturists unless they possess adoctorate level degree.202In general, states do not designate the titles of "doctor" or "physician" for optometrists orpodiatrists, although one Arkansas does allow the title "optometric physician," and sixstates allow "podiatric physician," podiatric surgeon" or "doctor of podiatric medicine."Nonetheless, it is common practice for clinicians in all three of these fields to use the titleof "doctor" in daily practice. Thus, although referred to as NPC's in this analysis,practitioners in many of the NPC disciplines are, in fact, "physicians," and the term NPChas been used to distinguish them from the larger group of MD and DO physicians.TRAINING PROGRAMSThere has been an explosive growth in the number and/or size of the programs trainingNPC's in many of the disciplines, particularly those associated with primary care.Traditional NPC's: The numbers of training programs for traditional NPC's haveincreased substantially over the past few years. NP programs grew from 88 in 1990 to132 in 1995, a 50% increment, and additional programs continue to open. CNMprograms more than doubled over the past seven years, from 19 in 1990 to 50 in 1997;and PA programs increased by more than 50%, from 48 in 1990 to 76 in 1997.The number of graduates from these traditional NPC programs has increased to an evengreater extent (Figure 5). Between 1992 and 1997, NP graduates more than tripled, from2,348 to 8,972; and further increases in the number of NP graduates can be anticipatedbased on the enormous increase in enrollment in recent years (Figure 6). CNM and PAgraduates increased, as well. Graduating CNM's grew almost four-fold between 1992 and1997, from 110 to 430. DUling this same period, there was a doubling in the number ofgraduating PA's, from 1,362 to 2,802. Collectively, the number of graduates in thesethree disciplines grew from just under 4,000 in 1992 to more than 12,000 in 1997.Alternative NPC's: Between 1990 and 1997, there was a 50% growth in the number ofschools of acupuncture and herbal medicine (from 22 to 33), while the number ofchiropractic colleges remained constant (at 16), and two new naturopathy schools wereadded (one last year), bringing the total to four. The number of graduates of schools ofacupuncture and naturopathy increased in proportion to the increasing numbers trainingprograms (Figure 7). However, despite the lack of a change in the number of chiropractic203colleges, the number of chiropractic graduates more than doubled, from 1,700 in 1992 to4,100 in 1997.SpecialtyNPC's: In contrast to the traditional and alternative disciplines, the number oftraining programs for specialty NPC's has remained relatively constant since 1990,although a new optometry school opened in 1997. Nonetheless, the number of graduateshas had an upward bias (Figure 8). Graduating optometrists increased by 5% (from 1,170in 1992 to 1,235 in 1997), and psychologists by 10% (from 2,081 in 1992 to 2,300 in1997). The number of graduating podiatrists increased by 25% (from 513 in 1992 to 643in 1997). Although the number of nurse anesthetist graduates more than doubled between1992 and 1996 (from 160 to 387), this number declined slightly to 369 in 1997, and acontinued decline is anticipated.CURRENT AND PROJECTED NUMBERS OF PRACTITIONERSReasonably accurate estimates can be made of the number of practitioners in most NPCdisciplines. However, estimating the number of NP's presents certain challenges. The1996 National Sample Survey of Registered Nurses indicated that there were 71,000nurses in 1995 who had formal preparation as NP's. Data from both the Sample Surveyand the National Council of State Boards of Nursing indicated that approximately 53,300of them had national and/or state certification but that only 77% of those who werecertified (40,100) were working in positions with the title of NP. The current andprojected size of the NP workforce has been estimated based on this smaller number ofcertified NP's who are presumed to be working as NP's. These figures for NP's do notinclude the 4,100 CNS's and 6,800 psychiatric CNS's in 1995 who serve different rolesand who are accounted for separately.Workforce estimates for all of the NPC disciplines other than NP's include the totalnumber of trained and active practitioners. Projections of the future number ofpractitioners began with base year 1995 data. These data were adjusted over time forboth the entry of new graduates into the profession and the loss of practitioners throughdeath and retirement, based on the age cohorts of graduates and practitioners. It wasassumed that, in all NPC professions other than nurse anesthetists (whose enrollment isdeclining), the numbers of graduates would continue to increase annually until 2000-2002,after which they would plateau.204Traditional NPC's: In recent years, the numbers of both PA's and CNM's increasedsubstantially (Figure 9). However the greatest increase was among NP's, whose numbershave grown from 21,500 in 1990 to 41,000 in 1995. NP's are projected to increasefurther to 133,000 in 2005 and 230,000 in 2015.The number ofPA's has steadily increased from 1l,000 in 1980 to 19,000 in 1990 and to29,000 in 1997. The number of PA's is projected to almost double again to 54,500 by2005, and, if high levels of enrollment continue, PA's will increase to 83,000 in 2015.CNM's also have grown in numbers, doubling from 3,000 in 1990 to 6,000 in 1997, andthey are projected to almost double again to 10,500 in 2005 and continue to increase to17,000 by 2015.Collectively, these three groups of traditional NPC's will grow from 1995 levels of73,000to more almost 200,000 in 2005 and more than 300,000 in 2015. This increment in thenumber of traditional NPC's is also large when expressed in per capita terms (Figure 10).Considering PA's, CNM's and only those NP's who are likely to be employed as NP's,there will be an increase from 321100,000 of population in 1995 to 701100,000 in 2005and, if current levels of production continue, to more than 100/100,000 in 2015.Alternative NPC's: Chiropractic is the most established of the alternative disciplines.Although there has been no change in the number of chiropractic colleges since the 1980s,the number of chiropractors has increased from 49,500 in 1990 to 56,400 in 1997 (Figure9). Because of recent increases in enrollment, the pace of growth in the number ofchiropractors is expected to increase, and the number of chiropractors is projected togrow to 97,000 in 2005 and, if the current high levels of enrollment continue, to 146,000in 2015.In 1990, there were 5,300 acupuncturists (excluding other professionals who useacupuncture), and this number almost doubled to 9,800 in 1997. However, the pace ofgrowth is increasing, and the number of acupuncturists will more than double again to22,000 by 2005. If the current pace of training continues, acupuncturists will double againto more than 42,000 by 2015.There are many fewer naturopaths, and many of those in practice received their trainingnot from the current four-year colleges but from more limited programs that existed in thepast. In 1990 there were 800 naturopaths, and this number grew to 1,300 in 1997.205Growth in the number of naturopaths is projected to accelerate, and the total number ofpractitioners is projected to grow to almost 4,000 by 2005 and more than 7,000 by 2015.The growth of the numbers of practitioners in these three alternative disciplines is alsolarge when expressed in per capita terms (Figure 10). Collectively, they will grow from241100,000 in 1995 to 401100,000 in 2005 and to 58/100,000 in 2015.Specialty NPC's: In 1990 there were approximately 26,000 optometrists, 12,500podiatrists, 21,000 nurse anesthetists and 8,000 CNS's. Unlike the general caredisciplines discussed above, there has been relatively little growth in the number ofspecialty NPC's (Figure 11). Between 1990 and 1997 the number of optometrists grewby 15%, nurse anesthetists grew by only 8%, and the number of podiatrists remainedrelatively constant. However, the number of CNS's actually declined, from 8,000 in 1990to 6,600 in 1995, coincident with a sharp increase in NP's (although the number ofpsychiatric CNS's grew significantly during that same period of time from 3,200 in 1990to 6,800 in 1997). When expressed in per capita terms, the numbers of optometrists per100,000 of population grew only modestly between 1990 and 1995, while the number ofnurse anesthetists remained relatively constant, and the numbers of podiatrists and CNS'sdecreased (Figure 12).As in the recent past, continued slow growth is projected in these specialty disciplines overthe coming decades. Podiatrists are projected tc increase from 14,300 in 1995 to 19,400in 2005 and to 24,000 in 2015 (Figure 11), but in per capita terms this represents anincrease from only 5/100,000 in 1995 to 6/100,000 in 2005 and 7/100,000 in 2015(Figure 12). Optometrists also will continue to increase in numbers, from 29,000 in 1995to 38,000 in 2005 and to 45,000 in 2015 (Figure 11), but, like podiatrists, these changesare quite small in per capita terms (111100,000 in 1995; 13/100,000 in 2005; and14/100,000 in 2015) (Figure 12). CNS's are projected to grow from 6,600 (2.5/100,000)in 1995 to 20,000 (71100,000) in 2005 and to 27,000 (8/100,000) in 2015. However, thenumber of nurse anesthetists is actually projected to decrease from 22,800 in 1995 to20,200 in 2005 and to only 15,400 in 2015. In per capita terms, nurse anesthetists willdecrease from 9/100,000 in 1995 to 71100,000 in 2005 and 5/100,000 in 2015.Data Modifications: To create an integrated workforce model that considers physiciansand NPC's, a number of modifications must be made to the data presented above.2061 Patient care practitioners: With the exception of NP's, the data presented aboveinclude all practitioners in each discipline who are trained and qualified for practice,whether or not they actually are practicing. A determination must me made of thepercentage of the workforce actually engaged in practice. This would exclude clinicianswho work at other professional tasks (e.g., teaching, research, supervision andadministration).2. Work effort: Not all practitioners work full-time, and workforce estimatesmust be adjusted to reflect full time equivalent (FTE) practitioners. In studies ofphysicians, it has been observed that, over the course of their careers, women physicianswork fewer hours than men, older physicians have fewer patient contacts than youngerphysicians, and salaried physicians work less than self-employed physicians. It is uncertainwhether these patterns also apply to NPC's. However, many of the NPC disciplines areskewed to women (Figure 13). For example, women account for more than 90% ofNP's, CNM's and nurse anesthetists and approximately 60% ofPA's, acupuncturists andnaturopaths. In addition, while the average ages of graduates of optometry, podiatry andchiropractic schools are in their late 20s, other NPC disciplines attract older students, and,therefore, the age spectrum of practitioners is skewed to older individuals (Figure 13).Moreover, a high percentage of NPC's work as employees of hospitals or clinics.3. Roles of practitioners: Practitioners in certain of the NPC disciplines are ableto move between roles in general care and specialty care, and their impact on workforceplanning depends not only on their total numbers but also on the specialty choices theymake. For example, most PA' s are trained as primary care providers, but approximatelyhalf serve in specialty roles. NP's have similar opportunities. Similarly, NPC's may shiftthe range of services that they provide. For example, chiropractors may limit theirpractices to manipulation for musculoskeletal disorders or expand their practices toencompass general primary care, and optometrists may limit their services to fitting lensesor expand them to therapeutic optometry.4. Overlap with physician services: Many of the services provided by NPC'ssubstitute for the services that physicians provide, whereas other services (e.g.,acupuncture and manipulation) may supplement rather than displace physician services.Similarly, only some of the roles played by NP's and PA's are within the arena of"physician services," whereas other roles are of a nursing or adjunctive character. Muchmore work is needed to understand the tasks and services provided by NPC's and to207appreciate the extent to which they displace the need for physicians or are supplementaryto the services normally provided by physicians.SCOPE·OF·PRACTICE and AUTONOMYThe scope-of-practice of licensed professionals is defined in a number of ways.First, in legal terms, scope-of-practice is defined by the states through formallegislation and regulation.Second, since regulations often are broad and ambiguous, scope-of-practice isdefined informally by the latitude of practice that is accepted by the involved stateregulatory boards and/or attorneys general.Third, scope-of-practice in practical terms is defined by insurers who often limitthe range of covered services to fewer than are permitted by law.Fourth, scope-of-practice is defined by practice organizations, such as clinics,group practices and managed care organization, that determine both the range ofpractices that NPC's will be permitted to undertake and the degree to whichpractice authority that is vested in physicians will be delegated to them.Finally, scope-of-practice is defined by practitioners themselves, whocommonly limit their activities to the range of practices they feel most competentto provide.For physicians, the regulated scope-of-practice spans the range of medicine, as taught inmedical school. However, scope-of-practice for other professions is more narrowlydefined. The characteristics of practice that are generally regulated are those that have thepotential of doing harm, either physical or emotional. They include:1) Making and communicating a diagnosis;2) Performing, ordering or interpreting diagnostic tests;3) Entering an orifice or interior space of the body;4) Setting or manipulating a bone or joint;5) Administering a substance by injection or inhalation;6) Prescribing a drug or device; and2087) Applying energy.Traditional NPC's and Alternative NPC's: Figure 14 summarizes the scope-of-practiceof the six NPC disciplines that provide general care. Most states permit the practitionersin each of these disciplines to perform a physical examination and make a diagnosisthroughout the range of disease and dysfunction that falls within each practitioner's area oftraining and expertise. Practitioners in all disciplines other than acupuncture also havebroad latitude in performing and interpreting laboratory tests and X-rays. Midwives andnaturopaths are permitted to care for normal pregnancies, participate in normal deliveriesand provide non-pregnant gynecologic care in almost all states, and to care forcomplicated pregnancies in many. In addition, suturing and minor invasive procedures arepermitted for NP's, CNM's and PA's in most states and for naturopaths in many.However, there is a considerable degree of variation in most of these practice prerogativesamong the states.The degree to which NPC's may practice independently from physician supervision ordelegation also varies considerably (Figure 15). NP's have independent practice authorityin almost half of the states. In most of the other states, authority may be delegated toNP's by physicians. However, the delegating physician need not be on site and directphysician involvement can often be at widely spaced intervals. In only two states is thepresence of a physician required for NP' s. Midwives have a similar degree ofindependence. In contrast PA's generally practice under physician direction, although, aswith NP's, the "direction" may be intermittent and distant. Two states allow PA's topractice independently, but through the use of approved protocols. Recent changes inMedicare, discussed below, have greatly expanded the independent settings in whichMedicare will reimburse NP's, PA's, and CNS's and greatly reduced the amount ofphysician supervision required. This is certain to have an important effect on theautonomy of traditional NPC's.In general, the practices of altemative NPC's are independent of physician supervision ordelegation. However, in 8 of the 33 states that license acupuncture, a referral from aphysician is required before patients may undergo therapy, and three states require aphysician on site.Specialty NPC's: Optometrists have long had privileges that include eye examination,diagnosis, prescribing lenses and other vision devices and removing superficial foreign209bodies from the eye. All states grant these privileges. Newer privileges that only somestates grant include treatment of glaucoma (41 states), laser surgery (three states). Inmany states, special certification is required to practice "therapeutic optometry," and theextension of this right to other states is a focus of much of the recent legislation dealingwith optometry.The range of practice of podiatrists also is expanding. Historically, their privileges have,centered on the diagnosis and treatment of diseases and disorders of the foot (below theankle), including medical treatment, fitting prosthetic devices and surgical treatment usinglocal anesthetics. Other therapeutic privileges include mechanical treatment in 46 states,manipulative treatment in 29 states and electrical treatment in 28. However, the range ofpractice of podiatrists is increasing and now extends to areas above the ankle but belowthe knee in 29 states, to upper muscles of the leg in nine states, and to the hand in fourstates. In addition, ten states allow podiatrists to amputate toes.Nurse anesthetists have included in their scope-of-practice the range of prerogativesnecessary to treat pain and perform anesthesia in a manner similar to physicians, and inmany smaller communities they are the sole practitioners capable of administeringanesthesia. The scope-of-practice ofCNS's is defined by their training and supervision.PRESCRIPTIVE AUTHORITY (Figure 16)Traditional NPC's: As with scope-of-practice, prescriptive authority varies widelyamong disciplines and among states. Most states permit NP's, CNM's and PA's toindependently prescribe non-controlled drugs and, to varying degrees, controlled drugs.However, a high proportion of states stipulate that this prescriptive authority must bedelegated by physicians. In practice, this delegation often is with little of no directphysician involvement.Alternative NPC's: Naturopathic physicians have the independent ability to prescribenon-controlled drugs in 2/3 of the states in which they are licensed, and they have theadded authority to prescribe dietary and herbal remedies in all states. These latterprerogatives also are available to chiropractors in almost all states and to acupuncturists inmore than half of the states in which they are licensed. However, the prescriptiveauthotity of chiropractors and acupuncturists is limited to these natural products, and210neither discipline has the authority to prescribe either controlled or non-controlledsubstances.Specialty NPC's: Optometrists have prescriptive privileges in all states, but there isvariation in the breadth of that authority among the states. In 19 states they are limited totopical medications, but in 32 states optometrists may prescribe oral drugs, as well, and in24 states they are permitted to prescribe controlled substances, although there arelimitations in some states concerning the range of schedules that they may use. Inaddition, four states permit optometrists to administer drugs by injection.Podiatrists have independent prescriptive authority in all of the states, although in severalstates this authority is not codified in the state's practice act. The prescriptive authority ofnurse anesthetists and CNS's follows the pattern ofNP's, but more states limitprescriptive privileges for these disciplines than for NP's (Figure 16)REIMBURSEMENTMedicare and MedicaidTraditional NPC's: Until the late 1970s, the reimbursement of NP's, CNM's and PA'sby Medicare and Medicaid was governed by the "incident to" provision, which allowedthose who were employed by physicians to be reimbursed by means of payments to theemployer. In 1977, the Rural Health Clinics Act extended Medicare and Medicaidreimbursement to NP's and CNM's working in freestanding, physician-directed ruralclinics located in health professions shortage areas (HPSA's). This subsequently wasexpanded to include other practice settings (e.g., homebound, nursing facilities, ruralpractices outside of HPSA's and rural practices within HPSA's but not in "rural clinics."In addition, periodic on-site physician supervision was not required, if such a practice waspermitted by the state.Recently passed federal legislation has further expanded Medicare reimbursement to NP's,PA's and CNS's to include all "physician services," regardless of the geographic area inwhich these services are provided and without any requirement for physician direction.Payment to NP's and CNS's will be direct, but payment for PA services will continue tobe to the employer (Figure 17).211The rate of Medicare "incident to" reimbursement for NP' s has been at 100% of physicianrates, and PA reimbursement has varied from 65% to 100% of physician fees, dependingon the site of service. The recent Medicare legislation allows for the reimbursement ofNP's and PA's in all settings (except hospitals) at a uniform rate of 85% of physician fees,with a 10% bonus in HPSAs.The rates of Medicaid reimbursement for traditional NPC' s vary widely among the states.Like other insurers, states have the option to cover only some of the services that areprovided by NPC's, and, therefore, state Medicaid plans may cover fewer services thanare permitted in the state's practice act. Similarly, Medicaid policy regarding on-sitesupervision by physicians may be more restrictive than the state's practice act requires.The federal mandate for Medicaid allows Family and Pediatric NP's and CNM's to billdirectly to the limits established by the state, if they are allowed to practice independentlyin that state. Many states also reimburse NP's other than Family/Pediatric, and somereimburse CNS's. In some states, reimbursement for CNM's is restricted to pregnancycare, but in others it covers any services within the CNM's scope-of-practice (AppendixI). Medicaid reimbursement for PA's in most states is to the employer, although Montanaand South Dakota allow PA's to bill Medicaid directly. Some states have uniqueidentifiers and separate fee schedules for PA's.Alternative NPC's: Chiropractors are allowed to bill Medicare, and new billing codesapplicable to chiropractic have facilitated this process (Figure 17). In 26 states, they alsoare allowed to bill Medicaid. However, neither Medicare nor Medicaid currently coversacupuncture or naturopathy.Specialty NPC: Optometrists and podiatrists are allowed to bill Medicare and Medicaiddirectly in a manner similar to physicians. Nurse anesthetists and CNS's also arepermitted to bill Medicare, CHAMPUS and other federal programs directly, and they areallowed to bill Medicaid directly in 36 states, although billing for both is commonlythrough the employer.State Mandates Affecting Private InsurersIn many states, legislation has mandated that health insurers reimburse NPC' s for services.covered by their plans (Figure 17). State mandates have directed private health plans toreimburse NP's for covered services in 27 states, CNM's in 37 states and PA's in three212states. Among the alternative NPC's, mandated coverage of chiropractic services byprivate health plans exists in 44 states. In addition, all states mandate the coverage ofchiropractic services by Workman's Compensation. Coverage for acupuncture ismandated in nine states and for naturopathy in two. State mandates also affect thespecialty NPC, with mandated coverage for optometrists