UBC Faculty Research and Publications

Squaring the circle : reconciling fee-for-service with global expenditure control Evans, Robert G., 1942- Aug 31, 1988

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

Item Metadata


52383-Evans_RG_Squaring_the_circle.pdf [ 1.2MB ]
JSON: 52383-1.0048511.json
JSON-LD: 52383-1.0048511-ld.json
RDF/XML (Pretty): 52383-1.0048511-rdf.xml
RDF/JSON: 52383-1.0048511-rdf.json
Turtle: 52383-1.0048511-turtle.txt
N-Triples: 52383-1.0048511-rdf-ntriples.txt
Original Record: 52383-1.0048511-source.json
Full Text

Full Text

SQUARING THE CIRCLE:RECONCILING FEE-FOR-SERVICEWITH GLOBAL EXPENDITURE CONTROLHPRU 88:8Health Policy Research UnitDivision of Health Services Research and DevelopmentOffice of the Coordinator of Health SciencesThe John F. McCreary Health Sciences CentreThe University of British ColumbiaVancouver, B.C.V6T lZ6R.G. EvansAugust 1988AcknowledgementsAdvice and encouragement is gratefully acknowledged from M.L. Barer,M.D. Low, the B.C. Medical Manpower Advisory Board, and the alumni of theUBC Health Services Planning Program. The author's research is assistedby a Research Scientist Award from the National Health Research andDevelopment Program.B~r~aining over Pllvsicians' Fees\\l1en Canada' s public health insurance programs were extended in the1960s to cover physicians' services, they adopted the process ofreimbursement which had been developed by their private, physician­sponsored predecessors, the not-for-profit members of Trans-Canada MedicalPlans (Shillington, 1972). Each provincial program pays physicians fortheir services according to a uniform fee schedule specific to thatprovince. Under the public plans, howe ve r , the schedule is negotiatedbe twe en an agency of the province ruin i s t ry of health or commission - andthe provincial medical association. In earlier years the medicalassociations simply promulgated their "Schedules of Minimum Fees 'lindependently.As a general principle, the medical associations set the relativefees of items in t be s chedu I.e . Negotiations focus primarily on theoverall percentage increase in total outlays, the weighted averageincrease in fees. which is implicit in the pattern of increases inindividual fee items. In the simplest description, medical associationsdetermine the relative values of the different items, and bargain over theconversion factor for translating relative value units into dollars.This description, however, neglects a critical aspect of the process.The overall increase in payments to physicians implicit in a fee schedulechange is inevitably an estimate, based on the application of the newschedule to past data on the numbers of different fee items billed.Actual provincial outlays I whi ch represent the gross receipts of the fee-practice medical profession, are the product of both the level of fees andthe rate of billings or utilization of services during the period ofapplication of the schedule.Thus even with a binding fee schedule, a provincial government'sfinancial liability remains open~ended, An estimate of outlays will beprepared as part of the government budgeting process; but unlike theallocation of funding for hospitals or for the other public healthactivities of government, the total payments to be made to physicians are,strictly speaking, beyond the control of provincial governments. Thebills must simply be pRid as they come in. In this limited sense,expenditures are indeed Ilout of control ".Total outlays are not, hcweve r , '-.711011y arbitrary or unpredictable"They are determined, in an accounting sense, by the numbers of activepractitioners in the province, and the activity level or volume ofbillings of each. The latter, in turn, depends on the time and effortwhi ch each physician puts into his/her practice, But it also depends onthe specific charRcteristics of the fee schedule, and the way in whichthese interact with the patient flow and practice styles of practitionersto yield an implicit rate of I'billings per practitioner-hour ",Depending upon its structure and associated rules for payment, a feeschedule may provide greater or lesser opportunities for practitioners toexpand their gross receipts by changes in practice style or billingbehaviour. 1 Thus, while fee-for~service reimbursement represents, inprinciple, an open~ended commitment of public funds, the degree of tl open -endedness H is itself influenced by the negotiation process. In general,provincial governments have attempted to ensure that fee schedules werestructured so as to minimize the opportunities for practitioners to expandtheir billings without a corresponding increase in time and effort.The success of these attempts has been variable. Over most of thehistory of the public plans the average billings per physician reimbursedhave escalated at about one to two percent per year faster than averagefees. But there have been dramatic examples of ll ut i l i z at i on creep 11through procedural relabelling and multiplication, most notably in Quebecbetween 1971 and 1975 (Contandriopoulos, 1986; Gabel and Rice, 1985).Physicians there were responding to a freeze on nominal fees during aperiod of rapid general inflation. Average gross receipts per physicianrose 17 percent over this period, or 4 percent per year (Barer, Evans andLabelle, 1988).2Until recently, however, most provincial governments have been able1 Schedules which permit "il la carte" billing for minor diagnosticand therapeutic procedures performed in the course of an office visit, forexample, allow physicians to increase their incomes by performing suchprocedures more frequently. Schedules which draw fine distinctionsbetween different types of visits or procedures provide an opportunity forphysicians to reclassify their services into higher-paying categories.2 'The increase in total expenditures on physicians' services inQuebec, divided by the reported number of "active civilian physicians" inQuebec, rose 30.5 percent over this period, or 6.9 percent per year. Thenumber of physicians reported as being reimbursed by the RAMQ rose muchmore rapidly than the total Quebec physician stock.to contain "utilization creepll within acceptable limits through the feenesotiation process. The open-ended nature of fee-for-servicereimbursement has been a significant, but not an explosive, factor inoverall cost escalation. But that process is a dynamic one, an on-goinggame rather than a once-for-all settlement. Changes in medicaltechnology, in physician attitudes and practice styles, or in the generalpolitical climate, all have the potential to upset the balance.In recent years, provincial governments have become increasinglyconcerned about the ever~present potential for large and unforeseen costoverruns in one of their largest budgetary items. Furthermore, theprobability of such overruns may be perversely related to the generalfiscal situation. ~~en provincial budgets are tight, negotiations withphysicians are tougher. But a very hard-nosed bargain is more likely toresult in an unexpected bounce in utilization, as physicians look harderfor ways to exploit the billing opportunities in a given schedule. Asprovincial governments give a higher priority to cost containment inhealth care, the open-ended nature of fee-far-service is becoming moreproblematic.Even if large I!utilization break-outs!! do not occur, provincialgovernments are still exposed to the cumulative effects of annualincreases in billings per physician which regularly exceed negotiated feeincreases. Over a period of years, and on a very large expenditure item,relatively small slippages compound to large sums.And finally, in addition to increases in fees and in (fee-adjusted)billings per physician, provincial outlays are driven upward by theincreasing supply of physicians, While physicians argue over whethertheir increased numbers actually cause increases in utilization, andparticularly bitterly over whether such increases may represent"unnecessary servicing"! two facts are indisputable.First, the supply of physicians is rising faster than the population(whether or not adjusted for changing age structure), by one to twopercent per year, and is projected to do so for the indefinite future(Barer, Gafni and Lomas, 1988), And second, the average volume ofutilization per physician - fee-adjusted billings - is continuing to rise,There is no sign of a saturated market; utilization and costs of medicalservices are rising in line with the numbers of physicians (Lomas, Barerand Stoddart, 1985; Barer and Evans, 1986; Lomas and Barer, 1986),Bargaining over Medical UtilizationAccordingly, provincial governments are becoming increasinglyconcerned about the factors underlying the utilization of physicians'services, and the prospects for influencing or managing them (Anderson andLomas, 1988; Rachlis and Fooks, 1988), But they are also trying to limitthe open-endedness of the reimbursement system, Quebec adopted thisstrategy over a decade ago, in response to the experiences of the early1970s, negotiating agreements limiting payments at both the aggregate andthe individual physician level (Contandriopoulos, 1986), B,C, begannegotiating caps or "givebacks" in 1982 (Barer et a1., 1987; Barer, Evansand Labelle, 1988).In effect, provincial governments are shifting towards bargainingover total outlays, rather than just over fee level increases. As notedabove, in earlier years bargaining over fee schedule structures and rulesof payment was always carried out with a close eye on implicit (expected)total provincial outlays (or, from the physicians' point of view, implicitaverage gross incomes). But the global constraints are becoming moreexplicit and binding.The logical limit of such a process is contemporaneous prorating.Under prorating, gross outlays for a province and time period arenegotiated in advance as a fixed amount. If total claims for servicesprovided during that period differ from the predetermined amount, the rateof reimbursement per claim is scaled up or down so as to equate totalclaims and agreed outlays. In effect l the conversion factor fortranslating billing dollars into reimbursement dollars becomes a floatingvalue, and the fee schedule is explicitly a relative value scale. Thisprocess is rather different from what is currently going on in B.C. andQuebec, but the actual outcome is becoming increasingly sirnilar. 33 There is a difference in principle. Under current arrangements, ifutilization increases are not large enough to raise total outlays to the"c ap ll amount, no adjustment is made in favour of physicians. Theprorating scheme is symmetrical - physicians will receive the total amountbargained for, even if they bill for less. So long as the "caps" arebinding, however, this distinction is without practical significance.1Physician reimbursement systems in the Canadian provinces thus appearto be moving toward an indirect form of prorating. This paper does notargue for or against that development, or attempt to describe in detailthe present state of bargaining. Rather, it takes the trend as given, andassumes that the underlying forces are likely to persist. In thiscontext, we suggest a more explicit and sophisticated form of proratingwhich may mitigate some of the inequities inherent in simpler schemes,direct or indirect. It also embodies some incentives for modifications inpatterns of medical practice which may be considered desirable on othergrounds, and provides practitioners, not individually but in local groups,with more scope for control over their own practice patterns.The Basic Structure of a Regional Prorating SystemUnder simple prorating, the negotiations between provincial agencyand medical association yield an explicit value, $B, for budgetary outlayson physicians' services during a given year. They would also establish afee schedule to govern the item~specific reimbursement process. In thecourse of the year, actual claims submitted would total $G. The ratio RBIG is then the scaling factor to be applied to all fees, to determineactual amounts paid. In practice, of course, payments are being madethroughout the year, so the scheme would require on-going monitoring ofoutlays against targets, and some form of end-of-year adjustment, up ordown, in reimbursements per practitioner.The potential problems with this approach are fairly well known. Itmay be acceptable as a temporary measure I but if applied over a period ofyears it places great pressure on all practitioners to change towardprocedure~intensive styles of practice. The reimbursement system is aprocess of "beggar-my~neighbourl1, or at least professional colleague, andthe practitioner who tries to avoid exploiting to the full the billingopportunities in a given fee schedule loses out economically to those whoare less inhibited. As this lesson is learned l total claims will escalateand the conversion ratio will fall; eventually the whole scheme may losecredibility.Unfortunately, it seems logically unavoidable that §flY scheme to tryto reconcile fee~for-service reimbursement and overall cost caps must havethis beggar-my-neighbour aspect, although the process of negotiation andapplication may, as at present, hide the fact for a time from some of theparticipants, The proposal outlined here shares this characteristic I butit provides a fe,v more degrees of freedom in the process.The first change from global prorating is regionalization. Budgetarytargets are set for each major region in the province. But they are notbargained separately, nor are they linked to the physician numbers in eachregion. Rather the overall negotiated amount for the province, $B, isallocated according to the population in each region, taking account oftheir age and sex mix. Thus regions are, in effect, capitated, and(subject to age and sex adjustment) the same amount of reimbursement isallocated on behalf of each resident of the province, regardless of wherehe or she lives." This allocation then reimburses all the servicesprovided to regional residents, by physicians located anywhere in theprovince (or out of it).The population~based allocation, however, represents a significantpolicy decision. One could, instead, choose to allocate shares of theprovincial total to regions on the basis of the numbers and billings ofphysicians currently located there. For example, one could designate eachphysician's office address, and then allocate the total of $B for thecurrent year in the proportions in which total outlays were split amongregions last year, using the office location of the physician reimbursedas the indicator of the region to which the funds were sent.The resident-based approach implies a judgement that the purpose ofthe financial allocations is to reimburse services on behalf of residents,rather than to provide support to particular physicians. It treats allprovincial residents equally, regardless of where they are located.The derivation of regional allocations of funds proceeds by firstcalculating age-sex specific relative rates of utilization of medical4 There is no reason in principle why age and sex should be the onlyfactors taken into account in determining the relative per capitaallocations. If other descriptive data on regional populations areavailable which show a clear-cut correlation with utilization ofphysicians' services, such other factors could easily be built into therelative values, without changing the total provincial allocation $B. (Itwould not, however, be appropriate to include the region-specificphysician to population ratio as one such variable! The data must bedescriptive of the population, not the health care system.)services. The total provincial population is partitioned into age and sexgroups, which might be ages 0-1, 1-4, 5-9, and so on by five year blocksup to 100.' This partition yields 42 mutually exclusive and collectivelyexhaustive age-sex categories.For each of these categories, we can compute a relative rate ofutilization of (or at least expenditure on) medical services. This isequal to the average per capita expenditure on physicians' services forall persons in the province in that age-sex group, divided by the averageper capita expenditure for the whole population. These relative rates arewell below one for males aged 10-14, for example, and well above one forfemales (or males) aged 70-74. They can be derived from the most recentavailable billing data, and can be updated annually.6The province is divided into regions, like the present regionalhospital districts in B.C. Each region has, not only a differentpopulation, but a different age-sex distribution of population, such thateach will have a different average per capita need for, or at leastutilization of, services, associated with that age-sex mix.5 A large number of categories make the allocation of funds quitesensitive to variations in utilization across age groups. Blocks such as0-5, or 65 and over, hide a good deal of variation in utilization, as say,between 65 year olds and 90 year olds (Barer and Wong Fung, 1987). Theycan therefore yield unfair results across regions. On the other hand, toofine a grid may result in instabilities if the numbers in a particular agecategory are quite small.6 The relative rates would presumably be computed for a period oftime long enough to average out use patterns, and would be based onexpenditures derived from a constant fee schedule.IIUsing the age-sex specific per capita utilization rates, however, wecan define a weighted population for each region which will reflect th~sedifferences in needs, The numbers of people in each age-sex group in aparticular region are simply multiplied by the corresponding relative rateof utilization of services, and then the products are totalled across age-sex groups, This yields a synthetic regional population which will beabove or below the actual, according to whether the region has apopulation of above or below average users of services. The sum ofsynthetic regional populations aggregates to the actual provincialpopulation,The negotiated budgetary total for the province, $B, is then splitamong the different regions in proportion to their synthetic populations,in equal amounts per weighted capita. This automatically gives relativelymore to regions with elderly populations, and less to regions with youngerpopulations, in exact proportion to their expected differences in medicalservices utilization. 7 If we refer to a representative region as regionj, its budgetary allocation will be $Bj ,The algebra is straightforward. The total provincial population, P,is subdivided into 42 age-sex groups. We can let Pi be the population ina representative age-sex group i, where i runs from 1 to 42. Let E betotal expenditure on physicians' services in the most recent available7 These "expected" differences, however, are established by applyingthe provincial pattern of utilization by age-sex class to each region.There is no allowance for regional variations in this pattern.period, and e i the expenditure on services for the ith age-sex group.Then the weight applied to persons in the ith group is wi ~ (ei/Pi)/(E/P).If then Pij is the number of people in region j who fall into the ith age-sex group, the synthetic population of region j is ZiWi*Pij qj' Thismay be greater or less than ZiPij Pj' the actual population of region j,but the sum of the qj across all regions equals the sum of the Pj.' i , e.total population P. Then the budgetary allocation for region j is simply(qjP)*B ~ $Bj.These $Bj of reimbursement are then available to pay the claimssubmitted by physicians for services rendered to the residents of regionj. we can designate as $Cj the total of all claims submitted, on behalfof residents of region j, by all the physicians in B.C. (or out of it).Note that Cj is not equal to the claims submitted by phvsicians residentin region j, although there will be considerable overlap, becausephysicians in region j may also provide services to residents of regionsk, 1, etc., and at the same time residents of region j may receiveservices from physicians in regions k, 1, etc.The total of claims submitted for the care of residents of region jwill not in general equal the budgetary allocation for region j. Theirratio will be the prorating factor for region j; that is, Rj - Bj/Cj. Feeclaims submitted for services rendered to residents of region j arereimbursed at Rj times the negotiated rate in the fee schedule; they carrya premium or a discount according to whether Rj is greater or less thanone. Looked at from the ~ractitioner's point of view, services providedto residents of (relatively) 11under s e r v i c e d l l areas will automatically bereimbursed more highly than those provided to residents of (relatively)lloverserviced!1 areas - no matter where the service was provided or.thepractitioner works or resides.Of course, a practitioner choosing to practice i~ a relativelyunder serviced area can expect that services provided to most of his/herpatients will carry a premium. Conversely, if s/he locates in arelatively overserviced area, fee claims will be more likely to bediscounted. But it is the rate of servicing received by the regionalresidents which is decisive, not the number of physicians per capitareported as available or the services which they provide.It may be viewed as a bit peculiar, that a physician seeing differentpatients for the sa~le problem may receive different fees, depending on thepatient's residence. But proposals to pay differential fees according towhere the physician is located also amount to paying different fees forthe same service. This approach merely extends the concept of regionalfee differentials to reflect the fact that what is at issue is not,fundamentally, where physicians locate, but whether or not patients areserved.Under this approach, the incentives for physicians to relocate inunderserviced regions are automatic, and require no explicit adjustment ofrelative fees, or l1 po l i c i ng l1 of where the physician is actually located.The physician may locate wherever slhe likes, but the provincial servicingpatterns will determine how much slhe is paid (per service). From theprovincial government's perspective, the "ove r suppLy " of physicians in' aparticular region is no longer a financial problem, because an increase inphysician numbers does not add to overall expenditures on behalf ofresidents of that region. The argument over whether increased numbers ofphysicians lead to increased rates of servicing then becomes moot. Ifthey do, the Rj value in the region of physician increase simply falls. 8The location of the physician is not, however, a matter ofindifference. In remote regions there may be a need to provide I1stand~by"capacity for emergencies. It may be appropriate to reimburse physicianssimply for being available, over and above the actual services theyprovide. But this should and can be provided separately from the feeschedule itself, through modification of existing lIallowancetl programs.Sec~ndly, people in lt uncte r - ctoc t or e d11 areas may not, in fact, receivefewer services than people in comparable areas - they are not l1 unde r -served" (Horne, 1987) - but they may have to go to more trouble andexpense to reach care, The premium for services supplied to persons in"uride r v do c t.o r e d " regions may not be sufficient to induce "enough" (by someexternal criterion) physicians to locate there - additional inducements8 The provincial government, being thus protected from the financialconsequences of increases in physician supply, might no longer wish tolimit the issuance of geographically unrestricted billing numbers (Barer,1988). The practitioner community might also look at that issuedifferently. Both government and practitioners should still ho"ever beconcerned, on medical grounds} if particular populations are grossly over­or under-served. But these concerns would no longer be entangled withfinancial issues.may be required. But this approach has the virtue of an automaticincentive - it can always be further adjusted.Finally! reimbursement of emergency services may require specialtreatment in a province like B.C. where a number of people have residencesin (well-serviced) urban areas but travel to remote (and often dangerous)job sites. It would obviously be inequitable for the physician working inthe remote interior, who provides emergency care to a logger or fisherman,to have his/her fee discounted because the patient's address of record isVancouver or Victoria. 9There is also a possible source of friction in the referralrelationships between physicians in smaller communities, and in urban andmetropolitan centres. If a general practitioner in a small co~munity inregion j refers a patient to Vancouver, and that patient receivesextensive and expensive physician care, the claims for that care becomepart of the claims against the reimbursement allocation for region j - asindeed they should. But the process may be interpreted as lt out - of - t own 11specialists tapping the pool of funds available for reimbursement of localphysicians - as indeed they are. The billing activities of themetropolitan specialists, ceteris paribus, reduce the value of Rj for the9 The problem may be broader than emergency care, for persons who areaway from home for long periods of time. Their normal source of care maynot coincide with their official residence. It seems unlikely, however,that care for this generally healthy sub-population will generatesignificant volumes of billings. If it does turn out to be problematic,the appropriate response seems to be to adjust the definition ofresidence.region j from wh i ch the patient was referred.Prorating by Region and CLass of ServiceThe scheme can be modified to limit this degree of financial conflictof interest, or at least make it a controllable parameter, in thefollo"ing "ay:The total budgetary allocation for physicians' services for theprovince, $B, can be divided up among different types of physicians'services. These might, at a minimum, be general practitioners I services,and services of medical, surgical, and diagnostic specialists. Then:B BG + BM + BS + BDwhere BG is the budget allocation for general practitioners' services, andso on. The distribution of $B among these components is part of thenegotiation process, although historical patterns "ould probably exercisea dominant influence. At this point, again, there is an important policydecision. Does one define these allocations in terms of particular typesof services, i.e. fee schedule items, or in terms of payments toparticular types of physicians? Again, consistent "ith the philosophythat the basic purpose of the health insurance program is to pay forservices, not to provide incomes for physicians, we shall assume that thecomponents of B are allocated by type of service, rather than by type ofphysician providing the service.12We defined 8 i l above, as the average per capita expenditure ofpersons in the ith age-sex group in the provincial population. We can nowpartition ei into eg i , emi , 85 i , and ed i • the average per capitaexpenditures of people in the ith age-sex group on general practiceservices, medical specialty services, etc. As above, these are used toderive weights for the age-sex components of the population, which in turnare combined with regional population patterns to yield region-specificand type-of-service specific allocations of reimbursement, BG j l BH j l BS j land BD j . ' OThe claims submitted on behalf of residents of region j would then besimilarly categorized into groups CG j , CM" CS j , and CD j . The matching oftotal claims by category against total reimbursements would yield a set ofratios RG j , RM j , RS j , and RD j , which would be the premiums or discounts tobe applied to fees for services of the corresponding types to residents ofregion j .This would have the effect of insulating the receipts ofpractitioners of one type from the level of activity of those of another,within or across regions, except insofar as practitioners cross specialtyboundaries. The general practitioner in a remote region who becomes a defacto general surgeon or anesthetist, because those services are notreadily available to the population s/he serves, will also find that s/he10 The synthetic populations in each region will be different foreach of the types of services considered, 'insofar as the age-sex patternsof utilization differ across types of services. But this poses noconceptual or computational difficulties.receives a premium fee for such services, because relatively few suchclaims are being submitted on behalf of that regional population.On the other hand, the medical sub-specialist in an area in which thesupply of general practitioners is rising rapidly, and services per capitalikewise, will not find his/her reimbursement rate lowered in consequence.The value of RG j will fall, as the volume of general practitioner billingsrises, but not that of RM j .Nor will a concentration of specialists in urban and metropolitancentres necessarily lead to their fees being discounted, insofar as theyprovide referral services to patients from a number of regions in theprovince. The medical sub-specialist above, if s/he receives referrals ofcomplex clinical problems from allover the province, is effectivelydr awi ng on the reimbursement allocations in every region. On the otherhand, the specialist who does not have a true referral practice, but isinstead competing with general practitioners in a relatively over-suppliedmetropolitan area, will find that services to most or all of his/herpatients are reimbursed from a. Upot ll that is stretched quite thin, and thefees are correspondingly discounted.To the extent that residents in metropolitan regions receive moreservices from specialists than do residents in the rest of the province,the specialists who provide those extra services will be reimbursed atlower rates. The specialist who moves to a community where such serviceswere previously unavailable, however, or more likely just hard to get to,should do rather well on both volume of work, and fees.Creating Controllable Financial Interdependence Among SpecialtiesThis "multiple pot" approach implies complete separation ofreimbursement rates for different specialties, or at least different typesof services. The global allocation of $Bj for all services provided toresidents of region j, on the other hand, implies complete integration.Every physician's activity, of whatever type, lowers the potentialreimbursement of every other. An intermediate approach, with a"controllable '! parameter, links the different specialty ITpotS!' orallocations for each region. Such linkage recognizes explicitly the"gate-keeper" role of the physician, and particularly the generalpractitioner, in shaping the pattern of medical care.To illustrate this intermediate approach, suppose that in region jthe allocation for diagnostic services BD j is insufficient to cover allthe claims for diagnostic services for the population of region j, CD j .Instead of adjusting the value of RD j , the ratio of reimbursements toclaims, so as to equate total payments to total allocations, part of thedifference is made up by drawing on the allocations for the otherservices. As an example, when claims exceed the budget allocation, lethalf of the difference be paid from the budget allocations forreimbursement of the other services.This will still leave total claims greater than the amount allocated,and so diagnostic services will still be reimbursed at a discount from thefee schedule. But this discount, RD'j' will be only half as great as ifno funds had been withdrawn from the allocations for other services: RD'j- (BDj + CD j)/2CD" as compared with RDj - BDj/CDj.The corresponding amount, (CD, - BDj)/2,is subtracted from thereimbursement allocations, BG j , BM j , and BSj , available for otherservices. The distribution process could be simple, lowering each of thebudgetary allocations by the same percentage, or sophisticated, trying toallocate responsibility for different diagnostic tests to differentspecialties.The basic idea, however, remains simple. To the extent that theactual billings for diagnostic services over- or under~run the amountsbudgeted for their reimbursement, some proportion (in this example 50%) ofthe discrepancy, positive or negative, is shifted to the reimbursement forother clinical services (and thereby for other practitioners who orderedthe tests). If practitioners in region j make less use of tests (percapita, age-sex adjusted) than those in region k, then (adjusted forpopulation) CDj < CDk . But BDJ and BDk will not be affected. Ceterisparibus, then, RG j > RGk , and the same for medical and surgical.specialists. There is a direct monetary incentive for more restrainedtest ordering behaviour.Of course there is an even stronger qirect incentive when all theclaims and reimbursements in a given region are pooled in one pot, sinceall the over- and under-runs in each service area have a direct impact onthe total Rj for the region. But the relationship is much less apparentfor the practitioners of a particular type. Furthermore, the "linked pot"approach permits the size of the sharing ratio to be varied to "fine-tune"the incentive.Practitioners may feel that 50 percent represents too much of anincentive, and would lead to skimping on appropriate tests. Perhaps thesimple existence of a link is sufficient to attract attention; perhaps 25percent sharing would be sufficient. The relationship is obviously opento further discussion. Indeed, it can be made part of the negotiationprocess; and it may be that physicians in different regions would wish tochoose different sharing ratios.Financially, the provincial government should be indifferent to theratios chosen, although like physicians themselves, the provincialgovernment must be concerned if the financial incentives threaten to leadto inappropriate patterns of practice. In this respect, however, provinceand physicians have a shared objective, at least in principle. Thecapping of the overall outlays at $B removes from consideration thosefinancial questions over which payer and payee must inevitably disagree.The practitioners in each community, however, will have a direct andidentifiable financial stake in the patterns of practice in thatcommunity. Insofar as they economize on the use of other resources inproviding care, their own gross receipts are increased. And conversely.The principle can be extended. The over- or under-runs in thespecialty service allocations can likewise be split in half, or in someother proportion, and part allocated back to the general practitionerpool. Then the total amount allocated over the claims for generalpractitioner services would be the amount originally allocated on behalfof residents of the region, BG;, plus one half of any under-runs, or lessone half of any over-runs, in the specialty service pools, after adjustingfor the allocation of over- or under-runs in the diagnostic service pool.To be specific, the adjusted reimbursement ratios for the differentservice areas are computed as follows:(1) RD' j [ BD; O.5(BD; - CD;> l/CD;(2) RJ·1' j IBM; + m(BD; CD;) + CM;)12CM;(3) RS'j [ BSj + s (BD; CD;) + CM,)/2CS;(4) RG' j [BG; + g(BD; CD;> + O.5(BM; - CM; + m(BD; - CD;> +BS; - cs; + s (BD, - CD;» J/CG jwhere as before, the RG'j' etc., are the ratios by which the fees for eachservice category are marked up or down; the BGj , etc. I are the totalallocations of reimbursement by service type and region before adjustmentfor over~ and under-runs in other service categories; the CG j , etc. aretotal claims for reimbursement, by region and service type, valued at thefee schedule currently in effect before prorating; and the m, s, g, arethe proportions of the diagnostic services under- or over-run which areallocated to each of the other three service categories - medicalspecialties, surgical specialties, and general practice, respectively.Since in these equations under~runs and over-runs in each categoryare assumed to be divided in half, with half borne within the category andhalf spread to another category, it follows that:(5) m + s + g ~ 1/2But this need not be so. One could define the sharing ratiosdifferently. As one moves away from one half, the algebra gets a littlemore involved. But it would be a straight-forward exercise to set upequations (1) to (5) in terms of general parameters for the sharingratios.To recapitulate, the allocation described in these equations firstcompares total billings for diagnostic services provided to residents ofregion j with the total amount budgeted for such services. Half of anyover- (under-) run is then debited (credited) to the budgetary allocationsfor medical, surgical, and general practice services in the proportions m,s, and g (which may be set ad lib but must sum to one half). Thereimbursement rate for diagnostic services provided to residents of regionj is then marked down (up) so as to equate the total claims to theoriginal budgetary allocation augmented (reduced) by the amounts charged(credited) to the other service categories. The effect of the transferwill be to cut in half the the size of the mark-up or mark-down on feesfor diagnostic services.The budgetary allocations for each of the classes of specialtyservices, reduced or augmented by their share of the Qver- or under~run indiagnostic services, are then compared with the total claims for theseservices in the region. Once again, half of any over~ (under-) runs aredebited (credited) to another allocation, in this case to the allocationfor general practitioner services. The fees paid for medical or surgicalspecialty services for residents of region j are then marked down (up) byratios &"'1 ' j and RS' .i ' sufficient to equate total claims of each type tothe budgetary allocations for each, reduced or augmented by amountscharged or credited to either diagnostic or general practitioner services,The amounts transferred to or from diagnostic services, however,depend on the over- or under-runs in that budget, while the amountstransferred to or from general practitioner services are determined by theover- or under-runs in the specialty sectors themselves.Finally, the allocation for general practitioners, reduced oraugmented by the amounts transferred to or from each of the other threeservice budgets, is compared with the total claims for generalpractitioner services, to determine the value of RG'j' the amount by whichfees for general practitioner services are to be marked up or down fromthe overall fee schedule in region j .The "residual status" of general practitioner services in thefinancial computation corresponds to a view of the general pr~ctitioner asthe basic gate-keeper. in the system. Billings for other types of services"depend ultimately on the referral decisions of the general practitioner.Accordingly some proportion of the financial implications of each of thosedecisions, for good or ill, comes back to the original gate-keeper.But the general practitioner is not the only gate-keeper; s/he sharesthis role with the specialists who also order diagnostic tests andinitiate and/or carry out diagnostic and therapeutic manoeuvres. Thepattern of transfers of funds among the different service allocations isintended to reflect the performance of ~his role.Accordingly the m, s, and g values which determine the impact ofdiagnostic services over- or under-runs on the different referringpractitioners could be adjusted in a more detailed and test- (orprocedure-) specific manner to reflect the extent to which the ordering ofcertain types of tests is linked with particular practice specialties. Ifsub-sets of tests can be identified which are clearly ordered almostexclusively by providers of medical specialty services, for example, over­or under-runs for those tests should be charged or credited back to thespecific allocation for those services. This extra element ofsophistication might become particularly important if, as discussed below,allocations for hospital care are also linked to the practitionerallocations.One could further strengthen the financial incentives associated withthis "gate-keeper" role, by changing slightly the way in which the short­falls or savings in diagnostic services are flowed through topractitioners. The process is expressed in the following modifications toequations (1) to (4).(lA) RD' j-[BD; a.S(BD; - CD)]/CDJ(2A) RH' . - [ BM; + CM; + 2m(BD; CD) )/2CM;J(3A) RS' ;-[BS; + CM; + 2s (BD; CD)] /2CS;(4A) RG';-[BG; + g (BD; - CD; ) + a.S( (BM; - CM) + (BS; -CS)l]/CG;These equations embody a different treatment of the over- and under~runs in the diagnostic services budget. It is still the case that a shareof these discrepancies (for the sake of example, one half) is transferredto the other service allocations, and divided among them in theproportions m, s, and g. But amounts transferred from (to) the diagnosticservices allocation ·are no longer treated equivalently with the initialallocations BM; and BS;. They are not pooled in with those allocationsfor the purpose of computing the specialty service over- (under-) runs tobe shared with the general practice allocation. This implies that suchdiscrepancies are not flowed through to the general practice servicesbudget, but have' a dollar for dollar impact on the payments to medical andsurgical specialists.Looked at another way, any over- or underrun in the reimbursement forspecialty services is shared half and half with providers of generalpractitioners' services (although of course the sharing ratio need not bechosen at that level). This is intended to reflect the role of thegeneral practitioner in making referrals. But sums allocated to or fromthe diagnostic services budget are not shared, they affect the payments tospecialists in their entirety. Hence the consequences of test-orderingbehaviour by specialists (as expressed in the ratios of the parameters mand s) fall directly on specialists.Example: The Allocation Scheme in Hvpothetical OperationFor concreteness, we provide a numerical example of how thereimbursement program would operate. (In this example we use the sharingpattern embodied in equations (1) to (4), without the variant justoutlined above.) Consider a region within B.C. which has a population of100,000 people. If the total allocation to reimburse physicians in B.C.through the Medical Services Plan were, say, $900 million, for a totalpopulation of three million, this would imply a per capita allocation of$300. Let us assume that the population of this region has the same agedistribution as the provincial population, so the total amount availableto reimburse services provided to this group of people is $30 million.Let us suppose that, on the basis of last year's billing patterns andnegotiations, 40 percent of this is allocated for reimbursement of generalpractitioners' services, 20 percent for diagnostic services, and 20percent for each of medical and surgical specialists' services, or:BG j - $12 million, BMj - $6 million, BS j - $6 million, and BD j - $6million.Now suppose the claims for medical services provided to thispopulation - by physicians anywhere in B.C. - come in at $7 million fordiagnostic services (CD j ) , $5 million for surgical services (CSj ) , $7million for medical specialists' services (CM j ) , and $11 million forgeneral practitioners' services (CG j ) (defined in all cases by the type ofservice provided, not the specialty of the provider). In total, thisregion is right on its allocation, and under a simple prorating system,all claims for services to its residents would be paid at 100 cents on thedollar.But under this proposal, the amounts actually paid will reflect thediscrepancies between the budgeted amounts, and the claims made. Onaverage, the total of claims made will be fully reimbursed, but some willcarry premiums, and others discounts. The million dollar over~run indiagnostic services will be cut in half, and only half borne by thesuppliers of diagnostic services. Their claims will be prorated by(6. 5/7) ~ .92857, and reimbursed at 92.857 cents for 'each dollar - adiscount of 7.1 percent in response to an overrun of total claims by 16.7percent. The remaining $500,000 of over-run in diagnostic services willbe allocated equally (for simplicity) among the other three practitionergroups, $166,667 being withdrawn from each."11 This implicitly assigns providers of each service group equalresponsibility for the over-run. One might argue for equiproportionateassignment, which given the budgetary amounts in the example would haveled to a deduction of $250,000 from the GP services and $125,000 from eachof the specialty groups. The numbers in the example would have beenchanged, with higher rates of reimbursement for each of the specialtyservices, and lower for the GP services.The medical and surgical specialists' services would each be leftwith an allocation of $5,833,333, to cover claims of $7 million and $5million respectively. But again, the discrepancies would be cut in half,and half of each transferred to the general practice allocation. Thesurgical specialties would have a surplus of $833,333, of which $416,667would be credited to GP services. The remainder would support a premiumreimbursement of $1.08333 reimbursement per dollar of claims. From themedical specialties, half of the short-fall of $1,166,667, or $583,333,would be debited to the allocation for general practice services, and theremaining shortfall would result in their claims being discounted to91.667 cents on the dollar.The budgetary allocation for general practice services would bereduced by $166,667 as their share of the diagnostic services over-run,then increased by $416,667 as half of the surgical services under-run,then reduced by $583,333 as a share of the medical services over-run, fora net adjustment (reduction) of $333,333. This would still, however,leave them with an excess of reimbursement allocation over claims;$11,666,667 compared with $11 million, so they would be reimbursed at$1.06061 per dollar of claims. (The extent of rounding applied in actualpractice will depend on the precision of the computers.)The outcome of the process is summarized as follows:Type of Budgetary Claims Amount ProratingService Allocation Received Paid FactorCeneralPractice: $12 million $11 mn. $11,666,667 1. 06061MedicalSpecialty: $6 million $7 mn. $6,416,667 .91667SurgicalSpec ialty: $6 million $5 mn. $5,416,667 1. 08333DiagnosticProcedure: $6 million $7 mn. $6,500,000 .92857$30 million $30 mn. $30,000,001 1.0 (avg. )The full amount of the negotiated budget is paid out, and types ofservices which exceed (fall short of) their initial allocation. areprovided with additional resources. But they are not fully reimbursed fortheir over·runs, or l1penalizedlT for their under-runs, as would happen inan across·the-board prorating system - or for that matter, in this case,by a simple reimbursement of all claims. Restraint i~ rewarded, and over-runs are discounted, whether it be in billings for ones' Olqn services, orin referrals to other specialists or for diagnostic services. And ofcourse the flexibility exists for regional groups of practitioners tonegotiate higher or lower sharing ratios.Extensions and AnalogiesNor need the process stop at this point. Since the categories G, M,S, and D are defined as types of services, not as types of practitioners,there is no reason why particular sub-sets of services, such asobstetrical services, should not be isolated for different treatment orgiven their own "pots ll • The finer the differentiation of specificservices, however, the more important it will become to adjust capitationrates across regions to take account of characteristics other than age andsex which may affect needs for and utilization of medical services.Furthermore, the interdependencies among different classes of services maymake the appropriate sharing formulae for over- and under-runs morecomplex. But there is no reason why the system cannot be fine-tuned fromyear to year while it is in operation.A substantial extension which is quite within the general spirit ofthe approach, would be to include expenditures on hospital services as anadditional category. In terms of the algebra above, one would introducebudgetary allocations and claims $BH and $CH, and again link over- andunder-runs to the amounts allocated for reimbursement of different typesof physicians' services. This linkage would create incentives similar tothose in capitated plans in the United States, in which changes inpatterns of hospital use have direct consequences for the reimbursement ofphysicians.The process would be an order of magnitude more complex for hospitalservices in Canada, ho\vever, because present accounting and budgetingsystems do not permit the attribution of expenditures to the care ofparticular patients. Thus one can readily establish whether utilizationrates for a particular regional populatipn, measured in age-sex adjusteddays or separations per capita, are above or below the averages for theprovince as a whole. But one can only very crudely translate these intocorresponding variations in per capita expenditures. Per diem costs byhospital are notoriously inadequate for this purpose.Nevertheless, one could go part way by comparing regional age-sexadjusted hospital utilization rates with the provincial averages. Suchrelative utilization rates could then be linked to the budgetaryallocations for physicians' services, augmenting these allocations forresidents of regions with low hospital utilization rates and conversely.In fact the Ontario Ministry of Health has done something similar foryears in reimbursing Health Service Organizations, providing them with anadditional budgetary allocation based on estimates of their '[saved '1hospital patient days.The translation of hospital utilization patterns into regionalallocations for physician reimbursement would open up some verycontentious areas for negotiation, over both the relative costs ofdifferent institutions, and the regional patterns of 'Ineed'l insofar asthey may not be reflected in age-sex distributions. Nor is such linkagenecessary, if the objective is simply to reconcile fee-for-servicereimbursement of physicians with predictability and capping of overalloutlays on physicians' services.But if one looks forward to a broader objective, that of providingphysicians with opportunities to benefit from finding ways to manage thecare of patients more efficiently, rather than simply more intensively, asat present, then at some point the integration between the physician andthe hospital must be addressed explicitly. It is, after all, a standardfinding that the principal economies in the American capitation-basedHealth Maintenance Organizations have in the past corne from more carefuluse of hospitals and hospital-based services (Luft, 1981; Manning et a1.,1984). To date, Canadian physicians have had no way to participate, atleast not directly, in any savings which their behaviour might generate inthe hospital sector.It is this latter objective, of generating incentives for providers toadopt more cost-effective styles of care and for patients to seek out suchproviders, which motivates the current policy of promoting capitated carein the United States. Under the U.S. Medicare system for the elderlypopulation, Health Maintenance Organizations (HMOs) which undertake toprovide (or contract for) "all necessary care" for a defined group ofpeople are reimbursed at 95 percent of the Adjusted Average Per Capita Cost(AAPCC), an estimate of what the cost of providing physician and hospitalservices to that group would have been, in that local area, under fee-far­service reimbursement (Anderson et a1., 1986).The AAPCC is adjusted for the age, sex, welfare and institutionalstatus of the covered group, as well as for historical patterns of costdifference in the local county area. (The system is applied across theentire U.S.) A principal difference between the U.S. system and theproposal advanced here, however, is that patients self-select into and outof HMOs. It is observed that past utilization patterns are a powerfulpredictor of future costs, for individual patients I resulting in strongincentives for HMOs to seek out previous low users. Further, it appearsthat more healthy patients tend to choose HMOs. In consequence, settingcapitation rates at 95 percent of regional AAPCC may actually increaseoverall costs, and in a dynamically unstable way (Ellis and McGuire, 1987;Muldoon and Stoddart, 1987). This has led to increasing interest inadjustment of the AAPCC for the past utilization pattern of the enrolledgroup.But the counter-argument is equally clear. Any lI c api t a t i on lt systemwhich adjusts the reimbursement rate to present or past utilization rates,generates incentives for increased servicing which undermine the wholepoint of capitation as a method of encouraging more conservative styles ofpractice. The dilemma seems unavoidable in eapitated systems with self­selected membership. An alternative would adjust the AAPCC, but only forpast "non-discretionary" utilization (Anderson et a1., 1986), if thiscould be defined unambiguously.A more radical suggestion is a geographically-based capitation systemsOMcalled Hearrier at risk ll - in which insurance carriers would bid ongeographically defined populations. The successful bidder would undertaketo pay all the (covered) expenses for that population, in return for aglobal payment from the Medicare administration, and would then undertaketo manage the care process within that region (U.S. Congress, Office ofTechnology Assessment, 1986; U.S. Congress, Congressional Budget Office,1986).Under our proposal, the medical community of the province becomes the"carrier at risk ll . But for the province as a whole, that risk is notfinancial, because the total allocation $B, predetermined by negotiation,is the payment both to the "ca r r i e r " and to the providers. Itsdistribution among regions is also determined; the only financial riskrelates to the distribution of those totals among different practitioners.Unlike the U.S. proposals, however, the local groups of practitioners atrisk also have predominant control over styles of practice, and aprofessional responsibility for quality of care.The U.S. geographic carriers, as purely financial agencies, would beinvolved in a cost-minimizing struggle with the practitioners in theirregions. They would find themselves in some form of continuousnegotiation with providers - rather like small scale provincialgovernments, but without the bargaining leverage which comes with coveringthe whole population (and controlling the legislature!).The principal "risk" in the present proposal, apart from possibleperverse patterns of intra-profession behaviour, arises if the regionalallocations of reimbursement do not adequately reflect the "needs" of thepopulations covered. If some regions have populations which are "sicker"than others, the practitioners who serve those populations will beinadequately reimbursed relative to their colleagues in healthier areas.They will have to work harder, for less. Should the regional allocationstherefore respond to factors other than age and sex?This is essentially the problem addressed by the Resource AllocationWorking Party (RAWP) process of regional allocation in the U.K. NationalHealth Service (Carr-Hill, 1987, n.d. [1988]), which likewise attempts todefine region-specific allocations of resources on the basis of thel'needs" of residents of each region. Unlike the U.S. discussion ofadjustments to the AAPCC, the RAWP allocations have in generaldeliberately avoided linking resources to historical patterns ofutilization and expenditure, for reasons which Carr-Hill spells outclearly.At the regional level, and without the problems of patient andcarrier selection which bedevil the U.S. approach to capitation, historicpatterns of use and cost reflect the location and activities of providersrather than. the needs of patients· highly unequal patterns which wereexactly what the RAWP program was intended to correct. From this point ofview - which is implicitly accepted in proposals like that of Anderson eta1. (1986) to relate the AAPCC only to past "non-discretionary"utilization - even the current adjustment of the American AAPCC forrelative per-beneficiary cost patterns in the local county isinappropriate. 12On the other hand, if the facilities and personnel are centralized -in London or Vancouver - and the resources are attached to the individualpersons wherever they may be, it is essential that the reimbursementsystem provide for transfer of resources from one region to anotherwhenever people cross regional boundaries to obtain services. Under theproposal in this paper, a transfer occurs automatically. In the U.K.,proposals to create "internal markets" within the National Health Serviceaddress this sort of implicit or explicit inter-regional contracting forservices (Enthoven, 1985; Institute of Health Services Management, 1988).But in the absence of a fee-for-service system, the structuring of suchl'internal market~ll is a more complex task.The RAI,P allocation formulae in the U.K. allocate resources by regionon the basis of age, sex, and socio-economic status, recognizing thegenerally accepted fact that poorer people have greater health needs.(The American AAPCC does likewise, in that it includes welfare status asan adjustment factor.) Unfortunately that general agreement does notextend either to how socia-economic status should be measured, for thepurposes of this allocation, or, equally important, how resource12 It is a fortiori inappropriate to adjust for the institutionalstatus of the covered population, since on a geographic basisinstitutionalization rates are so closely associated with the availabilityof facilities. But to fail to do so in a selected population would be toinvite disaster, because of the very large differentials in expectedhealth care costs between institutionalized persons and those living inthe community.distributions should be adjusted in response to whatever differences aremeasured (Carr-Hill, n.d. [1988J). In the circumstances, it seems prud~ntto leave this as an area for futur~ research. 13Indeed, a significant positive feature of this proposal may beprecisely that it could create a demand for better information on thehealth status and needs of people in different regions, as part of theprocess of regional competition for funds. It has frequently been notedthat Canadians spend a great deal on health care, but devote even lessattention than most other countries either to measuring its impact> or tostudying the underlying needs of the population. Under this reimbursementsystem, a region (or its practitioners) who felt that they were beingunder-reimbursed, would have a strong incentive to try to develop reliabledata on population health status which could be inserted into thefinancial formula. 14Some Practical ConsiderationsOne need not, however, solve all the problems at once. The proposaloutlined here is relatively straight-forward conceptually. Its principal13 It would not be difficult, however, to calculate differences inhealth care utilization and costs for (status) native populations, andadjust regional allocations for these. If there were significant regionalvariations in identified welfare populations, this too could be includedin the formula; but one would not expect this factor to be very importantin a geographically pooled population.14 Carr-HilI's (1987) very useful review of the need for and problemswith SES adjustments is a response to just such a request from a region inthe NHS.disadvantages, at least in the short run, seem to be the requirement forsubstantially more data on the populations served by provincial medicalinsurance systems, and the potential difficulty of communicating itsessential features to the practitioner and the general population. Theshift in reimbursement focus from practitioners to beneficiaries requiresa complete and up to date computer file of beneficiary data, includingage, sex, address, and any other information to be used in dete~miningpopulation weightings for regional reimbursement allocation. This is notnow available in all provinces, but is spreading.Understanding and acceptance by the practitioner community may bemore problematic, but will depend on their recognition that regional andspecialty prorating actually provides more incentives and opportunitiesfor practitioners collectively to manage the pattern of medical careprovision. By separating the determination of total provincial outlaysfrom the negotiation of fee schedule structure, this system reduces therange of issues in conflict between reirnburser and practitioners, andexpands the areas of possible common interest and co-operation.The main question is whether the simultaneous maintenance of fee-for-service reimbursement for physicians, and global capping of expenditures,is in fact a plausible and acceptable objective. 'S If it is, and15 A more complex system has been in operation for over a decade inthe Federal Republic of Germany, involving independent prorating by eachof the 'numerous sickness funds (Brenner, 1988). The prorating mechanismapplies only to ambulatory care; physicians' services to hospitalinpatients, and drugs, are not included. In this system, physicians earndifferent fees for the same service according to the sickness fund apatient happens to be enrolled with, which in turn depends on occupation,particularly if we are in the process of moving to it in a back-handedmanner anyway, the allocation process outlined above seems to have certaindesirable features.i) From the provincial government's point of view, it has the majoradvantage of making outlays predictable in advance, and eliminatingthe combination of utilization slippage, and contentious bargainingover l1 g i v e -b ac ks ll which nO\\I occurs periodicallly in B. C.ii) The regional, resident-based system mimics the behaviour of aI1 marke t l1 in that relative fees automatically rise (fall) inrelatively under~ (over-) serviced areas. Practitioners can choosetheir locations freely, but accept the consequences for their feesand incomes.iii) Allocation of reimbursement by type of service provides a similarautomatic fee adjustment, encouraging the more equal distribution ofspecialty services.iv) Of particular importance, groups of physicians in each region have astrong incentive to examine and manage their own patterns ofpractice. Physicians serving patients from regions with relativelyhigh rates of referral for diagnostic work or specialty consultationemployer, or area of residence. Funds with I1 s i cke r " enrollee populations,however, must collect a higher proportion of the payrolls of their coveredpopulation, and/or discount the fees paid on their behalf. The equity ofsuch an approach is obscure.41(or hospital use) may wish to consider whether these patterns areappropriate; in this consideration the provincial government will beable to co-operate rather than acting as an adversary.v) Last, but by no means least, the linkage of regional financialallocations to (estimates of) beneficiary needs, rather thanpractitioner activity levels, may finally stimulate some seriousinterest in the differences in regional patterns of medical need, andtheir correlates within and outside the health care system.Until now the Canadian health insurance plans, while comparativelysuccessful in world terms, have tended to focus the attention of bothproviders and reirnbursers almost exclusively on financial issues, andalmost always in conflict. The system of reimbursement outlined heremight not only move several important policy issues out of the adversarialarena, but also create 11c l i ent s l l for better information on both theeffectiveness of medical interventions, and the needs of the populationsserved. There would also be an obvious channel for the application ofthat information Uat the coalface l1 ~ in the determination of actualbudgetary allocations and patterns of medical care utilization.APPENDIXSome Issues of Implementation11111111111111111111111111111111111111111SOME ISSUES OF IMPLEMENTATIONThe process of physician reimbursement proposed in this paper - acombination of capitation and prorating on a regional and service-specificbasis - requires information on the region of residence of the enrolledpopulation. At present this information is not available in allprovinces. In B.C., for example, many people are enrolled in the MedicalServices Plan through employee groups, and the plan records contain onlythe address of the employer.Other provinces do maintain address information for each enrollee.For that matter, address information is also kept current for holders ofdrivers' licences. Thus the information requirements of thisreimbursement system are by no means infeasible or unreasonable. But itsfull application is clearly restricted to jurisdictions in which suchcomputerized address information is part of the operational data base.A somewhat less sophisticated form of the proposal, however, can beput into operation, pending the availability of enrollee address data.This more limited system has many of the same features as the full scheme,but the incentives which it embodies are not as precisely targeted. Itdoes, however, preserve the main characteristics of combining fee-for­service with overall budgetary control, and of providing physicians withsome incentives for more conservative practice styles.As in the more detailed system, the reimbursing agency and thephysicians' association bargain over a total allocation of funds forreimbursement, $B, and its allocation into sub-components $BG, $BM, $B5,and $BD, which are budgetary allocations for general practice services,medical specialty services, surgical specialty services, and diagnosticservices respectively. As in the original proposal, these allocations areto reimburse types of services l not types of practitioners.The total $BG is then broken up into regional allocations $BG,$BG j $BGn l as before, over the n regions of the province, on the basisof the age-sex adjusted population of each region. The populationadjustment factors are based on the relative rates of utilization ofgeneral practice services l by age and sex class l province-wide. Forspecialty services, however, no sub-provincial allocation of funds ismade. For purposes of reimbursing such services, the province is treatedas one large service area. Thus total claims for reimbursement, SCM, $C5,$CD, are aggregated at the total province level. A single proratingfactor will apply to all specialty services of a given type, providedanywhere in the province.These prorating ratios can be adjusted to reflect the inter­connectedness of medical practice, just as in the more detailed proposalabove. Over- or under-runs in the diagnostic services allocation can bemet in part (or, if desired, whole) by transfers from or to theallocations for the other services, and similarly specialty over- (under-)runs Can be reflected in the prorating factor for general practiceservices. The difference is that net allocations for general practiceservices in each region are affected by over- or under-runs in specialtyservices for the whole province, rather than responding only to levels· ofservices provided to the residents 'of that region.For general practice services, the (adjusted) regional allocationsBG j will reimburse all such services provided by practitioners in regionj. Thus total claims for general practice services, CG j ' , will be theclaims submitted by the practitioners in the area, regardless of theresidence of the patient. In contrast, CG j in the original proposal wastotal claims for such services submitted on behalf of the residents ofregion j, by practitioners anywhere in the province.The fees reimbursed to practitioners in each region j will be scaledup or down according to whether this value CG j ' falls short of or exceedsthe original budgetary allocation for the region, BG j , as augmented orreduced by transfers from or to the other specialty services. But theadjustment to these prorating factors to take account of these transferswill have to be in the same proportion in each region. This follows fromthe fact that, in ,the absence of patient residence data, one cannotattribute over- or under runs in specialty service use to any particulargroup of referring practitioners.Any services falling into the specialty categories, provided bypractitioners in region j, will be scaled up or down for reimbursementpurposes by the ratios defined at the provincial level.This process of reimbursement does not depend on any informationabout the residence of particular patients; it requires only estimates ofthe total population in each of the regions, and its age-sex distribution.Implicit in the process is the assumption that general practice servicesare provided within relatively compact service areas, and that patients donot travel long distances, or cross regional boundaries, for suchservices. (Or rather, such effects are assumed to be small.)To the extent that this assumption does not hold, however, regionswhich are "net expor t e r s " of general practice (their practitioners provideservices to patients from other regions) are placed at a disadvantagerelative to "net importers I', The funds allocated to reimbursepractitioners in each region are based only on the persons resident inthat region, and are not adjusted for border·crossing.The inability to adjust for such border-crossing, however, is simplya consequence of the lack of data on patient residence - in the absence ofsuch data one cannot even reliably identify llexporting" and l1irnporting"regions. (Note that it is the status of the region which matters - theindividual practitioner who sees a patient from another region is notthereby disadvantaged so long as there is some corresponding patient fromhis/her region travelling to a physician in another region. The proratingfactor which applies in region j depends on the net flows.)To the extent that su!h inter-regional flows are perceived as aproblem, the obvious answer is to speed up the generation of patientresidence data, and the shift to the process of reimbursement in the mainproposal. That fully residence-based system automatically adjusts for anyborder crossing which may occur. The justification for moving to aninterim, practitioner-based system is simply that in the case of generalpractice services, border-crossing by patients is unlikely to be a majorproblem.The interim proposal does, however, preserve the financial incentivefor general practitioners to locate in relatively under-serviced areas.The total amount of reimbursement allocated to region j is dependent onthe population size and structure in that region, independently of thenumber of practitioners in it, so prorating factors will tend to be aboveunity in regions with relatively few providers of such services, andconversely.As emphasized in the main proposal, however, a patient-based systemcreates an incentive to provide services to residents of under~servicedareas. This interim system creates an incentive to locate in underservedareas. For general practice services, the distinction may not be verysignificant. But it could yield some perverse results, if practitionersbegin moving their office addresses of record. Could practitioners spendone or two days a week in a shared office in northern B.C., report that asan office address, and then come back to Vancouver to work for the rest ofthe time? Under this scheme, such a manoeuvre might enable one to havethe northern B.C. prorating factor applied to all of ones' Vancouverbillings.The answer is probably yes; if a scheme has a loophole, someone willtake advantage of it. But the problem is unlikely to be large, or to'develop overnight. If the system of reimbursement based on practitionerlocations is clearly identified as an intermediate stage, pending thedevelopment of'a residence data base for patients, any such distortionswill be corrected automatically by the shift to a patient-based system.With respect to specialty services, the interim proposal has theweakness that it treats the whole province as one service area. Thus itprovides no incentive for specialists to locate in one area than inanother ~ if regional concentration of specialists is seen as a problem,this proposal will not address it.Furthermore, the incentive effects of the main proposal are diluted.Over- and under-runs of allocations for diagnostic and specialty servicescan be identified only at the level of the entire province, so the linkagebetween "po t s " of funds is also at the provincial level. If practitionersin one region become more conservative in their diagnostic testingbehaviour, for example, this tends to raise the prorating factor acrossthe whole province, but obviously by a relatively small amount.On the other hand, it is possible that the mere creation of thisexplicit financial linkage will have the effect of drawing physicians'attention to the questions of practice style and utilization patterns. Inany case, with a shift to the residence-based system of reimbursement, theincentives would become regionally targetted within a few years.To increase the incentive effects of the proposal, it would bepossible (and indeed probably essential, for political acceptability) toreport to physicians individually the components of their gross receipts,on an annual basis. Such a report should show their gross billings byservice category, the adjustments for transfers among different serviceallocations, the prorating factors applied, and the adjusted grossreceipts. 16 This would enable each physician to see at a glance the wayin which billings for different classes of services interacted with eachother, and the financial consequences of different practice patterns.Such an annual report implies an annual payment period, withbudgetary allocations and prorating factors set over that time. It wouldbe equally possible in principle to establish the scheme over a quarter,or even a month, depending on the administrative processes involved. Oneshould, however, distinguish between the contract or budgetary period, andthe period(s) over which rates of reimbursement could be adjusted.Several of the latter could be nested in the former.For example, suppose the negotiation period was a year, and thevarious budgetary allocations were determined over that period. Clearlyone cannot know what all the relevant inter~service transfers and regionalprorating factors are until all the claims are in and settled for thatyear. One could simply pay claims as they came in, but then there will be16 A combination of explanation in text, algebra, and examples mightalso help - for a program of some complexity, the more explanation thebette.r.a closing adjustment in which additional payments will have to be made tosome practitioners, and amounts recouped from others. The latter is, ofcourse, problematic. The delay for end-of-period settlement will ofcourse be greatly reduced as the province moves to on-line bulk billing,but it cannot be eliminated entirely.One possibility is simply to pay all claims as they corne in at, say,80 percent of the negotiated fee schedule, and then make end of yearsupplementary payments (with interest) as appropriate. This assumes thatno prorating factors will fall as low as 80 percent, which seemsplausible. Alternatively, one could pay claims at 100 percent, or closeto it, and then recoup or augment funds as necessary by a discount orsupplement to next period's claims (starting at the beginning of the nextpayment period). This would be closer to the after-the-fact adjustmentswhich now take place, but the difference is that the amounts recouped orpaid would have to calculated separately for each practitioner. Therewould obviously be some problems if practitioners left the province, orretired, problems which do not arise in the 80 percent approach.With the increased computerization of the payment process, however,it may be possible now or soon to monitor outflows on a monthly or bi­weekly basis. This would permit reimbursement in each bUdgetary period tobegin at 100 percent of negotiated fees. Then, within the period,observed outflows could be compared with budgeted targets, by region andspecialty. Since there is a distinct seasonal pattern to medical servicebillings, and even (within short periods) a response to the pattern ofstatutory holidays, these targets would have to be set to reflect suchnormal patterns.Adjustments could be made to the proportion of claims beingreimbursed, on a monthly or bi-weekly basis, according to whether thesetargets were being over- or under-run. This would not give an advantageto practitioners who billed early (or late) in the year, because therewould still be an end-of-period settlement to ensure that eachpractitioner's reimbursement corresponded to the formulas above. But thesize of this adjustment would be significantly reduced, insofar asadjustments were taking place all along the way.Practitioners would not be required to accept either discounts ontheir fees from the beginning of the budgetary period (which assumesoverruns until proven otherwise) or placed at risk of being required topay back substantial sums (also rather unpopular). Some subsequentadjustments are inevitable - and require interest compensation - but theycan be minimized by continuous monitoring and adjustment. On the otherhand the payment process becomes much more complex administratively thanthe simple approach of paying at 80 percent or 90 percent and then payinga one-time end-of-year adjustment. The latter approach may also have thenon-trivial advantage of making it easie~ for the individual practitionerto understand how the system is working, and how his/her reimbursement isbeing determined.ReferencesAnderson, G.F., J.C. Cantor, E.P. Steinberg and J. Holloway (1986),"Capitation Pricing: Adjusting for Prior Utilization and PhysicianDiscretion", Health Care Financin£ Review 8:2:27-34Anderson, G.M. and J. Lomas (1988), "The Development of UtilizationAnalysis: How,'Why, and Where it's Coi.ngv, paper presented at the FirstAnnual Health Policy Conference, Centre for Health Economics and PolicyAnalysis, McMaster University, Hamilton, May 27Barer, M.L. (1988), "Regulating Physician Supply: The Evolution of BritishColumbia's Bill 41", Journal of Health Politics, Policy and Law, 13:1:1-25Barer, M.L., R.G. Evans andR.J. Labelle (1988), "Fee Controls as CostControl: Tales from the Frozen North", The Milbank Ouarterlv 66(1):( forthcoming)Barer, M.L. and R.G. Evans (1986), "Riding North on a South-bound Horse?Expenditures, Prices, Utilization and Incomes in the Canadian Health CareSystem", in R.G. Evans and G.L. Stoddart (eds.), Medicare at Maturity:Achievements, Lessons and Challenges, Calgary: University of CalgaryPress, pp. 53-163Barer, M.L., A. Gafni and J. Lomas (1988), "Accommodating Rapid Growth inPhysician Supply: Lessons from Israel, Warnings for Canada", InternationalJournal of Health S~rvices (forthcoming)Barer, M.L., R. Labelle, S. Morris, R.G. Evans and G.L. Stoddart (1987),"The Impact on Medical Services Utilization of British Columbia's 1982/83Physician Fee "Giveback"; Preliminary Results ll , Canadian Journal of PublicHealth 78:1:37-42Barer, M.L. and P. Wong Fung (1987), Fee Practice Medical ServiceExpenditures Per Capita, and Full-Time-Equivalent Physicians in BritishColumbia, 1985-86, HMRU 87:1, Vancouver: Division of Health ServicesResearch and Development, University of British ColumbiaBrenner, G. (1988), "Negotiated Ceilings for Ambulatory Health CareExpenditures and Other Measures Undertaken In the Context of the FederalRepublic of Germany's Concerted Action in the Health Field", paperpresented to the International Symposium 1988, "Controlling Costs WhileMaintaining Health: The Experience of Canada, the U.S. and the F.R. ofGermany with Alternative Cost Containment Strategies", Bonn, F.R.G., June27-28Carr-Hill, R.A. (1987), Health Status, Resource Allocation and Socio­Economic Conditions, (Interim Report of Health Needs Research Study forWolverhampton Borough Council and District Health Authority) York:University of York Centre for Health EconomicsCarr-Hill, R.A. (n.d. [1988J), Revising the RAWP Formula: IndexingDeprivation and Modelling Demand, (Discussion Paper #41) York: Universityof York Centre for Health EconomicsContan dr i opouLos , A. P. (1986), "Cost Containment Through PaymentMechanisms: The Quebec Experience", Journal of Public Health PolicySummer:224-238Ellis, R.P. and T.G. McGuire (1987), "Setting Capitation Payments inMarkets for Health Services", Health Care Financing Review 8(4):55-64Enthoven, A.C. (1985), Reflections on the Management of the NationalHealth Service, London: The Nuffield Provincial Hospitals Trust(Occasional Papers 5)Gabel, J .R. and T.H. Rice (1985), "Reducing Public Expenditures forPhysician Services: The Price of Paying Le55 11 , Journal of Health Politics.Policv and Law 9(4):595-609Horne, J.N. (1987), "Searching for Shortage: A Population-Based Analysisof Medical Care Utilization in 1!Underdoctored l 1 and 11Undoctored 11Communities in Rural Manitoba l1 , in J.M. Horne (ed.), Proceedings of theThird Canadian Conference on Health Economics 1986, Winnipeg: Universityof Manitoba, Department of Social and Preventive Nedicine, pp. 173-98Institute of Health Services Nanagement (1988), Alternative Delivery andFunding of Health Services: Final Report, London: IHNSLomas, J. and M.L. Barer (1986), "And Who Shall Represent the PublicInterest? The Legacy of Canadian Health Nanpower Policy", in R.G. Evansand G.L. Stoddart (eds.), Nedicare at Maturity: Achievements, Lessons andChallenges, Calgary: University of Calgary Press, pp. 221-286Lomas, J., M.L. Barer and G.L. Stoddart (1985) Phvsician ManpowerPlanning: Lessons from the Macdonald Report, Toronto: Ontario EconomicCouncil, 129 pp.Luft, H.S. (1981), Health Maintenance Organizations: Dimensions ofPerformance, New York: Wiley-InterscienceManning, W.G., A. Leibowitz, G.A. Goldberg, W.H. Rogers, and J.P. Newhouse(1984), "A Controlled Trial of the Effect of a Prepaid Group Practice onUse of Services", New England Journal of Medicine 310(23):1505-10Muldoon, J.M. and G.L. Stoddart (1987), "The Potential Contribution of aCompetitive Approach to Controlling Health Care Expenditures for aHypothetical Community: A Simulation Model", in J.M. Horne (ed.),Proceedings of the Third Canadian Conference on Health Economics 1986,Winnipeg: University of Manitoba, Department of Social and PreventiveMedicine, pp. 19-51Rachlis, M. and C. Fooks (1988), "Utilization Analysis: CurrentInitiatives Across Canada ll , paper presented at the First Annual HealthPolicy Conference, Centre for Health Economics and Policy Analysis,McMaster University, Hamilton, May 27Shillington, C.H. (1972), The Road to Medicare in Canada, Toronto: DelGraphicsUnited States Congress, Office of Technology Assessment (1986), Paymentfor Physician Services: Strate~ies for Medicare, OTA-H-294 (WashingtonD.C.: U.S. Government Printing Office, February), pp. 205-208United States Congress, Congressional Budget Office (1986), PhysicianReimbursement Under Medicare: Options for Chan~e, (Washington, D.C.: CBO,April), pp. 95-97


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items