UBC Faculty Research and Publications

British Columbia Hospitals: examination and assessment of payment reform (B-CHeaPR) Sutherland, Jason M; McGrail, Kimberlyn M; Law, Michael R; Barer, Morris L; Trafford Crump, R Jun 24, 2011

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

Item Metadata


52383-12913_2011_Article_1668.pdf [ 240.69kB ]
JSON: 52383-1.0132657.json
JSON-LD: 52383-1.0132657-ld.json
RDF/XML (Pretty): 52383-1.0132657-rdf.xml
RDF/JSON: 52383-1.0132657-rdf.json
Turtle: 52383-1.0132657-turtle.txt
N-Triples: 52383-1.0132657-rdf-ntriples.txt
Original Record: 52383-1.0132657-source.json
Full Text

Full Text

STUDY PROTOCOL Open AccessBritish Columbia Hospitals: examination andassessment of payment reform (B-CHeaPR)Jason M Sutherland*, Kimberlyn M McGrail, Michael R Law, Morris L Barer and R Trafford CrumpAbstractBackground: Accounting for 36% of public spending on health care in Canada, hospitals are a major target forcost reductions through various efficiency initiatives. Some provinces are considering payment reform as a vehicleto achieve this goal. With few exceptions, Canadian provinces have generally relied on global and line-itembudgets to contain hospital costs. There is growing interest amongst policy-makers for using activity based funding(ABF) as means of creating financial incentives for hospitals to increase the ‘volume’ of care, reduce cost,discourage unnecessary activity, and encourage competition. British Columbia (B.C.) is the first province in Canadato implement ABF for partial reimbursement of acute hospitalization. To date, there have been no formalexaminations of the effects of ABF policies in Canada.This study proposal addresses two research questions designed to determine whether ABF policies affect healthsystem costs, access and hospital quality. The first question examines the impact of the hospital funding policychange on internal hospital activity based on expenditures and quality. The second question examines the impactof the change on non-hospital care, including readmission rates, amount of home care provided, and physicianexpenditures.Methods/Design: A longitudinal study design will be used, incorporating comprehensive population-baseddatasets of all B.C. residents; hospital, continuing care and physician services datasets will also be used. Data will belinked across sources using anonymized linking variables. Analytic datasets will be created for the period between2005/2006 and 2012/2013.Discussion: With Canadian hospitals unaccustomed to detailed scrutiny of what services are provided, to whom,and with what results, the move toward ABF is significant. This proposed study will provide evidence on theimpacts of ABF, including changes in the type, volume, cost, and quality of services provided. Policy- and decision-makers in B.C. and elsewhere in Canada will be able to use this evidence as a basis for policy adaptations andmodifications. The significance of this proposed study derives from the fact that the change in hospital fundingpolicy has the potential to affect health system costs, residents’ access to care and care quality.BackgroundAccounting for $46-billion - or 36% of public spendingon health care [1] - hospitals are a major target for costreductions through efficiency initiatives in Canada.Some provinces are considering payment reform as avehicle to achieve this goal. The use of financial incen-tives to increase hospital efficiency is now widespread inEurope and has occurred in the United States (US)since 1983 [2-4]; however, the approach remains largelyuntested in Canada [5].In April 2010, the British Columbia (B.C.) provincialgovernment implemented activity based funding (ABF)for hospitals, marking a fundamental change to themethod of funding acute hospital care. ABF will directup to 20% of available funds to acute hospitals on thebasis of types and volume of services. This change is sig-nificant in an industry accustomed to historically-basedglobal budgets and unaccustomed to detailed scrutiny ofwhat services are provided to B.C.’s 4.5 million residentsand at what cost.While this change in hospital funding is being imple-mented in B.C., a debate over the relative merits ofhospital ABF is unfolding across Canada [6,7]. For* Correspondence: jsutherland@chspr.ubc.caUniversity of British Columbia, Faculty of Medicine, Centre for Health Servicesand Policy Research, 201 - 2206 East Mall, Vancouver, B.C., V6T 1Z3, CanadaSutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150© 2011 Sutherland et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.example, the Canadian Medical Association and the B.C. Medical Association have recently staked positionssupportive of ABF [8-10], although support for ABFamong the medical community is mixed [11,12]. TheCanadian Nurses Association supports an increasingemphasis on output measurement [13] but falls shortof endorsing ABF, while other groups are calling formore comprehensive baseline data [14]. Althoughthere has been no formal examination of the effects ofABF policies in Canada - in part because B.C. is thefirst province to adopt ABF for acute hospitals on alarge scale - several other provinces are consideringadopting similar funding models.This project will provide an evidence base on theimpacts of ABF, including changes in the type, volume,cost, and quality of services provided. Policy- and deci-sion-makers in B.C. and elsewhere in Canada will beable to use this evidence as a basis for policy adaptationsand modifications.RationaleWhat is Activity Based Funding?In B.C., regional Health Authorities (HA) are responsi-ble for providing health services to their region’spopulation. The HAs derive their operating revenuefrom multiple sources, though the vast majority isprovided by the provincial government [15,5]. Provin-cial funds are allocated to HAs through a mix of: 1)block operating grants (which are historically- andpopulation-based, independent of the volume and typeof activity provided to the region’s residents), 2) tar-geted line items linked to specific policy objectives,and 3) to a very minor degree, pay for performanceinitiatives [16]. Health authorities then allocate fromtheir block grants to the various community-basedproviders (excluding physicians) in their regions, witha significant portion of those resources going to acutecare hospitals.ABF is a variant of fee-for-service in which funds areallocated based on the volume and type of servicesprovided, including considerations of patient case-mix.ABF is an alternative to the traditional Canadian blockoperating grant funding for hospitals but has beenwidely used outside of Canada for some time now [2-4].A small-scale implementation in Canada was abandonedby the Ontario Ministry of Health and Long-Term Care[17] in 2007 due to poor data quality [18,19] and lack ofsupport across affected stakeholders. In the B.C.program, as an addition to a base block grant, a portionof hospital funding will flow based on the number ofcases, with remuneration adjusted for the mix of patientdiagnoses and the services and procedures provided tothose patients.Information Base for Supporting ABFThe success of ABF is critically linked to the ability tomeasure ‘weighted’ hospital output accurately. Methodsfor characterizing hospital output according to thepatients’ mix of clinical conditions and interventionshave become common and are known as case mixadjustments. The most popular method is diagnosisrelated groups, or DRGs [20], though variants of DRGhave been developed, each customized to address localpolicy and health delivery characteristics and objectives[21,22]. In Canada, all acute inpatient discharge datasubmitted to the Canadian Institute for Health Informa-tion’s (CIHI) Discharge Abstract Database (DAD) arecase-mix-adjusted using their Case Mix Group (CMG+)methodology [23]. Case mix adjustment is importantsince it weights hospital discharges according to theirexpected relative use of resources (and therefore cost).Each CMG+’s expected relative cost is represented by aresource intensity weight (RIW). As the length of stayhas no bearing on the RIW value assigned (with minorexceptions for patients with exceptionally long lengthsof stay), the incentive is to shorten lengths of stay inorder to keep actual costs below the RIW-based case-specific relative amounts.The veracity of weighting hospital output by RIW is con-tingent on accurately reported clinical data. CIHI’s clinicalchart re-abstractions have shown that Canadian hospitalshave modified their clinical coding behaviour to maximizeRIWs even in the absence of direct financial incentives todo so [18,19]. This behavior introduces an additional chal-lenge to our study and requires us to be able to differenti-ate actual changes in activity mix as a result of thechanging financial incentives from apparent changes inactivity mix that are merely shifts in coding practices.Gaps in knowledgeThe introduction of payment tied to particular types ofactivity represents a major change in hospital paymentpolicy in Canada. We lack a knowledge base in Canadaregarding the actual effects of these policies; and the(also relatively sparse) lessons from other health caresystems cannot be necessarily generalized to the Cana-dian context. The natural experiment that is unfoldingin B.C. thus provides an important opportunity to exam-ine whether the theoretical incentive effects of ABF forhospitals, tied to efficiencies sought by B.C.’s Ministry ofHealth Services (MoHS), will actually materialize.This project also provides a first-ever opportunity toexamine the subsequent impact of ABF on the non-hos-pital sector. From a policy perspective, our results willprovide as close to real-time impact information to B.C.decision makers as is possible in the world of health ser-vices research regarding the effects of ABF on costs,quality and access to care for residents.Sutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 2 of 8Study ObjectiveThe primary objective of this study is to examine theimpact of ABF on acute care hospitals and related ser-vices in B.C. This objective will be fulfilled by meetingtwo specific aims.Aim 1: To measure internal (to the hospital) changesresulting from the shift toward ABF. In a longitudinalanalysis of observational data, we will measure whetherand how hospitals respond to the changed financialincentives. Related hypotheses include:Hypothesis 1.1: Lengths of stay will decrease andcase mix adjusted volume of patients will increasefrom baseline levels. We will test for changes intrend over time.Hypothesis 1.2: There will be changes in in-hospitalquality measures (mortality and adverse events) forspecific conditions that are characterized by theirvariability in clinical utilization patterns. We will testfor changes in trend over time.Hypothesis 1.3: Hospitals’ expenditures will increasefrom baseline levels. We will construct a model todetect changes in trend over time of hospitalexpenditures.Aim 2: To measure external (to the hospital) changes.In a longitudinal analysis of observational data, we willmeasure whether other components of the health caresystem are affected by the shift toward ABF. Relatedhypotheses include:Hypothesis 2.1: There will be an increase in 30 dayreadmission rate, rate of admissions from emergencydepartment (ED) and rate of admissions for ambula-tory sensitive conditions, all indicators of sub-opti-mal acute care. We will test for an increase in therates over time.Hypothesis 2.2: There will be an increase in thenumber of new home care patients subsequent tobeing discharged from acute care. We will constructa model to detect changes in the trend of: 1) newhome care patients over time and 2) readmissions toacute care from home care over time.Hypothesis 2.3: There will be an increase in expen-ditures on physician services. We will test for anincrease in physician expenditures over time.Methods/DesignDesignThis study has been approved by the University of Brit-ish Columbia’s Research Ethics Board.Aim 1We do not expect ABF to affect all patient types uni-formly. Conditions for which patterns of care are welldefined are expected to be invariant to hospital fundingincentives. Accordingly, we will map all hospitalizationsinto one of three 3 care types; supply sensitive care, pre-ference sensitive care and evidence-based care. The ratio-nale for stratifying analyses into these 3 categories isthat utilization intensity has been shown to vary by con-dition, region and clinician preference [24]. For eachstratum, the change in the average length of stay andnumber of discharges is expected to vary (as shown inthe 2nd column of Table 1).In addition to overall case mix adjusted volume andday surgery volume, we will calculate average lengths ofstay for case mix adjusted volume of patients for eachcondition listed in the third column of Table 1. We willuse ICD-10-CA codes underlying the CMG; a strategyfor identifying hospitalizations that ensures that the con-ditions identified in Table 1 are the primary reason forhospitalization (and removes patients whose hospitaliza-tion is for reasons other than those listed).In-hospital quality and outcomes as a result of a shiftto ABF will be monitored in B.C. and shared with theMoHS. Based on broadly accepted in-hospital qualityindicators [25], our hospital quality indicators will bemortality rates for selected medical and surgical patients(shown in Table 2), overall mortality rates plus thenumber of patient safety and adverse events [26].Hospital mortality rates will be calculated as the num-ber of inpatient deaths relative to the number of separa-tions for the conditions listed in Table 2. Methods forcalculating mortality rates for surgical patients are basedon the Leapfrog Group’s evidence-based hospital referral(EBHR) standards for surgery, while mortality rates formedical conditions are based on low mortality patientTable 1 Expected response of hospitals to ABF by type ofcare providedCare Type ExpectedResponseReason forHospitalizationSupply Sensitive Some Response Congestive heart failureCancerChronic lung diseasePreferenceSensitiveUnknown Benign prostatichypertrophyHip replacementKnee replacementEvidence-basedCareNone AppendectomyLeg fractureSutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 3 of 8groups [27] for which mortality is an unexpected out-come. We will adjust hospital mortality rates for differ-ences between hospitals of age, sex and risk profiles ofpatients [28,29].We will also calculate in-hospital event rates,representing unintended injuries or complications ofcare [30] per volume of surgical separations and ana-lyze changes in trend over time. These rates include:post operative sepsis (surgical site infection); postoperative thrombo-embolism; and repeat trips to theOR. These quality indicators are commonly appliedand can be readily computed with B.C. ’s hospitalseparation data.Finally, we will adjust hospital expenditure data tocontrol for factors associated with price inflation overtime, such as labour contract increases, in order to iso-late ABF-related changes in expenditures.Aim 2Since all segments of the health care system are inter-twined, even if funding streams are not, this study pro-vides a first-ever opportunity to develop knowledgeregarding the non-hospital consequences of ABF. Toexamine health system effects, we will calculate threecategories of indicators: early discharge from hospital(resulting in readmission to hospital); access to care; and(inappropriate) admissions for ambulatory care sensitiveconditions (see Table 3).Using hospital data for all three categories of indica-tors, we will measure the association between rates ofacute hospital readmissions and ABF. We will look forevidence of a change in 30 day hospital readmissionrates for all causes, acute myocardial infarction (AMI)and prostatectomy discharges. Risk adjusted AMI read-mission rate has been linked to the types of drugs pre-scribed at discharge, post-discharge therapies and thequality of follow-up care, whereas prostatectomy read-mission rates provide one measure of follow up care[31,32]. These indicators are often used to signal too-early discharge. We will use ICD-10-CA diagnosis andCanadian Classification of Health Interventions (CCI)procedure codes to identify patients.We will calculate indicators of system level effects onaccess: the rate of admissions from emergency depart-ment and the risk-adjusted rate of hip fracture inpatients aged 65 and older who underwent hip fracturesurgery on the day of admission or the next day.Increases in hospitals’ emergency department utilizationare linked to (less) availability of out-of-hospital carewhile delayed hip fracture surgery has been associatedwith a lack of resources, physician unavailability andother issues related to access to care [31].We will calculate rates of hospital admissions for twoambulatory sensitive conditions: dehydration [33] anddiabetes mellitus [33]. We believe changes in these ratesare important signals of care quality and potentiallynegative implications for quality of life [34]. We areselecting these two conditions due to their broad useplus our ability to derive these indicators from B.C.’shospital separation data. We will use the ICD-10-CAdiagnoses to identify these patients.Previous studies have not rigorously examined thenon-hospital effects of changing acute hospital paymentmethods, and it is not known how the change to ABFwill affect the demand for home care or the ability tomaintain high quality home care services. We willTable 2 Conditions for calculating age, sex and riskadjusted in-hospital mortality ratesType Title Method for Identifying CasesMedical Acute myocardial infarction ICD-10-CA codesCongestive heart failureStrokeGastrointestinal hemorrhagePneumoniaSurgical Esophageal resection CCI codesPancreatic resectionCarotid endarterectomyCraniotomyHip replacementTable 3 Indicators of early discharge, access and admissions for ambulatory care sensitive conditionsImpact Condition Level ofReportingReadmission All Cause HospitalAcute myocardial infarction HospitalProstatectomy HospitalAccess Emergency department admission rate Hip fracture surgery within 2 days ofadmissionHospitalHip fracture surgery within 2 days of admission HospitalAmbulatory care sensitiveconditionsDehydration: hospital admission rate Health AuthorityDiabetes mellitus: hospital admission rate Health AuthoritySutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 4 of 8calculate the number of new home care patients eachmonth using the continuing care dataset, looking retro-spectively to determine which patients are ‘new’ tohome care as opposed to existing patients. Changes inthe trend of the number of new home care patients willprovide valuable insight into the acuity of patients dis-charged from acute care. Based on the cohort of homecare patients, we will then calculate the number of read-missions to acute care from home care over time. Thisimportant signal of care quality may suggest increasedpressure on home care services attributable to ABF.To date, no published information describes the nat-ure of change in use of physician services after imple-mentation of ABF. Medical Services Plan data allow usto determine the amount of all fee-for-service physicianpayments. For each acute discharge, we will link theamount of physician payments in the 30-day period fol-lowing the incident discharge (including readmissions).Total physician expenditures will be linked to hospitaldata and registry data to enable our analytic dataset tobe able to be analyzed over time and for specific condi-tions (see Table 3). To isolate changes of physician utili-zation from physician payments over time, we will drawon work currently being undertaken at the Centre forHealth Services and Policy Research (CHSPR) on a sepa-rate study to remove the effects of fee changes fromphysician fees.Data Resources and Variable ConstructionThe analytic dataset for Aim 1 (internal to thehospital) will be derived from acute and outpatientDAD hospitalizations linked to the registry data. Thedataset will include the following variables: month andyear of discharge, CMG, CIHI-defined age group, hos-pital, region of residence (HA), length of stay (LOS),ICD-10-CA codes, indicator of death, and AdjustedClinical Groups (a measure of health status [28]). Thenumber of inpatient separations per year, for the con-ditions to be studied (Table 1), is expected to exceed5,000. We will analyze hospital separations over theperiod from fiscal year 2005/2006 through to, andincluding, 2012/2013, to ensure consistent applicationof CMG+ across all years.We will consider the effect of data quality and com-pleteness on the results. The most recent multi-stagerandomized studies of DAD data quality, which involveclinical chart re-abstractions, have been conducted byCIHI and have found the majority of data elements inthe DAD are accurately reported [35], with comorbid-ity reporting being an important exception. We willuse the results of these studies to exclude those ICD-10-CA codes susceptible to being affected by codingpractice changes. Further, we will link Medical ServicesPlan data to hospital separation data to evaluatecompleteness of reporting (a means to validate proce-dure information in the DAD).One comprehensive analytic dataset will be created forAim 2 (external to the hospital) for the years 2005/2006through 2012/2013. This analytic dataset will be theproduct of linking acute hospital, registry and continu-ing care datasets with summarized physician fee-for-ser-vice expenditure data. The hospital data will be used tocalculate rates of: readmission, admissions from theemergency department and ambulatory care sensitiveadmissions. The analytic dataset will include admissiondate, CMG, hospital, region of residence, ICD-10-CAcodes, procedure date and whether the patient wasadmitted from the ED. Readmissions at separate hospi-tals will be attributed back to the incident hospitaliza-tion. The analytic dataset will include an indicatorvariable to represent incident cases of new home care inthe period immediately following discharge from acutecare. A variable representing the amount of physicianservices will be determined from the physician fee-for-service Medical Services Plan (MSP) data linked topatient data.We recognize that approximately 20% of physiciansare not paid by on a fee-for-service basis, so we willlimit our inferences to physicians and expenditures paidby fee-for-service as we expect salaried physicians to beless likely to be affected by the incentives of ABF. Wewill identify those physicians paid primarily by fee-for-service using an information resource being developedat CHSPR for a separate project.AnalysisInitially, we will provide descriptive analyses of the over-all trend in provincial LOS and case mix adjustedvolume. We will then construct a multivariate zero-truncated Poisson regression model to identify factorsassociated with length of stay for each conditionincluded in Table 1. The dependent variable, length ofstay of patient i, is Yi. The regression model effectsinclude: hospital and patient level effects, Xi, (such asage and ACG). Hospital level clustering will be incorpo-rated into the model. The model is:Yi = Poisson(λi), Yi > 0log(λi) = α + Xiβ + γ · timei + θ · policyi + τ · policyi · timeiwhere time and policy (and interaction between timeand policy) represent study month (1,2,3...) and an indi-cator variable for ABF, respectively.If the MoHS changes ABF policy during the studyperiod (e.g., to 30% of acute funding), we will changethe parameterization of the model to reflect the change.The parameters of interest are θ and τ, where θ repre-sents an immediate change in the length of stay and τSutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 5 of 8represents a change in the trend in length of stay. Theresults will be presented as absolute and relative changesto facilitate interpretation. While we are confident thesemethods are appropriate for the data and the questionsbeing addressed, if small sample sizes inhibit analysis,we will explore composite measures (which combinemultiple conditions) for standardized rates [36-38].We will use a linear regression model to examine therelationship between ABF and hospitals’ case mixadjusted volume of cases (transformed for normality ifneeded). The dependent variable, Yi, is the hospital’smonthly number of case mix adjusted cases. Repeatedmeasures (clustering) on hospitals over time will beincorporated into the model through Xi, a vector of hos-pital effects. The regression equation is written:Yi = α + Xiβ + γ · timei + θ · policyi + τ · policyi · timei + ∈iwhere the parameters of interest are change in level(θ) and change in trend (τ). We will construct a regres-sion model for overall weighted cases plus a separateregression model for each condition included in Table 1.We will model the number of in-hospital medical andsurgical deaths, Yi, (as described in Table 2) relative tothe total number of separations, Ni, using a Poissonmodel for count data. The vector Xi will include hospi-tal effects and adjust for repeated measurements on thesame hospital. The regression model is written:Yi = Poisson(λiNi)Ni = Total separationslog(λi) = α + Xiβ + γ · timei + θ · policyi + τ · policyi · timeiThe parameters of interest are those associated withimmediate change in the number of deaths (relative tothe number of discharges, Ni) and a change in the trendof number of deaths. We will explore the use of zero-inflated Poisson (ZIP) models if we find an abundanceof counts of 0.Similar to the approaches described above, we will use alinear regression model (suitably transformed for normal-ity) to detect changes in hospital expenditures associatedwith ABF implementation (including adjustments foreffects of inflation) and controlling for repeated measureson hospitals.Following published methods for calculating hospitalreadmission rates [31], we will calculate overall readmis-sion rates using a multivariate Poisson model for countdata, controlling for repeated measures on hospitals. Wewill also construct separate models for the conditionslisted in Table 3. The regression models of monthlyhospital readmission rates will include effects for hospi-tal, time and a binary variable for ABF. The parametersof interest represent immediate change in readmissionand change in trend over time.We will use a linear regression model to examine therelationship between the number of (monthly) newhome care patients and ABF. The unit of observationwill be the number of new home care patients within anHA and we will control for repeated measures overtime. The effects of primary interest are those associatedwith immediate change in new home care patients and achange in the trend of home care patients. Given stablemonthly count data at the HA-level, we expect to bepowered able to detect moderate and large sized effects.Similar to the methods described above, we will modelthe number of monthly readmissions to acute care fromhome care using a Poisson model for count data andadjust for HA and hospital effects. The parameters ofinterest are those associated with immediate change in thenumber of readmissions to acute care from home care anda change in the trend of number of readmissions.We will also use a log-transformed model to examinethe association between patients’ fee-for-service physi-cian expenditures (fee-adjusted) over time, and imple-mentation of ABF. The regression equation for patients’fee-for-service, fee-adjusted, physician expenditures iswritten:Yi = α + Xiβ + γ · timei + θ · policyi + τ · policyi · timei + ∈iwhere the parameters of interest are change in level(θ) and change in trend (τ). The vector Xi will includeeffects to control for repeated measures on physiciansand hospitals plus will adjust for patient level effects,such as age and comorbidity burden (derived fromCMG).DiscussionThe examination of hospital and non-hospital effects ofthe implementation of ABF in B.C. is both critical andtimely given the high public profile of the “sustainabil-ity” of the health care system, the large cost footprintrepresented by acute care hospitals, and the mushroom-ing interest in ABF across Canada. For the MoHS, thepolicy rationale for ABF - to increase hospital efficiencywhile holding the line on aggregate expenditures - isclear: regional HA block operating grants fail to createthe incentives for hospital-derived efficiencies that ABFmay provide.If B.C.’s policy-makers are to assess the extent towhich their objective of increasing hospital efficiencieswithout sacrificing quality has been achieved, it iscritical that they be able to monitor effects of ABF on atimely basis. While from a policy perspective, this pro-posed work will help position the MoHS to ensure itsobjectives are met, the overall goal of this research is toexamine the impact of ABF not just on hospital activity,but more broadly on the health care system overall.Sutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 6 of 8Study LimitationsWhile we are leveraging the natural experiment of a sig-nificant change in hospital funding, we are limited in thetype of study design we can use. Since the MoHS’ ABFinitiative includes all hospitals, it is not possible to ‘ran-domize’ hospitals to ABF and non-ABF groups as to iso-late the effect of the implementation of ABF. We are notincluding another province’s hospitals to act as a ‘control’group due to the potential confounding from reportingdifferences between provinces, hospital funding policychanges in other provinces and the potentially insur-mountable problems associated with, but undoubtedlylengthy time required to, securing data access in multipleprovinces. Given these constraints, a longitudinal designis the strongest possible design. Our study will be limitedto administrative datasets, precluding the use of detailedclinical data that might inform our analyses.Potential Contributions and SignificanceThis project represents a unique opportunity to examinethe health care system during a time of significant policychange in B.C. that will have implications for health systemfunding across Canada. This project will lay a solid founda-tion upon which to build future projects, including thedevelopment and integration of outcome and quality indica-tors into health system performance measurement in B.C.This study represents a unique opportunity to leveragethe natural policy experiment occurring in B.C. and willcontribute to understanding the dynamics underlyingthe most significant and expensive segments of ourhealth care system. By the end of the study, we expectto propose a series of policy recommendations as theyrelate to monitoring health system use within a new fra-mework for hospital funding. The insight gained fromthese activities will be of high value to all Canadians.Funding AcknowledgementsThis study is supported by an operating grant from the Canadian Institutesof Health Research (MOP-222280-HS1-CAAA-115583), “British ColumbiaHospitals: Examination and Assessment of Payment Reform (B-CHeaPR)”, P.I.Jason Sutherland)Authors’ contributionsJS is responsible for the intellectual content of the study proposal; heconceived, developed, and articulated the conceptual framework, studydesign and statistical analysis of data. JS was the primary author of thefunded study proposal and gave his final approval of the version to besubmitted. KM made substantial contributions to the study design and studyproposal drafts. ML made substantial contributions to the study design andstudy proposal drafts. MB made substantial contributions to the studydesign, study proposal drafts and gave his final approval of the version tobe submitted. TC compiled and edited the study proposal for publication.All authors have read approved the final manuscript.Competing interestsThe authors declare that they have no competing interests.Received: 2 May 2011 Accepted: 24 June 2011 Published: 24 June 2011References1. Canadian Institute for Health Information: National Health ExpenditureDatabase Ottawa: Canadian Institute for Health Information; 2009.2. Moreno-Serra R, Wagstaff A: System-wide impacts of hospital paymentreforms, evidence from central and eastern Europe and central Asia,Policy research paper 4987 Washington: Development Research Group,Human Development and Public Services Team, The World Bank;2009.3. Street A, Vitikainen K, Bjorvatn A, Hvenengaard A: Introducing activity-basedfinancing: a review of experience in Australia, Denmark, Norway and SwedenYork: Centre for Health Economics; 2007.4. Busse R, Schreyögg J, Smith P: Hospital case payment systems in Europe.Health Care Management Science 2006, 9(3):211-213.5. McKillop I, Pink GH, Johnson LM: The financial management of acute care inCanada, A review of funding, performance monitoring and reporting practicesOttawa: Canadian Institute for Health Information; 2001.6. Kondro W: CMA transformation initiative scrutinized. Canadian MedicalAssociation Journal 2009, 181(8).7. Kondro W: Full speed ahead for CMA’s transformation train. CanadianMedical Association Journal 2009, 181(8).8. Collier R: Activity-based hospital funding: boon or boondoggle? CanadianMedical Association Journal 2008, 178(11).9. Kondro W: CMA president-elect focuses on patient-centred care.Canadian Medical Association Journal 2009, 181(8).10. Maynard A, Bloor K: Analysis: Will financial incentives and penaltiesimprove hospital care? British Medical Journal 2010, 340:c88.11. Canadian Doctors for Medicare: Position on activity-based funding inCanadian hospitals and other surgical facilities.[http://www.canadiandoctorsformedicare.ca/images/stories/ABF_Position_Paper_15OCT08.pdf].12. Climenhaga DJ: David Climenhaga’s Alberta Diary.[http://www.albertadiary.ca/2009/11/whats-hurry-activity-based-funding-for.html].13. Canadian Nurses Association: Position Statement: Financing Canada’s HealthSystem Ottawa: Canadian Nurses Association; 2009.14. Coutts J, Thornhill J: Service-based funding and pay for performance: willincentive payments give Canadian healthcare the quality boost itneeds? Healthcare Quarterly 2009, 12(3):42-49.15. Canadian Institute for Health Information: Exploring the 70/30 split: howCanada’s health care system is financed Ottawa: Canadian Institute forHealth Information; 2005.16. Providence Health Care. [http://www.providencehealthcare.org/EDP_Success.htm].17. Joint Policy and Planning Committee: Integrated population based allocationformula Reference document RD#9-8 Toronto: Joint Policy and PlanningCommittee; 2001.18. Preyra C: Coding response to a case mix measurement systembased on multiple diagnoses. Health Services Research 2004,39(4):1027-1046.19. Canadian Institute for Health Information: Coding variations in the DischargeAbstract Database (DAD) Data; FY 1996-1997 to 2000-2001 Ottawa: CanadianInstitute for Health Information; 2003.20. Fetter RB, Shin Y, et al: Case mix definition by diagnosis related groups.Medical Care 1980, 18:1-53, S(2).21. France FHR: Case mix use in 25 countries: a migration success butinternational comparisons failure. International Journal of MedicalInformatics 2003, 70(2-3):215-219.22. Reid B, Sutch S: Comparing diagnosis-related group systems to identifydesign improvements. Health Policy 2008, 87:82-91.23. Canadian Institute for Health Information: DAD resource intensity weights(RIW) and expected length of stay (ELOS) for CMG+ 2008 Ottawa: CanadianInstitute for Health Information; 2008.24. The Dartmouth Institute for Health Policy and Clinical Practice: Tracking theCare of Patients with Severe Chronic Illness: The Dartmouth Atlas of HealthCare 2008 Lebanon: The Dartmouth Institute for Health Policy and ClinicalPractice; 2008.25. Agency for Healthcare Research and Quality: AHRQ Quality Indicators.[http://www.qualityindicators.ahrq.gov/].26. De Vos M, Graaffmans W, Kooistra M, Meijboom B, Van Der Voort P,Westert G: Using quality indicators to improve hospital care: a review ofthe literature. International Journal for Quality in Health Care 2009,21(2):119-12.Sutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 7 of 827. Agency for Healthcare Research and Quality (AHRQ): AHRQ quality indicatorsGuide to patient safety indicators [version 3.1] Rockville: Agency forHealthcare Research and Quality; 2007.28. Johns Hopkins University: The Johns Hopkins University ACG Case-Mixsystem.[http://www.acg.jhsph.edu/].29. Reid RJ, MacWilliam L, Verhulst L, Roos N, Atkinson M: Performance of theACG case-mix system in two Canadian provinces. Medical Care 2001,39:86-99.30. Baker GR, Norton PG, Flintoft V, Blais R, Brown A, Cox J, Etchells E, Ghali WA,Hébert P, Majumdar SR, O’Beirne M, Palacios-Derflingher L, Reid RJ, Sheps S,Tamblyn R: The Canadian Adverse Events Study: the incidence ofadverse events among hospital patients in Canada. Canadian MedicalAssociation Journal 2004, 170(11):1678-86.31. Canadian Institute for Health Information (CIHI): Health indicators 2008:definitions, data sources and rationale Ottawa: Canadian Institute for HealthInformation; 2008.32. Jencks SF, Williams MV, Coleman EA: Rehospitalizations among patients inthe Medicare fee-for-service program. New England Journal of Medicine2009, 360:1418-1428.33. Agency for Healthcare Research and Quality: AHRQ quality indicators Guideto patient safety indicators [version 3.1] Rockville: Agency for HealthcareResearch and Quality; 2007.34. Gillick MR, Serrell NA, Gillick LS: Adverse consequences of hospitalizationin the elderly. Social Sciences and Medicine 1982, 16(10):1033-1038.35. Canadian Institute for Health Information (CIHI): CIHI Data Quality Study ofthe 2005-2006 Discharge Abstract Ottawa: Canadian Institute for HealthInformation; 2009.36. Carriere KC, Roos LL: Comparing standardized rates of events. AmericanJournal of Epidemiology 1994, 140(5):472-482.37. Staiger DO, Dimick JB, Baser O, Fan Z, Birkmeyer JD: Empirically derivedcomposite measures of surgical performance. Medicare Care 2009,47(2):226-233.38. Dimick JB, Welch HG, Birkmeyer JD: Surgical mortality as an indicator ofhospital quality: The problem with small sample size. Journal of theAmerican Medical Association 2004, 292:847-851.Pre-publication historyThe pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/11/150/prepubdoi:10.1186/1472-6963-11-150Cite this article as: Sutherland et al.: British Columbia Hospitals:examination and assessment of payment reform (B-CHeaPR). BMCHealth Services Research 2011 11:150.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitSutherland et al. BMC Health Services Research 2011, 11:150http://www.biomedcentral.com/1472-6963/11/150Page 8 of 8


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