医院管理学英文翻译2.pdf

Health Policy 92 2009 158 164 Contents lists available at ScienceDirect Health Policy journal homepage Hospital response to a global budget program under universal health insurance in Taiwan Shou Hsia Cheng Chi Chen Chen Wei Ling Chang Institute of Health Policy and Management College of Public Health National Taiwan University Taiwan a r t i c l ei n f o Keywords Global budget Health insurance Hospital Taiwan a b s t r a c t Objectives Globalbudgetprogramsareutilizedinmanycountriestocontrolsoaringhealth care expenditures The present study was designed to evaluate the responses of Taiwanese hospitals to a new global budget program implemented in 2002 Methods Using data obtained from the Bureau of National Health Insurance NHI and two nationwide surveys conducted before and after the global budget program changes in the lengthofstay treatmentintensity insuranceclaims andout of pocketfeeswerecompared in 2002 and 2004 The analysis was conducted using the Generalized Estimating Equations GEEs method Results Regressionmodelsrevealedthatimplementationoftheglobalbudgetwasfollowed bya7 increaseinlengthofstayanda15 increaseinthenumberofprescribedprocedures and medications per admission The claim expenses increased by 14 and out of pocket fees per admission increased by 6 Among the hospitals no coalition action was found during the study period Conclusions In the present study it appears that hospitals attempted to increase per case expenseclaimstoprotecttheirreimbursementfrompossiblediscountsunderaglobalbud getcap HowTaiwanesehospitalsrespondtothischallengeinthefuturedeservescontinued long term observation 2009 Elsevier Ireland Ltd All rights reserved 1 Introduction Soaring healthcare expenditures have become a major concern for the national healthcare insurance programs of manycountries Consequently manycountrieshaveimple mented global budget programs in order to control costs 1 3 Within a few years of being implemented in 1995 Taiwan s universal health insurance program faced severe fi nancial crises 4 As a result Taiwan s Bureau of National Health Insurance NHI began implementing global budget GB programs beginning fi rst with dental services in 1998 andthenextendingGBprogramstohospitalsin2002 After theimplementationoftheGBprograms theoverallgrowth rateofnationalhealthcareexpenditureshasdecreased The Corresponding author at Room 618 17 Hsu Chow Road Taipei 100 Taiwan Tel 886 2 33668057 fax 886 2 23414634 E mail address shcheng ntu edu tw S H Cheng present study examined the response of hospitals in Tai wan to the GB program during the fi rst two years of its implementation 1 1 Global budget program in Taiwan Taiwan s universal health insurance program began with a cost plus reimbursement system The majority of its healthcareserviceswerereimbursedusingafee for service payment scheme for which approximately 50 procedures were reimbursed by fi xed payments per case 4 The GB program adopted in Taiwan is an expenditure cap system witha fl oatingpointvalue mechanismthatisusedtopre cisely match the pre determined budget The formula is as follows expenditure service volume unit prices fl oating point value 0168 8510 see front matter 2009 Elsevier Ireland Ltd All rights reserved doi 10 1016 j healthpol 2009 03 008 S H Cheng et al Health Policy 92 2009 158 164159 The total NHI healthcare expenditure in a given year is negotiated and determined based on that of the previous year The unit prices for all service items listed on the NHI paymentschemearefi xed whiletheservicevolumeisvari able Healthcare providers fi le their monthly claims with the Bureau of NHI the expenditure claimed is calculated as the sum of the volume of services per item multiplied by the unit prices according to the payment schemes In order to meet the pre determined national expen diture the total expenditure claimed is reimbursed based on an adjusted amount the expenditure claimed multi plied by the fl oating point value The fl oating point value is retrospectively determined every three months when the review process for the hospital claims has been com pleted According to the formula the fl oating point value is negatively associated with the nationwide service volume which includes the number of patients seen and medical orders written a higher service volume will result in a lower fl oating point value that year for all hospitals con tracted nationwide Taiwan sGBprogramfordentalserviceswasintroduced in 1998 followed by a GB for Chinese Medicine in 2000 a GB for Western medicine community clinics in 2001 and fi nally by a GB for hospital services including outpatient and inpatient services in 2002 The hospital GB program is the most important of the four GB programs accounting forapproximately65 ofthetotalNHIexpenditurein2004 The total NHIbudget is distributed to six regions in Taiwan there is no pre determined budget for each hospital Indi vidual hospitals have to compete for patients and service volume with other hospitals 1 2 Previous studies evaluating GB programs SeveralstudieshaveexaminedtheeffectofGBprograms on controlling health care expenditure in Organization for Economic Co operation and Development OECD countries 2 3 These studies employed two different per spectives containment of health care expenditures at the macro level and the response of hospitals when facing fi nancial pressure at the micro level Atthemacro level severalstudiesevaluatedtheimpact ofGBprogramsonhealthexpendituresandhealthcareuti lizationusingtimeseriesanalyses Thesestudiesfoundthat the GB program was successful in slowing down expendi turegrowth 5 6 aswellasindecreasingutilizationgrowth 7 In addition to healthcare expenditure and utilization Detsky et al reported that the GB program might result in decreased hospital input and a lower per admission cost 8 At the micro level Chen et al found that the GB pro gramwasassociatedwithanincreaseinhospitaladmission due to hospital competition for market share within the GB pool 9 1 3 Theory explaining provider behavior under a GB program Two economics theories could potentially explain providers response to a GB program The fi rst game the ory may be an appropriate means by which to evaluate providers behavior under the GB program 10 According to this theory hospitals choose a strategy that maximizes their interests Thus two providers in a market will both be better off if they choose to reduce service quantities at the same time to pursue a higher fl oating point value How ever if a provider does not know its competitor s strategy choosing the high service volume strategy to ensure mar ket share tends to be safer in the limited GB pool When both providers decide to increase their service volume the fl oating point value subsequently decreases and reduces the hospital s reimbursement from NHI This scenario is referred to as the prisoner s dilemma The second theory the theory of common property resources with a focus on the aspect of social welfare can also be utilized to explain providers behaviors under a GB in health care markets 11 12 In this framework providers share a limited budget and tend to increase their mar ket share Chen et al reported that hospitals signifi cantly increased their service volume after the global budget program was implemented in Taiwan supporting the pre dictions of this theory 9 According to these two theories hospitals set a goal to pursueahighermarketsharewhilefacinglimitedresources under a GB Because reimbursement claims contain ser vice volumes and unit prices we posed the following question would hospitals respond to the GB program by increasing the unit prices of their services The present study hypothesized that hospitals would cope with the GB program by increasing the length of stay or treat ment intensity in an attempt to raise the claim expense In addition we hypothesized that under the universal health insurance scheme in Taiwan hospitals would shift costs to the patient s out of pocket fees 2 Methods 2 1 Data Taiwan s hospital GB program was introduced July 1 2002 In this study we tapped two major data sources to evaluate the impact of the GB program The fi rst was hos pital claim data fi les that were obtained from the Bureau of NHI These data fi les provided information about the service volume claimed medical expense and detailed treatment orders before and after the GB program was implemented The second included two nationwide sur veys of patients discharged from hospitals These surveys were conducted to gather information on patients out of pocket fees and their experience before and after the GB programwasimplemented Thefi rstsurveywasconducted to gather information from patients discharged between January 2002 and March 2002 the second survey was administered to patients discharged between March 2004 and May 2004 These different time periods enabled us to observe the changes in hospital behaviors while avoiding theimpactoftheSARSepidemicinTaiwan whichoccurred in the spring of 2003 Survey data were later linked with hospital claim datasets for analysis Becausediagnosticandtreatmentprocedures aswellas medical expenses vary signifi cantly according to the dis ease we surveyed only those patients who had received 160S H Cheng et al Health Policy 92 2009 158 164 specifi c diagnoses or procedures After consulting medical professionals and considering the prevalence of particular diagnoses wechosetosurveydischargedpatientswhohad one of four medical diagnoses stroke asthma pneumo nia or diabetes mellitus or one of two surgical procedures cesarean section or appendectomy For ease of adminis tration only patients discharged from hospitals accredited as medical centers regional hospitals or district teaching hospitalswereinterviewed 13 Theselarge volumehospi tals comprised more than 70 of hospital beds nationwide in 2002 14 Our sample was selected with the assistance of the Bureau of NHI All NHI contracted hospitals fi le monthly claims to one of six regional NHI branches For the fi rst survey the branch offi ces listed patients with the above mentioned major diagnoses that had been discharged from hospitals accredited as district teaching hospitals or higher between January 1 2002 and March 31 2002 Potential participants were selected from these lists Patients who expired during their stay in the hospital or patients with morethanonediagnosisforthesixdiseasecategorieswere excluded Systematic random sampling was conducted to select 50 of the listed patients from each hospital per month with an upper limit of 15 patients per diagnosis Thequestionnaireswerethenmailedtothepatientswithin two to three months of discharge At least three telephone calls were made by the employees of the six Bureau of NHI branches to encourage participants who had not returned their questionnaires to do so The second survey was con ducted in the same manner In 2002 134 hospitals were accredited as medical cen ters regional hospitals or district teaching hospitals in Taiwan in addition to 344 non accredited or non teaching hospitals A total of 14 408 questionnaires were mailed out to the randomly selected patients who had been treated for one of the six disease categories at these hospitals In total 6725 valid questionnaires were completed and returned yielding a response rate of 46 7 In 2004 16 233 questionnairesweremailedtoselectedpatientsdischarged from 107 hospitals smaller hospitals with fewer than 20 returned questionnaires in the fi rst wave survey were excluded in order to increase the effi ciency of our study In the second survey 6736 valid questionnaires were com pleted resulting in a response rate of 41 5 Among the six disease categories investigated the four medical diagnoses were reimbursed by fee for service FFS payment schemes The two surgical procedures however were reimbursed by DRG Diagnosis Related Groups like case payment Hospitals may adopt different strategies for treating patients under DRG or FFS payments tocopewiththeGBprogram Inordertoincreasethehomo geneityofthestudysample patientswhounderwentanyof the two surgical procedures were excluded from the analy sis In addition cases for which questionnaires could not be matched with the claim data and cases in which patients admission or discharge dates exceeded the study period were also excluded from our analysis The latter was car riedoutinordertoreducetheriskofobtaininganincorrect calculation for length of stay In total 4056 subjects from the fi rst survey and 4413 subjects from the second survey were included in our analysis 2 2 Variables of interest Because we suspected that hospitals might change the nature of their diagnosis and subsequent treatment strate gies for conditions in response to fi nancial pressure the present study fi rst examined changes in medical services We utilized length of stay and treatment intensity to mea sure service content Hospital claim data fi les provided detailed information about the patient s diagnostic treat ment or examination orders as well as medical expenses for each admission For every patient length of stay was calculated using the date of admission and date of dis charge available in the NHI claim data For inpatient stays the number of orders including medications laboratory tests or examinations performed during the hospital stay was calculated as a proxy variable for treatment intensity In addition to length of stay and treatment intensity two other variables were incorporated in order to exam ine the changes in the pricing behavior of hospitals 1 medical expenses claimed to the NHI per admission and 2 out of pocket fees charged to the patient per admis sion The medical expense incurred for each hospitalized episode was considered to be the amount submitted to the Bureau of NHI for reimbursement Due to the limitations of theclaimdata wewereunabletoobtaininformationonthe actual cost of individual treatment items for each hospital stay Out of pocket patient fees were the amounts charged and collected by the hospital including a 10 insur ance co payment requirement and other fees not covered by the NHI program i e single bed rooms meals and supporting medical equipments Data concerning the out of pocket amount paid were obtained from a structured questionnaire The NHI 10 co payment was deducted frompatient reportedout of pocketamountstorefl ectthe actual fee charged by the hospital The claimed medical expenses and out of pocket fees in 2004 were infl ation adjusted for comparison to those of the 2002 fi gures in the regression models Because the claimed expenses and out of pocket fees appeared to be right skewed their log transformed values were incorporated into the regression models Other variables included patient age gender diagnoses in the four disease categories and perceived severity upon admission Variables concerning hospital characteristics such as hospital ownership and level of accreditation were also included in the analysis 2 3 Statistical analyses Descriptive statistical analyses including the frequency percentage mean andstandarddeviationarepresentedfor related variables assessed in this study Chi square tests and t tests were performed to examine related variables in 2002 and 2004 surveys Our data were hierarchically structured i e patients were nested within hospitals level 1 patients level 2 hospitals Therefore to analyze the data we used the Generalized Estimating Equations GEEs method with proper distributions for each patient while taking into account the correlation among patients within thehospitals 15 Thecorrelationstructurewasmodeledas S H Cheng et al Health Policy 92 2009 158 164161 Table 1 Characteristics of the survey respondents before and after the hospital global budget in Taiwan Variables2002 N 4056 2004 N 4413 p Value N N Gender0 1133 Male224255 28251957 08 Female180744 55189442 92 Missing70 17 Age0 4059 35100324 73111125 18 35 6090822 39103023 34 60214552 88227251 48 Perceived severity X SD 2 700 692 660 680 0095 Disease categories0 0003 Diabetes Mellitus99024 41113625 74 Pneumonia135833 48129029 23 Stroke86521 33103723 50 Asthma84320 7895021 53 Hospital Accreditation level0 7233 Medical center90022 1995021 53 Regional Hospital246360 72271461 50 District Teaching Hospital69317 0974916 97 Hospital ownership0 1357 Public hospital132232 59149833 95 Non profi t hospital190446 94197644 78 Private hospital83020 4693921 28 an exchangeable structure According to the features of the variables a negative binominal distribution was fi tted for the length of stay and the number of orders and a normal distribution was fi tted for claimed expenses and out of pocketfees Allstatisticalprocedureswereperformedusing the SAS package 3 Results The characteristics of the sample subjects in the 2002 and 2004 are listed in Table 1 Comparison of the patient characteristics in the 2002 and 2004 surveys revealed that males comprised a greater percentage of partic ipants in 2004 than in 2002 In 2002 patients were more likely to display a higher perceived severity score than those in 2004 2 70 vs 2 66 respectively ranging from 0 to 4 p 0 0095 In both 2002 and 2004 the largest group of patients was admitted for pneumonia 33 48 and 29 23 The 2002 and 2004 surveys also showed limited change with regard to patient distribu tion among the three hospital levels which were in descending order medical centers regional hospitals dis trict teaching hospitals p 0 7233 and hospitals owned by different entities p 0 1357 Because differences were observed between the two samples patient characteristics andhospitalfeatureswereincorporatedintotheregression models Table 2 shows the means of the dependent variables for the two years A longer length of stay 9 12 days vs 8 71 days p 0 0091 and a larger number of medical orders 66 17 vs 58 45 0 0001 were found in 2004 than in 2002 The percentage changes were 4 71 and 13 21 respec tively With regard to cost the medical expenses claimed for an admission increased from 28 966 NT dollars in 2002 to 32 953 NT