HEALTH SERVICE COSTS IN EUROPE: COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT IN NINE COUNTRIES

Transcription

1 HEALTH ECONOMICS Helth Econ. 17: S9 S20 (2008) Published online in Wiley InterScience ( HEALTH SERVICE COSTS IN EUROPE: COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT IN NINE COUNTRIES TOM STARGARDT* Deprtment of Helth Cre Mngement, Fculty of Economics nd Mngement, Berlin University of Technology, Berlin, Germny SUMMARY This pper ssesses vritions in the cost of primry hip replcement between nd within nine member sttes of the Europen Union (EU). It lso compres the cost of service with public-pyer reimbursements. To do so, dt on cost nd reimbursement were surveyed t the micro-level in 42 hospitls in Denmrk, Englnd, Frnce, Germny, Hungry, Itly, The Netherlnds, Polnd, nd Spin. The totl cost of tretment rnged from h1290 (Hungry) to h8739 (The Netherlnds), with men cost of h5043 ðstd h2071þ: The min cost drivers were found to be implnts (34% of totl cost on verge) nd wrd costs (20.9% of totl cost on verge). A one-wy rndom effects nlysis of vrince model indicted tht 74.0% of vrition ws between nd only 26% of vrition ws within countries. In two-level rndom-intercept regression model, purchsing-power prities explined 79.4% of the explinble between-country vrition, while the percentge of uncemented implnts used nd the number of beds explined 12.1 nd 1.6% of explinble withincountry vrition, respectively. The lrge differences in cost nd reimbursement between Polnd, Hungry, nd the other EU member sttes shows tht primry totl hip replcement is highly relevnt cse for cross-border cre. Copyright # 2008 John Wiley & Sons, Ltd. KEY WORDS: hospitl costs; totl hip replcement; dignosis-relted groups; regression nlysis; Europen Union INTRODUCTION When other tretment options such s phrmcologicl therpy, weight loss, muscle-strengthening exercises, nd ssistive devices hve been exhusted, hip replcement surgery is considered to be the best pproch to mnging severe hip osteorthritis nd restoring mobility, provided tht the ptient is otherwise in good helth (Nilsdotter nd Lohmnder, 2002). The procedure entils replcing the disesed hip joint with n rtificil one. As prt of the hip prosthesis, n rtificil socket is implnted in the pelvis. The neck nd hed of the femur re replced by prosthesis tht enbles motion within the socket. The different components of the new joint cn be ttched to the femur nd cetbulum with or without bone cement. The surgery is performed in hospitl setting. The mjority of ptients re older thn 60 yers (Fulkner et l., 1998). Hip replcement is elective surgery nd prt of the benefit bsket in mny Europen countries. As such, it is predestined to become service eligible for cross-border cre. The surgery is performed with gret frequency (e.g /yer in Germny nd /yer in Englnd nd Wles) nd thus hs significnt budget impct (Dreinho fer et l., 2006). The demnd for hip replcement hs incresed over the pst 10 yers nd is likely to continue to do so s result of geing popultions nd becuse of the extension of the ge rnge for this tretment (Fulkner et l., 1998; Sochrt nd Porter, 1997; Upshur *Correspondence to: Deprtment of Helth Cre Mngement, Fculty of Economics nd Mngement, Berlin University of Technology, Secr. EB2, Strsse des 17. Juni 145, Berlin, Germny. E-mil: Tom.Strgrdt@tu-berlin.de Copyright # 2008 John Wiley & Sons, Ltd.

2 S10 T. STARGARDT et l., 2006). According to Orgnistion for Economic Co-opertion nd Development dt, the procedure is subject to witing times in some Europen countries such s Denmrk, Finlnd, Norwy, The Netherlnds, nd the United Kingdom (Sicilini nd Hurst, 2003). Differences in the cost of tretment, public reimbursement, nd tretment ptterns mong the Europen Union (EU) countries re thus of gret interest to helthcre decision mkers nd public pyers. The objective of this rticle is to ssess vritions in the cost of specific type of primry hip replcement between nd within nine member sttes of the Europen Union. The min gols re () to clrify whether differences in hospitl costs re due to differences in the ctul services provided, vritions in the use of resources, or differences in unit cost nd (b) to compre cost nd reimbursement. MEDICAL BACKGROUND AND SHORT LITERATURE REVIEW Hip replcement is indicted in the mngement of rthritis, frcture of the femur or femorl neck, rheumtic joint inflmmtion, or mechnicl problems rising from hip joint dysplsi (Ju ni et l., 2006). Initilly, the ptient s symptoms re generlly chrcterised by stiffness in the ffected joint. Pin is felt when putting weight on the joint. In lter stges, the joint hurts even t rest nd throughout the night. The pin is felt in the groin, trochnter region, nd buttocks, nd my spred to the upper thigh nd knee; mobility is restricted. Becuse the ptient rects to the flexion contrcture of the hip with hyperlordosis of the lumbr region of the spine, pins in the lower bck re not infrequent. Dignoses re bsed primrily on physicl exmintion, X-rys, or mgnetic resonnce imging (Pet et l., 2001). Generlly, the ffected joint is treted conservtively for s long s possible, e.g. by improving ptient self-mngement of the condition or reducing the lod on the joint through weight loss in overweight ptients or with the use of orthopedic ids. Physicl therpy, including cryopplictions, mssge, nd gymnstics, re lso common tretment strtegies. Therpy cn be supported by dministering nonsteroidl nti-inflmmtory drugs nd/or steroids (Ju ni et l., 2006). In the event tht conservtive pproch does not led to sufficient clinicl improvement, however, n endoprosthetic replcement joint cn improve qulity of life (Ju ni et l., 2006; Nilsdotter nd Lohmnder, 2002). Hip replcement surgery typiclly lsts between 45 min nd 2 h under full or prtil nesthesi nd involves hospitl sty of 5 10 dys. This is followed by rehbilittion mesures, which tke plce in hospitl or in the outptient setting nd vry widely ccording to individul needs (Munin et l., 1998). Estimtes mde using Medicre dt from 1999 to 2003 show tht the compliction rtes for primry hip replcement re 2.5% for 90 dys mortlity, 0.5% for pulmonry embolus, nd 1.2% for postopertive wound infection (Crm et l., 2007). The technology for hip replcement hs improved considerbly since the development of the first genertion of cemented prostheses in the 1960s. The longevity of implnts is incresing, nd mny designs nd ttchment methods re now vilble. In generl, it is possible to distinguish between three groups of hip prosthesis, ll of which differ in the wy the rtificil joint is ttched to the ptients bones: cemented, uncemented, nd prtilly cemented implnts. The three groups cn be divided into subgroups bsed on the type of cement, the cement-free methods of ttchment, nd the mteril used to produce the implnt itself nd the use of hip resurfcing methods (Fulkner et l., 1998). METHODS Dt collection nd cse definition As prt of the HelthBASKET project, reserchers from nine EU countries (i.e. Denmrk, Englnd, Frnce, Germny, Hungry, Itly, The Netherlnds, Polnd, nd Spin) collected dt on the inhospitl cost nd reimbursement for specific cse of primry totl hip replcement in 2005, s well s

3 COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT S11 dt on the type of implnt provided. The cse cn be described s follows: womn between 65 nd 75 yers of ge with hip osteorthritis, who requires hip replcement due to considerble impirment of function, is dmitted for her first hip replcement (unilterl). The ptient hs no co-morbidities tht require expensive drugs or other tretment. The surgeon uses the most common implnt for femle ptients, nd surgery is completed without severe complictions. The cse ends with the ptient being dischrged from hospitl either to home or to seprte rehbilittion fcility. To llow for stndrdised dt collection, methodologicl guidelines were developed nd greed upon mong ll reserchers. The guidelines specified which cost components would be surveyed, defined which cost ctegories should be included in overhed costs, nd set generl stndrds for the results. The hospitl smple in ech country ws bsed on the vilbility of costing informtion, willingness to prticipte in the study, nd bility to fulfil the requirements set forth in the methodologicl guidelines. It ws decided tht medium-sized hospitls representtive of ech country should be included in the hospitl smple, wheres very smll nd very lrge hospitls, s well s teching hospitls, should not. As the verge size of hospitls vried considerbly cross countries, hospitl sizes in the different smples rnge from 200 to 1200 beds. Dt on resource use were collected by ll countries through fce-to-fce interviews using questionnires with physicins nd finncil controllers t the respective hospitls. In ddition, ptient-level dt were cquired by nlysing medicl records (e.g. Polnd) or through hospitl informtion systems (e.g. Frnce, Germny). Cost dt were obtined from ntionl or regionl dtbses (e.g. Englnd nd The Netherlnds), from hospitls prticipting in the ntionl smple for dignosis-relted group (DRG) cost-weight clcultions (e.g. Frnce, Germny), or by interviews with finncil controllers (ll countries). In rre cses, unit cost hd to be clculted bsed on expert estimtes (e.g. Polnd), s some hospitls experienced difficulties clculting unit cost for specific items or medicl services described by medicl specilists during the interviews. Becuse of differences in the types of dt vilble in ech country (e.g. ptient level vs hospitl level), dt for ll countries were ggregted t the hospitl level. Averge exchnge rtes from 2005 were used to convert ntionl currencies into euros. Dt nlysis In order to explin whether differences in the cost of tretment within countries were lrger thn those between countries, one-wy rndom effect nlysis of vrince (ANOVA) ws estimted. In ddition, regression nlysis ws conducted, including fctors tht might influence the totl cost of tretment t the hospitl level. Becuse hospitl-level nd country-level vribles were included s independent vribles, two-level rndom-intercept model ws used to ccount for country-specific effects (Grieve et l., 2005; Singer, 1998). The hypothesis of the regression residuls following norml distribution when using cost of tretment s dependent vrible could not be rejected (Shpiro Wilk, P ¼ 0:8005). Dt nlyses were performed using SAS version 9.1. Bsed on the literture deling with the estimtion of hospitl cost functions nd in ccordnce with economic theory, the number of hospitl beds ws included in the regression nlysis to test whether economies of scope nd scle influenced totl cost (Adm nd Evns, 2006; Breyer, 1987; Vitlino, 1987). In ddition, the number of physicins per bed nd dummy vrible for urbn loction (i.e. if the popultion in the city where the hospitl ws locted/provided cre ws bove ) were included to ccount for structurl differences within the hospitl smple. To djust for differences in countries price levels, purchsing-power prities in euros ws included in the model s proxy. The verge length of sty ws lso included in the regression nlysis, becuse cost differences between countries my hve been prtly due to differences in the point of time t which hospitl tretment ended nd tretment in n inptient rehbilittion fcility begn. In ddition, cpcity use ws surveyed for ech hospitl nd included in the regression, s it my hve influenced the mount of overhed costs llocted to ech

4 S12 T. STARGARDT cse. Becuse of lck of dt on cpcity use in Polish hospitls, verges from the entire smple were employed to correct for missing vlues. To ccount for differences in the prosthesis cost nd the use of technology, the percentge of uncemented prosthetics used in ech hospitl ws dded to the regression nlysis. An interction vrible of length of sty nd percentge of uncemented prosthetics ws dded to nlyse whether higher costs for uncemented prosthetics re mde up for by reduced length of sty. The regression model ws specified by first including ll vribles described in the Methods section. Subsequently, bckwrd stepwise selection procedure ws pplied until only significnt vribles remined t the 5% level (Twisk, 2006). RESULTS Descriptive sttistics In totl, dt were collected from 42 hospitls. The number of providers contributing dt differed from country to country, rnging from two providers (Denmrk nd Englnd) to eight providers (Germny). The verge length of sty rnged from 5.9 dys (Denmrk nd The Netherlnds) to 16.2 dys (Germny) (see Tble I). In terms of verge length of sty, the rnk order of countries in our smple ws generlly comprble to the rnk order of countries in the WHO HFA dtbse (World Helth Orgniztion, 2007). The totl cost of tretment rnged from h1290 (Hungry) to h8739 (The Netherlnds), with men cost of h5043 ðstd h2071þ (see Figure 1). Evidence from previous studies suggests tht these results re within plusible rnge (Briggs et l., 1998; Iorio et l., 2001; Johnsson et l., 2006; Lupcis et l., 1994). Compring the totl cost of tretment between countries, there ws lrge gp between the former EU-15 member sttes nd the new EU member sttes, Hungry nd Polnd. The verge cost of tretment in Polnd nd Hungry ws less thn hlf of tht observed in ny other member stte, except for Spin. Excluding Polnd nd Hungry, the verge cost per cse in our smple incresed to h5778 ðh1523þ: For most countries, within-country cost vritions seemed to be greter thn betweencountry cost vritions. Before interpreting individul cost components, differences in the mount of costs prtly or totlly subsumed under overhed costs in ech country must be explined. These differences re prtly due to differences in the vilbility of dt, in the extent to which hospitl informtion systems supported cost-unit ccounting, nd in the methodology pplied for cost ccounting in the respective hospitls. For exmple, wheres Dnish reserchers experienced difficulties determining norml wrd costs nd the cost of surgery, they collected detiled cost informtion on mediction use t their hospitls. Germn reserchers, on the other hnd, hd ccess to ptient-level dt, but were unble to provide detiled informtion on mediction costs becuse the cost of phrmceuticl cre is only rrely linked to the ptient level in Germn hospitls. Thus, vribles such s bed dys were used s n lloction bse to relte phrmceuticl costs incurred in hospitl deprtment in Germny to ll ptients treted in tht deprtment, wheres phrmceuticl costs in Dnish hospitls were llocted bsed on individul ptients drug use. Therefore, the costs presented for ech cost component in Tble II cn only provide rough estimtes for ech country nd hve to be compred crefully by looking t the percentge of unexplined costs subsumed under overhed. Excluding Denmrk from the nlysis becuse of high percentge of unexplined overhed costs (77.5%), the min cost drivers were found to be implnts (34% of totl cost on verge), followed by wrd costs (20.9% of totl cost on verge), nd the cost of surgery (12.9% of totl cost on verge nd excluding the cost of the implnt). Drug costs (4.0% of totl cost on verge) nd dignostics (2.6% of totl cost on verge) were of only minor importnce, except in Englnd, which reported the use of very expensive drugs for nesthesi. On verge, 25.6% of totl cost remined unexplined nd ws llocted to ech cse s overhed. The cost of the implnt rnged from 11.6% of totl cost (Englnd)

5 COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT S13 Tble I. Smple chrcteristics ðþ= Std:Þ Denmrk Englnd Frnce Germny Hungry Itly The Netherlnds Polnd Spin Hospitls included (no.) Purchsing Power Prities in Euro djusted to EU-25 verge Hospitl chrcteristics Beds per hospitl (no.) Physicins per hospitl bed (no.) 0:53 0:04 0:34 0:03 0:27 0:06 0:22 0:07 0:19 0:01 0:59 0:03 0:22 0:08 0:27 0:04 0:67 0:09 Nurses per hospitl bed (no.) 1:67 0:08 1:92 0:35 1:03 0:41 0:58 0:13 0:71 0:02 1:26 0:14 1:91 0:75 0:73 0:21 1:10 0:04 Beds per deprtment (no.) 42:0 45:3 } 31:6 10:9 79:5 28:4 34:0 19:8 33:2 7:79 25:2 11:8 35:5 15:5 } Physicins per deprtment bed (no.) 0:86 0:49 } 0:86 0:49 0:17 0:04 0:20 0:01 0:29 0:10 0:27 0:27 0:26 0:04 } Tretment chrcteristics Non-cemented implnt (%) 0:63 0:21 0:34 0:21 0:41 0:44 0:23 0:25 0:00 0:00 0:90 0:17 0:25 0:35 0:62 0:37 0:11 0:09 Prtilly cemented implnt (%) 0:20 0:28 0:00 0:00 0:00 0:00 0:20 0:20 0:00 0:00 0:00 0:00 0:14 0:38 0:00 0:00 0:00 0:00 Cemented implnt (%) 0:17 0:23 0:66 0:21 0:59 0:44 0:57 0:41 1:00 0:00 0:10 0:17 0:71 0:49 0:38 0:37 0:89 0:09 Length of sty (dys) 5:9 3:0 7:1 1:1 5:9 3:0 16:2 2:3 12:9 0:3 8:2 1:1 5:9 1:2 11:8 1:8 5:9 1:2

6 S14 T. STARGARDT 10,000 Sweden 2000: 9,740 in 8,000 6,000 8,482 6,982 7,622 6,364 5,466 6,782 6,101 5,683 7,853 5,932 6,925 8,739 5,691 5,605 US : 8,646 8,332 Cnd 1988: 6,927 6,754 UK 1996/1997: 5,380 4,000 4,524 4,011 4,457 4,070 4,126 3,599 3,190 2,431 2,000 2,125 1,509 1,298 1,294 1,290 0 Itly (N=5) Germny (N=8) Frnce (N=5) Denmrk (N=2) Englnd (N=2) Netherlnds (N=7) Spin (N=5) Polnd (N=6) Hungry (N=2) Results from other studies Figure 1. Men cost, minimum nd mximum cost of primry totl hip replcement, evidence from previous studies to 50.4% of totl cost (Spin); in terms of bsolute vlue, from h483 (Hungry) to h3416 (Itly). Differences in the cost of the implnt re relted, in prt, to differences in the use of expensive uncemented implnts, e.g. 90% (Itly) vs 0% (Hungry), nd to the underlying definitions on the lloction of costs to the cost components implnts nd mteril. In Itly, for exmple, costccounting systems llocted the cost of mteril used in connection with the implnt itself (e.g. screws) to the cost component implnt (mteril cost h22); however, ccounting systems in Germny llocted this cost to the cost component mteril (mteril costs h249). Wrd costs, which ccounted between 5.2% (Itly) nd 36.9% of totl cost (Hungry), rnged from h360 (Itly) to h1963 (Englnd) nd depended primrily on the cost of nursing stff. The extremely low wrd costs reported for Itly re prtly due to these hving been included in overhed costs. Averge provider reimbursement from the Ntionl Helth Service or public helth insurers rnged from h1771 to h9975; with smple verge of h6067 ðstd 2404Þ: Provider reimbursement for Spin ws not vilble, s hospitls in most Spnish regions receive retrospective budget pyments tht depend only prtly on the number of cses treted (S nchez-mrtı nez et l., 2006). Compred with the verge cost of tretment, verge reimbursement per country ws greter thn totl costs for ll countries except Polnd. The profit mrgin rnged from 10.5% (Polnd) to +32.5% (Denmrk). Dt nlysis The results of the one-wy rndom effects ANOVA model indicted tht 74% of vrition ðh1909þ ws between countries, wheres only 26% of vrition ðh1132þ ws within countries. Excluding Hungry nd Polnd from the nlysis (i.e. becuse they hd less thn hlf of the costs of ny other member stte except for Spin) resulted in smller mount of between-country vrition (38.4% of totl vrition) compred with within-country vrition (61.6% of totl vrition). The finl regression model contined the following vribles: number of hospitl beds, percentge of uncemented implnts used, nd purchsing-power prity in euros. While the use of uncemented implnts nd purchsing-power prity were positively correlted with totl cost per cse, the number of

7 COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT S15 Tble II. Totl cost, cost components, nd reimbursement of totl hip replcement Denmrk Englnd Frnce Germny Hungry Itly The Netherlnds Polnd Spin Dignostic procedures Imging h h87.95 h60.01 h79.83 h7.82 h63.37 h32.90 h33.80 h42.53 Lbortory h35.01 h5.74 h h h10.02 h58.42 h45.12 h14.00 h54.62 Other h6.22 h0.00 h h2.87 h18.06 h19.07 h15.30 h2.52 Norml/intensive wrd Physicin h18.04 h h88.80 h h h h h Nursing h h h h h h h h h Other stff h h h h h0.51 h78.00 h h45.97 h0.00 Mteril h6.40 h h5.78 h16.75 h1.27 Opertion (including wke-up room) Anesthetist/surgeon h h h h h93.25 h h h52.08 h Nursing h h h h h18.53 h99.57 h h9.64 h Other stff h42.52 h0.00 h44.75 h h11.42 h h0.00 h0.00 Implnt h h h h h h h h Mteril h h h h h22.31 h35.00 h0.18 Drugs h59.63 h h60.99 h h72.50 h74.30 h h h46.20 Overhed h h h h h h h h h % Overhed of totl h77.5% h28.7% h36.2% h26.3% h10.0% h37.7% h32.2% h15.1% h18.9% Totl cost h h h h h h h h h Totl cost (djusted by PPP) h h h h h h h h h Reimbursement h h h h h h h h b Subsumed in overhed costs. b Hospitls receive retrospective budget pyments tht depend only prtly on the number of cses treted.

8 S16 T. STARGARDT Tble III. Results from two-level rndom-intercept model Independent vrible Coefficient S.E. t-vlue p-vlue Fixed effects Intercept Number of beds per hospitl n PPP n % Non-cemented implnts n Rndom effects Intercept Residul n Significnt 50:05: hospitl beds ws negtively correlted with the totl cost of tretment, suggesting economies of scope nd scle (see Tble III). Accounting for price differences between countries by using purchsing-power prities in euros explined 79.4% of the explinble between-country vrition, wheres the percentge of uncemented implnts nd the number of beds explined 12.1 nd 1.6% of explinble within-country vrition. However, the results of this pseudo R 2 -pproch must be treted with cution nd cn only serve s rough indictor (Singer, 1998). Multicollinerity dignostics were conducted nd confirmed tht multicollinerity ws not relevnt to the model. DISCUSSION Discussion Results from the regression nlysis show tht the totl cost of primry hip replcement depended primrily on the purchsing-power prity in euros, which ws used s proxy for price nd wge level. Totl cost lso depended on the type of implnt, which ws used s proxy for level of technology, nd the size of the hospitl, which indicted the existence of economies of scle nd scope. Economies of scle nd scope will, mong other things, relte to individul nd hospitl experience. There ws cler difference in cost between Polnd nd Hungry, on the one hnd, nd Denmrk, Englnd, Frnce, Germny, Itly, nd The Netherlnds, on the other, while Spin ws between the two groups of countries. Besides depending on price/wge level, the remining cost vrition between countries my hve been due to orgnistionl spects of the helthcre systems themselves (e.g. the incentive set by provider reimbursement for erly dischrge) or the orgnistion of cre (e.g. the trnsfer between providers tht re involved in tretment before nd fter the hospitl episode). Wheres, in some countries, dignostic procedures re conducted on n outptient bsis nd subsequently mde vilble to hospitl physicins (e.g. Denmrk), in other countries ll dignostic procedures re performed on n in-hospitl bsis (e.g. Germny). The ltter system leds to greter costs for dignosis in some countries. The sme is true for expenditures on physiotherpy (included in the cost component other stff in the norml/ intensive wrd section in Tble II). Wheres some hospitls strt rehbilittion mesures while the ptient is still in hospitl, others leve this ctivity minly to rehbilittion provider. However, it hs to be noted tht the mount spent on physiotherpy lso vries gretly within countries. Other unexplined cost differences between hospitls my hve been due to differences in tretment ptterns, e.g. the surgicl technique used, surgeons experience, the operting thetre environment, the intensity of nursing cre, the number of beds in hospitl room (Fulkner et l., 1998), or differences in qulity, the lst of which goes beyond the scope of this study. In ddition, country-specific fctors (e.g. the existence of witing lists) my hve lso influenced totl cost per cse. Although one study

9 COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT S17 demonstrted tht witing lists do not hve n impct on the outcome of hip replcement (Nilsdotter nd Lohmnder, 2002), witing lists might hve influenced cpcity utilistion for totl hip replcement nd, thus, the mount of overhed llocted per cse. The results of our study lso indicte tht the cost of primry hip replcement using uncemented prosthetics is greter thn the cost of primry hip replcement using cemented prosthetics nd tht these cost differences re not mde up for by reduced length of sty. However, no sttement on the most costeffective type of prosthesis cn be mde, s the cse ends t the hospitl door nd does not mesure outcomes such s the durbility of the prosthesis, the compliction rte fter surgery, or the cost of revision surgery, which is usully much more expensive thn the cost of primry hip replcement (Briggs et l., 1998; Engeseter et l., 2006; Gillespie et l., 1995). Guidelines published by NICE in the yer 2000 rgue tht there is currently more evidence of the long term vibility of cemented prostheses, which, in mny cses, occupy the lower end of the rnge of prostheses cost, thn there is for uncemented nd hybrid prosthesis (Ntionl Institute for Helth nd Clinicl Excellence, 2000). However, other studies clim tht cemented hips hve proved to be superior only in short-term medicl outcomes nd tht evidence is lcking for their superiority over the long term (Ni et l., 2005). The minly positive difference between reimbursement nd costs cn be explined by the use of DRG or DRG-like systems (HRGs, DBCs) for reimbursement in combintion with the cse definition. DRG systems ttempt to clssify cses ccording to medicl criteri (min dignosis) nd economic severity (costs), llowing for certin mount of cost vrition within DRG (Schreyo gg et l., 2006). As the definition for this cse of totl hip replcement excludes ny comorbidity, it cn be ssumed tht the selected cses re less severe nd therefore less costly thn the verge cse within the respective DRG. Thus, reimbursement system tht is bsed on setting reimbursement for the verge severe cse within DRG will led to reimbursement bove totl cost for uncomplicted cses. Polnd, the only country whose verge cost exceeded reimbursement in this study, does not use DRGs for reimbursement. The deficit incurred for totl hip replcement cn be explined by the fct tht deficits of public hospitls re usully mde up for by subsidies provided by the region in which the hospitl opertes (Kozierkiewicz et l., 2006). In ddition, the smple size nd composition of hospitl smples in our study my not hve led to representtive cost estimte. Although it ws the im of ll prticipting reserchers to provide dt bsed on hospitl smple tht ws representtive of their respective country, the low number of providers tht prticipted in ech country limits the generlisbility of our results. This is lso true of the tretment ptterns observed, s these my hve differed from hospitl to hospitl within countries. In ddition, differences in the use of cost-ccounting systems between countries nd hospitls, especilly in the wy tht overhed costs were clculted nd llocted to cses, my hve contributed to some of the cost differences, s well. Alloction bses used to llocte costs to cses, the level of detil vilble from hospitl informtion systems, nd the ccounting methods used to determine unit costs could not lwys be influenced by reserchers. On the one hnd, using ntionl cost dtbses (e.g. Englnd nd The Netherlnds) or working with dt clculted ccording to stndrds set by the ntionl DRG institute (e.g. Frnce nd Germny) reduced vritions in ccounting prctices within countries, but t the sme time incresed differences between countries becuse of the use of slightly different costing methodology. Hrmonising cost ccounting nd setting disclosing stndrds for hospitls would therefore gretly improve the dtbse for future reserch in this re. Evidence from other studies Although some of these studies were conducted more thn 5 yers before our study nd they do not present systemtic review, evidence suggests tht our results re plusible. In rndomised controlled tril on cemented prosthetics vs internl fixtion in 2000, the cost of primry hip replcement including n initil outptient visit in Sweden ws estimted to be h9740: The severity of disese in this popultion

10 S18 T. STARGARDT smple ws, however, greter thn ours in this study, becuse the men ge of the study popultion ws 84 yers. In ddition, 38.4% of the ptients prticipting in the study hd mentl illness, which considerbly incresed tretment cost compred with ptients without mentl illness (Johnsson et l., 2006). A study conducted in the UK in using cost dt from the Nuffield Orthopedic Centre in Oxford nd dt on resource use bsed on clinicl estimtes of n verge cse clculted the cost of primry hip replcement with cemented prosthesis to be h5375 ð 4052Þ: The cost of prosthesis including the cost of cement ws h487 ð 367Þ (9.1% of totl cost), which is comprble to the cost of cemented prosthesis for UK ðh658þ in our study. The min cost driver ws found to be wrd costs, which ccounted for 59.5% of the totl cost (Briggs et l., 1998). In the US, study on the optiml tretment for displced femorl neck frctures in elderly ptients conducted between 1993 nd 1996 estimted the hospitl cost of cemented totl hip replcement t h8332 (US$10 490). The hospitl cost of uncemented totl hip replcement ws clculted to be h8646 (US$10 886). The cost of implnt ws 17 nd 20% of totl cost. After hospitl tretment (verge length of sty ws 5.5 dys), ptients were sent to nother inptient rehbilittion fcility (verge length of sty ws 19 dys), which led to n dditionl cost of h7511 (US$9457) for both types of implnts. The study lso indicted tht the cost of hospitl sty depends to lrge extent on the point in time t which the ptient is trnsferred from the hospitl to the rehbilittion fcility (Iorio et l., 2001). Another US study clculted the cost of primry hip replcement in 2003 to be h (US$24 170). Results from multivrible regression nlysis indicted tht higher severity of disese ws ssocited with cost increses (Boznic et l., 2005). A rndomised controlled tril conducted in Cnd in 1988 compred the cost of cemented prostheses with the cost of uncemented prostheses nd found virtully no difference in costs between the two types of implnts. The cost of primry hip replcement with cemented prosthesis ws h6754 (Cn$9233) nd h6927 (Cn$9470) for n uncemented prosthesis. Similr to our study, the min cost drivers were found to be wrd costs nd the cost of implnt (Lupcis et l., 1994). CONCLUSION The lrge difference in costs nd reimbursement between Polnd, Hungry, nd other countries shows tht primry totl hip replcement is highly relevnt cse for cross-border cre. However, in the very long run, s these differences re minly driven by differences in unit costs, it cn be expected tht moving towrds cross-border cre nd competing mong the scrce resource of physicins within Europe will led to convergence of costs nd prices. Even though the comprbility of cost per cse in our study is limited due to the smll hospitl smple nd vritions in study conditions (e.g. cost-ccounting stndrds) between countries, our results provide good estimte for policy mkers nd cn serve s strting point for future reserch in this re. Our results show tht compring costs t the micro-level is fesible nd cn serve s n instrument of helthcre systems nlysis. The study design is expndble to other interventions nd might be conducted for benchmrking resons on regulr bsis. However, the methodology cn still be improved by enhncing the scope beyond hospitl doors or by including mesures for the qulity of cre provided. Future studies in this re would lso gretly benefit from stndrdistion of costccounting systems in Europen hospitls. ACKNOWLEDGEMENTS The results presented in this rticle re bsed on the project Helth Benefits nd Service Costs in Europe HelthBASKET, which ws funded by the Europen Commission within the Sixth Frmework Reserch Progrmme (grnt no. SP21-CT ).

11 COST AND REIMBURSEMENT OF PRIMARY HIP REPLACEMENT S19 CONFLICT OF INTEREST No conflicts of interest declred. REFERENCES Adm T, Evns D Determinnts of vrition in the cost of inptient stys versus outptient visits in hospitls: multi-country nlysis. Socil Science & Medicine 63: Boznic K, Ktz P, Cisterns M et l Hospitl resource utiliztion for primry nd revision totl hip rthroplsty. The Journl of Bone nd Joint Surgery 87: Breyer F The specifiction of hospitl cost function. Helth Economics 6: Briggs A, Sculpher M, Britton A et l The costs nd benefits of primry totl hip replcement. Interntionl Journl of Technology Assessment in Helth Cre 14: Crm P, Vughn-Srrzin M, Wolf B et l A comprison of totl hip nd knee replcement in specility nd generl hospitls. The Journl of Bone nd Joint Surgery 89: Dreinho fer K, Dieppe P, Stu rmer T et l Indiction for totl hip replcement: comprison of ssessments of orthopedic surgeons nd referring physicins. Annls of the Rheumtic Diseses 65: Engeseter L, Espehug B, Lie S et l Does cement increse the risk of infection in primry hip rthroplsty? Act Orthopedic 77: Fulkner A, Kennedy L, Bxter K et l Effectiveness of hip prostheses in primry totl hip replcement: criticl review of evidence nd n economic model. Helth Technology Assessment 2: Gillespie W, Pekrsky B, O Connell D Evlution of new technologies for totl hip replcement. The Journl of Bone nd Joint Surgery 77: Grieve R, Nixon R, Thompson T et l Using multilevel models for ssessing the vribility of multintionl resource use nd cost dt. Helth Economics 14: Iorio R, Hely W, Lemos D et l Displced femorl neck frctures in the elderly. Clinicl Orthopedics nd Relted Reserch 383: Johnsson T, Bchrch-Lindstro m M, Aspenberg P et l The totl costs of displced femorl neck frcture: comprison of internl fixtion nd totl hip replcement. Interntionl Orthopedics 30: 1 6. Ju ni P, Reichenbch S, Dieppe P Osteorthritis: rtionl pproch to treting the individul. Best Prctice & Reserch in Clinicl Rheumtology 20: Kozierkiewicz A, Stmirski M, Stylo W et l The definition of prices for inptient cre in Polnd in the bsence of cost dt. Helth Cre Mngement Science 9: Lupcis A, Bourne R, Rorbeck C et l Costs of elective totl hip rthroplsty during the first yer. The Journl of Arthroplsty 9: Munin M, Rudy T, Glynn N et l Erly inptient rehbilittion fter elective hip nd knee rthroplsty. The Journl of the Americn Medicl Assocition 279: Ntionl Institute for Helth nd Clinicl Excellence Guidnce on the Selection of Prostheses for Primry Totl Hip Replcement. Ntionl Institute for Helth nd Clinicl Excellence. Ni G, Lu W, Chiu K et l Cemented or uncemented femorl component in primry totl hip replcement? A review from clinicl nd rdiologicl perspective. Journl of Orthopedic Surgery 13: Nilsdotter A, Lohmnder L Age nd witing time s predictors of outcome fter totl hip replcement for osteorthritis. Rheumtology 41: Pet G, Croft P, Hy E Clinicl ssessment of the osteorthritis. Best Prctice & Reserch in Clinicl Rheumtology 15: S nchez-mrtı nez F, Abell n-perpin n M, Mrtı nez-pe rez J et l Cost ccounting nd public reimbursement schemes in Spnish hospitls. Helth Cre Mngement Science 9: Schreyo gg J, Strgrdt T, Tiemnn O, Busse R Methods to determine reimbursement rtes for dignosis relted groups (DRG): comprison of nine Europen countries. Helth Cre Mngement Science 9: Sicilini L, Hurst J Explining witing times vritions for elective surgery cross OECD countries. OECD Helth Working Ppers, vol. 7. Singer J Using SAS PROC MIXED to fit multilevel models, hierrchicl models, nd individul growth models. Journl of Eductionl nd Behviorl Sttistics 24: Sochrt D, Porter M Long-term results of totl hip replcement in young ptients who hd Ankylosing spondylitis. Eighteen to thirty-yer results with survivorship nlysis. The Journl of Bone nd Joint Surgery 79: Twisk J Applied Multilevel Anlysis. Cmbridge University Press: Cmbridge, UK

12 S20 T. STARGARDT Upshur R, Moineddin R, Crighton E et l Sesonlity of service provision in hip nd knee surgery: possible contributor to witing times? A time series nlysis. BMC Helth Services Reserch 22: 1 5. Vitlino D On the estimtion of hospitl cost functions. Helth Economics 6: World Helth Orgniztion Europen Helth for ll Dtbse. World Helth Orgniztion.