Trialling diagnosis-related groups classification in the Iranian health system: a case study examining the feasibility of introducing casemix

Transcription

1 EMHJ Vol. 16 No Eastern Medterranean Health Journal La Revue de Santé de la Médterranée orentale Trallng dagnoss-related groups classfcaton n the Iranan health system: a case study examnng the feasblty of ntroducng casemx S. Ghaffar, 1, C. Doran, 1 A. Wlson 1 and C. Asbett 3 اختبار تصنيفي ملجموعات مرتبطة بالتشخيص يف النظام الصحي اإليراين: دراسة حالة جدوى إدخال مزيج احلاالت شهرام غفاري كريستفر دوران آندرو ويلسون كريستفر آيزبت اخلالصة: تدرس هذه املقالة جودة املعلومات التي جتمع بشكل روتيني يف إحدى املستشفيات اإليرانية يف جتربة لتصنيف مزيج احلاالت. وقد استخدم الباحثون أداة املجموعات املرتبطة بالتشخيص واملعدلة يف أستراليا لتصنيف النوبات املرضية لدى املرىض. وتعرف الباحثون عىل 37 جمموعة مرتبطة بالتشخيص منها %0 لدهيا حالة واحدة كام تعر ف برنامج حتديد املجموعات إىل سجالت غري صحيحة لدى %4 من جممل السجالت املنفصلة. وقد صن ف الباحثون %4.5 من احلاالت تقريبا ضمن املجموعات اخلاطئة املرتبطة بالتشخيص فيام كان %3.4 من احلاالت غري قابلة للتصنيف يف جمموعات. ومل يتعرفوا عىل أية مضاعفات أو مراضة مرافقة لدى %93 من جممل احلاالت. وبلغت قيمة R )التفاوت يف فترة املكث التي يمكن تفسريها( %44 يف احلاالت غري املشذ به وقد زادت لتصل إىل %63 عند تشذيبها بطريقة L3H3 وبمقدار %57 عند هتذيبها بطريقة QR1 وعند هتذيبها بالرشحية املئوية العارشة إىل 95. ABSTRACT Ths paper examnes the qualty of routnely collected nformaton n an Iranan hosptal n a tral of casemx classfcaton. Australan Refned Dagnoss Related Groups (AR-DRG) were used to classfy patent epsodes. There were 37 DRGs dentfed, of whch 0% had only 1 case. The grouper program dentfed nvald records for 4% of total separatons. Approxmately 4.5% of cases were classfed nto error DRGs and 3.4% were ungroupable. No complcaton and comorbdty effects were dentfed wth 93% of total cases. R (varance n length of stay explaned) was 44% for untrmmed cases, ncreasng to 63%, 57% and 58% after trmmng by L3H3, IQR and 10th 95th percentle methods respectvely. Essa de classfcaton par Groupes Homogènes de Malades dans le système de santé ranen : étude de cas analysant la fasablté de l ntroducton du «case-mx» RÉSUMÉ Cet artcle étude la qualté des nformatons recuelles de manère systématque dans un hôptal ranen dans le cadre d un essa de classfcaton par «case-mx». La verson australenne affnée des Groupes Homogènes de Malades (GHM) a été utlsée pour classer les épsodes clnques des patents. Au total, 37 GHM ont été détermnés, parm lesquels 0 % n ncluaent qu un seul cas. Le programme qu effectue le groupage a dentfé des enregstrements non valdes pour envron 4 % de l ensemble des séparatons. Envron 4,5 % des cas étaent classés comme erreur GHM et 3,4 % comme ngroupables. Aucune conséquence lée à des complcatons ou à une comorbdté n a été dentfée dans 93 % des cas. R (varaton de la durée de séjour explquée) état de 44 % pour les cas non classés, et passat à 63 %, 57 % et 58 % après classement à l ade des méthodes L3H3, IQR (écart nterquartle) et du 10 e au 95 e percentle, respectvement. 1 School of Populaton Health, Unversty of Queensland, Brsbane, Australa (Correspondence to S. Ghaffar: sghaffar000@yahoo.com). Socal Securty Organzaton, Tehran, Islamc Republc of Iran. 3 Laeta Pty Ltd, Randwck, New South Wales, Australa. Receved: 6/11/07; accepted: 5/05/08 460

2 املجلة الصحية لرشق املتوسط املجلد السادس عرش العدد اخلامس Introducton The Islamc Republc of Iran s preparng tself to mplement the casemx budgetng system, commonly used n many developed countres, n ts hosptals. Casemx uses the dagnoss-related group (DRG) system to classfy acute npatents and was ntally desgned for qualty assurance but s now extensvely used for fundng purposes. DRG s a system for lnkng the acute npatents that a hosptal treats to the costs ncurred by the hosptal [1]. Epsodes of care are classfed nto the dfferent DRGs accordng to the prncpal and sgnfcant secondary dagnoses, man surgcal procedure, types of separaton (patent dscharge, death or transfer), brth weght, age and sex of patent []. The Australan refned DRG (AR- DRG) s a non-propretary and welldocumented system used n Germany, Ireland, New Zealand, Sngapore, Slovena [3], and to some extent n some of our neghbourng countres such as Turkey and Saud Araba. Clncans nvolvement n the development of AR- DRG, and ts regular updatng as clncal practce changes, dstngushes t from other systems. Despte the sgnfcant mprovement acheved n some areas over past decades, the Islamc Republc of Iran as a low mddle-ncome country stll has unresolved neffcences and nequaltes n ts health care system [4]. Total health expendture per capta was estmated at US$ 498, whch s 6.5% of the total gross domestc product, n 003 [5]. The country has 1.7 beds per 1000 populaton, wth an occupancy rate of 56% n state-owned hosptals. Hosptals are funded based on annual budgetng n whch neffcency s a major problem, attrbutable to poor manageral systems. Accordng to the Iranan Natonal Health Account, hosptals consume approxmately 36% of total annual health expendture n the country [4]. However, the true level of expendture s estmated by health managers to be hgher than ths. Ths paper studes the feasblty of applyng the AR-DRG classfcaton n hosptals run by the Iranan Socal Securty Organzaton. It examnes the adequacy and qualty of routnely collected hosptal nformaton and dentfes problems assocated wth DRG classfcaton and recommends requred mprovements. The study hosptal was comparable to other larger Iranan hosptal n terms of nformaton systems. Methods Study settng Patents demographc and clncal nformaton for the year were obtaned from Kashan hosptal, a well establshed hosptal n Tehran, Islamc Republc of Iran. It has 16 beds n use and provdes both outpatent and npatent servces. Inpatent servces are provded by 1 wards, ncludng surgcal (general, eye, orthopaedc, urology and ear, note and throat) and nternal medcne, paedatrcs, maternty, coronary care and ntensve care. In , there were outpatent and npatent occasons of servces reported n ths hosptal (.e. number of admssons or number of admtted patents). Codng Whle procedures are coded by physcans n the Iranan system, clncal coders are responsble for assgnng dseases to the approprate Internatonal Classfcaton of Dsease (ICD) codes. The ICD 10th revson, Australan modfcaton (ICD-10-AM) and the ICD 9th revson, clncal modfcaton (ICD-9-CM) codes are used for codng dseases and procedures respectvely. A mappng algorthm was used to map Iranan procedures nto the Australan verson. The code mappng was smlar to that done for the evaluaton of AR- DRG for Irsh hosptals and was further refned by the Australan experence n changng from ICD-9-CM to ICD-10- AM, n changes to versons of AR-DRG and n the use of Natonal Centre for Classfcaton n Health maps between edton of ICD-10-AM [6]. The collected data were nput nto a grouper devsed by Laeta, a specalst health nformaton company. The grouper s a computer-based software program that assgns patent epsodes nto DRG classes and assesses the qualty and adequacy of the documentaton system through dentfyng nvald or mssng data, ncludng patent age, sex, length of stay, prncpal and secondary dagnoss codes, procedure codes, etc. [7]. The AR-DRG system uses 4 alphanumerc characters and classfes patent epsodes nto 665 DRGs and 3 major dagnostc categores (MDC) and error DRGs by sequental steps as follows [1]: demographc and clncal edts; assgnment of MDC usng prncple dagnoss; pre-mdc processng whch ncludes records for very hgh cost casetypes; parttonng of MDC n whch patents are classfed nto medcal, surgcal or other parttons; assgnment of adjacent-drg whch classfes patents based on the resource consumpton level; assgnment of the complcatons and comorbdty level and patent clncal complexty level; and fnally assgnment of DRG. Cases that have very hgh and varable cost or cases that cannot be classfed nto any MDC based on prncpal dagnoss are grouped nto the pre-mdc class. Coeffcent of varaton The coeffcent of varaton, whch s the standard devaton dvded by the mean, often multpled by 100 to gve a percentage [8], was used to measure the varaton n length of stay for ndvduals wthn each DRG [9]. A coeffcent of varaton less than 100 reflects acceptable wthn-group homogenety [9] and measures the meanngfulness of the classfcaton system. Reducton 461

3 EMHJ Vol. 16 No Eastern Medterranean Health Journal La Revue de Santé de la Médterranée orentale n varance (R ) was used to measure the extent to whch the dsperson of length of stay could be explaned by ths groupng. R s an overall measure of how well patents are classfed nto acceptable groups on the bass of resource consumpton [10] and how well the classfcaton system performs n our settng. Values of R range from 0 (no reducton) to 1 (perfect match). Stata, verson 9. was used to calculate the coeffcent of varaton and R. R was computed as follows [10]: where y s the value of the varable (.e. length of stay) for the th patent, A s the average value for the varable n the database and A g s the average value of the varable n DRG g. The square of the dfference between the actual (y ) and the predcted value (A or A g ) s a measure of the varaton n the data. Trmmng ( y A) ( y A) Trmmng, whch s a method of excludng outlers (unusual length of stay or cost), was appled to approxmate a normal dstrbuton. We used 3 dfferent trmmng methods to dentfy outler cases: In the L3H3 method, the low- and hgh-stay trm-ponts for every DRG equal the average length of stay for the DRG dvded and multpled by 3 respectvely [11]. In the nterquartle range (IQR) method, low and hgh trm-ponts are calculated as: Q1 1.5 (Q3 Q1) and 1.5 (Q3 Q1) + Q3 respectvely, where Q1 and Q3 refer to the 1st and 3rd quartles of the dstrbuton [1]. In the 10th and 95th percentle method, the outlers are dentfed by the 10th percentle where at least 90% of patents would have a length of stay greater than or equal to that pont [13]. The pont at whch 95% of pa- ( y A ) g tents have a length of stay less than or equal to t the 95th percentle ndcates the hgh trm-pont [14]. The patents who fall between low and hgh trm-ponts are known as nlers. The percentage of outler cases was used as an ndcator to measure the approprateness of the classfcaton algorthms and trmmng methods n our settng. Results The man fndngs of classfyng npatent occasons of servce are presented n ths paper. Further detals are avalable from the correspondng author on request. Qualty of hosptal records Table 1 provdes an overvew of the vald and nvald or mssng nformaton dentfed by the grouper. The code 00 shows a normal groupng condton (96.6%) and all other codes show some sort of problem dentfed by the grouper, ncludng mssng or nvald prncpal dagnoss (code 01), nvald age (code 04), nvald sex (code 05), nvald length of stay (code 08) and nvald same day separaton (code 09). MDC assgnment The hghest volume MDC, whch encompassed 19% of total hosptal separatons, was dseases and dsorders of the dgestve system (MDC 06) (Fgure 1). Only a few cases fell nto MDC 17 (neoplastc dseases), MDC 19 (mental dseases), MDC 0 (alcohol/drug use and alcohol/drug-nduced organc mental dsorders) and MDC 3 (factors nfluencng health status and other contacts wth health servces). There were no cases n MDC (burns). Pre-MDC processng and MDC parttonng DRG A06Z (tracheostomy or ventlaton > 95 hours) was the only DRG dentfed durng pre-mdc processng. There were no cases of lver, lung, heart or renal transplant at ths hosptal. Almost 54% of total separatons were classfed as surgcal and 46% of them were classfed nto the medcal partton. Only 18 cases fell nto the other partton (cases that had no operatng room procedure but had at least 1 nonoperatng room procedure). Table 1 Vald and nvald or mssng nformaton about cases dentfed by the grouper program at Kashan hosptal, Code Descrpton No. of cases % 00 Normal groupng, ncludng assgnment to all error DRGs except 960Z Invald or mssng prncpal dagnoss Dagnoss code cannot be used as prncpal dagnoss Record does not meet crtera for any DRG Invald age Invald sex Invald mode of separaton Invald admsson weght a 08 Invald LOS Invald same day status 4 0. Total a Admsson weght was not recorded. DRG = dagnoss-related group; LOS = length of stay. 46

4 املجلة الصحية لرشق املتوسط املجلد السادس عرش العدد اخلامس Complcaton and comorbdty level and assgnment of patent clncal complexty level The majorty of the separatons n the study hosptal (93%) were assgned a value of 0, whch means that dagnoss codes for the specfc separaton were not dentfed as complcaton and comorbdty codes or, f they were, they were closely related to the prncpal dagnoss. The remanng 7% of the patent records, wth a varyng degree of clncal complexty, were classfed nto patents clncal complexty levels 1 to 4 (Table ). Almost 63% of total separatons (7355) were dscharged wth a sngle dagnoss code and only 7% (818) of them were dscharged wth 3 or more dagnoss codes. DRG assgnment There were 37 DRGs dentfed n the study, 0% of whch had only 1 case and 47% had less than 5 cases. DRG C16A, lens procedure (8%), was the hghest volume DRG dentfed n ths study. About 3.4% of the hosptal separatons fell nto DRG 960Z as ungroupable. Error DRGs Approxmately 4.5% of separatons were classfed nto the error DRGs. DRG 960Z, whch contans records wth nvald or mssed prncpal dagnoss or other nvald essental nformaton such as age, sex or admsson weght-for-age < 1 year, comprsed 78% of the total error DRGs. DRG 901Z and 90Z (13%) nclude all procedures Table Patents clncal complexty level (PCCL) at Kashan hosptal, PCCL Descrpton No. of cases % 0 No complcaton and comorbdty effect Mnor complcaton and comorbdty effect 7 0. Moderate complcaton and comorbdty effect Severe complcaton and comorbdty effect Catastrophc complcaton and comorbdty effect 8 0. Total wth codes rrelevant to the prncpal dagnoss. DRG 961Z and 963Z (8%) nclude patent records wth unacceptable prncpal dagnoses. Length of stay The range of length of stay vared from 1 to 60 days, and the hghest proporton of patents (34%) separated on the same day or the day after admsson. Length of stay was not recorded for about % of the total separatons. Average length of stay for untrmmed cases was 3.09 and for trmmed cases were 3.06,.9 and.93 days respectvely usng the L3H3, IQR and percentle methods. Approxmately 5.%, 5.5% and 4.% respectvely of total separatons were dentfed as outlers after trmmng by L3H3, IQR, and 10th 95th percentle methods. DRG X07B (skn graft for njures) had the hghest average length of stay across all DRGs, at 3 days. Excludng DRGs wth less than 5 cases, DRG G03C (stomach, oesophageal and duodenal procedure wthout malgnancy) had the hghest average length of stay (1.4 days). The hghest volume DRG, C16A (lens procedures), had an average length of stay of 1.8 days. In general, MDC 05 wth 6.70 and MDC 14 wth 1.6 had the hghest and lowest average length of stay, respectvely. The average lengths of stay were 4.6, 4.0 and.3 days for the other, medcal and surgcal parttons, respectvely. Table 3 shows a summary dstrbuton of DRGs wth a coeffcent of varance < 100 and varance n length of stay explaned (R ) for untrmmed and trmmed data. It shows that wthngroup homogenety ncreased from 90% for untrmmed data to 99% for trmmed data by the L3H3 method. The results show that the value of R (varance n length of stay explaned), whch was 44% for untrmmed data, ncreased to 63%, 57% and 58% after trmmng by L3H3, IQR and 10th 95th percentle methods respectvely. As our objectve was not a comprehensve evaluaton of the performance of the AR-DRG Table 3 Summary of dagnoss-related groups (DRGs) wth coeffcent of varaton < 100 and varance explaned (R ), for trmmed and untrmmed data at Kashan hosptal, Trmmng method Total number of DRGs a % DRGs wth coeffcent of varaton < 100 Varance explaned (R ) Untrmmed data Trmmed by L3H Trmmed by IQR Trmmed by 10th 95th percentles a DRGs wth 1 case were excluded, as coeffcent of varaton cannot be calculated n ths case. L3H3 = long stay trm-pont 3 tmes average length of stay of DRG; IQR = nterquartle range. 463

5 EMHJ Vol. 16 No Eastern Medterranean Health Journal La Revue de Santé de la Médterranée orentale system, and our sample comprsed only 1 hosptal, we dd not go nto a detaled analyss to evaluate R for DRGs at MDC level. Dscusson Effectve mplementaton of any casemx classfcaton system requres accurate and thorough recordng and codng of patents demographc, clncal and fnancal nformaton. Although ths nformaton s usually avalable n a hosptal s dscharge system, the qualty of nformaton and ts avalablty through a computerzed system are problematc n low-resource countres. Despte a natonal movement toward usng ICD-10 codes n the Islamc Republc of Iran, many hosptals stll do not apply ICD-10 or apply t partally. Surgcal procedures n the study hosptal were coded usng ICD-9-CM. Although ICD codes were not prmarly desgned for DRG and casemx purposes, they are consdered to be the basc ngredent of casemx recpe [15]. Applyng a codng system compatble wth the verson of DRG whch s to be employed for classfcaton purposes s an mportant step toward a successful DRG classfcaton tral. In ths exploratory study, we used AR-DRG whch requres ICD-10-AM codng (the Australan verson of ICD-10). Mappng Iranan hosptal data from ICD-9 to ICD-10 and ICD-10-AM would requre addtonal effort and techncal expertse whch s probably not avalable across the country. Mode of separaton, whch s a compulsory varable for completng DRG classfcaton [16], was recorded n the hosptal but not ncluded n the software language of the grouper we employed. The complexty of the mappng process for medcal record codes arose from the nconsstency n the Iranan codng, but ths was a modest problem as we revewed only 1 hosptal. Mappng tres to facltate groupng and s usually not a serous problem when the separaton codes are used consstently. However, n the long term, t would be better to avod t through upgradng the documentaton system and/or choosng an approprate verson of DRG. The accuracy of DRG assgnment depends on the qualty of data, whch, n turn, ha a drect mpact on the usefulness of the nformaton produced by the casemx system, whether for management or fundng purposes [17]. In the Islamc Republc of Iran, procedures are recorded by physcans and matchng the procedure and dsease codes wth approprate ICD codes s the responsblty of coders. Poor codng practce, ncludng choosng the rght prncpal dagnoss and recordng all secondary dagnoses and man procedures, was the man shortcomng dentfed durng ths study. Whle the grouper has the ablty to use up to 30 dagnoss and procedure codes, no more than 4 secondary dagnoses and 3 procedures were recorded at the study hosptal. Secondary dagnoses and man procedures reflect the severty of the llnesses and are essental for correct groupng. Accuracy and completeness n documentaton and profcency of morbdty coders are essental for achevng a meanngful groupng []. Although general practtoners are employed to control codng accuracy, there s no standard qualty control to secure the accuracy and consstency of codng ether at the physcan or coder level and qualty of codng s always questonable. There are stll some coders n Iranan hosptals who have not been formally traned for codng clncal records. The ICD codng of patent clncal nformaton was 1 year behnd. Informaton was stored n out-of-date programmng languages such as DOS and FoxPro. It s dffcult to prepare data from these sources for commonly used contemporary data management software compatble wth the grouper program (such as Mcrosoft Excel or Access). Nether qualfed nor experenced staff was avalable to upgrade the systems to fully match the study requrements. Inaccurate and low qualty data results n error DRGs. We dentfed 6 error DRGs n the AR-DRG ncludng 901Z, 90Z, 903Z, 960Z, 961Z, 963Z, whch all contaned nvald or atypcal nformaton [18]. Errors n DRG classfcaton occur ether because of poor qualty data due to nvald prncpal dagnoss, mssng codes or data entry naccuraces. Other problems dentfed by the grouper arose due to nvald data entry. Not all nvald nformaton affects the groupng qualty n the same way. For example, the grouper dentfed 137 nstances of nvald age and 133 of nvald sex, whch affected groupng of 136 and 14 cases respectvely. There were also 46 nvald length of stay and 109 nvald prncpal dagnoses that affected qualty. Invald nformaton can be classfed wth a warnng and fatal flag, dependng on the sze of the effect. A fatal flag dentfes problems leadng to error DRGs 960Z, 961Z, 963Z, for nstance, conflct between prncpal dagnoss and sex for the obstetrc DRGs [1]. Dealng wth factors leadng to error DRGs s crtcal for hgh qualty classfcaton. Tranng and educaton are mportant n reducng codng errors [19] and are central to casemx mplementaton when the majorty of staff has ether no or very lmted knowledge of casemx [0]. Admsson weght and age n days for patents wth age < 1 year, whch are essental for a DRG classfcaton, were not collected n the system. The Laeta grouper default and nternatonal admsson weght growth curve data were used to classfy such patents nto these DRGs, but demographc dfferences between natons suggest the need for the development of a country-specfc classfcaton system. Leave days, whch s an optonal but useful feld n dentfyng length of stay outsde the acute settng, and a same day flag needs to be 464

6 املجلة الصحية لرشق املتوسط املجلد السادس عرش العدد اخلامس drectly recorded. Identfyng the source of admsson emergency or electve admsson also provdes useful nformaton for utlzaton revew and qualty assurance programmes [1]. Electvely admtted patents have a hgher chance of gettng standard care than emergency ones. There s no specfc code n the study hosptal to ndcate patents admsson status. MDC and DRG provde worthwhle nformaton about hosptal actvty whch s useful for polcy-makng. Informaton provded by MDC s even more worthwhle where the number of separatons fallng nto the DRG groups s small. The volume of the cases classfed nto each DRG or MDC group s drectly affected ether by the poor qualty of data or choosng an ncompatble verson of DRG. Too many DRG classes mean that there are too few observatons wthn each class, whch n turn makes t hard to understand actual varatons between hosptals. On the other hand, too few classes mean that there are too many heterogeneous cases wthn each class. In ths case, the possblty of placng large number of dssmlar cases nto 1 group wll cause a dffculty n fndng real varaton between doctors, nurses and hosptal output []. In our sngle-hosptal study 50% of the DRGs dentfed had too few observatons,.e. fewer than 5 observatons wthn each DRG. Inaccurate cost weght could be the man problem arsng due to low-volume DRGs [9], and a larger sample would be needed to estmate cost weghts. Approxmately 5% of total separatons were dentfed as outlers after trmmng by L3H3 and IQR, and 4% as outlers after trmmng by the 10th 95- th percentle method. It s normally accepted that an outler proporton of more than 10% s too hgh, reflectng ether napproprate classfcaton algorthms or trmmng problems []. The overall hgh R values of 0.63, 0.57 and 0.58, after trmmng usng L3H3, IQR and 10th 95th percentle respectvely, and the hgh proporton of DRGs wth coeffcent of varance less than 100 (more than 90%), suggest that AR-DRG provde good explanatory power and wthn-group homogenety n ths hosptal. The value of R (0.63 trmmed by L3H3) was comparable to that reported by other studes [1,3]. However, ths requres replcaton n other Iranan hosptals due to the small sample sze, the large proporton of DRGs wth only 1 case (0%) and the low qualty of the data. Concluson Our study shows that DRG classfcaton s achevable n ths hosptal run by the Iranan Socal Securty Organzaton usng routnely collected nformaton. However, to acheve a classfcaton system to nform fundng and management decsons, the followng changes would need to occur: upgradng of the current computerzed system ncludng documentaton systems at admsson and dscharge ponts; usng ICD-10 for prncpal dagnoses across the hosptal system; recordng age n days and admsson weght for patents < 1 year old; flaggng same-day separatons and recordng admsson type (emergency versus electve); and employng experenced, profcent and sklled coders. The problems dentfed durng the data edtng and classfcaton processes for ths tral are lkely to be relevant to other Iranan hosptals and other countres wth smlar hosptal documentaton nfrastructure when plannng and mplementng casemx for ether management or fundng purposes. Although the result of R was acceptable by the usual benchmarks, studes wth larger data sets and dfferent classfcaton systems are recommended f measurement of DRG performance s desred. Acknowledgements The authors gratefully acknowledge the valuable assstance of the staff at the Iranan Socal Securty Organsaton and Kashan hosptal, partcularly that of Dr Amr Abbas Manochehr, Majd Hasanan, Al Mohammda Sanjar and Aboulfazl Taher n collectng the requred nformaton. We also acknowledge the support of the Iranan Socal Securty Organzaton for gvng permsson to publsh these data. References 1. Australan refned dagnoss related groups: verson 5.1: defntons manual. Canberra, Commonwealth Department of Health and Ageng, Eagar K, Hndle D. Casemx n Australa: an overvew. Canberra, Department of Human Servces and Health, Hndle D. Implementng DRGs n Slovena: why the Australan varaton was selected. Australan health revew, 003, 6: Iran natonal health accounts. Tehran, World Bank, 004 ( accessed 9 November 009). 5. Core health ndcators database, 007. World Health Organzaton Statstcal Informaton System [onlne database] ( apps.who.nt/whoss/database/core/core_select.cfm, accessed, 9 November 009). 6. Asbett CW et al. Measurng hosptal case mx: evaluaton of alternatve approaches for the Irsh hosptal system. Dubln, Ireland, Economc and Socal Research Insttute, 007 (Paper WP19) 7. Natonal hosptal cost data collecton: cost report, round 7 (00 003). Canberra, Commonwealth Department of Health and Ageng,

7 EMHJ Vol. 16 No Eastern Medterranean Health Journal La Revue de Santé de la Médterranée orentale 8. Bland M. An ntroducton to medcal statstcs. New York, Oxford Unversty Press, Red B, Palmer G, Asbett C. The performance of Australan DRGs. Australan health revew, 000, 3: Averll R et al. The evoluton of casemx measurement usng dagnoss related groups (DRGs). Studes n health technology and nformatcs, 1994, 14: Duckett SJ. Casemx fundng for acute hosptal npatent servces n Australa. emja, 1998, 19:s17 1. Gong Z et al. Descrbng Chnese hosptal actvty wth dagnoss related groups (DRGs): a case study n Chengdu. Health polcy, 004, 69: Publc hosptal cost benchmarks Techncal paper. Brsbane, Queensland Health, Cole S, Stomfay B. R.I.P L3H3? In: Proceedngs of the 15th Casemx Conference n Australa. Sydney, Commonwealth Department of Health and Aged Care, Roberts RF, Innes KC, Walker SM. Introducng ICD-0-AM n Australan Hosptals. Medcal journal of Australa, 1998, 169:S Asbett C. Access grouper: a user frendly mplementaton of AR- DRG, verson 5.0. Sydney, Laeta, Red B, Palmer GR, Asbett C. Under-codng n Australa lmts the performance of DRG groupers. Health nformaton management, 1999/000, 9: Australan refned dagnoss related groups, verson 5.0. Defntons manual. Canberra, Commonwealth Department of Health and Ageng, 00. Hay PJ, Pearce T. Casemx fundng n psychatry: Some problems and common ptfalls. Australan health revew, 1996, 19: Ghaffar S, Doran C, Wlson A. Casemx n the Islamc Republc of Iran: current knowledge and atttudes of health care staff. Eastern Medterranean health journal, 008, 14(4): Lchtg LK. Hosptal nformaton systems for casemx management. New York, Wley, Palmer G, Red B. Evaluaton of the performance of dagnossrelated groups and smlar casemx systems: methodologcal ssue. Health servces management research, 001, 14: Jackson T. ANDRG3 and ARDRG4: how do they compare on resource homogenety? In: Proceedngs of the Annual Conference of Patent Classfcaton System/Europe. Gronngen, Netherlands, Patent Classfcaton System/Europe, 000. Management Effectveness Intatves The management of health care s a pvotal factor n the delvery of effectve health servce wth growng recognton of the key role that non-clncal actvtes play n the way that health care s delvered. Management effectveness s crucal n all health care settngs: hosptals, prmary health care clncs, moble unts, laboratores and pharmaces. The WHO Regonal Offce for the Eastern Medterranean (EMRO) works n partnershp wth mnstres of health of the Regon to strengthen the way n whch health care facltes and professonals are managed. The ultmate am s to mprove ther functonng by workng towards greater effectveness, effcency, qualty and coverage of servces whch lead to better health outcomes. EMRO offers techncal assstance n developng key health management tools and approaches. Further nformaton on Management Effectveness Intatves n EMRO can be found at: 466