Differences in prevlence of pre-existing moridity etween injured nd non-injured popultions Cte M. Cmeron, 1, 2 Dvid M. Purdie, 3 Erich V. Kliewer, 4 & Roderick J. McClure 1 Ojectives To identify nd exmine differences in pre-existing moridity etween injured nd non-injured popultion-sed cohorts. Methods Administrtive helth dt from Mnito, Cnd, were used to select popultion-sed cohort of injured people nd smple of non-injured people mtched on ge, gender, originl sttus nd geogrphicl loction of residence t the dte of injury. All individuls ged 18 64 yers who hd een hospitlized etween 1988 nd 1991 for injury (Interntionl Clssifiction of Diseses, Ninth Edition, Clinicl Modifiction (ICD-9-CM) code 800 995) (n = 21 032), were identified from the Mnito dischrge dtse. The mtched non-injured comprison group comprised individuls rndomly selected 1:1 from the Mnito popultion registry. Moridity dt for the 12 months prior to the dte of the injury were otined y linking the two cohorts with ll hospitl dischrge records nd physicin clims. Results Compred to the non-injured group, injured people hd higher Chrlson Comoridity Index scores, 1.9 times higher rtes of hospitl dmissions nd 1.7 times higher rtes of physicin clims in the yer prior to the injury. Injured people hd rte of dmissions to hospitl for mentl helth disorder 9.3 times higher, nd physicin clims for mentl helth disorder 3.5 times higher, thn tht of non-injured people. These differences were ll sttisticlly significnt (P < 0.001). Conclusion Injured people were shown to differ from the generl non-injured popultion in terms of pre-existing moridity. Existing popultion estimtes of the ttriutle urden of injury tht re otined y extrpolting from oserved outcomes in smples of injured cses my overestimte the mgnitude of the prolem. Keywords Wounds nd injuries/epidemiology/complictions; Comoridity; Cuslity; Helth services/utiliztion; Cost of illness; Retrospective studies; Cohort studies; Cnd (source: MeSH, NLM). Mots clés Plies et trumtismes/épidémiologie/compliction; Moridité ssociée; Cuslité; Services snté/utilistion; Coût mldie; Etude rétrospective; Etude cohorte; Cnd (source: MeSH, INSERM). Plrs clve Herids y lesiones/epidemiologí/complicciones; Comorilidd; Cuslidd; Servicios de slud/utilizción; Costo de l enfermedd; Estudios retrospectivos; Estudios de cohortes; Cndá (fuente: DeCS, BIREME). Bulletin of the World Helth Orgniztion 2005;83:345-352. Voir pge 351 le résumé en frnçis. En l págin 351 figur un resumen en espñol. Introduction WHO hs predicted tht injury will e the second leding cuse of the world disese urden y the yer 2020 (1, 2). A limittion in the current process of deriving popultion estimtes of the urden ttriutle to injury is the filure to tke into ccount pre-existing moridity. If injured people differ from the generl popultion in terms of pre-existing moridity, then oserved outcomes in injured smples tht re ttriuted to injury my in prt e due to pre-existing moridities rther thn to the injury in question. The Austrlin Burden of Disese study cknowledged the importnce of co-existing conditions in estimting the ttriutle urden of prticulr conditions (3). Mthers et l. (3) concluded tht severl methodologicl issues relting to comoridity remin to e ddressed if urden of disese models re to e dvnced. These issues include how comoridities ffect long-term disility; which comoridities re relevnt; nd how to del with the logistics of modelling lrge numers of comintions of comoridities (3). Although some ttempts hve een mde to look t differences in helth sttus in ptients pre- nd post-injury using self-reported retrospective recll (4), investigtors using this method cknowledge inevitle ises (5 9). Bis is etter mnged y scertining pre-injury moridity t point in time efore the injury ws sustined. Although rre in injury outcome studies (10, 11), the use of comoridity indices sed on dministrtive clims dt recorded prior to the index event is well estlished in other fields of reserch such s cncer, crdiovsculr disese nd dietes (12 14). 1 School of Popultion Helth, University of Queenslnd, Myne Medicl School, Herston Rod, Herston, Brisne, Queenslnd, Austrli 4006. Correspondence should e sent to Dr Cmeron t this ddress (emil: cmeronc@uq.edu.u). 2 Centre of Ntionl Reserch on Disility nd Rehilittion Medicine, University of Queenslnd, Queenslnd, Austrli. 3 Northern Cliforni Cncer Center, Cliforni, USA. 4 CncerCre Mnito, Mnito Helth, Winnipeg, Cnd. Ref. No. 04-011130 (Sumitted: 6 Jnury 2004 Finl revised version received: 13 July 2004 Accepted: 10 Novemer 2004) Bulletin of the World Helth Orgniztion My 2005, 83 (5) 345
Reserch Glol urden of disese estimtes currently ssume tht the distriution of moridity in the community is independent of injury sttus. Outcome studies of clinicl cse series cnnot test this ssumption nd few popultion-sed studies tht compre injured nd non-injured people hve een reported in the literture (10). The present study used dministrtive helth dtses to compre the frequency nd distriution of moridity in people in the 12 months prior to their sustining n injury with the helth sttus of the generl non-injured popultion. Methods Study design The study descried in this pper is n exmintion of the prevlence of pre-existing disese in two smples drwn from dministrtive helth dt from Mnito, Cnd. The two smples were originlly identified on the sis of exposure to injury, for the purposes of conducting mtched, popultionsed retrospective cohort study, with follow-up period of 10 yers. The University of Mnito Reserch Ethics Bord nd the Helth Informtion Privcy Committee of Mnito Helth pproved this study. Dt extrctions were completed y Mnito Helth nd ll identifying vriles were removed from the dt efore the study investigtor ws grnted ccess to them. Dt sources The province of Mnito provides universl helth-cre coverge for popultion of 1.14 million residents (15). Mnito Helth mintins dtses of clims mde y helth providers for reimursement of services (hospitl, physicin nd extendedcre services), s well s popultion registry of those eligile for helth coverge (16). Virtully every resident of Mnito is covered y the provincil helth-cre pln (15). The dtses hve een used extensively in helth reserch nd re descried in detil elsewhere (17, 18). Setting nd prticipnts A cohort of injured people (n = 21 032) ws identified s ll persons ged 18 64 yers resident in the province of Mnito, who hd een hospitlized for tretment of n injury etween 1 Jnury 1988 nd 31 Decemer 1991. The cohort memers included ll individuls who hd n Interntionl Clssifiction of Diseses, Ninth edition, Clinicl Modifiction (ICD-9-CM) code 800 995 (excluding lte effects from injury 905 909, nd llergies from within 995), in the first or second dignostic fields. For individuls who hd more thn one injury-relted hospitl dmission during the study period, the first dmission ws designted s the index cse record. For ech injured suject, non-injured person ws rndomly selected from the Mnito popultion registry nd mtched on ge, gender, originl sttus nd geogrphicl loction of residence t the dte of dmission of the index cse. Excluded from oth the injured nd non-injured cohorts were residents of nursing homes, ptients in extended hospitl cre nd people who hd not een resident in the province for 12 months prior to the dmission dte of the index record (for the purposes of extrcting seline dt). Helth service utiliztion nd mesurement of comoridity Informtion on helth service utiliztion ws extrcted from hospitl dischrge dt nd physicin services clims for oth Cte M. Cmeron et l. the injured nd non-injured cohorts. Dtes of dmissions, services, tretment, dignostic nd dischrge informtion were extrcted for the 12-month period prior to the dte of the index record for ll mtched pirs (19). Physicin clims were restricted to multory clims, including outptient, emergency deprtment (where ville) nd visits to medicl prctitioners in the community (20, 21). For individuls with multiple hospitl records ssocited with single episode (e.g. when interhospitl trnsfers or redmissions hd tken plce), they were integrted into one summry record. Frequency nd types of pre-existing comorid conditions for the two cohorts were determined from the extrcted hospitl nd physicin clims during the pre-injury period. For the purpose of identifying pre-existing comoridities, only the primry dignostic field ws used, nd conditions were ctegorized ccording to the 18 disese chpters of ICD-9-CM. The numer of comorid conditions ws counted y summing the numer of different disese ctegories present. Dt on comorid conditions from physicin clims were derived using the sme method. The Drtmouth-Mnito version of the Chrlson Comoridity Index (CCI) (22) ws lso used to quntify preexisting comoridity. The CCI ws computed for the injured cohort sed on 12 months of hospitl dt for the yer preceding the index injury record. For the non-injured cohort, the CCI ws clculted for the 12-month period prior to dte of injury of the mtched cse. If no comorid conditions were found, or where no hospitl records existed, the CCI score ws set to zero. Injury clssifiction The injured were nlysed s the totl cohort nd y the nture of injury codes (ICD-9-CM 800 995). Seven sugroups (rin injury, spinl injury, urns, long-one frctures, poisonings, internl injuries nd other) were creted cross suchpter hedings to enle more comprehensive exmintion of those injury types commonly studied in injury outcomes reserch. ICDMAP-90 softwre from Johns Hopkins University ws used to generte n Injury Severity Score (ISS) for injured cses. These ISS scores were grouped into minor (ISS 1 8), moderte (ISS 9 15) nd severe (ISS 16) ctegories in ccordnce with the convention (23). ICDMAP-90 mps severity score for only proportion of the totl Injury nd Poisonings ICD-9-CM codes. Therefore not ll injured cses were scored. Anlysis Anlysis of dt involved univrite nd ivrite sttistics. The sttisticl significnce of differences etween groups nd sugroups ws ssessed y chi-squred sttistics for ctegoricl dt nd with the Mnn Whitney U test for continuous dt ecuse of non-norml distriutions. Rte rtio confidence intervls were djusted for mtching vriles using Poisson regression methods. All tests were two sided with 5% level of significnce. Rtes of helth service utiliztion were reported using person-yers (PYs) of exposure time. Anlysis ws conducted using SAS version 8.2 sttisticl softwre. Results Smple chrcteristics The chrcteristics of the 21 032 injured cses re shown in Tle 1. The men ge t the strt of follow-up for oth injured nd non-injured cohorts ws 35.7 yers. Mles represented 346 Bulletin of the World Helth Orgniztion My 2005, 83 (5)
Cte M. Cmeron et l. lmost two-thirds of the cohort nd more injuries occurred in individuls in the younger ge ctegories (18 34 yers). Of the injury sugroups, frctures of long ones (12%), poisonings (10.3%) nd rin injury (6.1%) were the most common. An ISS ws generted for 62% of the totl injured (n = 17 002). Over 85% of scored cses were of minor severity, 10% of moderte severity nd lmost 4% were clssified s mjor injuries. Pre-existing moridity for injured smple Memers of the injured cohort hd 1.9 times higher rte of hospitl dmissions in the 12 months efore the dte of the index record thn memers of the non-injured cohort (injured 276/1000 PYs nd non-injured 148/1000 PYs). The men numer of hospitl dmissions per person in the pre-injury yer ws 0.12 in the injured (rnge 0 19) nd 0.09 in the non-injured cohort (rnge 0 12). The men length of sty in hospitl differed significntly etween the two cohorts in the pre-injury yer (Tle 2). Injured people were less likely to e dmitted s dy ptients (28.6% versus 36.3%) nd more likely to hve totl length of sty greter thn 14 dys (13.3% versus 5.7%). The injured cohort hd 1.7 times higher rte of physicin clims in the pre-injury period (670/100 PYs, verge 3.9 clims per person) thn the non-injured cohort (387/100 PYs, verge 2.2 clims per person). A greter percentge of the injured cohort (65%) thn of the non-injured cohort (48%) hd three or more physicin clims in the 12-month pre-injury period (Tle 2). Significntly more people from the injured cohort (5.9%) thn from the non-injured cohort (1.2%) hd CCI scores of 1 or more (Tle 3). Overll, greter percentge of injured people hd comoridities, nd they hd more comoridities per person thn the memers of the non-injured cohort, sed on hospitliztions nd physicin clims. Injured people hd n verge of 2.2 different conditions (rnge 0 14) for which they hd consulted physicin in the pre-injury yer, wheres in the non-injured people this figure ws 1.5 (rnge 0 13). There were 1498 people in the injured group nd 423 in the non-injured cohort, who hd moderte or severe pre-existing mentl helth condition, s indicted y helth service use in the pre-injury period. The memers of the injured cohort hd higher rtes of hospitl dmissions nd physicin clims for ll cuses in the pre-injury period thn those in the non-injured cohort (Tle 4 nd Tle 5). The rte rtios were 1.5 or greter for 14 of the 18 chpters for hospitl dmissions nd for six chpters for physicin clims. When rnked y rte rtio, for oth hospitl dmissions nd physicin clims, the gretest differences were seen for mentl helth disorders nd previous injuries. Injured people hd 9.3 times higher rte of dmissions to hospitl for mentl helth disorder (injured 42/1000 PYs nd non-injured 4.5/1000 PYs) nd 3.5 times higher rte of mentl helth physicin clims (injured 98/100 PYs) thn the non-injured people (28/100 PYs). Mentl disorders were the most frequent cuse of ll hospitl dmissions for the totl injured cohort. Almost hlf of the mentl helth dmissions were for lcoholic psychoses, ffective psychoses nd schizophrenic disorders. Over 80% of ll mentl helth physicin clims for the injured cohort were for non-psychotic or personlity disorders, more specificlly for such conditions s pnic, nxiety or depression. During the pre-injury period, the injured cohort hd 3.7 times higher rte of dmissions to hospitl for previous injury (injured 9.9/1000 PYs nd non-injured 2.7/1000 PYs) Reserch Tle 1. Demogrphic chrcteristics for injured nd noninjured cohorts, injury sugroups nd injury severity scores t the time of the cse index injury dmission Injured Non-injured (n = 21 032) (n = 21 032) n % n % Gender Mle 13 441 63.9 13 441 63.9 Femle 7 591 36.1 7 591 36.1 Age in yers 18 24 5 410 25.7 5 422 25.7 25 34 6 014 28.6 5 990 28.6 35 44 3 959 18.8 3 972 18.8 45 54 2 805 13.3 2 799 13.3 55 64 2 844 13.5 2 849 13.5 Plce of residence Urn 8 687 41.3 8 799 41.9 Rurl 8 167 38.8 8 208 39.0 Remote 4 178 19.9 4 025 19.1 Injury sugroups Brin injury 1 290 6.1 Spinl injury 95 0.5 Burns 524 2.5 Frctures of long ones 2 515 12.0 Poisonings 2 169 10.3 Internl injuries 593 2.8 Other injuries 13 846 65.8 Injury Severity Score (ISS) Minor (ISS 1 8) 14 599 85.9 Moderte (ISS 9 15) 1 746 10.3 Severe (ISS 16) 657 3.8 Age-mtched on yer of irth nd plce of residence on prtil postcode, thus there re smll differences in ctul numers of injured nd non-injured. ISS not computed for 4030 cses. nd 2.7 times higher rte of physicin clims for injury-relted mtters (117/100 PYs) thn the non-injured group (43/100 PYs). Nineteen per cent of the previous injury-relted dmissions for the injured cohort (nd 10% for the non-injured) were for self-inflicted hrm or suicide ttempts y poisoning. Nineteen per cent of the previous physicin injury clims for the injured group (nd 28% for the non-injured) were for sprins nd strins. Pre-existing moridity y injury type Helth service use in the 12-month pre-injury period ws similr cross the injury types nlysed. However, memers of the injured cohort hospitlized for poisonings were notle in tht they only ccounted for 10% of the injured, ut ccounted for 24% of ll dmissions nd 20% of ll physicin clims in the pre-injury period. Furthermore, this group hd 59 times higher rte of hospitl dmissions (95% confidence intervl (CI), 27.75 123.64) nd n 11 times higher rte of physicin clims (95% CI, 10.38 12.0) for mentl helth disorders thn their mtched counterprts in the uninjured group. Over 50% of the hospitl dmissions for mentl helth prolems nd 78% of physicin clims for the poisonings group in the pre-injury Bulletin of the World Helth Orgniztion My 2005, 83 (5) 347
Reserch Cte M. Cmeron et l. Tle 2. Hospitl dmissions, length of sty nd physicin clims in the 12-month period prior to the index dmission for injured nd non-injured cohorts Helth service use for the 12 months prior to injury Injured (21 032) Non-injured (21 032) Significnce n % n % Totl hospitl dmissions 5 804 100 3 111 100 Admissions per person No dmissions 17 316 82.33 18 665 88.75 1 dmission 2 569 12.21 1 879 8.93 P < 0.0001 2 dmissions 698 3.32 360 1.71 3 dmissions 236 1.12 78 0.37 4 or more dmissions 213 1.01 50 0.24 Men numer of dmissions (rnge) c 0.12 (0 19) 0.09 (0 12) P < 0.0001 Totl length of sty in hospitl (dys) Dy ptient 1 062 28.58 859 36.29 1 2 625 16.82 363 15.34 3 4 575 15.47 473 19.98 P < 0.0001 5 7 543 14.61 329 13.90 8 14 416 11.19 208 8.79 15 29 281 7.56 79 3.34 30 dys or more 214 5.76 56 2.37 Men length of sty in dys (rnge) c 0.36 (0 248) 0.18 (0 258) P < 0.0001 Totl physicin clims 140 986 100 81 302 100 Clims per person No clims 2 746 13.06 5 137 24.42 1 2 clims 4 685 22.28 5 864 27.88 3 5 clims 5 328 25.33 5 082 24.16 P < 0.0001 6 10 clims 4 342 20.64 3 246 15.43 11 20 clims 2 736 13.01 1 396 6.64 21 50 clims 1 064 5.06 286 1.36 51 or more clims 131 0.62 21 0.10 Men numer of clims (rnge) c 3.9 (0 295) 2.2 (0 127) P < 0.0001 Determined y χ 2 sttistics. Determined y Mnn Whitney U test. c Geometric men clculted due to non-norml distriutions. yer were for non-psychotic, neurotic or personlity disorders often for specific nxiety or depressive disorders. The poisonings group hd significntly more hospitl dmissions in the pre-injury yer for previous injuries nd poisonings (rte rtio (RR) = 11.5; 95% CI, 4.14 31.95), circultory diseses (RR = 6.2; 95% CI, 3.22 11.52) nd ill-defined conditions (RR = 4.7; 95% CI, 2.82 7.26) thn the mtched sugroup. This injury group lso hd higher rtes of physicin clims for ill-defined conditions (RR = 3.4; 95% CI, 3.12 3.60), neoplsms (RR = 2.9; 95% CI, 2.27 3.77), digestive diseses (RR = 2.7; 95% CI, 2.39 2.96), thn the mtched sugroup. Excluding poisonings from the nlysis hd little effect on the sttisticl significnce of differences etween helth service utiliztion y the memers of the injured nd non-injured cohorts during the pre-injury period. Discussion To the est of our knowledge, this is the first study to quntify difference in pre-existing moridity etween injured nd non-injured popultions. Compred to non-injured people, injured people were lmost five times s likely to hve CCI score of one or more, were dmitted to hospitl t lmost twice s often, nd hd 1.7 times the rte of physicin clims. The injured cohort hd fewer memers with no comorid conditions nd significntly more memers with greter numers of different conditions present thn did the mtched smple from the generl popultion. The results of this study support the concern tht current urden of injury estimtes my e inccurte if the influence of pre-existing ill-helth is not tken into ccount. Although few studies reported in the literture hve een similr in scope nd methods to the current study (3), one key popultion-sed study of elderly people with nd without hip frctures, did find similr results (10). This study found tht, prior to their injury, the study sujects hd significntly higher comoridities using the CCI, higher disility mesures nd were more likely to e residents of nursing homes thn controls (10). Severl studies of clinicl injury outcome hve mesured comoridities using numer of different methods, mostly relying on the dignostic fields of the index dmission record (7, 8, 24). These studies hve found consistent ssocition etween the presence of comorid conditions nd incresed risk of mortlity following injury. Smller nd less consistent ssocitions were found in the few studies tht hve considered comoridity nd non-ftl outcomes (25 27). 348 Bulletin of the World Helth Orgniztion My 2005, 83 (5)
Cte M. Cmeron et l. Reserch Tle 3. Comorid conditions in the 12-month period prior to the index dmission for injured nd non-injured cohorts Comorid conditions for the 12 months prior to injury Injured (21 032) Non-injured (21 032) Significnce n % n % Chrlson Comoridity Index No comoridity 19 797 94.12 20 778 98.79 P < 0.0001 1 or more comoridity 1 235 5.87 254 1.21 Comorid conditions per person (sed on hospitl dmissions) No comoridities 17 316 82.33 18 665 88.75 1 comoridity 3 004 14.28 2 110 10.03 P < 0.0001 2 comoridities 571 2.71 224 1.06 3 or more comoridities 141 0.67 33 0.16 Men numer of conditions (rnge) c 0.15 (0 6) 0.09 (0 6) P < 0.0001 Comorid conditions per person (sed on physicin clims) No comoridities 2 746 13.06 5 137 24.42 1 comoridity 3 838 18.25 4 750 22.58 2 comoridities 4 018 19.10 3 959 18.82 P < 0.0001 3 comoridities 3 341 15.89 2 904 13.81 4 comoridities 2 604 12.38 1 944 9.24 5 or more comoridities 4 485 21.32 2 338 11.12 Men numer of conditions (rnge) c 2.2 (0 14) 1.5 (0 13) P < 0.0001 Determined y χ 2 sttistics. Determined y Mnn-Whitney U test. c Geometric men clculted due to non-norml distriutions. Tle 4. Hospitl dmissions per 1000 person-yers for ll ICD-9-CM disese chpters during the 12-month period prior to the injury, for the injured nd mtched non-injured cohort, rnked y rte rtios Rte of hospitl dmissions per 1000 PYs ICD-9-CM chpters Injured Non-injured Adjusted rte rtio 95% confidence n = 21 032 n = 21 032 intervls Mentl helth disorders 42.03 4.52 9.31 7.53 11.50 Injury nd poisonings 9.98 2.71 3.68 2.73 4.90 Blood diseses 1.76 0.52 3.36 1.71 6.57 Endocrine nd metolic 7.56 2.71 2.79 2.05 3.75 Musculoskeletl diseses 18.59 7.13 2.61 2.16 3.15 Nervous system diseses 12.31 5.23 2.35 1.88 2.94 Ill-defined conditions 18.50 8.04 2.30 1.91 2.74 Respirtory diseses 16.78 8.46 1.98 1.65 2.36 Circultory diseses 13.12 7.70 1.70 1.40 2.07 Supplementry clssifiction 15.12 9.56 1.58 1.32 1.89 Congenitl nomlies 0.81 0.52 1.55 0.73 3.31 Skin diseses 12.41 8.04 1.54 1.27 1.87 Digestive diseses 32.05 21.11 1.52 1.34 1.71 Infectious diseses 3.57 2.38 1.50 1.04 2.13 Genitourinry diseses 22.97 15.74 1.46 1.26 1.67 Complictions of pregnncy 36.80 33.00 1.12 1.01 1.24 c Neoplsms 11.60 10.56 1.10 0.91 1.32 Conditions of perintl period 0.00 0.00 NA NA PYs = person yers. Adjusted for ge, gender nd plce of residence. c Undjusted confidence intervl. Bulletin of the World Helth Orgniztion My 2005, 83 (5) 349
Reserch Cte M. Cmeron et l. Tle 5. Physicin clims per 100 person-yers for ll ICD-9-CM disese chpters during the 12-month period prior to the injury, for the injured nd mtched non-injured cohort, rnked y rte rtios Rte of physicin clims per 100 PYs ICD-9-CM chpters Injured Non-injured Adjusted rte rtio 95% confidence n = 21 032 n = 21 032 intervls Mentl helth disorders 98.15 28.05 3.50 3.40 3.61 Injury nd poisonings 117.41 43.23 2.72 2.65 2.78 Ill-defined conditions 61.21 33.72 1.82 1.76 1.87 Musculoskeletl diseses 63.74 36.18 1.76 1.71 1.81 Digestive diseses 30.29 18.72 1.62 1.55 1.68 Blood diseses 3.48 2.27 1.53 1.36 1.72 Nervous system diseses 43.69 30.85 1.42 1.37 1.46 Congenitl nomlies 0.67 0.47 1.41 1.09 1.83 Respirtory diseses 71.26 51.52 1.38 1.35 1.42 Endocrine nd metolic 23.67 17.14 1.38 1.32 1.44 Genitourinry diseses 40.66 29.67 1.37 1.33 1.42 Skin diseses 33.17 24.85 1.33 1.29 1.38 Infectious diseses 15.37 11.72 1.31 1.24 1.38 Neoplsms 5.75 4.52 1.27 1.17 1.39 Circultory diseses 29.11 24.10 1.21 1.16 1.25 Supplementry clssifiction 31.49 28.44 1.11 1.07 1.15 Complictions of pregnncy 1.20 1.08 1.11 0.92 1.32 c Conditions of perintl period 0.03 0.04 0.75 0.26 2.18 PYs = person-yers. Adjusted for ge, gender nd plce of residence. c Undjusted confidence intervl. This study identified two res of importnce for estimting the ttriutle urden of injury. The first is tht injured people hve significntly more hospitl dmissions nd physicin clims prior to their injury, cross lmost every disese clssifiction, thn do the generl popultion. The differentil ws most pronounced for hospitliztions for which 14 of the 18 chpters hd rte rtios of 1.5 or greter, wheres six of the 18 chpters for physicin clims hd rte rtios of 1.5 or greter. When rnked y the rte rtios, two disese chpters, mentl helth disorders nd previous injuries, consistently hd the highest rtes of hospitl dmission nd physicin clims. The finding regrding mentl helth is consistent with previous study of trum ptients, which found tht more thn 30% of trum ptients were ffected y psychitric moridity efore the trum event, nd tht psychitric conditions represented 18% of the totl moridity (9). Outcome studies of the mentl helth consequences of injury should consider controlling for pre-existing conditions efore scriing mentl helth disorders s consequence of the injury. Second, there ws no single injury sugroup tht ccounted for the overll differences in the prevlence of preexisting disese. However, the poisonings sugroup hd greter solute nd reltive moridity in lmost ll ICD-9-CM disese chpters nd hd the highest rte rtio for mentl helth dmissions when compred with the findings in non-injured counterprts. The poisonings sugroup tended to e younger, femle nd hd mde greter use of helth services for tretment of non-psychotic, neurotic or personlity disorders. Nevertheless, excluding this group from the nlysis hd little effect on the overll findings. Helth service use following n dmission for poisoning, if considered solely s consequence of tht injury dmission does not tke ccount of the high seline service use tht would most likely hve continued irrespective of the poisoning event. Aspects of the methodology need to e considered s possile study limittions. First, helth service use ws used s proxy for helth sttus. Although previous studies conducted in this study popultion hve supported the vlidity of this prctice (21, 28), it does entil severl ssumptions out the equity of ccess to helth services, ptterns of referrl, nd service delivery eing dependent on helth sttus nd consistent cross ll susections of the community. Second, the generlizility of the findings my e compromised y injured cses eing defined s those hospitlized during the study period. This excludes injuries tht did not led to hospitliztion nd pre-hospitl ftlities, nd my e ised y issues of service ccess nd referrl pthwys (29). Third, is the question of whether incident cses could e ccurtely identified from the dministrtive dtses used to select the injured cohort. Of the 21 032 people in the injured smple only 60 hd een dmitted in the previous 12 months for n injury with the sme ICD-9 injury group code. Of these, 24 were cses of poisoning with repet dmissions mostly occurring fter numer of months, nd re likely to hve een new episodes rther thn repet dmissions for the originl injury, leving only 36 cses (i.e. 0.17%) which could potentilly hve een misclssified s incident cses. Finlly, exmintion of the six ctegories of injury type, while justified on the sis tht they re the groups most relevnt to injury outcomes reserch, ccounts for only 35% of ll the cses. Further rekdown of the other ctegory for the purposes of nlysis creted numer of smll ctegories nd did not contriute informtion relevnt to the ims of the study. 350 Bulletin of the World Helth Orgniztion My 2005, 83 (5)
Cte M. Cmeron et l. There re mny strengths which set this study prt from previous studies tht hve included considertion of comoridity in their exmintion of injury outcomes. The injured group in this study is popultion-sed incidence smple nd s such, voids the ises inherent in most clinicl study smples. This is one of the few studies to hve included popultion-sed non-injured comprison group. The smple size of the injured group ws sufficiently lrge to enle considerle degree of nlysis of injury sugroups, which will provide rich context for future work in the field. Of principl vlue hs een the ility to mesure helth sttus prior to injury in the injured smple, nd previous helth sttus in the noninjured group. Through the use of dministrtive dt, nlysis of the 12-month pre-injury period circumvents the recll is noted previously, which hs hmpered ttempts to identify pre-injury Reserch helth sttus. This study hs shown tht clrifying the contriution of pre-existing moridity, to enle the refined clcultion of popultion ttriutle-urden estimtes, is vitlly importnt in urden-of-disese pproches to pulic helth policy. O Acknowledgements We grtefully cknowledge the provision of the dt used in this study y Mnito Helth, Winnipeg, Cnd. We lso wish to thnk Mr André Wjd for his invlule ssistnce with dt extrction nd project consulttion. This reserch ws funded y the Centre of Ntionl Reserch on Disility nd Rehilittion Medicine, University of Queenslnd, Austrli. Competing interests: none declred. Résumé Comprison de l moridité pré-existnte entre des popultions de personnes ynt été victimes d un trumtisme et ne l ynt ps été Ojectif Identifier et étudier les différences de moridité préexistnte entre des cohortes en popultion de personnes ynt Résultts Pr rpport u groupe de personnes n ynt ps sui de remoursement de soins médicux. été victimes d un trumtisme et ne l ynt ps été. trumtisme, les individus victimes d un trumtisme présentient Méthodes Les uteurs ont utilisé les données snitires dministrtives des vleurs plus fortes de l indice de comoridité de Chrlson, du Mnito, Cnd, pour sélectionner une cohorte en popultion de des tux d hospitlistion 1,9 fois plus élevés et des tux de personnes ynt sui un trumtisme et un échntillon de personnes remoursement de soins médicux 1,7 fois plus élevés dns l nnée n en ynt ps sui, ppriées selon l âge, le sexe, l origine utochtone précédnt le trumtisme. Les personnes ynt sui un trumtisme ou non et le lieu de résidence à l dte du trumtisme. Ils ont présentient un tux d hospitlistion pour troules mentux 9,3 fois identifié, à prtir de l se de données de déchrge hospitlière du plus élevé et des tux de demnde de remoursement concernnt Mnito, tous les individus âgés de 18 à 64 ns ynt fit l ojet d une des troules mentux 3,5 fois plus élevés que les personnes n ynt hospitlistion pour trumtisme entre 1988 et 1991 [Clssifiction ps été victimes d un trumtisme. Ces différences étient toutes interntionle des mldies, neuvième édition, Modifiction clinique sttistiquement significtives (p < 0,001). (ICD-9-CM) codes 800 à 995] (n = 21 032). Les uteurs vient Conclusion Les uteurs ont montré que les personnes victimes de constitué le groupe de comprison, composé de personnes n ynt trumtisme différient de l popultion générle n ynt ps sui ps sui de trumtisme ppriées, en choisissnt u hsrd, dns un de trumtisme pr l moridité pré-existnte. Les estimtions en rpport 1/1, des individus dns le registre de popultion du Mnito. popultion disponiles de l chrge de trumtisme ttriule à Ils ont otenu les données de moridité pour les 12 mois précédnt l moridité pré-existnte, otenues pr extrpoltion des résultts l dte du trumtisme en mettnt en reltion les deux cohortes vec oservés sur des échntillons de cs de trumtismes, peuvent l ensemle des registres de déchrge hospitlière et des demndes de surestimer l mpleur du prolème. Resumen Diferencis en l prevlenci de morilidd previ entre polciones de trumtizdos y no trumtizdos Ojetivo Identificr y nlizr ls diferencis en l morilidd con todos los registros de lts hospitlris y ls fcturs de preexistente entre cohortes polcionles de persons trumtizds los médicos. y no trumtizds. Resultdos En comprción con el grupo de no trumtizdos, Métodos Se usron los dtos dministrtivos snitrios de pr ls persons trumtizds se hllron puntuciones Mnito (Cndá) pr seleccionr un cohorte polcionl myores del Índice de Comorilidd de Chrlson, tss de ingreso de persons trumtizds y un muestr de persons no hospitlrio 1,9 veces myores, y tss de fcturción de médicos trumtizds preds por edd, sexo, originlidd y 1,7 veces superiores durnte el ño previo l trumtismo. Ls lugr geográfico de residenci en l fech del trumtismo. Se persons trumtizds presentn tss de ingreso hospitlrio identificó en l se de dtos de lts de Mnito todos por trstornos de slud mentl 9,3 veces superiores, y en su cso los individuos de 18 64 ños que hín sido hospitlizdos ls fcturs médics por tles trstornos ern 3,5 veces myores entre 1988 y 1991 cus de trumtismos (Clsificción que ls de los no trumtizdos. Tods ess diferencis ern Interncionl de Enfermeddes, 9ª revisión, Modificción estdísticmente significtivs (P < 0,001). Clínic (CIE-9-MC) códigos 800 995) (n = 21 032). El grupo de Conclusión Ls persons trumtizds diferín de l polción comprción de no trumtizdos predos est integrdo no trumtizd en cunto l morilidd preexistente. Ls por individuos selecciondos l zr, en proporción 1:1, prtir ctules estimciones polcionles de l crg triuile de del registro de polción de Mnito. Se otuvieron los dtos trumtismos, otenids extrpolndo prtir de los resultdos de morilidd correspondientes los 12 meses previos l fech oservdos en muestrs de persons trumtizds, podrín estr del trumtismo, pr lo cul se relcionron ls dos cohortes sorestimndo l mgnitud del prolem. Bulletin of the World Helth Orgniztion My 2005, 83 (5) 351
Reserch Cte M. Cmeron et l. References 1. Murry CJ, Lopez AD. The glol urden of disese: comprehensive ssessment of mortlity nd moridity from diseses, injuries nd risk fctors in 1990 nd projected to 2020. Cmridge (MA): Hrvrd School of Pulic Helth; 1996. 2. Stone DH, Jrvis S, Pless B. The continuing glol chllenge of injury. BMJ 2001;322:1557-8. 3. Mthers CD, Vos ET, Stevenson CE, Begg SJ. The urden of disese nd injury in Austrli. Bulletin of the World Helth Orgniztion 2001;79:1076-84. 4. Amertung SN, Norton RN, Whitlock G, Mcmhon S, Coggn, C, Jckson RT, et l. The New Zelnd Blood Donors Helth Study: seline findings of lrge prospective cohort study of injury. Injury Prevention 2002;8:66-9. 5. Whitlock JA, Hmilton BB. Functionl outcome fter rehilittion for severe trumtic rin injury. Archives of Physicl Medicine nd Rehilittion 1995; 76:1103-12. 6. Morris JA Jr, McKenzie EJ, Edelstein SL. The effect of preexisting conditions on mortlity in trum ptients. JAMA 1990; 263:1942-6. 7. Scco WJ, Copes WS, Bin LW Jr, McKenzie EJ, Frey CF, Hoyt DB, et l. Effect of preinjury illness on trum ptient survivl outcome. Journl of Trum 1993; 35:538-42; discussion 542-3. 8. Milzmn DP, Boulnger BR, Rodriguez A, Soderstrom CA, Mitchell KA, Mgnnt CM. Pre-existing disese in trum ptients: predictor of fte independent of ge nd injury severity score. Journl of Trum 1992;32:236-43; discussion 243-4. 9. Wrdle TD. Co-morid fctors in trum ptients. British Medicl Bulletin 1999;55:744-56. 10. Leison CL, Tosteson AN, Griel SE, Rnsom JE, Melton LJ. Mortlity, disility, nd nursing home use for persons with nd without hip frcture: popultion-sed study. Journl of the Americn Geritrics Society 2002;50:1644-50. 11. Rozentl TD, Brns CC, Bozentk DJ, Beredjiklin PK. Survivl mong elderly ptients fter frctures of the distl rdius. Journl of Hnd Surgery 2002;27:948-52. 12. Klunde CN, Wrren JL, Legler JM. Assessing comoridity using clims dt: n overview. Medicl Cre 2002;40 Suppl:IV 26-35. 13. Reid BC, Alerg AJ, Klssen AC, Rozier RG, Grci I, Winn DM, et l. A comprison of three comoridity indexes in hed nd neck cncer popultion. Orl Oncology 2002;38:187-94. 14. Schneeweiss S, Mclure M. Use of comoridity scores for control of confounding in studies using dministrtive dtses. Interntionl Journl of Epidemiology 2000;29:891-8. 15. Bernstein CN, Blnchrd JF, Leslie W, Wjd A, Yu BN. The incidence of frcture mong ptients with inflmmtory owel disese. A popultionsed cohort study. Annls of Internl Medicine 2000;133:795-9. 16. Young TK, Kliewer E, Blnchrd J, Myer T. Monitoring disese urden nd preventive ehvior with dt linkge: cervicl cncer mong Aoriginl people in Mnito, Cnd. Americn Journl of Pulic Helth 2000;90:1466-8. 17. Roos LL, Nicol JP. A reserch registry: uses, development, nd ccurcy. Journl of Clinicl Epidemiology 1999;52:39-47. 18. Burchill C, Roos LL, Fergusson P, Jemni L, Turner K, Dueck S. Orgnizing the present, looking to the future: n online knowledge repository to fcilitte collortion. Journl of Medicl Internet Reserch 2000;2:E10. 19. Finn J, Holmn D, Jcos I. The use of linked mulnce dt to estimte the effect of comoridity on determinnts nd outcomes of out-of-hospitl crdic rrest in Perth, Western Austrli. Symposium on Helth Dt Linkge: Its vlue for Austrlin helth policy development nd policy relevnt reserch, Sydney, Austrli, 20 21 Mrch, 2002. Adelide, South Austrli: Adelide University; 2002. p. 72-77. Aville from: http:// www.pulichelth.gov.u/symposium.htm 20. Ttryn DJ, Roos NP, Blck CD. Utiliztion of physicin resources for multory cre. Medicl Cre 1995;33 Suppl:DS84-99. 21. Cohen MM, McWillim L. Mesuring the helth of the popultion. Medicl Cre 1995;33 Suppl:DS21-42. 22. Roos LL, Strnc L, Jmes RC, Li J. Complictions, comoridities, nd mortlity: improving clssifiction nd prediction. Helth Services Reserch 1997;32:229-38; discussion 239-42. 23. Stevenson M, Segui-Gomez M, Lescohier I, Di Scl C, McDonld-Smith G. An overview of the Injury Severity Score nd the New Injury Severity Score. Injury Prevention 2001;7:10-3. 24. Morris JA Jr, McKenzie EJ, Dmino AM, Bss SM. Mortlity in trum ptients: the interction etween host fctors nd severity. Journl of Trum 1990;30:1476-82. 25. Rochon PA, Ktz JN, Morrow LA, McGlinchey-Berroth R, Ahlquist MM, Srkrti M, et l. Comorid illness is ssocited with survivl nd length of hospitl sty in ptients with chronic disility. A prospective comprison of three comoridity indices. Medicl Cre 1996;34:1093-101. 26. Richmond TS. An explntory model of vriles influencing postinjury disility. Nursing Reserch 1997;46:262-9. 27. Nordin M, Hieert R, Pietrek M, Alexnder M, Crne M, Lewis S. Assocition of comoridity nd outcome in episodes of nonspecific low ck pin in occuptionl popultions. Journl of Occuptionl nd Environmentl Medicine 2002;44:677-84. 28. Roos NP, Blck CD, Frohlich N, Decoster C, Cohen MM, Ttryn DJ, et l. A popultion-sed helth informtion system. Medicl Cre 1995;33 Suppl:DS13-20. 29. Lngley J, Stephenson S, Cryer C, Bormn B. Trps for the unwry in estimting person sed injury incidence using hospitl dischrge dt. Injury Prevention 2002;8:332-7. 352 Bulletin of the World Helth Orgniztion My 2005, 83 (5)