The association of socioeconomic disadvantage with postoperative complications after major elective cardiovascular surgery

Transcription

1 Researh report An additional appendix is published online only at jeh.bmj.om/ontent/vol62/ issue10 1 Epidemiology Department, Loal Health Authority Rome E, Rome, Italy; 2 Epidemiology Department, Harvard Shool of Publi Health, Boston MA, USA; 3 Epidemiology Unit, Loal Health Authority, Milan, Italy; 4 Regional Health Servie, Emilia Romagna, Bologna, Italy; 5 Epidemiology Unit, Piedmont Region, Turin, Italy; 6 Epidemiology Unit, Loal Health Authority, La Spezia, Italy; 7 Epidemiology Unit, Loal health Authority, Bologna, Italy Correspondene to: Dr N Agabiti, Department of Epidemiology, Loal Health Authority Rome E, Via di S Costanza Rome, Italy; agabiti@asplazio.it Members of the Italian Study Group on Inequalities in Health Care: Carla Anona, Valeria Belleudi, Giulio Bugarini, Giulia Cesaroni, Marina Davoli, Patrizia Shifano, Carlo A. Perui, (Roma), Luigi Bisanti, Antonio Russo, Carlo Zohetti (Milano), Niola Carani, Claudia Galassi, Stefano Mattioli (Bologna), Giuseppe Costa, Chiara Marinai, Teresa Spadea (Torino), Patrizia Vittori (Valle d Aosta). Aepted 1 January 2008 This paper is freely available online under the BMJ Journals unloked sheme, see jeh.bmj.om/info/unloked.dtl The assoiation of soioeonomi disadvantage with postoperative ompliations after major eletive ardiovasular surgery N Agabiti, 1 G Cesaroni, 1 S Piiotto, 1,2 L Bisanti, 3 N Carani, 4,5 G Costa, 5 F Forastiere, 1 C Marinai, 5,6 P Pandolfi, 7 A Russo, 3 C A Perui, 1 on behalf of the Italian Study Group on Inequalities in Health Care ABSTRACT Bakground: Understanding the mehanism by whih both patient- and hospital level fators at in generating disparities has important impliations for liniians and poliy-makers. Objetive: To measure the assoiation between soioeonomi position (SEP) and postoperative ompliations after major eletive ardiovasular proedures. Design: Multiity hospital-based study. Subjets: Using Hospital Disharge Registries (ICD-9-CM odes), patients were identified undergoing five ardiovasular operations (oronary artery bypass grafting (CABG), valve replaement, arotid endarteretomy, major vasular bypass, repair of unruptured abdominal aorta aneurysm (AAA repair)) in four Italian ities, Measures: For eah patient, a five-level median inome index by ensus blok of residene was alulated. Inhospital 30-day mortality, ardiovasular ompliations (CCs) and non-ardiovasular ompliations (NCCs) were the outomes. Odds ratios (ORs) were estimated with multilevel logisti regression adjusting for ity of residene, gender, age and omorbidities taking into aount hospital and individual dependenies. Main results: In-hospital 30-day mortality varied by type of surgery (CABG 3.7%, valve replaement 5.7%, arotid endarteretomy 0.9%, major vasular bypass 8.8%, AAA repair 4.0%). Disadvantaged people were more likely to die after CABG (lowest vs highest inome OR 1.93, p trend 0.023). For other surgeries, the relationship between SEP and mortality was less lear. For ardia surgery, SEP differenes in mortality were higher for publily funded patients in low-volume hospitals (lowest vs highest inome OR 3.90, p trend 0.039) than for privately funded patients (OR 1.46, p trend 0.444); however, the differene in the SEP gradients was not statistially signifiant. Conlusions: Disadvantaged people seem partiularly vulnerable to mortality after ardiovasular surgery. Efforts are needed to identify strutural fators that may enlarge SEP disparities within hospitals. Soioeonomi disadvantage is assoiated with higher prevalene of ardiovasular risk fators, 1 2 morbidity and mortality from ardiovasular disease, 3 4 redued aess to speialist are, 5 6 and less uptake of appropriate treatment and proedures. 7 8 Inidene of postoperative ompliations after oronary artery bypass graft surgery (CABG) an effetive treatment for severe ishaemi heart disease is higher in deprived patients Some evidene exists on assoiation between rae/ethniity and worse prognosis after valve surgery and other ardiovasular surgery To our knowledge no studies have expliitly investigated the influene of soioeonomi position (SEP) on postoperative ompliations after ardiovasular operations other than CABG in ontexts outside the USA. The relative ontributions of patient and hospital fators may have important impliations for addressing soial disparities in health outomes. 14 Raial minorities are more likely to be treated by lower quality providers for CABG and for other surgeries. In the USA uninsured people and blak people are less likely to reeive are for omplex surgery at high-volume hospitals, where hospital volume is aepted as a strutural proxy for quality On the other hand, the SEP gradient for hospital mortality among eletive surgery patients treated in intensive are units (ICUs) was partially explained by diagnosti delays and severe omorbidity although no evidene of aess to lower quality ICUs emerged. 19 We aimed to evaluate the extent to whih SEP is assoiated with the ourrene of postoperative ompliations after major eletive ardiovasular proedures and to determine whether any assoiation differs by type of surgery. Among ardia surgery patients, we also examined whether hospital strutural harateristis at as effet modifiers of the assoiation. METHODS Soure of data and ohort seletion We examined hospital reords of residents in four Italian ities, Rome, Milan, Turin and Bologna, between 1997 and Disharge abstrats are routinely olleted by Regional Information Systems and ontain patient soiodemographi data, inluding ensus blok (CB) of residene, date of admission and disharge, up to six disharge diagnoses (International Classifiation of Disease, 9th revision, Clinial Modifiation (ICD-9-CM)), up to six hospital proedures (ICD-9-CM), two of them with date, type of disharge (alive, dead, transferred to other hospital). In order to define single episodes of are, we traed patients who were transferred to other hospitals and assessed patient disharge status at the end of the episode. We onsidered all reords between 1 July 1997 and 30 September 2000 of 882 J Epidemiol Community Health 2008;62: doi: /jeh

2 Table 1 Charateristis of patients by type of surgery, four Italian ities, Coronary artery bypass grafting Valve replaement patients aged years undergoing isolated ardiovasular operations (CABG, valve replaement, arotid endarteretomy, major vasular bypass, and repair of unruptured abdominal aorta aneurysm (AAA repair)). Eah ondition has been seleted on the basis of speifi ICD-9-CM proedure odes resulting in five separate ohorts. For the AAA repair ohort, we exluded those patients a with a diagnosis of aorti aneurysm dissetion. Few patients (3.5% of the total) with multiple episodes were represented multiple times. Details and odes are reported in the web-only Appendix. Comorbidities Following the enhaned Elixhauser AHRQ-Web-ICD-9-CM oding algorithm, 20 we defined eight seleted omorbidities that an play a role in the outomes of surgery: ardia and irulatory disease, vasular disease (inluding erebrovasular), hypertension, pulmonary disease, renal disease, liver disease, Carotid endarteretomy Major vasular bypass Disharges Subjets Subjets with multiple disharges City of residene Bologna Milan Rome Turin Soiodemographi harateristis (%) Women years age years age Area-based inome index (quintiles) I high II III IV V low Comorbidities (%) Cardia and irulatory disease Vasular disease inluding erebrovasular Hypertension Pulmonary disease Diabetes Renal disease Liver disease Tumours Organisational harateristis of hospital are (%) Publi: high volume Publi: low volume Private Outomes (%) In-hospital 30-day mortality Cardiovasular ompliations Non-ardiovasular ompliations Repair of unruptured abdominal aorta aneurysm Researh report tumours. For eah surgery, we did not onsider as omorbidity the diagnosis that ould reflet the primary surgial indiation. We identified onditions on the basis of ICD-9-CM odes registered in hospital admissions in the previous 6 months. Details and odes are reported in the web-only Appendix. Soioeonomi position We used a small area inome index based on the ensus blok (CB) of residene, developed by the Italian Study Group on Inequalities in Health Care, and already used in other settings Briefly, the ities were divided into CBs with the mean number of inhabitants per CB ranging from 200 in Bologna to 500 in Rome. A reord linkage between the 1998 Tax Register and the Population Registers onneted family status information to inome data for eah subjet, then the net family inome and equivalised per apita inome were alulated. Data were aggregated at the CB level, and the median value for eah J Epidemiol Community Health 2008;62: doi: /jeh

3 Researh report Table 2 Charateristis of patients by inome level for the five different surgeries Area-based inome index (quintiles) I high II III IV V low Coronary artery bypass grafting Disharges (n) Women (%) years (%) years (%) Hypertension (%) Pulmonary disease (%) Diabetes (%) Valve replaement Disharges (n) Women (%) years age (%) years age (%) Hypertension (%) Pulmonary disease (%) Diabetes (%) Carotid endarteretomy Disharges (n) Women (%) years (%) years (%) Hypertension (%) Pulmonary disease (%) Diabetes (%) Major vasular bypass Disharges (n) Women (%) years (%) years (%) Hypertension (%) Pulmonary disease (%) Diabetes (%) Repair of unruptured abdominal aorta aneurysm Disharges (n) Women (%) years (%) years (%) Hypertension (%) Pulmonary disease (%) Diabetes (%) CB was obtained; quintiles of the inome distribution by CB were alulated for eah ity (I, high inome; V, low inome). Using the CB at disharge, the index was attributed to all individuals under study. Hospital level measures For CABG and valve ohorts, we studied two organisational harateristis of hospital servies: hospital payment struture and volume of surgery. In Italy universal aess and omprehensive overage exist in a publily funded health system. 23 Aording to payment struture of the hospital where the intervention was performed, hospitals were lassified as follows: (1) publi, referring to publily finaned hospital are; (2) private, referring to private are provision. The first group inludes both publi and private hospitals with full-ost reimbursement and the seond group inludes all hospitals in whih patients must pay (partially or fully). In the publi group we also distinguished two ategories of proedure volume for ardia surgery: (1a) publi low volume, (1b) publi: others. We determined hospital proedure volumes by alulating the total number of speifi proedures in our sample for eah hospital in the 4-year study period. Aording to their volume, we ontrasted those hospitals under the lowest 20th perentile of distribution (publi low volume) with all others (publi: others). The proedures ut-off points were 50 per year for CABG, and 43 per year for valve replaement. Outome measures The main outome was in-hospital mortality within 30 days after the intervention. Less than 2% of all reords (with the exeption of 4% for endarteretomy) were missing the date of major intervention. In those ases, we alulated mortality within 30 days from the date of admission plus the mean waiting time for the seleted proedure. We defined two groups of ompliations as seondary outomes: (1) ardiovasular ompliations (CCs) inluding aute myoardial infartion, arrhythmia, ardiogeni shok, ardia arrest, erebrovasular ompliations, arterial aute diseases, and (2) non-ardiovasular ompliations (NCCs) inluding ompliations of anaesthesia, deubitus uler, sepsis, deep vein thrombosis/pulmonary embolism, foreign body left during proedure, seleted infetions due to medial are, pneumonia, postoperative haemorrhage or haematoma, postoperative physiologial and metaboli derangement, postoperative wound dehisene, transfusion reation, gastrointestinal ompliations. We identified these onditions by ICD-9-CM odes in the index (any position) or in any potential subsequent admission within 30 days after surgery (main diagnosis) on the basis of the oding algorithm for Patient Safety Indiators reently developed by the US Ageny for Health Care Researh and Quality. 24 Definitions and details on ICD-9-CM odes are reported in the web-only Appendix. Statistial analysis To examine the assoiations between inome and outomes, we estimated odds ratios (OR) separately for the five ohorts. In order to take into aount the hierarhial struture of the data, we performed multilevel modelling using mixed logit models with random interepts for hospitals and individuals and fixed interepts for ities of residene. Individual ovariates were age (linear), gender, omorbidities and inome. Bakward stepwise proedures were used to disard those variables that were not assoiated with the speifi outome (p.0.20). Quintiles of inome were onsidered as a ategorial variable using the first quintile (high inome) as the referene group, and p values for linear trend were alulated using the Wald test. Unobserved hospital level fators affeting treatment an be related to patient harateristis. 25 We ould not exlude the possibility that disadvantaged people in our sample were systematially admitted to hospitals that provided low quality of are. Therefore, as an additional analysis, we performed a single-level logisti model with a fixed effet for hospital (with dummy variables for all hospitals) and all the other ovariates. In that way, we tried to isolate the within-hospital SEP effet on the outomes. In this ase, we adjusted estimates of variane for lustering and used robust estimates. Finally, we hypothesised that the inome/mortality relationship may vary aording to hospital strutural harateristis. Therefore, to further larify the role of hospital level fators, we tested the assoiation between inome and mortality among 884 J Epidemiol Community Health 2008;62: doi: /jeh

4 Table 3 Assoiation between inome and in-hospital 30-day mortality after eletive ardiovasular surgery Area-based inome index (quintiles) I high II III IV V low Coronary artery bypass grafting % OR* (95% CI) (0.77 to 1.91) 1.42 (0.91 to 2.22) 0.89 (0.55 to 1.43) 1.93 (1.23 to 3.05) OR{ (95% CI) (0.74 to 1.72) 1.26 (0.84 to 1.89) 0.84 (0.54 to 1.30) 1.69 (1.15 to 2.48) Valve replaement % OR* (95% CI) (0.56 to 2.20) 1.34 (0.70 to 2.56) 1.41 (0.74 to 2.69) 1.65 (0.86 to 3.18) OR{ (95% CI) (0.57 to 2.05) 1.28 (0.70 to 2.36) 1.30 (0.72 to 2.37) 1.52 (0.83 to 2.77) Carotid endarteretomy % OR* (95% CI) (0.98 to 7.75) 2.46 (0.87 to 6.91) 0.87 (0.26 to 2.88) 0.97 (0.30 to 3.10) OR{ (95% CI) (0.96 to 9.50) 2.61 (0.84 to 8.09) 0.69 (0.18 to 2.59) 0.96 (0.27 to 3.43) Major vasular bypass % OR* (95% CI) (0.05 to 3.21) 0.71 (0.11 to 4.73) 0.35 (0.05 to 2.72) 0.51 (0.08 to 3.47) OR{ (95% CI) (0.26 to 1.37) 0.81 (0.37 to 1.76) 0.49 (0.21 to 1.12) 0.58 (0.26 to 1.29) Repair of unruptured abdominal aorta aneurysm % OR* (95% CI) (1.03 to 3.80) 1.40 (0.70 to 2.80) 1.48 (0.74 to 2.95) 2.03 (1.03 to 3.97) OR{ (95% CI) (0.90 to 3.62) 1.24 (0.61 to 2.55) 1.22 (0.58 to 2.58) 1.68 (0.82 to 3.43) *OR, two-level logisti regression (hospital and subjet). OR adjusted for ity of residene, gender, age and omorbidities. {OR single level logisti regression. OR adjusted for ity of residene, gender, age, omorbidities and hospital. those with CABG and valve replaement stratifying by organisational harateristis of hospital servies. Effet modifiation was tested using an interation term in the regression model and the likelihood ratio test. A sensitivity analysis was performed for the ity of Rome to evaluate whether the results would be different using as outome 30-day mortality regardless of the plae of death, in ase of an assoiation between inome and plae of death. We obtained vital status-linking reords to the regional mortality information system. In addition, a final sensitivity analysis was onduted exluding all those patients with repeated episodes of are. Datasets were prepared using ORACLE Database 10g, and all statistial analysis was performed using the software STATA version 10. All tests of signifiane are at the 5% level, and all p values reported are two sided. RESULTS Charateristis of patients by type of surgery and by SEP The distribution of surgeries in the ities reflets the number of inhabitants (table 1). The largest ohort was CABG and the smallest was major vasular bypass. Exept for valve replaement, the ardiovasular operations were more ommon among men. Among those with valve replaement and CABG there was a higher prevalene of young people, whereas the opposite held among those with arotid endarteretomy and AAA repair. In all the ohorts with the exeption of AAA repair the proportion of people in the lowest inome levels was higher than 20%. The prevalene of omorbidities and the inidene of outomes varied widely aross the ohorts. Lower inome was assoiated with higher proportions of women (with exeption of major vasular bypass) and of younger people (with the exeption of AAA repair) (table 2). The prevalene of omorbidities was generally higher in lowinome groups, partiularly among ardia patients. Researh report p Trend Assoiation between SEP and outomes after surgery For CABG there was evidene of an assoiation between inome and mortality (OR 1.93 lowest vs highest inome; p trend 0.023). For valve replaement, suh an assoiation was suggested (OR 1.65, p trend 0.090). In the AAA repair ohort, an inrease in the risk of mortality was observed (OR 2.03), although the trend was not statistially signifiant. No signifiant assoiations were found for arotid endarteretomy and major vasular bypass. When we adjusted for the hospital effet using a single-level analysis with a fixed effet for hospitals, the assoiations between inome and mortality persisted (for CABG, OR 1.69, p value 0.046) indiating that the estimated SEP disparity reflets within-hospital differenes among low- and high-inome patients (table 3). There was no evidene of statistially signifiant assoiation between inome and CCs and NCCs in any ohort (table 4). Cardia patients were treated in 27 publi low-volume (proedure volume range in the 4-year period: CABG 1 194, valve 1 65); 10 publi: others (proedure volume range: CABG , valve ) and 8 private hospitals. Mortality tended to be lower in private than in publi hospitals (2.6% vs 4.6%, OR 0.54, 95% CI 0.40 to 0.71 adjusted for ity of residene, age, gender and omorbidities). The inome/mortality assoiation was stronger among ardia patients in publi hospitals (lowest vs highest inome OR 1.65, p trend 0.022) in omparison with those in private are (OR 1.46, p trend 0.444) (table 5). However, a formal test for interation was not statistially signifiant (p = 0.465). This assoiation was even higher for those patients treated in publi low-volume hospitals (OR 3.90, p trend 0.039). Again, a formal test for interation was not statistially signifiant (p = 0.230). In all analyses, omorbidities seemed not to onfound the assoiations. In the sensitivity analysis onduted on the data set of Rome, similar results were found using 30 day mortality regardless of the plae of death. Finally, very similar results were obtained when we exluded those patients with multiple episodes. J Epidemiol Community Health 2008;62: doi: /jeh

5 Researh report Table 4 Assoiation between inome and outomes after eletive ardiovasular surgery Cardiovasular ompliations Non-ardiovasular ompliations Area-based inome index (quintiles) Area-based inome index (quintiles) I high II III IV V low p Trend I high II III IV V low p Trend Coronary artery bypass grafting % OR 1 (95% CI) (0.71 to 1.47) 1.02 (0.67 to 1.38) 0.96 (0.85 to 1.72) 1.21 (0.96 to 1.69) (0.81 to 1.61) 0.99 (0.70 to 1.41) 0.90 (0.63 to 1.28) 1.02 (0.72 to 1.44) OR 2 (95% CI) (0.62 to 1.23) 0.93 (0.67 to 1.30) 0.89 (0.64 to 1.24) 1.08 (0.79 to 1.49) (0.81 to 1.53) 0.97 (0.70 to 1.33) 0.88 (0.63 to 1.22) 0.99 (0.71 to 1.36) Valve replaement % OR* (95% CI) (0.64 to 1.91) 0.93 (0.54 to 1.59) 1.02 (0.60 to 1.72) 1.54 (0.94 to 2.53) (0.63 to 1.91) 0.93 (0.54 to 1.61) 1.19 (0.70 to 2.02) 1.26 (0.73 to 2.16) OR{ (95% CI) (0.63 to 1.88) 0.88 (0.51 to 1.51) 0.97 (0.57 to 1.65) 1.47 (0.89 to 2.42) (0.62 to 1.72) 0.89 (0.54 to 1.47) 1.13 (0.70 to 1.83) 1.18 (0.72 to 1.92) Carotid endarteretomy % OR* (95% CI) (0.58 to 1.59) 1.23 (0.76 to 2.00) 1.07 (0.66 to 1.72) 1.00 (0.62 to 1.61) (0.39 to 1.95) 0.86 (0.39 to 1.89) 0.84 (0.39 to 1.81) 1.02 (0.49 to 2.14) OR{ (95% CI) (0.56 to 1.36) 1.09 (0.71 to 1.66) 0.93 (0.61 to 1.42) 0.88 (0.58 to 1.33) (0.51 to 1.85) 0.95 (0.50 to 1.79) 0.93 (0.50 to 1.72) 0.99 (0.54 to 1.82) Major vasular bypass % OR* (95% CI) (0.03 to 1.39) 0.36 (0.07 to 1.89) 1.26 (0.25 to 6.29) 0.79 (0.17 to 3.73) (0.14 to 0.95) 0.88 (0.43 to 1.79) 0.61 (0.28 to 1.31) 0.54 (0.25 to 1.19) OR{ (95% CI) (0.28 to 1.14) 0.59 (0.31 to 1.13) 0.99 (0.52 to 1.89) 0.84 (0.44 to 1.59) (0.17 to 0.74) 0.80 (0.45 to 1.43) 0.56 (0.30 to 1.02) 0.58 (0.33 to 1.01) Repair of unruptured abdominal aorta aneurysm % OR* (95% CI) (0.75 to 2.35) 1.10 (0.61 to 1.99) 1.36 (0.77 to 2.39) 1.56 (0.88 to 2.75) (0.54 to 1.29) 0.92 (0.60 to 1.42) 1.07 (0.70 to 1.62) 0.78 (0.49 to 1.22) OR{ (95% CI) (0.65 to 2.22) 1.01 (0.54 to 1.89) 1.16 (0.63 to 2.12) 1.38 (0.76 to 2.49) (0.50 to 1.22) 0.88 (0.57 to 1.36) 0.96 (0.63 to 1.47) 0.72 (0.44 to 1.15) *OR two-level logisti regression (hospital and subjet). OR adjusted for ity of residene, gender, age and omorbidities. {OR single-level logisti regression. OR adjusted for ity of residene, gender, age, omorbidities and hospital. 886 J Epidemiol Community Health 2008;62: doi: /jeh

6 Table 5 Assoiation between inome and in-hospital 30-day mortality after CABG and valve replaement, by organisational harateristis of hospital are Area-based inome index (quintiles) I high II III IV V low p Trend Publi: low volume n (% death) 282 (2.1) 280 (4.3) 323 (3.4) 333 (3.3) 345 (6.1) OR (95% CI) (0.83 to 7.95) 1.71 (0.56 to 5.22) 1.79 (0.59 to 5.40) 3.90 (1.23 to 12.4) Publi: high volume n (% death) 1042 (4.4) 1300 (4.1) (5.4) (3.9) (6.0) OR (95% CI) (0.55 to 1.37) 1.17 (0.76 to 1.80) 0.80 (0.51 to 1.25) 1.39 (0.92 to 2.12) Overall publi n (% death) 1324 (3.9) (4.1) (5.0) (3.8) (6.0) OR (95% CI) (0.69 to 1.58) 1.26 (0.85 to 1.88) 0.91 (0.60 to 1.37) 1.65 (1.12 to 2.43) Private n (% death) 590 (1.5) 460 (2.8) 469 (3.0) 432 (3.0) 387 (2.8) OR (95%CI) (0.73 to 4.21) 1.66 (0.70 to 3.96) 1.83 (0.75 to 4.43) 1.46 (0.57 to 3.72) Two-level logisti regression (hospital and subjet). OR adjusted for ity of residene, gender, age and omorbidities. DISCUSSION This study desribes the ourrene of postoperative ompliations after major ardiovasular surgery in a large multiity Italian sample. Soioeonomi disadvantage is assoiated with worse outomes after ardia surgery, but no lear indiation of a similar relationship for vasular surgery has been found. Disparities in outomes after ardia surgery are strongest in publi low-volume hospitals. There is still little researh on the relationship between SEP and healthare outomes and the results are inonsistent. Our study onfirms previous results for CABG and ontributes to the knowledge on other types of ardiovasular proedures Higher deprivation sore has been assoiated with younger age, more omorbidities, and more postoperative ardiovasular ompliations after CABG It is of note that the majority of available studies have foused on raial disparities among patients over 65 years in the USA, whereas less information is available on other indiators of soial status It has been suggested that disadvantaged people who underwent surgery might have higher baseline risks than well-off patients, resulting in worse prognosis. The greater presene of omorbid onditions in low-inome people in our study supports this hypothesis. Lak of knowledge about the proedure and its benefits might be higher among soial disadvantaged groups. In the USA blak people tend to postpone eletive surgery and experiene more advaned disease In our study, evidene of disparity in outomes is less lear for proedures other than CABG; although results are not statistially signifiant, possible assoiations annot be exluded. The reason is diffiult to explain. Heterogeneous SEP differenes in the prevalene of omorbidities among the different ohorts support the hypothesis that poor people undergoing ardia surgery had worse baseline onditions than poor people with vasular surgery in omparison with their respetive rih ounterparts. For that reason low SEP patients who had ardia surgery would have been more suseptible to ompliations than low SEP vasular patients. Alternatively, ardia surgery itself may present more dangers to those who are vulnerable than does vasular surgery. On the other hand, ardia and vasular surgery tend to share many proesses of are related to anaesthesia, intensive are and postoperative are. Finally, unknown individual (ie, perioperative risk) or system fators (ie, surgeon s speialty or skill) that we were not Researh report able to measure in this study might have ontributed to our findings. Few studies have examined the role of hospital level fators on soial disparities in health outomes, and those available have mainly evaluated rae/ethniity. Higher rates of surgial ompliations among blak people are largely explained by differential use of high-quality hospitals However, a large proportion of raial differenes in post-proedure mortality has been found to be unrelated to hospital volume, an aepted marker of quality. Similarly, differenes in health outomes after aute myoardial infartion aording to SEP have been found to be mainly explained by ardiovasular risk fators than by aess to appropriate are, adding more to the debate on mehanisms of disparities. 27 We onsidered both patient and hospital level harateristis by applying reently developed methods and we suggest that hospital strutural fators (in our ase publi low-volume hospitals) may at as effet modifiers. Better organisation and proesses of are, more skilled surgeons and adequate numbers of nurses in private failities ould be partially responsible for the observed homogeneous outomes aross soial groups. However, the lower mortality among those treated in private than in publi hospitals in our study suggests less severe preoperative status, the main risk fator for outome after surgery Worse preoperative status among patients in publi hospitals ould be related to the shorter waiting times for surgery in privately funded hospitals. 37 Its different distribution among patients in publi and private hospitals ould partially explain the results. The population-based design, the number of operations and the validated algorithms are the strengths of this study. Our study is one of the first studies in Europe to test the feasibility of AHRQ ICD-9-CM-based surgial are indiators, although administrative datasets have been previously used to identify postoperative ompliations. 38 Despite their wide use as a valuable soure for healthare researh, hospital disharge data have several limitations, whih have been repeatedly reognised. 39 In our ase, the datasets do not inlude information on relevant prognosti fators suh as operative priority. Underreporting is also possible, as proven by a validation study of CABG patients in Rome. 40 However, it is unlikely that different reporting aross hospitals and mislassifiation errors of omorbidity or ompliations are assoiated with SEP. It is more probable, in ontrast, that true inidene of ompliations and their severity may be higher than reported, weakening the J Epidemiol Community Health 2008;62: doi: /jeh

7 Researh report What is already known on this subjet Disparity in health outomes is a omponent of the omplex piture of inequity in health. Disadvantaged people tend to experiene worse outomes after are even in ountries with universal overage. Few studies have examined the role of hospital level fators on soial disparities in outomes after surgery, and those available have mainly evaluated rae/ethniity. What this study adds Poor people who undergo ardia surgery are more likely to die after operation than their rih ounterparts in Italy; for vasular surgery, a relationship between soioeonomi position (SEP) and outomes is also suggested. The assoiation between SEP and outomes after ardia surgery is partiularly evident in low-volume publi hospitals whereas differenes narrow in private hospitals. Interventions aimed at improving the quality of surgial are espeially where the are is suboptimal are reommended as a means to redue health disparities. evidene for existing soioeonomi disparities. Different omorbidity measurements based on ICD-9-CM or ICD-10- CM ode have been tested in ardiovasular researh despite their lower ability in prediting outomes in omparison to linial data; the Elixhauser method allows a omprehensive definition of omorbidity and has proven superior to others in prediting mortality after seleted onditions. 43 Some limitations should be underlined. First, we are onerned that our measure of volume might not aurately reflet experiene with the proedure beause it is based on the studied ohorts. It is likely, however, that our surrogate measure is well related to the true proedure volume. Seond, attributing an aggregated indiator may not aurately represent the individual s true SEP, and the assoiation may be distorted. 44 Our area-based inome index inludes eonomi resoures provided by work, pension, real estate and investments whih ontribute to defining the material well-being and standards of living of all family members. However, inome might not be a good marker for other soial onditions relevant for health, and risk fators suh as oupation or level of eduation. 45 Although we did not have information on other important soial determinants of health, other studies have shown that even after adjustment for suh variables, the assoiation between eonomi resoures and health persists. 46 Moreover, at the small area level, the preditive power of eonomi poverty indiators has been showed to be omparable with that related to omposite SEP indies. 47 We observed a orrelation oeffiient higher than 0.7 even among elderly people between the inome index and more omposite SEP indiators available both for Rome and Turin. 48 These results made us onfident in using the inome indiator as a good proxy for the omplex onstrut of SEP. In onlusion, this study found that SEP influenes postoperative status after seleted major ardiovasular surgery in the Italian hospital are system, orroborating the evidene from other ountries. Low SEP persons undergoing surgial proedures may be more vulnerable to adverse events and should be monitored arefully. Expliit efforts should be made to identify systemi fators that amplify inequities. Aknowledgements: We thank Giovanni Casali for helpful omments on preliminary analysis and ritiism on the manusript draft. We thank Roberta Mai and Sandra Magliolo for their support in finding the ited artiles. Preliminary results of this study were presented at the IEA-EEF European Congress of Epidemiology 28 June to 1 July 2006, Utreht, The Netherlands. Funding: This study was partially finaned by the Ministry of Health (Riera Finalizzata ; ex art.12, omma 2, lett b), del D.Lgs. 502/92). Competing interests: None. REFERENCES 1. Van Lenthe FJ, Makenbah JP. Neighbourhood and individual soioeonomi inequalities in smoking: the role of physial neighbourhood stressors. J Epidemiol Community Health 2006;60: Giskes K, Turrell G, van Lenthe FJ, et al. A multilevel study of soio-eonomi inequalities in food hoie behaviour and dietary intake among the Duth population: the GLOBE study. Publi Health Nutr 2006;9: Piiotto S, Forastiere F, Stafoggia M, et al. Assoiations of area based deprivation status and individual eduational attainment with inidene, treatment, and prognosis of first oronary event in Rome, Italy. J Epidemiol Community Health 2006;60: Gerward S, Tyden P, Hansen O, et al. Survival rate 28 days after hospital admission with first myoardial infartion. Inverse relationship with soio-eonomi irumstanes. J Intern Med 2006;259: Anona C, Ara M, Saitto C, et al. Differenes in aess to oronary are unit among patients with aute myoardial infartion in Rome: old, ill, and poor people hold the burden of ineffiieny. BMC Health Serv Res 2004;4: Morris RW, Whinup PH, Papaosta O, et al. Inequalities in oronary revasularisation during the 1990s: evidene from the British regional heart study. Heart 2005;91: Anona C, Agabiti N, Forastiere F, et al. Coronary artery bypass graft surgery: soioeonomi inequalities in aess and in 30 day mortality. A population-based study in Rome, Italy. J Epidemiol Community Health 2000;54: Bradley EH, Herrin J, Wang Y, et al. Raial and ethni differenes in time to aute reperfusion therapy for patients hospitalized with myoardial infartion. JAMA 2004;292: Taylor FC, Asione R, Rees K, et al. Soioeonomi deprivation is a preditor of poor postoperative ardiovasular outomes in patients undergoing oronary artery bypass grafting. Heart 2003;89: Fisella K, Franks P, Meldrum S, et al. Raial disparity in surgial ompliations in New York State. Ann Surg 2005;242: Taylor NE, O Brien S, Edwards FH, et al. Relationship between rae and mortality and morbidity after valve replaement surgery. Review. Cirulation 2005;111: Sheikh K, Jiang Y, Bullok CM. Effet of omorbid and fatal oexistent onditions on sex and rae differenes in vasular surgial mortality. Ann Vas Surg 2007;21: Trivedi AN, Sequist TD, Ayanian JZ. Impat of hospital volume on raial disparities in ardiovasular proedure mortality. J Am Coll Cardiol 2006;47: Fisella K. Soioeonomi status disparities in healthare outomes: seletion bias or biased treatment? Med Care 2004;42: Liu JH, Zingmond DS, MGory ML, et al. Disparities in the utilization of high-volume hospitals for omplex surgery. JAMA 2006;296: Luas FL, Stukel TA, Morris AM, et al. Rae and surgial mortality in the United States. Ann Surg 2006;243: Rothenberg BM, Pearson T, Zwanziger J, et al. Explaining disparities in aess to high-quality ardia surgeons. Ann Thora Surg 2004;78:18 24; disussion Konety SH, Vaughan Sarrazin MS, Rosenthal GE. Patient and hospital differenes underlying raial variation in outomes after oronary artery bypass graft surgery. Cirulation 2005;111: Huthings A, Raine R, Brady A, et al. Soioeonomi status and outome from intensive are in England and Wales. Med Care 2004;42: Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining omorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43: Agabiti N, Piiotto S, Cesaroni G, et al. for the Italian Study Group on Inequalities in Health Care. The influene of soioeonomi status on utilization and outomes of eletive total hip replaement: a multiity population-based longitudinal study. Int J Qual Health Care 2007;19: Forastiere F, Stafoggia M, Taso C, et al. Soioeonomi status, partiulate air pollution, and daily mortality: differential exposure or differential suseptibility. Am J Ind Med 2007;50: Frane G, Taroni F, Donatini A. The Italian health-are system [review]. Health Eon 2005;14(Suppl 1):S Ageny for Health Researh and Quality (AHRQ). Patient safety indiators. Overview. (aessed 9 May 2008). 888 J Epidemiol Community Health 2008;62: doi: /jeh

8 25. Barnato AE, Luas FL, Staiger D, et al. Hospital-level raial disparities in aute myoardial infartion treatment and outomes. Med Care 2005;43: Alter DA, Naylor CD, Austin P, et al. Effets of soioeonomi status on aess to invasive ardia proedures and on mortality after aute myoardial infartion. N Engl J Med 1999;341: Pilote L, Tu JV, Humphries K, et al. Soioeonomi status, aess to health are, and outomes after aute myoardial infartion in Canada s universal health are system. Med Care 2007;45: Conrad MF, Shepard AD, Pandurangi K, et al. Outome of arotid endarteretomy in Afrian Amerians: is rae a fator? J Vas Surg 2003;38: Morris AM, Wei Y, Birkmeyer NJ, et al. Raial disparities in late survival after retal aner surgery. J Am Coll Surg 2006;203: Mahomed NN, Barrett JA, Katz JN, et al. Rates and outomes of primary and revision total hip replaement in the United States mediare population. J Bone Joint Surg Am 2003;85: Lutf R, Torquati A, Sekhar N, et al. Preditors of suess after laparosopi gastri bypass: a multivariate analysis of soioeonomi fators. Surg Endos 2006;20: Burneo JG, Blak L, Martin R, et al. Rae/ethniity, sex, and soioeonomi status as preditors of outome after surgery for temporal lobe epilepsy. Arh Neurol 2006;63: Press R, Carrasquillo O, Nikolas T, et al. Rae/ethniity, poverty status, and renal transplant outomes. Transplantation 2005;80: Peterson ED, Coombs LP, DeLong ER, et al. Proedural volume as a marker of quality for CABG surgery. JAMA 2004;29: Ekstein HH, Brukner T, Heider P, et al. The relationship between volume and outome following eletive open repair of abdominal aorti aneurysms (AAA) in 131 German hospitals. Eur J Vas Endovas Surg 2007;34: Toumpoulis IK, Anagnostopoulos CE, DeRose JJ, et al. Does EuroSCORE predit length of stay and speifi postoperative ompliations after oronary artery bypass grafting? Int J Cardiol 2005;105: Researh report 37. Istituto Nazionale di statistia (ISTAT). Condizioni di salute, fattori di rishio e riorso ai servizi sanitari (aessed 19 May 2008). 38. White RH, Zhou H, Romano PS. Inidene of symptomati venous thromboembolism after different eletive or urgent surgial proedures. Thromb Haemost 2003;90: Iezzoni LI. Assessing quality using administrative data. Ann Intern Med 1997;127: Ageny for Publi Health, Lazio Region (Italy). La valutazione della qualità della ompilazione e odifia della sheda di dimissione. (aessed 18 May 2008). 41. Baldwin LM, Klabunde CN, Green P, et al. In searh of the perfet omorbidity measure for use with administrative laims data: does it exist? Med Care 2006;44: Pine M, Jordan HS, Elixhauser A, et al. Enhanement of laims data to improve risk adjustment of hospital mortality. JAMA 2007;297: Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of omorbidity measurement in administrative data. Med Care 2004;42: Galobardes B, Shaw M, Lawlor DA, et al. Indiators of soioeonomi position (part 2). J Epidemiol Community Health 2006;60: Braveman PA, Cubbin C, Egerter S, et al. Soioeonomi status in health researh: one size does not fit all. JAMA 2005;294: Daly MC, Dunan GJ, MDonough P, et al. Optimal indiators of soioeonomi status for health researh. Am J Publi Health 2002;92: Krieger N, Chen JT, Waterman PD, et al. Geooding and monitoring of US soioeonomi inequalities in mortality and aner inidene: does the hoie of areabased measure and geographi level matter? Am J Epidemiol 2002;156: Marinai C, Spadea T, Biggeri A, et al. The role of individual and ontextual soioeonomi irumstanes on mortality: analysis of time variations in a ity of northwest Italy. J Epidemiol Community Health 2004;58: J Epidemiol Community Health: first published as /jeh on 12 September Downloaded from on 19 August 2018 by guest. Proteted by J Epidemiol Community Health 2008;62: doi: /jeh