Study of the Relationship Between Social Deprivation and Outcome After Stroke

    the Department of Clinical Neurosciences, Western General Hospital, Crewe Road, Edinburgh, Scotland.

    Abstract

    Background and Purpose— Although the incidence and mortality of stroke are known to be inversely related to socioeconomic status (SES), the relationship between SES and recovery after stroke has been little-studied. This study has investigated the relationship between SES and case fatality, "death or dependency," and "death or institutional care" at 6 months after stroke.

    Methods— Patients with acute stroke (n=2709) were identified using routine hospital discharge data and SES was measured using Carstairs scores (an ecological index of social deprivation). Case mix and treatment data were collected by medical chart review, case fatality by record linkage, and functional status and place of residence by questionnaire. Logistic regression was used to adjust the association of social deprivation and outcome for case mix and selected treatment variables.

    Results— With increasing social deprivation, patients were younger, more likely to live alone, and, on admission, more likely to need help to walk. Social deprivation was not associated with case fatality or with "death or institutional care" in any analysis. However, patients residing in the most deprived areas (deprivation categories 6 and 7) were significantly more likely to be dead or dependent than patients from more affluent areas. This association was weakened but remained after adjusting for case mix and treatment variables.

    Conclusions— These findings contribute to growing evidence of an inverse social gradient in disability after stroke. Institutionalization, as a proxy for functional outcome, may not reflect this fact. A marked social gradient in case fatality after stroke seems unlikely.

    Key Words: outcome  socioeconomic status  stroke

    Introduction

    Several studies have shown that socioeconomic status (SES) is an important risk factor for ischemic heart disease and stroke, with the incidence and mortality of each increasing as SES declines.1,2 Only some of this social gradient can be attributed to the greater prevalence of traditional cardiovascular risk factors in lower socioeconomic groups,3 and alternative explanations have been suggested.4,5 The impact of SES on survival and recovery after myocardial infarction and stroke has been addressed more recently. After myocardial infarction, it seems clear that lower SES is associated with increased case fatality6 and morbidity.7 After stroke, matters are less certain. Some studies show an association between lower SES and increased case fatality8,9 but others do not.10,11 Only 2 studies have reported the impact of SES on disability after stroke,9,12 and studies on residential outcome disagree.8,10,12 The aim of this study, therefore, is to further explore the relationship between SES and outcome after stroke.

    Materials and Methods

    This report is a secondary analysis of a study that investigated the quality of stroke services at 5 Scottish hospitals between 1995 and 1997.13 We identified patients first using hospital discharge data (cases with any International Classification of Diseases cerebrovascular disease code listed as the principal diagnosis) and then by inspecting their medical charts. We accepted a diagnosis of stroke if that was the most senior recorded opinion (excluding cases of subarachnoid hemorrhage), and we abstracted data describing SES, case mix, and treatment (Tables 1 and 2). We measured SES using Carstairs scores, which are calculated for each postal code sector (usually 5000 households) using small area census data.14 Each postal code sector is assigned to a deprivation category ranging from 1 (most affluent) to 7 (most deprived).

    At 6 months, we collected case fatality by linkage to death certificate data and functional outcome and place of residence by questionnaire. We measured dependency in activities of daily living using the simple dependency question15 and, in the last 9 months of the study, using the modified Rankin scale.16 We used this to define dependency (score 3 to 5) if the simple dependency question was not answered. We defined institutional care as residence in a hospital or nursing/residential home.

    We have reported the association of deprivation with case fatality, death or dependency, and death or institutional care. We tested for univariate associations across social strata using the 2 test for trend (for binary data) and the Jonckheere–Terpstra test (for continuous data). We adjusted for differences in important baseline risks using a set of externally validated logistic regression models, each accounting for the same 6 predictor variables (Table 1).17 When a significant relationship remained, we also adjusted for remaining case mix and selected treatment variables (Table 2). We performed all analyses using SPSS for Windows (12.0).

    Results

    We identified 4223 hospital admissions, audited the medical record of 4017 (95%), and identified 2724 cases of acute stroke. We were unable to collect postal code and hence deprivation data in 15 cases. This study therefore refers to a cohort of 2709 patients (range, 386 to 745 cases per hospital). We were unable to collect all 6 covariates used by the prognostic models in another 19 cases but retained them by inserting pessimistic values (analyses with optimistic values gave similar results). We measured case fatality in all cases, but because only 71% of survivors responded to follow-up, we can report "death or dependency" and "death or institutional care" for only 82% of patients (2209 and 2215 cases, respectively). Deprivation categories 1 and 2 contained only 69 and 198 cases, respectively, and therefore have been combined in all analyses.

    As social deprivation increased, patients with acute stroke were significantly more likely to be admitted at a younger age, to live alone, to have had an ischemic rather than hemorrhagic stroke, and, on examination, to be unable to walk without assistance (Table 1). Patients from more deprived areas were significantly less likely to be discharged from the care of a specialist stroke physician and more likely to experience delay to computed tomography head scan or admission to a stroke rehabilitation unit (Table 2).

    Social deprivation was associated with lower response to follow-up and, except for deprivation categories 1 and 2, the predicted risk of death or dependency (derived from the prognostic model) was higher in nonresponders than in responders within each social stratum. Hence, except for patients from the most affluent areas, it is likely we have underestimated the proportion with poor functional outcome in each social stratum (Table 3).

    Neither case fatality nor "death or institutional care" at 6 months was associated with deprivation in univariate analyses or after adjusting for baseline risks (Table 4). However, death or dependency at 6 months showed a strong inverse association with deprivation. This finding was only significant for deprivation categories 6 and 7 and remained after adjusting for baseline risk using the prognostic model. The association was weakened by adjusting for the remaining case mix and treatment variables but remained significant. Similar results applied when men and women were analyzed separately and for analyses restricted to patients still alive at 6 months (results not shown).

    Most patients (59%) in deprivation categories 1 and 2 came from one hospital, whereas most patients (71%) in deprivation categories 6 and 7 came from another. After excluding each hospital in turn, the same social gradients in adjusted outcome remained, although the increased risk of death or dependency in deprivation categories 6 and 7 (compared with deprivation categories 1 and 2) was no longer significant (Table I, available online only at http://www.strokeaha.org).

    Discussion

    This study has shown that across 5 Scottish hospitals, patients from the most socially deprived areas were significantly more likely than patients from more affluent areas to be dead or dependent 6 months after admission for an acute stroke. Because case fatality was not associated with residence in deprived areas, this finding reflects an increase in the proportion of survivors who were dependent in activities of daily living. There was no association between deprivation and institutional care at 6 months. Differences in case mix and in access to selected treatments explained some but not all of the association between social deprivation and the likelihood of dependent survival.

    Before drawing any conclusions, it is necessary to consider certain problems with our study. Most importantly, it is possible that our findings may be biased because our sample refers only to patients who were admitted to hospital and identified using routine data, and because patients from the most deprived areas were mostly admitted to one hospital and patients from the most affluent areas were mostly admitted to another. Our finding of an increased prevalence of dependency in socially deprived groups therefore may reflect local factors at one or both of these hospitals (eg, in their populations, referral patterns, admission thresholds, diagnostic habits, coding systems, and treatment practices) rather than any general underlying truth. Ideally, we should have accounted for this by adding a hospital term to our regression analyses before adding the social deprivation term. However, the small number of hospitals studied and the uneven distribution of deprivation between them meant that the hospital term effectively incorporated the gradient in social deprivation. Instead, therefore, we tested the stability of our findings after excluding the hospitals that contributed the majority to each social extreme. The patients from the most deprived areas remained more likely to be dead or dependent, suggesting that our findings might be real (provided the lack of statistical significance reflects only the small numbers in one or other social extreme in each analysis).

    Second, because we collected case mix and treatment data from the medical chart, it is possible that their accuracy may have varied between hospitals and so between social strata. However, we have previously shown that the variables used in our prognostic models were reliable across all 5 hospitals and likely to be valid when retrospectively collected,18 and our treatment variables were simple and important, and therefore also likely to have been accurately recorded. Third, our measurements of dependency were probably biased by nonresponse. However, because we probably under rather than overestimated dependency, it is likely that we have underestimated rather than overestimated its association with deprivation. Nonetheless, nonresponse may have led us to underestimate the proportion in institutional care in less affluent groups. Last, our use of an area-based measure of social deprivation might be criticized because not all residents of deprived areas are themselves deprived, and because the census data used to construct the measure may be out of date. However, the former problem would likely dilute rather than amplify the association of deprivation with dependency,19 and we calculated Carstairs scores using recent (1991) census data.14

    Our finding of a worse functional outcome after stroke in patients from socially deprived areas is in keeping with the 2 other studies that have addressed the topic directly. Thus, in a large (n=6903) Finnish population-based study, patients with low incomes were more likely to be dependent in activities of daily living at 28 days after stroke9; and in a small (n=465) Dutch hospital-based study, patients with low SES (defined by years of education) were significantly more likely to be disabled or handicapped at 6 months, and at 5 years, after stroke, despite adjusting for baseline demographics and case mix.12 Similarly, 3 studies have shown an inverse relationship between SES and residence at home after stroke over the short-term8,9 and longer-term.12 However, a large (n=2026) study of patients admitted to a single Scottish hospital did not show any relationship between SES (defined by prestroke area of residence) and residence at home at 1 or 3 months or at 2 years.10 The failure of this study and our own to find an association between SES and residential outcome in the face of an association between SES and disability is difficult to explain. A local explanation is possible because both studies included patients from Glasgow, Scotland and, as noted, our residential data may be biased by nonresponse. However, it should also be borne in mind that dichotomized residence data are a crude marker of disability that may also be influenced by ill-understood cultural, social, and economic factors and that discrepancies between the 2 measures are recognized.20

    Only one study has shown a strong association between lower SES and increased case fatality after stroke.9 Our findings regarding case fatality after stroke are more in keeping with the remaining studies in the field, which suggest a less pronounced8 or uncertain10,11 social gradient. Given that we used an area-based index of deprivation (that likely diminishes differences between social extremes19), the possibility that we failed to identify a shallow social gradient cannot be excluded. Similarly, it remains possible that consequent on the increased prevalence of dependency, social deprivation may be associated with increased case fatality over the longer-term.21 On balance, however, the impact of SES on case fatality after stroke would appear to be modest. As such, it is tempting to suggest that the marked social gradient in stroke mortality found in most developed societies2 relates mostly to differences in incidence and less to differences in case fatality between social strata.

    There are likely several explanations of the association between low SES and poor functional outcome after stroke. There is some evidence that patients with low SES tend to have more severe strokes, greater levels of traditional cardiovascular risk factors, and/or demographic profiles that, when taken into account, explain some of the social gradient in outcome.10,12 Similarly, there is evidence that patients with low SES have reduced access to key items of acute stroke care,8,9 a result perhaps of the reduced ability of deprived areas to attract skilled staff and medical resources. Some of the social gradient in disability after stroke may therefore be mediated (rather than confounded) by differences in acute treatment. The findings of this study are in keeping with both of these hypotheses. However, it should also be noted that, in this study, adjustment using our externally validated prognostic model explained little of the relationship between deprivation and dependency. Hence, the apparently greater "explanation" achieved by also adjusting for the remaining case mix and treatment variables may simply be a biased result derived from over-fit models. Regardless, our findings suggest that factors other than those entered into our models must account for the worse functional outcome of patients from deprived areas. These other factors might include: (1) acute treatment variables that we did not take into account; (2) an increased prevalence at the time of the stroke or, subsequently, of disorders that increase the likelihood of dependent survival, eg, recurrent stroke, dementia, depression, chronic disease, alcohol dependency, etc; (3) reduced psychological and material resources, personally or within the local environment, with which to soften the impact of disability; (4) reduced access to and compliance with rehabilitation once back in the community; and (5) perhaps even a cultural difference between social strata in the meaning of dependency.

    In summary, although the findings of this study are certainly not conclusive, they are in keeping with a growing body of evidence that patients with low SES are not only more likely to have a stroke but also more likely to be disabled by one. Measurement of residential status as a proxy for functional outcome may not reflect this fact. These observations merit further investigation with regard to how improved access to treatments and social support for less affluent groups might diminish the social gradient in disability after stroke, and how the overall burden of stroke might be reduced by reductions in social deprivation.

    Acknowledgments

    We are grateful to the staff and patients who participated in the Scottish Stroke Outcomes Study, on which this report is based. This study was supported by grants from the Chief Scientist’s Office and the Clinical Resource and Audit Group of the Scottish Office. N.U.W. was supported by a Wellcome Trust Health Services research-training grant.

    Footnotes

    Dr Martin Dennis is the stroke consultant at one of the hospitals included in this study.

    References

    Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation. 1993; 88: 973–1998.

    Avendao M, Kunst AE, Huisman M, van Lenthe F, Bopp M, Borrell C, Valkonen T, Regidor E, Costa G, Donkin A, Borgan JK, Deboosere P, Gadeyne S, Spadea T, Andersen O, Mackenbach JP. Educational level and stroke mortality. A comparison of 10 European populations during the 1990s. Stroke. 2004; 35: 432–437.

    Hart CL, Hole DJ, Davey Smith G. The contribution of risk factors to stroke differentials, by socioeconomic position in adulthood: the Renfrew/Paisley study. Am J Public Health. 2000; 90: 1788–1791.

    Hart CL, Hole DJ, Davey Smith G. Influence of socioeconomic circumstances in early and later life on stroke risk among men in a Scottish cohort study. Stroke. 2000; 31: 2093–2097.

    Macintyre S, Ellaway A, Cummins S. Place effects on health: how can we conceptualise, operationalise and measure them Social Sci Med. 2002; 55: 125–139.

    Morrison C, Woodward M, Leslie W, Tunstall-Pedoe H. Effect of socio-economic group on incidence of, management of, and survival after myocardial infarction and coronary death: analysis of community event register. BMJ. 1997; 314: 541–546.

    Barakat K, Stevenson S, Wilkinson P, Suliman A, Ranjadayalan K, Timmis AD. Socioeconomic differentials in recurrent ischaemia and mortality after acute myocardial infarction. Heart. 2001; 85: 390–394.

    Kapral MK, Wang H, Mamdani M, Tu JV. Effect of socioeconomic status on treatment and mortality after stroke. Stroke. 2002; 33: 268–273.

    Jakovljevi; D, Sarti C, Sivenius J, Torppa J, Mhnen M, Immonen-Rih P, Kaarsalo E, Alhainen K, Kuulasmaa K, Tuomilehto J, Puska P, Salomaa V. Socioeconomic status and ischemic stroke. The FINMONICA Stroke Register. Stroke. 2001; 32: 1492–1498.

    Aslanyan S, Weir CJ, Lees KR, Reid JL, McInnes GT. Effect of area-based deprivation on the severity, subtype and outcome of ischemic stroke. Stroke. 2003; 34: 2623–2628.

    Peltonen M, Rosén M, Lundberg V, Asplund K. Social patterning of myocardial infarction and stroke in Sweden: incidence and survival. Am J Epidemiol. 2000; 151: 283–292.

    van den Bos GAM, Smits JPJM, Westert GP, van Straten A. Socioeconomic variations in the course of stroke: unequal health outcomes, equal care J Epidemiol Commun Health. 2002; 56: 943–948.

    Weir N, Dennis M for the Scottish Stroke Outcomes Study Group. Towards a national system for monitoring the quality of hospital based stroke services. Stroke. 2001; 32: 1415–1421.

    McLoone P. Carstairs Scores for Scottish Postcode Sectors From the 1991 Census. Glasgow: Public Health Research Unit; 1994.

    Dennis M, Wellwood I, Warlow C. Are simple questions a valid measure of outcome after stroke Cerebrovasc Dis. 1997; 7: 22–27.

    van Swieten JC, Koudstaal PJ, Visser HJ, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988; 19: 604–607.

    Counsell C, Dennis M, McDowall M, Warlow C. Predicting outcome after acute and subacute stroke. Development and validation of new prognostic models. Stroke. 2002; 33: 1041–1047.

    Weir NU, Counsell CE, McDowall M, Gunkel A, Dennis M. Reliability of the variables in a new set of models that predict outcome after stroke. J Neurol Neurosurg Psychiatry. 2003; 74: 447–451.

    Davey Smith G, Neaton JD, Wentworth D, Stamler R, Stamler J, for the MRFIT Research Group. Mortality differences between black and white men in the USA: contribution of income and other risk factors among men screened for the MRFIT. Lancet. 1998; 351: 934–939.

    Rudd AG, Irwin P, Rutledge Z, Lowe D, Wade D, Morris R, Pearson M. Regional variations in stroke care in England, Wales and Northern Ireland: results from the National Sentinel Audit of Stroke. Clin Rehab. 2001; 15: 562–572.

    Engstad T, Viitanen M, Arnesen E. Predictors of death among long-term stroke survivors. Stroke. 2003; 34: 2876–2880.