Long-Term Mortality in Cerebrovascular Disease
Background and Purpose— Stroke is the third leading cause of death in the United States, yet data are limited about the temporal pattern of mortality among patients with cerebrovascular disease. The objectives of this study were to identify predictors of 6-month mortality and to evaluate 5-year mortality in patients with cerebrovascular disease.
Methods— Our population included fee-for-service Medicare beneficiaries aged ≥65 years who were discharged with an acute ischemic stroke, transient ischemic attack (TIA), or carotid stenosis (International Classification of Diseases, Ninth Revision, Clinical Modification codes 433 to 436) from Connecticut acute care hospitals in 1995. This cohort was followed through 2000 by means of part A Medicare claims and Social Security Administration mortality data.
Results— Among 5123 patients, 4781 survived their hospitalization and were followed for an average of 3.4 years; 670 (14.0%) died within 6 months of discharge, and 2517 (52.6%) died within 5 years. Predictors of 6-month mortality included older age, male sex, increasing comorbidity, discharge not to home, and prior admission within a year of the index hospitalization. The annual mortality rates for year 1 after discharge differed depending on the discharge diagnosis of the index hospitalization: carotid stenosis, 10.6%; TIA, 14.8%; and acute ischemic stroke, 26.4%. The 5-year cumulative mortality rates were as follows: carotid stenosis, 38.3%; TIA, 49.6%; and acute ischemic stroke, 60.0%.
Conclusions— Mortality after acute ischemic stroke, TIA, and carotid stenosis is substantial. Rates and patterns of mortality differ according to patients’ discharge diagnoses.
Cerebrovascular disease is a major medical problem in the United States, with more than 730 000 strokes and 1 million transient ischemic attacks (TIAs) each year.1,2 Although stroke is the third leading cause of death in adults, prior research in stroke prognosis focuses mainly on in-hospital and short-term mortality. Recently, investigators have begun to examine long-term mortality in patients with stroke.3–12 For example, mortality rates of 7% to 37%5,7,8,10,13 after ischemic stroke have been reported in the first year after a stroke, with a 5% to 11%8,10 annual risk of death for each year thereafter and a 5-year mortality of 53% to 60%.5,7,8 This prior research, however, only included patients with first strokes,5 younger stroke patients,6 cohorts from past eras,7,9 participants enrolled in clinical trials,13 cohorts from a single city or center,3,4,7 or cohorts from outside of the United States.3–5,8–10
Accurate estimates of both the absolute mortality rates and the pattern of mortality in cerebrovascular disease over time are required by clinicians caring for patients, researchers designing clinical trials, and policy makers allocating limited healthcare resources. The purposes of this study were to identify the predictors of 6-month mortality and to examine the patterns of mortality over a 5-year period in a cohort of patients aged ≥65 years who survived hospitalization for acute ischemic stroke, TIA, and carotid stenosis in the United States.
Subjects and Methods
A cohort of patients with cerebrovascular disease was assembled from fee-for-service Medicare beneficiaries aged ≥65 years who had been discharged with a primary discharge diagnosis of acute ischemic stroke, TIA, or carotid stenosis from any acute care hospital in Connecticut during the period January 1, 1995, through December 31, 1995. The index hospitalization was defined as the first or only hospitalization the patient had in 1995 with a principal discharge diagnosis from one of the following International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes: 433 (carotid stenosis), 434 (cerebrovascular occlusion), 435 (TIA), and 436 (acute cerebrovascular disease). Given the evidence that ICD-9-CM codes 434 and 436 are the most specific for acute ischemic stroke,14,15 we combined the cerebrovascular occlusion (ICD-9-CM code 434) and acute cerebrovascular disease (ICD-9-CM code 436) patients into a group that we refer to as “acute ischemic stroke” patients.
Medicare is the largest health insurance program in the United States and provides health insurance to people aged ≥65 years, patients with end-stage renal disease, and patients with disabilities. The Medicare program does not limit eligibility on the basis of socioeconomic status. In the present study we did not include Medicare beneficiaries enrolled in health maintenance organizations, nor did we include out-of-state discharges for patients whose primary residence was in Connecticut. Patients who were transferred from one Connecticut acute care hospital to another were included.
The annual part A Medicare claims files for those patients with cerebrovascular disease were linked to follow the cohort forward in time from their index hospitalization in 1995 through December 31, 2000. These Medicare data were also linked to Social Security files to determine each patient’s mortality status with the use of the date of death from the master beneficiary records of the Social Security Administration. The primary outcome measure was all-cause mortality.
Zero-time is the point from which prognostic predictions can be made16 and for this study was chosen as the date of discharge of the index hospitalization in 1995. Any information that is available before zero-time can be included to make prognostic predictions; information that becomes available after the zero-time should not be used prognostically. Accordingly, we defined zero-time at hospital discharge so that we could include the data from the hospital stay (eg, length of stay, discharge disposition) as potential prognostic predictors.
Each patient’s comorbidity at the time of the index hospitalization was assessed by means of the Deyo Score,17 a scale based on the Charlson Comorbidity Index18 that can be applied to administrative data. To determine whether a patient with an index hospitalization in 1995 had a hospitalization before 1995, we linked the 1995 files to 1994 Medicare files using the unique Medicare beneficiary number. We recorded the number of previous hospitalizations as well as the primary and secondary ICD-9-CM codes and Current Procedural Terminology procedure codes associated with those hospitalizations. We specifically examined whether patients had a previous diagnosis of stroke, carotid endarterectomy, coronary artery disease, or peripheral vascular disease.
Descriptive statistics were used to summarize baseline characteristics. We examined the mortality rates for each of the 4 stroke discharge diagnoses and generated 5-year survival curves for each discharge diagnosis using Kaplan-Meier methodology. For patients with acute ischemic stroke, we compared the baseline characteristics for those patients who died within 6 months of their index hospitalization with those who did not die within 6 months. We used χ2 tests to assess differences in mortality for binary independent variables. The baseline characteristics that were associated with 6-month mortality in the bivariate analysis (P<0.05) were entered into a backward logistic regression, modeling death within 6 months of the index hospitalization. The Hosmer-Lemeshow goodness-of-fit test was used to evaluate the models. All calculations were performed with the software program PC-SAS 8.0. (SAS Institute).
A total of 5123 patients were hospitalized in an acute care hospital in Connecticut with a diagnosis of acute ischemic stroke, TIA, or carotid stenosis during the calendar year 1995 (Table 1). The in-hospital mortality was 6.7% (342/5123), with 4781 patients surviving their index hospitalization. These 4781 patients were followed for an average of 3.4 years (median, 4.6 years; range, 1 day to 5 years); no patients were lost to follow-up; and 19 of 4781 patients (0.4%) were transferred to another acute hospital. Thirty-nine patients (0.8%) had missing secondary diagnosis code data so that a Deyo Score could not be calculated.
Overall, 259 (5.4%) of the 4781 patients who survived their index hospitalization died within the first month, and a total of 670 (14.0%) died within 6 months of discharge (Table 2). The annual mortality rate for the first year after discharge was 20.2% and decreased to a relatively constant 11.3% to 12.9% for years 2 through 5. The cumulative mortality rate for the entire cohort was 52.6% after the fifth year after discharge.
Annual Mortality by Discharge Diagnosis
The annual mortality rates varied by the discharge diagnosis code: 10.6% to 26.4% in the first year after discharge from the index hospitalization and 8.5% to 14.6% for years 2 through 5. Patients with the carotid stenosis (ICD-9-CM code 433; n=1099) had the lowest annual mortality (10.6%), and patients discharged with acute ischemic stroke (ICD-9-CM code 434 or 436; n=2603) had the highest annual mortality (26.4%; Table 2). Patients discharged with a diagnosis of TIA (ICD-9-CM code 435; n=1079) had an intermediate annual mortality (14.8%; Table 2).
Five-year survival curves for patients who survived their index hospitalization are shown in the Figure. The poorest survival was observed for patients with cerebrovascular occlusion (ICD-9-CM code 434).
Cumulative Mortality by Discharge Diagnosis
The 5-year cumulative mortality also varied by discharge diagnosis code. Patients with carotid stenosis had the lowest mortality (38.3%), and patients with acute ischemic stroke had the highest 5-year mortality (60.0%) (Table 2).
Predictors of 6-Month Mortality
To identify independent predictors of 6-month mortality, patients with cerebrovascular occlusion (ICD-9-CM code 434; n=2100) and acute cerebrovascular disease (ICD-9-CM code 436; n=503) were grouped together, and those patients who were alive within 6 months after discharge were compared with patients who died. In bivariate analysis, the patients who died were older, had higher comorbidity, were more likely to have been hospitalized within the past 12 months, were more likely to have a greater number of prior hospitalizations, and were less likely to have been discharged to home (Table 3). These factors remained important predictors of 6-month mortality in a multivariable analysis that adjusted for age, sex, comorbidity, discharge disposition, and prior admissions (Hosmer-Lemeshow goodness-of-fit test statistic, P=0.59; overall R2 for the model, 0.09) (Table 4).
Our study provides novel information regarding long-term mortality after hospitalization for cerebrovascular disease with the use of a geographically based US cohort. We found that the majority of patients (52.6%) who survive a hospitalization for cerebrovascular disease will die within 5 years. The absolute mortality rates that we observed are similar to those seen in previous research.9,19 For example, Retchin19 evaluated mortality in Medicare beneficiaries hospitalized with major stroke and found that 301 of 701 stroke patients (43%) who survived their hospital stay died during a 3-year follow-up period. In addition, the patterns of mortality that we observed are similar to those seen in other settings.7,8 For example, Petty and colleagues7 followed a cohort of first stroke patients from Rochester and found that 27% died within the first year after their stroke and 8% died during the period from year 1 to year 2 after the stroke. The Danish MONICA Study Group reported mortality data for a community cohort from Copenhagen with a first stroke and found 41% mortality in the first year after the stroke and a relatively constant 10% annual mortality risk for each year thereafter (follow-up period of 5.5 to 15.5 years).8 We have reported mortality rates and patterns in patients who survived their index hospitalization, and our mortality results are taken from the point of discharge from the index hospitalization. Therefore, comparisons with other studies need to be made with the knowledge that most other studies report mortality from the time of stroke symptom onset.
We observed that mortality rates varied according to the cerebrovascular disease discharge diagnosis code. The ICD-9-CM codes of 434 and 436 have the highest sensitivity (0.81) and specificity (0.90) for ischemic stroke.14 These codes were associated with the highest mortality in the present study. ICD-9-CM codes 433 and 435 include patients with asymptomatic carotid stenosis and TIA.14,15 In the present study, patients with a discharge diagnosis of carotid stenosis (ICD-9-CM code 433) had the lowest mortality, and those with TIA (ICD-9-CM code 435) had an intermediate mortality.
We found that increasing age, male sex, increasing comorbidity, discharge disposition to a site other than home (from the index hospitalization), and prior hospital admissions were important predictors of 6-month mortality in patients with acute ischemic stroke (ICD-9-CM codes 434 and 436). Prior research has identified age, male sex, recurrent stroke (as opposed to first), stroke subtype, comorbidity, neurological symptom severity, prestroke functional status, and place of residence as important predictors of mortality after stroke.3,5,7,9,11,12
The mortality observed among Medicare stroke survivors can be compared with Medicare acute myocardial infarction survivors. Krumholz and colleagues20 studied Medicare beneficiaries who survived hospitalization for an acute myocardial infarction in 1994–1995; they reported a 22% 1-year mortality rate. Therefore, cerebrovascular disease survivors appear to have 1-year mortality similar to that of acute myocardial infarction survivors. Although such comparisons are limited, they provide a framework in which to view the observed cerebrovascular disease mortality rates.
This observational study included a cohort with a diverse group of patients who represent the full spectrum of cerebrovascular disease treated at a variety of hospital types (academic and community); these findings should therefore have broad generalizability. This population included Medicare beneficiaries aged ≥65 years enrolled in fee-for-service programs, and therefore these results may not be applicable to young patients with stroke or those enrolled in managed care programs. The primary limitation of this study is that it used administrative data without detailed clinical information. The benefit of using the Medicare and Social Security data, however, is that it was possible to assemble a large, geographically based cohort and then follow the cohort through a 5-year follow-up period with complete outcome assessment and with no losses to follow-up.
In conclusion, we found that mortality after discharge from hospitalization for acute ischemic stroke, TIA, and carotid stenosis was high in the first year after discharge from an index event, depending on discharge diagnosis, and that it remains elevated in the years after an index stroke. These findings regarding long-term mortality after ischemic stroke should be useful to clinicians caring for patients with stroke, researchers designing clinical trials, and policy makers who must allocate limited healthcare resources.
Drs Ho, Scinto, and Meehan are employed by Qualidigm, the Connecticut peer review organization. Dr Bravata is supported by a career development award from the Department of Veteran Affairs Health Services Research and Development Service. The analyses upon which this publication is based were performed under contract 500-99-CT01, entitled “Utilization and Quality Control Peer Review Organization for the State of Connecticut,” sponsored by the Center for Medicare and Medicaid Services, Department of Health and Human Services. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does its mention of trade names, commercial products, or organizations imply endorsement by the US government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. This article is a direct result of the Health Care Quality Improvement Program initiated by the Center for Medicare and Medicaid Services, which has encouraged identification of quality improvement projects derived from analysis of patterns of care, and therefore required no special funding on the part of this contractor. Ideas and contributions to the authors concerning experience in engaging with issues presented are welcomed.
- Received May 2, 2002.
- Revision received August 19, 2002.
- Accepted September 23, 2002.
Broderick J, Brott T, Kothari R, Miller R, Khoury J, Pancioli A, Gebel J, Mills D, Minneci L, Shukla R. The Greater Cincinnati/Northern Kentucky Stroke Study: preliminary first-ever and total incidence rates of stroke among blacks. Stroke. 1998; 29: 415–421.
Chambers BR, Norris JW, Shurvell BL, Hachinski VC. Prognosis of acute stroke. Neurology. 1987; 37: 221–225.
Waltimo O, Kaste M, Fogelholm R. Prognosis of patients with unilateral extracranial occlusion of the internal carotid artery. Stroke. 1976; 7: 480–482.
Hankey GJ, Konrad J, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, Stewart-Wynne EG. Five-year survival after first-ever stroke and related prognostic factors in the Perth Community Stroke Study. Stroke. 2000; 31: 2080–2086.
Marini C, Totaro R, Carolei A. Long-term prognosis of cerebral ischemia in young adults. Stroke. 1999; 30: 2320–2325.
Petty G, Brown R, Whisnant J, Sicks J, O’Fallon W, Wiebers D. Survival and recurrence after first cerebral infarction: a population-based study in Rochester, Minnesota, 1975 through 1989. Neurology. 1998; 50: 208–216.
Bronnum-Hansen H, Davidsen M, Thorvaldsen P. Long-term survival and causes of death after stroke. Stroke. 2001; 32: 2131–2136.
Bonita R, Ford M, Stewart A. Predicting survival after stroke: a three-year follow-up. Stroke. 1988; 19: 669–673.
Staaf G, Lindgren A, Norrving B. Pure motor stoke from presumed lacunar infarct: long-term prognosis for survival and risk of recurrent stroke. Stroke. 2001; 32: 2592–2596.
Samsa G, Bian J, Lipscomb J, Matchar D. Epidemiology of recurrent cerebral infarction: a Medicare claims-based comparison of first and recurrent stroke on 2-year survival and cost. Stroke. 1999; 30: 338–349.
Kernan WN, Viscoli CM, Brass LM, Makuch RW, Sarrel PM, Roberts RS, Gent M, Rothwell P, Sacco RL, Liu R-C, Boden-Albala B, Horwitz RI. The Stroke Prognosis Instrument II (SPI-II): a clinical prediction instrument for patients with transient ischemia and nondisabling ischemic stroke. Stroke. 2000; 31: 456–462.
Williams G, Jian J. Development of an ischemic stroke survival score. Stroke. 2000; 31: 2414–2420.
Goldstein LB. Accuracy of ICD-9-CM coding for the identification of patients with acute ischemic stroke: effect of modifier codes. Stroke. 1998; 29: 1602–1604.
Derby C, Lapane K, Feldman H, Carleton R. Trends in validated cases of fatal and nonfatal stroke, stroke classification, and risk factors in southeastern New England, 1980 to 1991: data from the Pawtucket Heart Health Program. Stroke. 2000; 31: 875–881.