PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Stroke. Author manuscript; available in PMC 2010 August 1.
Published in final edited form as:
PMCID: PMC2830874
NIHMSID: NIHMS176083

Long-term functional recovery after first ischemic stroke: The Northern Manhattan Study

Abstract

Background and Purpose

Several factors predict functional status after stroke, but most studies have included hospitalized patients with limited follow-up. We hypothesized that ischemic stroke patients experience functional decline over 5 years independent of recurrent stroke and other risk factors.

Methods

In the population-based Northern Manhattan Study, incident ischemic stroke patients ≥40 years were prospectively followed using the Barthel index (BI) at 6 months and annually to 5 years. Baseline stroke severity was categorized as mild (NIH Stroke Scale <6), moderate (6–13), and severe (≥14). Follow-up was censored at death, recurrent stroke, or myocardial infarction. Generalized Estimating Equations provided odds ratios (OR) and 95% confidence intervals (95%CI) for predictors of favorable (BI≥95) versus unfavorable (BI<95) functional status, after adjusting for demographic and medical risk factors.

Results

Of 525 patients, mean age was 68.6±12.4 years, 45.5% were male, 54.7% Hispanic, 54.7% had Medicaid/no insurance, and 35.1% had moderate stroke. The proportion with BI≥95 declined over time (OR 0.91, 95% CI 0.84–0.99). Changes in BI by insurance status were confirmed by a significant interaction term (β for interaction=−0.167, p=0.034); those with Medicaid/no insurance declined (OR 0.84, p=0.003), whereas those with Medicare/private insurance did not (OR 0.99, p=0.92).

Conclusions

The proportion of patients with functional independence after stroke declines annually for up to 5 years, and these effects are greatest for those with Medicaid or no health insurance. This decline is independent of age, stroke severity, and other predictors of functional decline, and occurs even among those without recurrent stroke or myocardial infarction.

Keywords: disability, stroke, recovery

Introduction

Stroke is the leading cause of serious, long-term disability in the United States.1 Considering the staggering prevalence of stroke,2 the burden of post-stroke disability is of primary public health importance. Prior research on the natural history of disability after stroke has shown varying degrees of functional recovery within 6–12 months.36 Long-term functional outcomes are less clear, however.

Previous studies have identified demographic and stroke-specific characteristics that are associated with functional recovery after stroke.35, 714 However, most of these studies have included only hospitalized patients with limited follow-up, and few population-based studies have examined predictors of functional status with long-term follow-up. Furthermore, the effect of socioeconomic status on functional status after stroke has not been examined over the long-term.15

We sought, in a prospective, population-based, multi-ethnic, urban stroke cohort study, to determine the long-term natural history and predictors of functional status among participants who experienced a first ischemic stroke. We hypothesized that ischemic stroke patients experience functional decline over 5 years independent of recurrent stroke and other risk factors.

Subjects and Methods

The Northern Manhattan Study (NOMAS) includes a population-based incident ischemic stroke follow-up study designed to determine predictors of stroke recurrence and prognosis in a multi-ethnic, urban population, as previously described.16, 17 The race-ethnic mixture of the northern Manhattan community at the time of the study consisted of 63% Hispanic, 20% black, and 15% white residents.

Selection of NOMAS cohort

The methods of patient identification and enrollment have been described in previous publications.18, 19 Briefly, ischemic stroke patients were enrolled if they: (1) were diagnosed with a first stroke; (2) were age ≥40 years, and (3) resided in Northern Manhattan for ≥3 months in a household with a telephone. NOMAS was limited to ischemic stroke to restrict cases to a more homogeneous population and limit the underlying pathophysiologies. Case ascertainment occurred between July 1993 and June 1996, and assessments were completed in August 2001. Over 80% of patients with acute ischemic stroke in northern Manhattan are hospitalized at the Columbia University Medical Center (CUMC). Subjects hospitalized at other local hospitals were identified through active surveillance of admissions to those hospitals and through agreements with local physicians. Approximately 5% of incident ischemic stroke patients in northern Manhattan were not hospitalized and were also included.16, 19 Evaluation of patients was performed at the hospital; those subjects either not hospitalized or hospitalized elsewhere were evaluated in the outpatient research clinic. Of the participants who were not hospitalized at CUMC, only those who had an initial evaluation within 20 days of their first stroke were included in analyses involving the NIH Stroke Scale. The study was approved by the CUMC Institutional Review Board. All participants gave consent directly or through a surrogate when appropriate.

Index Evaluation of Subjects

Data were collected through interviews by trained research assistants, and physical and neurological examinations were conducted by study neurologists, as previously described.18, 19 Assessments were conducted in English or Spanish depending upon the primary language of the participant. Race-ethnicity was based upon self-identification. Standardized questions were adapted from the Behavioral Risk Factor Surveillance System20 by the Centers for Disease Control and Prevention regarding the following conditions: hypertension (HTN), diabetes mellitus (DM), hypercholesterolemia, peripheral vascular disease, transient ischemic attack, cigarette smoking, and cardiac conditions such as myocardial infarction (MI), coronary artery disease, angina, congestive heart failure (CHF), atrial fibrillation, other arrhythmias, and valvular heart disease. HTN was defined as a systolic blood pressure recording≥160 mmHg or a diastolic blood pressure recording ≥95 mm Hg (based on the average of two blood pressure measurements) or the patient’s self-report of a history of HTN or antihypertensive use. DM was defined by a fasting blood glucose level ≥126 mg/dL, the participant’s self-report of such history, or insulin or oral hypoglycemic use.18, 19 Presence or absence of urinary incontinence within 7–10 days of the index stroke was determined. Presence or absence of depressed mood in the week prior to stroke was assessed at enrollment. Physical activity was assessed using a questionnaire adapted from the National Health Interview Survey of the National Center for Health Statistics, and classified as light-moderate and heavy, as previously described.18

Stroke severity was assessed using the NIHSS score derived from a standardized neurological examination, and was categorized into mild (<6), moderate (6–13), and severe (≥14). This categorization was based on previous analyses of stroke severity in relation to stroke outcome from our population and a clinical trial.21, 22 Stroke diagnostic evaluation included computerized tomography and/or magnetic resonance imaging of the brain, ultrasound evaluation and/or magnetic resonance angiography of the extracranial and intracranial cerebral vessels, and transthoracic or transesophageal echocardiogram as appropriate. A consensus of stroke neurologists assessed stroke subtype using modified Stroke Data Bank criteria and all available information, as described in a previous publication.23 Stroke neurologists also determined brain side of index stroke, presence of aphasia, and parietal lobe dysfunction including neglect.

Follow-Up and Outcomes Assessment

Follow-up evaluations were conducted at six months and then annually for five years. Evaluations were conducted at the medical center or by telephone interaction with the patient, family member, or caregiver. There was also a telephone assessment at 1.5 years. Information on vital status, functional status as measured by the Barthel index (BI), and intercurrent symptoms, illness, or hospitalization was collected. Previous research has demonstrated the reliability of phone assessments of functional status using the BI.24 In-person assessments included interviews and assessment of functional status, as well as measurement of vital signs, physical and neurological examination.

Patients unable or unwilling to come to the medical center were visited by a member of the research staff, and the evaluation was conducted at home or in an alternative place of residence (e.g., nursing home). An ongoing surveillance system of admissions to the CUMC and other local hospitals, described in a previous publication,16 was also used to identify study participants who experienced recurrent stroke, MI, hospitalization, or death. When available, medical records were reviewed for all outcome events including death. All outcome events were initially reviewed by a specially-trained research assistant, and events were then adjudicated by study neurologists or cardiologists, as appropriate.

Statistical Analyses

Statistical analyses were conducted using SAS Version 9.1.3 (SAS Institute, Cary, NC). For descriptive purposes, means were calculated for continuous variables and proportions for categorical variables. Follow-up was censored at the time of death, recurrent stroke, or MI because of the effect of these events on functional status. We used Generalized Estimating Equations (GEE) to assess the relationship between several predictor variables, including time of follow-up at regular intervals, with functional outcome. The advantage of GEE is that it accounts for within-subject correlation and produces unbiased and more efficient regression estimates in studies that have a longitudinal, repeated-measures design.25 GEE with binary outcome and logistic link function provided parameter estimates, from which odds ratios and 95% confidence intervals (OR, 95% CI) were calculated for predictors of favorable (BI ≥95) versus unfavorable (BI<95) functional status, after adjusting for demographic (age, sex, race/ethnicity, education level, marital status, insurance status) and medical (HTN, coronary artery disease or MI [CAD], DM, hypercholesterolemia, smoking, physical activity) risk factors, as well as stroke characteristics known to be associated with functional recovery in previous studies (initial stroke severity, depressed mood prior to stroke, presence or absence of aphasia, presence of parietal lobe dysfunction including neglect, side of stroke, and presence of urinary incontinence). Insurance status was characterized as Medicare/private insurance versus Medicaid/no insurance. Multivariate models were constructed in a backward stepwise manner using those variables that were significant at p<0.10. In secondary analyses, we evaluated predictors for BI ≥60 versus <60.

Subgroup analysis was performed in those who had BI ≥95 at 6 months, in order to determine the natural history and predictors of functional status in those deemed “recovered” after the acute phase of recovery.

The interaction between time of the follow-up assessment and insurance status was included in multivariate regression models to assess whether the change in functional status over time differed by insurance status. After testing for interaction with insurance, the effect of follow-up time on functional status was also stratified by type of insurance.

Time of follow-up assessment was analyzed both continuously and discretely (at 1, 1.5, 2, 3, 4, and 5 year time points, with 0.5 year as reference). The analysis of discrete time points did not assume any particular functional form, whereas the continuous model assumed a linear time trend. Using the model with discrete time points, we tested whether the trend was linear. In addition, in the continuous model, appropriateness of linearity was determined by testing whether the quadratic term was significantly different from zero.

To explore whether there was a significant decrement in functional status at a single time point -- a change point -- maximum likelihood estimates of the change point were obtained. To assess whether differential length of follow-up was related to functional status, which would have been a source of bias, a series of logistic regression models was fitted for each time point using an indicator of whether each subject dropped out after each follow-up time point as the outcome, and BI≥95 and the same covariates in our final model as the predictors.

Results

There were 655 patients enrolled in the NOMAS stroke incidence and follow-up study. No follow-up functional data was available for participants who died within 6 months (n= 83) and for those who died between 6 months and 1 year (n=14). Five participants died after 1 year, but before functional data were collected. Twenty-four participants had recurrent stroke before first functional assessment and were excluded from analysis. There were no functional data and no indication of death for 4 participants, who were categorized as lost to follow-up. Five hundred twenty-five patients were thus available for this analysis. In addition, during follow up, 84 patients had recurrent stroke or MI, and there were 133 deaths. Table 1 lists baseline and demographic information for study participants.

Table 1
Baseline characteristics of study population*

There was an annual decline in the proportion of patients with BI≥95 in an unadjusted regression model considering time to follow-up assessment as a continuous variable (OR per year after stroke 0.96, 95% CI 0.92–1.01). When adjusted for demographic variables, the annual decline was significant (adjusted OR 0.94, 95% CI 0.89–1.00) and did not change further when adjusted for demographic variables and medical risk factors (adjusted OR 0.94, 95%CI 0.88–0.99). Further adjusting for stroke characteristics, such as stroke severity, neglect, side of stroke, and urinary incontinence, led to a slightly more pronounced decline (adjusted OR per year after stroke 0.91, 95% CI 0.84–0.99). Results for the cutoff of BI≥60 showed more marked declines; an unadjusted model showed OR of 0.93 (95% CI 0.87–0.98), and models adjusted for demographic variables (adjusted OR 0.90, 95% CI 0.84–0.96) and demographics and medical risk factors (adjusted OR 0.89, 95% CI 0.83–0.96) showed that with every year of follow-up, participants were less likely to have BI≥60. Table 2 reports the fully adjusted results.

Table 2
Multivariate generalized estimating equation (GEE) regression model of predictors of function after ischemic stroke*

In subgroup analysis for those participants who had BI≥95 at 6 months (n=245), there was still a significant annual decline in the proportion with BI≥95 over follow-up (unadjusted OR per year 0.81, p<0.0001), and there was little change when adjusted for demographic variables (adjusted OR 0.79, p<0.0001) and in the fully adjusted model (adjusted OR per year 0.76, 95% CI 0.65–0.89; Table 3).

Table 3
Multivariate generalized estimating equation (GEE) regression model of predictors of function after ischemic stroke, among those with BI ≥95 at 6 months*

Changes in BI over time differed significantly by insurance status (β for interaction=−0.167, p=0.034), with a decline in BI over time for those with Medicaid/no insurance (adjusted OR 0.84, 95% CI 0.75–0.94, p=0.003), but no definite decline among those with Medicare/private insurance (adjusted OR 0.99, 95% CI 0.90–1.11, p=0.92). No other two-way interactions between time to follow-up and other variables were significant. Figures 1 and and22 depict functional status stratified by insurance status, unadjusted for other risk factors.

Figure 1
Percentage of participants with Barthel ≥95, stratified by insurance status.
Figure 2
Percentage of participants with Barthel ≥60, stratified by insurance status.

In the final regression model, time was treated as a continuous variable in linear form because there was no evidence that an alternative model better fit the data (lack of linearity Chi-square test: df=4, p=0.65).

In order to evaluate at which time point the greatest change occurs, we conducted an analysis of the change point in functional decline. The maximum likelihood estimate of the change point was at 3 years, and the change in proportion of participants with BI≥95 before and after the 3 year time point was significantly different from zero (OR 0.76, p=0.0088), indicating that the greatest functional decline began to occur 3 years after stroke.

There was no significant difference in functional status when those with shorter follow-up were compared to those with relatively longer follow-up, confirming that differential follow-up does not depend on functional status.

Other predictors of BI≥95 among all participants were age at stroke, diabetes, marital status, stroke severity, and baseline urinary continence (Table 2). Results were similar for the BI cutoff of 60, with the exception that DM approached but did not reach significance as a predictor (p=0.0842) and marital status was not associated with outcome (p=0.94). Left-sided stroke (versus right) was associated with better functional outcome. Among participants with BI≥95 at 6 months, predictors were similar to those among all participants, except that Hispanic ethnicity was a significant predictor whereas severe stroke was not (Table 3).

Discussion

In this large, population-based, prospective cohort study of first-time ischemic stroke patients, we observed a modest annual decline in functional status over the long term after stroke, even after censoring those with recurrent stroke and MI and adjusting for risk factors and baseline stroke severity. There was a significant decline over years of follow-up even among those with early functional “recovery” (BI≥95 at 6 months). In the United States, there are 4 tiers of insurance access: private insurance (69% of the U.S. population in 200426), Medicaid (a state-administered insurance only available to low income residents that has been associated with limited access to health care in New York State27, 28; 13% of the population), Medicare (a federally-administered insurance primarily for residents aged ≥65 years; 14% of the population), and no insurance (16% of population). Functional decline in our study was seen particularly among those who were uninsured or insured with Medicaid. This decline began to be apparent at about 3 years when we used as a threshold for disability the BI threshold of 95, and at 2 years with a BI cutoff of 60.

Few community-based studies have examined long-term disability after stroke. One study reported that 37% of first stroke patients worsened in Rankin scores between 1 and 5 years after stroke in Rochester, Minnesota.29 Limitations of this study included comparison of mean Rankin scores without reporting of confidence intervals and the lack of multivariable analysis for predictors of outcomes. Furthermore, the GEE analysis that was used in our study is more sensitive to individual changes in disability over time. In another study, dependence in ADL tasks in 109 first stroke patients was analyzed, and a smaller percentage of women were independent at 5 years and there was a greater proportion of dependence among older patients.30 However, this study had a small sample, reported only percentages in each category without confidence intervals or significance tests, did not conduct multivariable analysis, and used a potentially insensitive ADL measurement scale. Another study examined modified BI at 5 years in 129 survivors of first stroke and found that 36% developed new major disability, with predictors including older age, drowsiness at baseline assessment, moderate hemiparesis, and recurrent stroke.31 This study was limited by small numbers of data available at 5 years and only one follow-up assessment of disability. Finally, none of these studies censored recurrent stroke or cardiac events, and hence it is unclear whether worsened disability over time was the result of recurrent stroke or long-term effects of a single index stroke. In contrast to these studies, our study collected data at multiple timepoints and excluded those with recurrent clinical vascular events and so was able to assess a linear change in function over time.

The delayed functional decline that we observed has not been well-described before and has several possible explanations. Non-stroke comorbidities that are known to affect functional status may contribute in the long-term, as prior research in non-stroke cohorts has shown.32 In particular, cognitive dysfunction has been shown to be associated with functional status after stroke.33, 34 In cognitive research, the concept of “cognitive reserve” refers to differing susceptibility to cognitive impairment that is related to variables such as education, literacy, intelligence quotient, and engagement in leisure activities.35 Considering the close correlation between cognitive status and functional status, there may be a similar phenomenon with regard to performing ADLs that we may term “functional reserve.” The deficit caused by a first stroke may result in a depleted functional reserve and a consequent failure to compensate for brain aging. We were unable to include detailed measurements of cognitive function in our analyses, however.

Another possible explanation for the observed delay in functional decline among uninsured and Medicaid patients is that the stroke caused a functional deficit that affected participants equally regardless of insurance status. Stroke, in other words, may serve as a sort of equalizer of function due to its direct biological effects on the brain, as well as the availability of acute rehabilitation and other therapies available to participants with all types of insurance. Over time after stroke, however, functional status among those with private insurance and Medicare and those with Medicaid or no insurance may diverge, due to disparities in care and more limited access to rehabilitative services, information about health, and ongoing management of risk factors and chronic conditions28, 36 that are known to have an impact on functional status.37

The delayed decline in functional status could also be due to the ceiling effect that has been observed with the BI.38 It is possible that there was a steady decline in functional status that began soon after ischemic stroke, but since the BI is insensitive to small changes in disability, this decline was not captured until 3 years after the stroke.

Another possible explanation for the observed decline in functional status is that participants may have experienced clinically silent recurrent strokes during follow-up. Studies have shown a prevalence of 6–18% of clinically silent strokes in different populations depending on risk factors and imaging protocols,39 and silent infarcts may be as much as 5 times as prevalent as symptomatic infarcts.40 A previous study in the Northern Manhattan population showed that 18% of 892 participants free of clinical stroke had subclinical infarcts.41 Despite the term “silent” stroke, these subclinical events may affect functional status.42

Previous studies with limited follow-up have shown predictors of functional status similar to the predictors that we found, in particular age,3, 4, 7, 8 DM,4, 8, 9 marital status (a marker of social support),5, 10 stroke severity,4, 7, 8, 11 side of stroke,11, 12 and urinary continence.13, 14 Most previous studies, however, have assessed these functional outcomes at single points in time, whereas in our analysis we were able to assess the associations for these risk factors with functional outcomes measured at multiple annual intervals. These risk factors thus continue to predict functional status over several years. The fact that these predictors were found to be significant in our study suggests that they represent robust associations with functional status, for selection bias is minimized in population-based studies and follow-up continued to 5 years post-stroke. The fact that DM was a significant predictor of functional outcome is notable considering the high prevalence of DM in this population, which is likely due to a combination of environmental (including dietary) and genetic factors.

In our multi-ethnic population-based study, race/ethnicity was not associated with functional outcome among all participants, in univariate or multivariate analyses. Socioeconomic status and access to care were primary mediating effects of functional status, perhaps mitigating the effect of race/ethnicity on outcome, which has been previously described.43

Limitations of this study include use of the BI to assess disability instead of a disability scale designed specifically for stroke patients. However, the BI is widely used in stroke research, which allows comparison of this study with prior studies. Second, detailed data on rehabilitation received after the stroke was not collected, and may have informed the relationship between late functional decline and insurance status. Third, since some of the data on baseline risk factors was obtained by self-report, uninsured participants who do not regularly get medical care may not be aware of undiagnosed conditions. Hence, in order to lessen bias, we used laboratory and blood pressure measurements and information about prescription medications to capture risk factors such as DM, HTN, and hypercholesterolemia. Fourth, neuropsychological data were not analyzed in this study, which could have informed the link between cognitive status and functional status. Future research will be directed towards clarifying the relationship between time and functional decline, comparing functional status over time in stroke patients and non-stroke controls matched for risk factors, and assessing the role of cognitive decline as a mediator of the change in functional status. Finally, the decline in functional status among those with less access to care highlights the need to improve the ways in which knowledge of risk factor control is translated in clinical practice to improve outcome following stroke.

Acknowledgments

Acknowledgments and Funding

This work was supported by grants from the National Institute of Neurological Disorders and Stroke (R01 NS48134, MSVE; R37 29993, RLS/MSVE).

References

1. Prevalence of disabilities and associated health conditions among adults--united states, 1999. MMWR Morb Mortal Wkly Rep. 2001;50:120–125. [PubMed]
2. American heart association. Heart disease and stroke statistics -- 2008 update. Dallas, Texas: American Heart Association; 2008.
3. Pan SL, Lien IN, Yen MF, Lee TK, Chen TH. Dynamic aspect of functional recovery after stroke using a multistate model. Arch Phys Med Rehabil. 2008;89:1054–1060. [PubMed]
4. Hankey GJ, Spiesser J, Hakimi Z, Bego G, Carita P, Gabriel S. Rate, degree, and predictors of recovery from disability following ischemic stroke. Neurology. 2007;68:1583–1587. [PubMed]
5. Johnston M, Pollard B, Morrison V, MacWalter R. Functional limitations and survival following stroke: Psychological and clinical predictors of 3-year outcome. Int J Behav Med. 2004;11:187–196. [PubMed]
6. Jorgensen HS, Reith J, Nakayama H, Kammersgaard LP, Raaschou HO, Olsen TS. What determines good recovery in patients with the most severe strokes? The copenhagen stroke study. Stroke. 1999;30:2008–2012. [PubMed]
7. Weimar C, Konig IR, Kraywinkel K, Ziegler A, Diener HC. Age and national institutes of health stroke scale score within 6 hours after onset are accurate predictors of outcome after cerebral ischemia: Development and external validation of prognostic models. Stroke. 2004;35:158–162. [PubMed]
8. Weimar C, Ziegler A, Konig IR, Diener HC. Predicting functional outcome and survival after acute ischemic stroke. J Neurol. 2002;249:888–895. [PubMed]
9. Jorgensen H, Nakayama H, Raaschou HO, Olsen TS. Stroke in patients with diabetes. The copenhagen stroke study. Stroke. 1994;25:1977–1984. [PubMed]
10. Glass TA, Matchar DB, Belyea M, Feussner JR. Impact of social support on outcome in first stroke. Stroke. 1993;24:64–70. [PubMed]
11. Glymour MM, Berkman LF, Ertel KA, Fay ME, Glass TA, Furie KL. Lesion characteristics, nih stroke scale, and functional recovery after stroke. Am J Phys Med Rehabil. 2007;86:725–733. [PubMed]
12. Robertson IH, Ridgeway V, Greenfield E, Parr A. Motor recovery after stroke depends on intact sustained attention: A 2-year follow-up study. Neuropsychology. 1997;11:290–295. [PubMed]
13. Massucci M, Perdon L, Agosti M, Celani MG, Righetti E, Recupero E, Todeschini E, Franceschini M. Prognostic factors of activity limitation and discharge destination after stroke rehabilitation. Am J Phys Med Rehabil. 2006;85:963–970. [PubMed]
14. Pettersen R, Dahl T, Wyller TB. Prediction of long-term functional outcome after stroke rehabilitation. Clin Rehabil. 2002;16:149–159. [PubMed]
15. Putman K, De Wit L, Schoonacker M, Baert I, Beyens H, Brinkmann N, Dejaeger E, De Meyer AM, De Weerdt W, Feys H, Jenni W, Kaske C, Leys M, Lincoln N, Schuback B, Schupp W, Smith B, Louckx F. Effect of socioeconomic status on functional and motor recovery after stroke: A european multicentre study. J Neurol Neurosurg Psychiatry. 2007;78:593–599. [PMC free article] [PubMed]
16. Sacco RL, Boden-Albala B, Gan R, Chen X, Kargman DE, Shea S, Paik MC, Hauser WA. Stroke incidence among white, black, and hispanic residents of an urban community: The northern manhattan stroke study. Am J Epidemiol. 1998;147:259–268. [PubMed]
17. Hartmann A, Rundek T, Mast H, Paik MC, Boden-Albala B, Mohr JP, Sacco RL. Mortality and causes of death after first ischemic stroke: The northern manhattan stroke study. Neurology. 2001;57:2000–2005. [PubMed]
18. Sacco RL, Gan R, Boden-Albala B, Lin IF, Kargman DE, Hauser WA, Shea S, Paik MC. Leisure-time physical activity and ischemic stroke risk: The northern manhattan stroke study. Stroke. 1998;29:380–387. [PubMed]
19. Sacco RL, Elkind M, Boden-Albala B, Lin IF, Kargman DE, Hauser WA, Shea S, Paik MC. The protective effect of moderate alcohol consumption on ischemic stroke. Jama. 1999;281:53–60. [PubMed]
20. Gentry EM, Kalsbeek WD, Hogelin GC, Jones JT, Gaines KL, Forman MR, Marks JS, Trowbridge FL. The behavioral risk factor surveys: Ii. Design, methods, and estimates from combined state data. Am J Prev Med. 1985;1:9–14. [PubMed]
21. Sacco RL, Boden-Albala B, Chen X, Lin IF, Kargman DE, Paik MC. Relationship of 6-month functional outcome and stroke severity: Implications for acute stroke trials from the northern manhattan stroke study. Neurology. 1998;50:A327.
22. Sacco RL, DeRosa JT, Haley EC, Jr, Levin B, Ordronneau P, Phillips SJ, Rundek T, Snipes RG, Thompson JL. Glycine antagonist in neuroprotection for patients with acute stroke: Gain americas: A randomized controlled trial. Jama. 2001;285:1719–1728. [PubMed]
23. Gan R, Sacco RL, Kargman DE, Roberts JK, Boden-Albala B, Gu Q. Testing the validity of the lacunar hypothesis: The northern manhattan stroke study experience. Neurology. 1997;48:1204–1211. [PubMed]
24. Shinar D, Gross CR, Bronstein KS, Licata-Gehr EE, Eden DT, Cabrera AR, Fishman IG, Roth AA, Barwick JA, Kunitz SC. Reliability of the activities of daily living scale and its use in telephone interview. Arch Phys Med Rehabil. 1987;68:723–728. [PubMed]
25. Hanley JA, Negassa A, Edwardes MD, Forrester JE. Statistical analysis of correlated data using generalized estimating equations: An orientation. Am J Epidemiol. 2003;157:364–375. [PubMed]
26. Bureau of labor statistics: Current population survey. 2005. http://www.bls.gov/CPS/
27. Giacovelli JK, Egorova N, Nowygrod R, Gelijns A, Kent KC, Morrissey NJ. Insurance status predicts access to care and outcomes of vascular disease. J Vasc Surg. 2008 [PMC free article] [PubMed]
28. Calman NS, Golub M, Ruddock C, Le L, Hauser D. Separate and unequal care in new york city. J Health Care Law Policy. 2006;9:105–120. [PubMed]
29. Dombovy ML, Basford JR, Whisnant JP, Bergstralh EJ. Disability and use of rehabilitation services following stroke in rochester, minnesota, 1975–1979. Stroke. 1987;18:830–836. [PubMed]
30. Kojima S, Omura T, Wakamatsu W, Kishi M, Yamazaki T, Iida M, Komachi Y. Prognosis and disability of stroke patients after 5 years in akita, japan. Stroke. 1990;21:72–77. [PubMed]
31. Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Anderson CS. Long-term disability after first-ever stroke and related prognostic factors in the perth community stroke study, 1989–1990. Stroke. 2002;33:1034–1040. [PubMed]
32. Stuck AE, Walthert JM, Nikolaus T, Bula CJ, Hohmann C, Beck JC. Risk factors for functional status decline in community-living elderly people: A systematic literature review. Soc Sci Med. 1999;48:445–469. [PubMed]
33. McGuire LC, Ford ES, Ajani UA. Cognitive functioning as a predictor of functional disability in later life. Am J Geriatr Psychiatry. 2006;14:36–42. [PubMed]
34. Mok VC, Wong A, Lam WW, Fan YH, Tang WK, Kwok T, Hui AC, Wong KS. Cognitive impairment and functional outcome after stroke associated with small vessel disease. J Neurol Neurosurg Psychiatry. 2004;75:560–566. [PMC free article] [PubMed]
35. Stern Y, Zarahn E, Habeck C, Holtzer R, Rakitin BC, Kumar A, Flynn J, Steffener J, Brown T. A common neural network for cognitive reserve in verbal and object working memory in young but not old. Cereb Cortex. 2008;18:959–967. [PMC free article] [PubMed]
36. Mold F, McKevitt C, Wolfe C. A review and commentary of the social factors which influence stroke care: Issues of inequality in qualitative literature. Health Soc Care Community. 2003;11:405–414. [PubMed]
37. Kiely DK, Morris JN, Morris SA, Cupples LA, Ooi WL, Sherwood S. The effect of specific medical conditions on functional decline. J Am Geriatr Soc. 1997;45:1459–1463. [PubMed]
38. Dromerick AW, Edwards DF, Diringer MN. Sensitivity to changes in disability after stroke: A comparison of four scales useful in clinical trials. J Rehabil Res Dev. 2003;40:1–8. [PubMed]
39. Das RR, Seshadri S, Beiser AS, Kelly-Hayes M, Au R, Himali JJ, Kase CS, Benjamin EJ, Polak JF, O’Donnell CJ, Yoshita M, D’Agostino RB, Sr, DeCarli C, Wolf PA. Prevalence and correlates of silent cerebral infarcts in the framingham offspring study. Stroke. 2008;39:2929–2935. [PMC free article] [PubMed]
40. Vermeer SE, Koudstaal PJ, Oudkerk M, Hofman A, Breteler MM. Prevalence and risk factors of silent brain infarcts in the population-based rotterdam scan study. Stroke. 2002;33:21–25. [PubMed]
41. Prabhakaran S, Wright CB, Yoshita M, Delapaz R, Brown T, DeCarli C, Sacco RL. Prevalence and determinants of subclinical brain infarction: The northern manhattan study. Neurology. 2008;70:425–430. [PMC free article] [PubMed]
42. Schmidt WP, Roesler A, Kretzschmar K, Ladwig KH, Junker R, Berger K. Functional and cognitive consequences of silent stroke discovered using brain magnetic resonance imaging in an elderly population. J Am Geriatr Soc. 2004;52:1045–1050. [PubMed]
43. Horner RD, Swanson JW, Bosworth HB, Matchar DB. Effects of race and poverty on the process and outcome of inpatient rehabilitation services among stroke patients. Stroke. 2003;34:1027–1031. [PubMed]