Prestroke Physical Function Predicts Stroke Outcomes in the Elderly

562 Prestroke Physical Function Predicts Stroke Outcomes in the Elderly Angela Colantonio, PhD, Stanislav V. Kasl, Phi), Adrian 2111. Ostfeld, hid, Lisa F. Berkman, PhD ABSTRACT. Colantonio A, Kasl SV, Ostfeld AM, Berkman LF. Prestroke physical function predicts stroke outcomes in the elderly. Arch Phys Med Rehabil 1996;77:562-6. Objective: To determine whether physical function before stroke is an independent predictor of physical function and institutionalization 6 months after discharge from hospital in elderly stroke patients. Design: Population-based prospective cohort design where incidence of stroke was monitored from 1982 through 1988. Baseline demographic and health information including prestroke function was collected prospectively. Eligible subjects who had a stroke were interviewed 6 months after discharge from hospital to assess outcomes. Setting: New Haven, Connecticut. Patients: Subjects were recruited from an initial sample of 2,812 older adults. Of 79 subjects who survived a first stroke at 6 months postdischarge, complete follow-up data were obtained on 63 subjects. Main Outcome Measure: Physical function as measured by the Katz scale and institutionalization. Results: Fewer limitations in activities of daily living before stroke were associated with fewer limitations in physical function after stroke controlling for stroke severity and other relevant health and sociodemographic conditions (p <.01). Fewer limitations in gross mobility function before stroke were also independently associated with a lower risk of institutionalization (p <.05). Conclusion: This study provides useful information in assessing the prognosis of elderly stroke patients upon admission to hospital. It also supports the concept of general frailty being a risk factor for poorer health and institutionalization overall in aged persons. Studies have shown that factors related to physical frailty, such as decline in muscle function, can be reversed. The effect of interventions aimed at improving the physical function of the elderly on stroke incidence, stroke outcomes, and allcause mortality, however, needs to be determined. 1996 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the Department of Occupational Therapy, Preventive Medicine and Bitstatistics (Dr. Colantonio), University of Toronto; the Department of Epidemiology and Public Health (Drs. Kasl, Ostfeld), Yale University; and the Department of Health and Social Behavior (Dr. Berkman), Harvard University. Submitted for publication June 27, 1995. Accepted in revised form December 13, 1995. Supported in part by the Robert Wood Johnson Foundation (no. 9923: Psychosocial Predictors of Recovery in the Elderly), the National Institute on Aging (NOIAG02105: Establishment of Populations for the Epidemiologic Study of the Elderly) and the Social Sciences and Humanities Research Council of Canada. No commercial party having a direct or indirect interest in the subject matter of this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint requests to Dr. Angela Colantonio, Department of Occupational Therapy, University of Toronto, 256 McCaul Street, Toronto, Ontario M5T IW5. 1996 by the American Congress of Rehahilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003-9993/96/7706-357353.0(3/0 S TROKE is the major cause of adult disability in the United States.t Because survival from stroke is more prevalent than ever before 2 better information is needed on determinants of stroke outcomes, particularly in elderly persons, who have different sets of social and health conditions than younger stroke victims. Physical function is an important means of assessing general health and disability in geriatric populations. A recent study showed that level of function before stroke is a predictor of stroke incidence. 3 The role of function before stroke as a predictor of stroke outcomes, however, has not been established. This study assessed the role of prestroke physical function as an independent predictor of two important stroke outcomes: physical function and institutionalization. Subjects METHODS The data for this study came from a larger longitudinal study, the Yale Health and Aging Project (YHAP). This survey is part of the Established Populations for Epiderniologic Studies of the Elderly (EPESE) Program. The EPESE program is a collaborative program funded by the National Institute on Aging and consists of epidemiological cohort studies in four locations: New Haven, East Boston, Iowa and Washington counties in Iowa, and Durham, NC. The purpose of these studies is to assess the general level of physical and mental health in a community population of older individuals and to determine which behaviors, socio-environmental conditions, and biological variables are predictive of morbidity and mortality. 4 This study is based on a probability sample of 2,812 noninstitutionalized men and women, 65 years of age and older, living in New Haven, CT in 1982. The sampling frame is based on samples drawn from three housing strata reflecting the three most common types of housing for those 65 years of age and older: (1) public elderly housing which is age- and incomerestricted; (2) private elderly housing which is age-restricted, and (3) private community housing and apartments. In both public and community housing, men were oversampled because of a preponderance of elderly women. A detailed description of this population is given elsewhere. 5 Some 208 respondents who reported having had a stroke or whose medical records indicated a stroke prior to the beginning of the study were eliminated, leaving a sample size of 2,604. These stroke-free cases were followed from baseline interview in 1982 until December of 1988. A trained research assistant regularly monitored admission records in the two New Haven general hospitals where 90% of all Yale Health and Aging Project hospitalizations of the cohort occurred. Medical records of patients discharged with ICD-9 codes for stroke (431, 432.9, 433.0 to 433.9, 434 to 434.9, 436, 437, and 437.1) were reviewed by a trained research assistant with a nursing background for accuracy of diagnosis and to abstract information from the charts (ie, stroke severity, comorbidity, complications, etc). All data from the charts were also reviewed by the study physician. One hundred thirty-one incident strokes were identified by

STROKE OUTCOMES IN THE ELDERLY, Colantonio 563 this method, and thus 131 patients were eligible for a face-toface interview at 6 months after discharge from hospital. Four of the 13 l cases were not followed up because they were identified shortly before the study had ended and an additional 48 had died by 6 months postdischarge. Stroke patients had similar sociodemographic characteristics to the overall study population except for being significantly older (p <.01). Stroke patients had significantly more premorbid comorbidity which was assessed prospectively such as the presence of hypertension (p <.05), angina (p <.001), diabetes (p <.001), intermittent claudication (p <.001), more cognitive impairment (p <.01) and more limitations in physical function (p <.01). Another 36 incident stroke cases were not included in the analysis because subjects died suddenly or were admitted to hospitals that were not within the surveillance area. They were identified by other methods: (1) death certificates, (2) Health Care Financing Administration data (HCFA), and (3) self-reported strokes from annual YHAP interviews that were confirmed through hospital records or HCFA data. These cases were not interviewed at 6 months postdischarge from hospital because they were identified only retrospectively. Tests for significant baseline differences were conducted between these 36 potential stroke cases and the remaining 127 subjects. No significant differences regarding the prestroke sociodemographic and health characteristics of interest were found between these two groups. Of the 79 subjects who survived a first stroke at 6 months after discharge, complete data on the physical function outcome were available on 63 subjects. Data on physical functioning were missing because of (1) loss to follow-up (n = 7), (2) refusals (n = 5), (3) rehospitalization (n = 2), and (4) 2 additional subjects were not interviewed because the study ended before their interview was scheduled. Tests for significant differences of baseline characteristics between the 63 individuals assessed for physical function and the 16 missing observations on the physical function outcome were carried out using t tests for continuous variables and chisquared analysis for categorical variables. None of the sociodemographic or health characteristics differed significantly between missing and nonmissing subjects (p <.05). The mean age of the missing group was higher (75.8 and 79.2; p =.07, based on t test). Predictor Variables One premorbid measure of physical functioning was based on Katz' Activities of Daily Living Scale. 6 The Katz scale, composed of six items pertaining to activities of daily living (ie, feeding, bathing, toileting, etc), was expanded to include walking. This modification conformed with a practice utilized in other surveys of elderly populations] Another scale of gross mobility was based on 3 items from the scale of Rosow and Breslau. 8 The latter scale measured higher levels of physical function than the Katz scale. These items included (1) the ability to do heavy work around the house, like washing windows, walls, or floors without help, (2) the ability to walk up and down stairs to the second floor without help, and (3) the ability to walk half a mile without help. Each of the items was adapted and used previously in the Framingham Disability Study. 9 A score of 0 was given to each of the items assessed if the respondent did perform the activity independently and 1 if the respondent needed assistance or was unable to perform the task. Thus, a score between 0 to 7 was possible for the modified Katz scale and a score of 0 to 3 was possible for the Rosow scale. Since the distribution of the Katz scores were skewed, we dichotomized this scale so that 0 equaled no limitations and 1 significant limitations in one or more items assessed. Control Variables: Sociodemographic and Health Conditions Baseline sociodemographic characteristics used included age, sex, years of education, race, and housing stratum. Housing stratum refers to the type of housing: public, or private elderly housing, and owned or rented housing in the community according to which this population was sampled. Health variables were obtained from standardized annual interviews (1982 to 1987) and medical records. The health measures drew on information collected through the closest interview before the stroke onset. Our variable measuring comorbidity was composed of the sum of all comorbid conditions. Information from the medical record abstract included: prior angina, arrhythmias, hypertension, congestive heart failure, diabetes, myocardial infarction, peripheral diseases of the arteries, cancer, and arthritis. This information was used to add to but not to replace, information gathered from the annual YHAP interviews. The presence of prestroke hypertension, however, was based on actual blood pressure readings taken by interviewers and by evidence of taking medication. 4 This scale was coded into 3 intervals ranging from 0 to 5 due to the skewed nature of this variable. Cognitive impairment was measured using a 10-item screening instrument similar to the Kahn and Goldfarb Short Portable Mental Status Exam. In this scale, "What is the name of this place?" was changed to "What is your address?" because the latter seemed appropriate for community-dwelling respondents. For this scale, refusals were scored as incorrect, j As an ordinal variable the cutoff points were a score of 1 or less for no impairment, 2 to 3 for mild impairment, and 4 or more for cognitive impairment. This three-level variable has been successfully used by other investigators. ~ Since cognitive impairment has been associated with physical function, it was included in the univariate analysis as a potential control variable to assess the independent effects of physical function. The scale of stroke severity used in our analyses was based on information from the medical record and was comprised of a cumulative score of (1) areas of weakness or paralysis such as arm and/or leg, (2) the presence of impaired mental status, (3) the presence of aphasia, and (4) urinary incontinence occurring subsequent to stroke. Assessment of these items was based on status at discharge. The maximum score possible on this scale was 5. The scale represents most of the major prognostic indicators tapping severity identified by previous investigators. 12-~5 The presence of complications from stroke, coded as a binary variable, included items such as decubitus ulcers, urinary tract infections, and pneumonia. This was based on information abstracted from the medical record. Table 1 provides information on sociodemographic and health characteristics of this sample of incident stroke patients. Outcome Measures Data regarding physical functioning at 6 months after discharge from hospital were obtained by personal interview administered by a trained interviewer. The Katz Activities of Daily Living Scale was used as the major outcome variable. The Rosow Scale was also examined as an outcome variable but these analyses yielded few predictors in comparison to the Katz Scale and are not present here. Institutionalization, measured as a dichotomous variable, was determined by a self-report question on institutionalization at

564 STROKE OUTCOMES IN THE ELDERLY, Colantonio TabLe 1: Sociodemographic and Health Characteristics of Study Sample of Incident Stroke Patients (N = 79) n % Age <75 38 48.1 75-84 27 34.2 85+ 14 17.5 Sex Male 31 39.2 Female 48 60.8 Race White 62 79.5 Nonwhite 16 20.5 Education <9 yrs 41 52.6 9 yrs+ 37 47.4 Housing stratum Public 19 24.1 Private 25 31.6 Community 35 44.3 Prestroke cognitive impairment None 41 54.7 Mild 21 28.0 Impaired 13 17.3 Comorbidity 0-1 conditions 25 32.9 2-4 conditions 35 46.1 5+ conditions 16 21.6 Stroke severity 0-1 mild 30 38.0 2-3 moderate 41 51,9 4-5 severe 8 10.1 Katz (prestroke physical function) 0 limitations 55 71.7 1-}- limitations 34 28.3 Rosow (prestroke gross mobility function) 0 limitations 36 41.1 1-2 limitations 28 38.4 3-1imitations 15 20.5 the poststroke interview. Additional information on institutionalization was obtained by proxy, through medical record discharge summaries, and through annual YHAP interviews for those unable/unwilling to answer. Twenty-six surviving stroke patients were institutionalized at 6 months after discharge. Statistical Analysis Linear regression was employed for the physical function outcome and logistic regression was used to analyze poststroke institutionalization. The strategy of the analysis was to establish whether prestroke physical function predicted stroke outcomes over and above sociodemographic and health predictors of these outcomes. All the sociodemographic and health predictors that were significant in the univariate analysis were placed in the model. Each of the two prestroke physical function variables of interest was then added to the model, one by one, to assess the contribution of each variable to the model to assess its impact. Table 2: Univariate Predictors of 6-Month Postdischarge Physical Functioning (Katz Scale) Coefficient SE p value Sociodemographic variables Age -.0415.4542.3647 Sex -1.0239.7079.1532 Race -2.1636.7946.0085 Education -1.0333.6905.1398 Housing ~ratum.2979*.4276.4886.9568.8140 Health variables Cognitive impairment.0998.4985.8401 Comorbidity index --.1759.1719.3061 Stroke severity.6612.2693,0169 Complications 1.3056.6802,0596 Prestroke physical functioning Katz scale 1.9863.7451.0099 Rosow scale.5771,2875.0496 * Reference category is public housing. variables to physical functioning are displayed in table 2. An examination of the sociodemographic variables revealed a significant relationship (p <.01) between race and independence in activities of daily living, with nonwhites tending to have lower functioning than whites. Of the health variables, there was a strong positive relationship between stroke severity and 6-month physical functioning measured by the Katz ADL scale. Of the baseline functioning variables, only the Katz ADL scale was a significant predictor. Lower levels of function before stroke were associated with more limitations after stroke. Figure 1 shows that having 1 or more limitations in activities of daily living before stroke is associated with more limitations after stroke as measured by the Katz Scale. Multivariate Analysis. Linear regression analysis described earlier was employed to identify the best set of sociodemographic and health conditions associated with risk of ADL limitations to serve as control variables. The variables that were retained were race and stroke severity because they were all significant at the.05 level. Housing stratum was retained in the model to control for study design. With the three control variables in the model, the variables of interest were added one at a time. The Katz Scale not only retained its significance but was the strongest predictor of poststroke function (p <.007). This model, displayed in table 3, yielded an r 2 of.32. In the multivariate model, the Rosow Scale was also predictive of poststroke function but was of borderline significance (p <.06). co E RESULTS Physical Function 6 Months After Discharge As stated previously, by 6 months postdischarge from hospital, complete data on physical function 6 weeks postdischarge were available on 63 subjects. Twelve of the 63 subjects (19%) were not able to perform any of the activities of daily living, whereas 15 individuals (23%) were able to perform all of them. Univariate Analysis. Univariate relationships of predictor 0 1+ Pre-stroke limitations (Katz Scale) Fig 1. The relationship between prestroke and poststroke limitations in activities in daily living as measured by the Katz Scale, Arch Phys Med Rehsbil Vol 77, June 1996

STROKE OUTCOMES IN THE ELDERLY, Colantonio 565 Table 3: Sociodemographic and Health Predictors of 6-Month Postdischarge Physical Functioning (Katz Scale)* (N = 59) Predictor Coefficient SE p value Race - 1,3936.8424.0026 Stroke severity.4737.2611.0387 Katz Scale 1.9329.6892.0070 r = =.32. * Controlling for housing stratum. Institutionalization 6 Months After Discharge By 6 months postdischarge, 26 individuals resided in nursing homes. The majority of the other stroke survivors were living at home or were living in someone else's home. Two individuals had entered a rehabilitation hospital who were living in homes before admission. Another had been rehospitalized the day before the 6-month postdischarge interview was to take place because of a fall. These three individuals were included in the groups of home-bound individuals. Univariate Analysis. Table 4 lists univariate logistic regression coefficient of predictor variables in relation to risk of institutionalization at 6 months after discharge from hospital. Age was positively associated with institutionalization (p <.01). Lower levels of education were predictive of institutionalization (p <.05). The only health variable that had a significant effect was prestroke cognitive impairment (p <.05), with higher levels of impairment being associated with institutionalization. In addition, more limitations in gross mobility function as measured by the Rosow scale was associated with institutionalization (p <.05). Multivariate Analysis. The multivariate analysis consisted of first placing age, education, cognitive impairment, and housing stratum in the model. Physical function variables were then added individually to this model to assess their independent effect. The Rosow Scale was still a significant predictor when the aforementioned sociodemographic and health effects were controlled (p <.05). The Katz Scale did not show a significant independent relationship. Figure 2 presents the probability for institutionalization as predicted by each level of prestroke Rosow scores adjusted for age, education, cognitive impairment, and housing stratum. The adjustment was obtained by using the mean value of the variables in the regression equation described in Table 5. Generally, the more limitations in gross mobility before stroke, the higher the probability of institutionalization after stroke. DISCUSSION Few studies to date have published data on predictors of stroke outcome in community-based populations exclusively Table 4: Univariate Sociodemographic and Health Predictors of 6- Month Postdischarge Institutionalization Coefficient SE p value Sociodemographic variables Age.1060.0352.0025 Sex.4046,2606.1205 Race -.0233.3020.9386 Education -.5530.2506.0273 Housing Stratum.3681".4672.4308.5482.3840 Health variables Cognitive impairment.9420.3384.0054 Comorbidity Index.0031.1718.9856 Stroke severity.0677.2052.7416 Complications -.2910.2428.2309 Physical function variables Katz Scale (prestroke),3466,2674.1950 Rosow Scale (prestroke),4865,2189.0262 * Reference category is public housing. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Number of limftations pre-stroke (Rosow Scale) Fig 2. Probability of institutionalization by level of gross mobility (Rosow Scores) adjusted for age, education, cognitive impairment, and housing stratum. composed of elderly individuals 65 years of age or older. The Yale Health and Aging Project data provided an opportunity to examine physical function prospectively as a predictor of stroke outcome in a sample of community-dwelling elderly. Most measures of physical function in the literature used to predict later function are usually taken on admission for stroke at the earliestj 6 However, this type of analysis can only be done properly prospectively because retrospective information on prior physical function can be biased or impossible to obtain after a stroke. Furthermore, this study includes uniform prestroke assessments of not only disability but also comorbidity and sociodemographic factors. The two outcomes studied are important because it has been estimated that up to 50% of elderly stroke victims experience permanent loss of functionj 7 This has profound implications for stroke patients and their caregivers. Institutionalization is an important outcome because it also signifies loss of independence. The role of functional disability after stroke, ie, during hospital admission or discharge, in predicting stroke outcomes such as institutionalization is well established. 16'18 However, this study shows that premorbid levels of physical function are also important in determining the health of the elderly and its role in predicting outcomes other than mortality even after a serious health event such as a stroke. We were able to disentangle the effects of disability from those of comorbidity and sociodemographic variables. Physical function has been shown to be predictive of stroke incidence 3 and all cause mortality in community-dwelling elderly. I9 Physical function may serve as a proxy for morbidity or, indeed, as a better indicator than the comorbidity index used in this study. Poorer physical function has been found to be predictive of long-term care utilization in general. 2 These findings may be reflecting a broader phenomenon related to nursing home admissions in general rather than a more specific picture of the consequences of stroke in the elderly. Limitations in activities of daily living and gross mobility function can Table 5: Predictors of Institutionalization 6 Months Postdischarge* (N = 62) Co-efficient S,E. p value Age.1292,0502.0100 Education - 1.1598.4427.0088 Cognitive impairment 1.3839.5710.0154 Rosow Scale.6451.3212.0446 * Controlling for housing stratum.

566 STROKE OUTCOMES IN THE ELDERLY, Colantonio be seen as new explanatory variables in this analysis, however, because the effects of other confounders such as race and severity of illness have been controlled. Furthermore, the analysis showed that there is a high correlation between prestroke and poststroke functional scales. It is not clear, however, why the Rosow scale would be the sole functional predictor of institutionalization given that both outcomes were significantly correlated. In a previous study, the Rosow scale, however, was a more dominant predictor of stroke incidence) It could be that this measure is more sensitive to a greater loss of independence. Other predictors of physical function and institutionalization were consistent with our findings examining these outcomes at 6 week postdischarge from hospital. Race is a stable predictor of poststroke activities of dally living. Our finding that severity of stroke is a predictor of poststroke disability has been found in large community-based prospective studies. Education and prestroke cognitive impairment were also stable predictors of institutionalization. A more detailed discussion is found elsewhere. 21 Contrary to a recent community-based study, we did not find that age was predictive of poststroke physical function but was associated with institutionalization. 22 In conclusion, this study underscores the importance of assessing functional status in the care of older persons. This study provides information useful in assessing the prognosis of elderly stroke patients upon their admission to the hospital. It also supports the concept of general frailty being a risk factor for poorer health and institutionalization overall in aged persons. Studies have shown that factors related to physical frailty, such as decline in muscle function, can be reversed) 3 The effect of interventions aimed at improving the physical function of the elderly on stroke incidence, stroke outcomes, and all-cause mortality,however, must be determined. References 1. Wolf PA, Kannel WB, Verter J. Current status of risk factors for stroke. Neurol Clin 1983; 1:317-43. 2. Posner JD, Gorman KM, Waldow A. Stroke in the elderly: epidemiology. J Am Geriatr Soc 1984;32:95-102. 3. Colantonio A, Kasl SV, Ostfeld A. Leyel of function predicts first stroke in the elderly. Stroke 1992;23:1355-7. 4. Cornoni-Huntley J, Brock DB, Ostfeld AM, Taylor JO, Wallace RB, editors. Established populations for the epidemiologic studies of the elderly. Washington, DC: National Institutes of Health, 1984. 5. Colantonio A, Kasl SV, Ostfeld A. Depressive symptoms and other psychosocial factors as predictors of stroke in the elderly. Epidemiol 1992; 136:884-94. 6. Katz S, Downs TD, Cash HR, Grotz RC. Progress in the development of an index of ADL. Gerontologist 1970; 10:20-30. 7. Lubben JE, Weiler PG, Chi I. Gender and ethnic differences in the health practices of the elderly poor. J Clin Epidemiol 1989;42:725-33. 8. Rosow I, Breslau N. A Guttman Health Scale for the aged. Gerontol 1966; 21:556-9. 9. Jeue AM, Branch LG. The Framingham Disability Study: II. Physical disability among the aging. Arrl J Public Health 1981;71:1211-6. 10. Pfeiffer E. A short portable mental status questionnaire tbr the assessment of organic brain deficit in elderly patients. J Geiiatr Soc 1975;23:433-41. 11. Moritz D, Kasl SV, Berkman LF. The health impact of living with eognitively impaired elderly spouse: depressive symptoms and social functioning. J Gerontol Social Sci 1989;44:$17-27. 12. Allen CMC. Predicting the outcome of acute stroke: prognostic score. J Neurol Neurosurg Psychiatry 1984;47:475-80. 13. Caronna JJ, Levy DE. Clinical predictors of outcome in ischemic stroke. Neurol Clin 1983; 1:103-16. 14. Dove HG, Scheider KC, Wallace JD. Evaluating and predicting outcome of acute cerebral vascular accident. Stroke 1984; 15:858-64. 15. Kelly-Hayes M, Wolf PA, Kannel WB, Sytknowski P, D'Agostino RB, Gresham GE. Factors influencing survival and need for institutionalization following stroke: the Framingham study. Arch Phys Med Rehabil 1988;69:415-8. 16. Jongloed L. Prediction of function after stroke. Stroke 1986; 17:765-76. 17. Silliman RA, Wagner EH, Fletcher RH. The social and functional consequences of stroke for elderly patients. Stroke 1987; 18:200-3. 18. Jorgensen HS, Nakayama H, Raaschou HO, Vive-Larsen J, Stoier M, Olsen TS. Outcome and time course of recovery in stroke. Part I: Outcome. The Copenhagen Stroke Study 1995;76:399-405. 19. Manton KG. A longitudinal study of functional change and mortality in the United States. J Gerontol 1993;48:S153-61. 20. Branch LC. Relative risk rates of nonmedical predictors of institutional care among elderly persons. Compr Ther 1984; 10:33-40. 21. Colantonio A, Kasl SV, Ostfeld AI Psychosocial predictors of stroke outcome in the elderly. J Gerontol 1993;48:$265-8. 22. Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. The influence of age on stroke outcome. Stroke 1994;25:808-13. 23. Fiatarone MA, Evans WJ. The etiology and reversibility of muscle dysfunction in the aged. J Gerontol 1993:48(Special Issue):77-83. Arch Phys Med Rehabil Vo177, June 1996