Background and Purpose Few studies have examined the utility of a new generic health status measure, the Short Form 36 health survey questionnaire (SF-36), in stroke patients. Our aim was to test the internal consistency and validity of the SF-36 in a cohort of long-term stroke survivors.
Methods The Australian version of the SF-36 was tested in 90 consecutive 1-year stroke survivors (mean age, 72 years) identified from our hospital discharge data. The instrument was administered by personal interview. Validity was assessed by comparing patients' scores on the SF-36 with those obtained for the Barthel Index, the 28-item General Health Questionnaire, and the Adelaide Activities Profile, an instrument developed from the Frenchay Activities Index.
Results The SF-36 was relatively quick and easy to use and had satisfactory internal consistency (Cronbach's α>0.7). For all eight SF-36 health scales, the mean scores for patients dependent in self care and with mental ill health were significantly different from patients without these disabilities, but the strength of the differences varied in a predictable manner. However, the SF-36 social functioning scale did not provide a valid measure of everyday activities relevant to many elderly patients as measured by the Adelaide Activities Profile.
Conclusions The SF-36 avoids the “ceiling effect” of most disability scales and provides a valid measure of physical and mental health after stroke, but it does not appear to characterize well social functioning. Thus, the instrument may need to be supplemented by other measures for a comprehensive assessment of stroke outcome.
Demographic changes, the increasing prevalence of chronic diseases and disability, and associated financial constraints on healthcare systems have prompted the use of broad measures of outcome to evaluate the effectiveness of care and medical interventions intended to improve both the quality and duration of life.1 These consist of disease-specific measures designed to be sensitive to the outcomes of particular disease processes and generic measures designed to be applicable across a wide range of medical conditions. A new generic health status measure gaining popularity is the SF-36, a relatively brief and simple questionnaire developed from the Medical Outcomes Study in the United States.2 3 The SF-36 comprises eight health scales: physical functioning (10 items), role limitations–physical (4 items), bodily pain (2 items), general health (5 items), vitality (4 items), social functioning (2 items), role limitations–emotional (3 items), and mental health (5 items). Two core dimensions of health, physical and mental, can be derived from these eight scales. There is also a single separate item that is used to assess any change in health from 1 year before.
Information on the validity and reliability of the SF-36 in ambulatory persons is rapidly becoming available, and reference values are being established for different populations.4 However, less is known regarding its performance in neurologically disabled patients, particularly those who are elderly.5 6 7 Although quality-of-life issues are considered an important aspect of stroke outcome, few studies have examined methods for collecting such data.8 Our aim was to test the internal consistency and validity of the SF-36 in a population of long-term survivors of stroke.
Subjects and Methods
We retrospectively identified 170 consecutive patients with acute stroke admitted to the Flinders Medical Centre over the 6-month period of February 1994 through July 1994. The Flinders Medical Centre is a 400-bed acute teaching hospital serving a population of 340 000 in the southern metropolitan region of Adelaide, South Australia. The hospital separations database was screened for the diagnosis codes of 430 through 438 for cerebrovascular disease from the ninth revision of the International Classification of Diseases (only the primary diagnosis). The diagnosis of acute stroke was confirmed by a neurologist (C.A.) who reviewed all medical records of identified patients using the World Health Organization criteria of “rapidly developing symptoms and/or signs of focal, and at times global, loss of cerebral function, with symptoms lasting more than 24 hours or leading to death with no apparent cause other than that of vascular origin.”9 The term “global” refers to subarachnoid hemorrhage, cases of which were included in the study. We aimed to assess the outcome of all surviving patients at 1 year after the index stroke event that required admission to hospital. Patients (or the next of kin) who gave written informed consent underwent a face-to-face interview in their own homes with a research nurse (S.L.). The study was approved by the Clinical Investigations (Ethics) Committee of Flinders Medical Centre.
Pretesting of the authorized Australian version of the SF-364 among patients in a stroke rehabilitation unit showed that a high proportion were unable to self-complete the questionnaire because of visual problems, confusion, and physical disability. Hence, interview administration was the standard method used in this study. The SF-36 was administered first as part of a long interview schedule. Scores for the SF-36 were calculated using the methods set out by Ware et al5 and range from 0 to 100, with higher scores indicating a better health state. Internal consistency, or the extent to which items on the eight scales of the SF-36 are correlated with each other, was assessed with Cronbach's α, an inter-item correlation statistic.10 Construct validity was assessed by examining the extent to which the SF-36 scores varied according to predefined hypotheses. We hypothesized that the scores should vary in a predictable manner among patients with physical disability, mental ill health, and varying levels of social activities according to the measures outlined below.
Physical disability was measured using the BI,11 a quick, reliable, and well-validated measure of dependence in mobility and ADL that gives a score between 0 and 20 in 1-point increments. The top score of 20 implies “independence” but not necessarily the absence of physical disability. Mental ill health was measured using the GHQ-28,12 a widely used screening instrument for the detection of psychiatric disorders covering the symptoms of anxiety, depression, somatic disturbance, and social dysfunction. Scores of 5 or more on the GHQ-28 have been shown to indicate “probable” cases of psychiatric disorder among stroke patients.13 For this study, patients who scored this way were classified as having “mental ill health.” Social activities were measured using the AAP,14 a new instrument for the measurement of the lifestyle activities of elderly people that was developed from the Frenchay Activities Index.15 The AAP focuses on both a patient's behavior and physical capacity to undertake a range of daily activities that can be grouped into four distinct and meaningful clusters: domestic chores (eg, preparing meals, washing clothes), household maintenance (eg, gardening, house/car maintenance), service to others (eg, attending religious services, voluntary or paid employment), and social activities (eg, outdoor recreation or sport, social activities at a center).
Data were stored on a personal computer, and analyses were conducted with SPSS for Windows software (version 6).16 The Mann-Whitney test was used for comparisons of ranked scores, and multiple regression analyses were used to control for age and sex where necessary. Probability values are two-tailed.
Of the 124 survivors (73%) of the original cohort of 170, 90 (73%) were able and willing to complete the SF-36 (mean time to completion, 8 minutes). Six patients were found to have moved interstate, 15 patients refused participation, and 13 patients were unable to communicate or otherwise not assessable. The ages of the patient group ranged from 36 to 92 years (mean±SD, 72±12 years), and 53% were men. CT-confirmed ischemic stroke was diagnosed in 80% of patients, and 75% had experienced their first-ever stroke at baseline. Overall, 23% of patients were living in a hostel or nursing home at the time of follow-up.
Fig 1⇓ and Table 1⇓ show the frequency distribution of scores obtained for the eight SF-36 scales. Low scores were obtained for physical functioning, general health, and vitality, whereas the scores for social functioning were similar to those of the general population aged 65 years and over.17 For all eight scales except vitality, internal consistency by Cronbach's α satisfied Nunnally's criterion of 0.7.18 There was a significant decline in the mean scores for the physical functioning scales with increasing age (r=−.27, P=.01), and women had higher mean scores for the scales of role limitations–emotions (P<.001), social functioning (P=.01), and mental health (P=.06).
Construct validity was demonstrated by clear differences across all eight SF-36 scales for patients with identified health problems. Fig 2⇓ shows significant differences in the mean scores on the SF-36 for patients with physical disability or dependence in ADL (defined by the BI) and mental ill health (defined by the GHQ-28) compared with patients without these disabilities, but the strength of the associations varied in a predictable manner. Among patients dependent in ADL, the difference in mean scores was greatest for the physical functioning and general health scales, whereas for patients with emotional ill health the strongest associations were for the social functioning, role limitations–emotions, and mental health scales. Controlling for age and sex in multiple regression analyses did not alter the associations between the physical functioning scale and the BI (β=−0.55, P<.001), and the role limitations–emotions and social functioning scales and the GHQ-28 (β=−0.41, P<.001).
The SF-36 may be an indirect measure of “quality of life,” since important correlations are reported between the scales and several specific dimensions of quality of life, including living arrangements, financial situation, and family life.5 We therefore tested the validity of SF-36 as a measure of social functioning. Table 2⇓ shows that there was no association between the social functioning scale and the distinct domains of everyday activities relevant to many elderly patients covered by the AAP: service to others, such as caring for others and voluntary employment, and social activities, such as outdoor recreation and sport and social activities at a community center. On the other hand, the physical functioning scale of the SF-36 was significantly associated with all these grouped activities and is thus a sensitive measure of mild functional losses relevant to independent living. This issue is further illustrated by comparing the scores on the physical functioning scale (mean±SD, 48±33; variance, 1081) with those obtained on the BI (mean±SD, 18±3; variance, 11). The ceiling effect of the BI is well apparent, with the frequency distribution skewed toward higher scores, whereas the physical functioning scale of the SF-36 showed a uniform distribution of scores that reflected a broad range of physical disability (Fig 1⇑).
In both clinical practice and research, health measures are broadly applicable for establishing a baseline description of a condition or illness, screening those patients in most need of care, setting goals, and monitoring the success of interventions. The aims of assessments can differ, however, and therefore what is required of the measures may also differ substantially between different settings. The present evaluation confirms the psychometric validity of the SF-36 for patients with stroke, but it also highlights some limitations in this patient group.
The majority of formalized instruments developed and used to measure health outcomes emphasize, or are limited to, the constellation of physical variables known as ADL. While ADL scales identify those patients in most need of care, they fail to detect mild functional losses or other factors that may result from disabling illnesses such as stroke. These issues are particularly relevant to the assessment of the outcome from stroke because the majority of survivors continue to live at home with relatively good recovery of basic tasks of self-care. Yet many patients fail to return to their previous lifestyles because of depression and agoraphobia, for example, while the high level of emotional distress among caregivers may also be prominent.19 In view of the important interrelationships between physical functioning, mental state, and social factors, especially in the context of rehabilitation and care of the elderly, multidimensional measures are often used to provide a broad assessment of health status.
Evidence of the validity of the SF-36 to provide an assessment of both physical and mental health was revealed by a decline in scores in a predictable manner across the eight scales for stroke patients with physical disability defined by the BI and emotional ill health defined by the GHQ-28. Moreover, the sensitivity of the SF-36 to detect higher levels of everyday physical functioning allowed a broader range of needs to be identified. Our finding of a decline in mean scores for the physical functioning scale with increasing age has also been well documented elsewhere.4 5 17 20 21 In contrast to most other field studies, however, we found that women tended to score more highly than men on the SF-36, suggesting that women had better stroke outcomes, at least in terms of social functioning and mental health, in this study.
Additional critical information is often needed to assess a patient's ability to “return to normal living” after stroke, including such areas of “non-ADL” activities or “social functioning” as their ability to return to work and undertake leisure or recreational pursuits. We found that low scores on the social functioning scale of the SF-36 were not associated with low levels of social activities on the AAP. We therefore recommend that, for comprehensive monitoring of the health of this patient group, the SF-36 be supplemented by another measure of social functioning, although the number of instruments available for such purposes is limited.
In terms of acceptability of the SF-36 among disabled elderly people, the rate of missing data was extremely low in our study because the SF-36 was used in an interview setting. Other methods of administering the questionnaire to this patient group are likely to be compromised by a higher rate of missing data.
In summary, this study shows that the SF-36, when used in an interview setting, is suitable for use in elderly patients with stroke-related disability and is not compromised by high rates of missing data or poor construct validity. Depending on the results of ongoing research into its sensitivity to change, and whether it is felt that the cost of collecting this information is justified, the SF-36 may well prove to be an outcome measure suitable for administration to patients in clinical trials as well as routine practice.
Selected Abbreviations and Acronyms
|AAP||=||Adelaide Activities Profile|
|ADL||=||activities of daily living|
|GHQ-28||=||General Health Questionnaire|
|SF-36||=||Short Form 36 health survey questionnaire|
We thank David Candler and Cathy Murphy for help with administration of the survey, Michael Clark and Sally Rubenach for comments on the paper, and the South Australian Health Commission who funded the study.
- Received February 29, 1996.
- Revision received June 20, 1996.
- Accepted June 20, 1996.
- Copyright © 1996 by American Heart Association
Ware JE, Brook RH, Williams KN, Stewart AL, Davies-Avery A. Conceptualization and Measurement of Health for Adults in the Health Insurance Study, Vol 1: Model of Health and Methodology. Santa Monica, Calif: Rand Corp; 1980.
Stewart AL, Ware JE, eds. Measuring Functioning and Well Being: the Medical Outcomes Study Approach. London, UK: Duke University Press; 1992.
McCallum J. The SF-36 in an Australian sample: validating a new, generic health status measure. Aust J Pub Health Med. 1995;19:160-166.
Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual and Interpretation Guide. Boston Mass: Health Institute, New England Medical Centre; 1993.
Lyons RA, Lo SV, Littlepage BN. Comparative health status of patients with 11 common illnesses in Wales. J Epidemiol Community Health. 1994;48:388-390.
Jenkinson C, Peto V, Fitzpatrick R, Greenhall R, Hyman N. Self-reported functioning and well-being in patients with Parkinson's disease: comparison of the short-form health survey (SF-36) and the Parkinson's Disease Questionnaire (PDQ-39). Age Ageing. 1995;24:505-509.
De Haan R, Aaronson N, Limburg M, Langton Hewer R, Van Crevel H. Measuring quality of life in stroke. Stroke. 1993;24:320-327.
Clark MS, Bond MJ. The Adelaide Activities Profile: a measure of the life-style activities of elderly people. Ageing Clin Exp Res. 1995;7:174-184.
Holbrook M, Skilbeck CE. An activities index for use with stroke patients. Age Ageing. 1983;12:166-170.
SPSS for Windows, version 6.0. Chicago, Ill: SPSS Inc; 1993.
Behavioural Epidemiology Unit. South Australian Population Norms for the Short Form 36 (SF-36) Health Status Questionnaire. Adelaide, South Australia: South Australian Health Commission; 1995.
Nunnally JC. Psychometric Theory. 2nd ed. New York, NY: McGraw-Hill Book Co; 1978.
Anderson CS, Linto J, Stewart-Wynne E. A population-based assessment of the impact and burden of caring for long-term survivors of stroke. Stroke. 1995;26:843-849.
Jenkinson C, Coulter A, Wright L. Short-form 36 (SF-36) health survey questionnaire: normative data for adults of working age. BMJ. 1993;306:1437-1438.
Hayes V, Morris J, Wolfe C, Morgan M. The SF-36 Health Survey Questionnaire: is it suitable for use with older adults? Age Ageing. 1995;24:120-125.