Evidence for the validity of the short-form 36 questionnaire in an elderly population - SF-36

Age and Ageing,  May, 1994  by Ronan A. Lyons,  Huw M. Perry,  Beverley N.C. Littlepage

• E-mail
• Print
• Link

The objective of this study was to determine whether the Short-form 36 Health Status Questionnaire (SF-36) is suitable for use in an elderly population. The SF-36 was administered by interview to a random sample of 827 adults from West Glamorgan. Among the 216 adults aged 65 and over the data were 98.8% complete. Evidence for a high degree of internal consistency was good with Cronbach's alpha statistic exceeding 0.8 for each parameter. The evidence for construct validity was also good with the SF-36 distinguishing between those with and without markers of poorer health. The SF-36 is suitable for use with an elderly population when used in an interview setting.

Introduction

Interest in assessing the health status of populations and individuals has been growing over recent years. One of the promising health status measures, recently promoted, is the Short-form 36 Questionnaire (SF-36), which was developed for use in the Medical Outcomes Study in America (1)(2). A version of this instrument has been validated for use in Britain in two population-based postal surveys (3)(4). The first, in Sheffield, included persons between the ages of 16 and 74 and reported a higher level of missing data in the 65-74-year-old group (data not presented) and suggested that the SF-36 may not be a suitable instrument for elderly people (3). The second survey, in Oxford, limited its scope to the 16-64-year-old group, but the authors repeated the above suggestion in their discussion (4). In view of its ability to distinguish between persons suffering from a variety of widely differing conditions, the SF-36 has been mooted as a potential patient outcome measure in the NHS (5). Since elderly patients are the main users of the health services, a health status measure which was not suitable for them would have limited its practical use.

In 1993, an interview-administered health status survey, which included the SF-36, took place in West Glamorgan. This study contained a sufficient number of elderly persons to test the validity of the SF-36 when used in this group. The purpose of this paper is to determine whether the extent of missing data, or deficiencies in internal consistency or construct validity, limit the usefulness of the SF-36 in elderly populations.

Methods

The SF-36 is a 36-item questionnaire which measures eight parameters of health status: physical functioning (10 items), role limitations due to emotional problems (3 items), role limitations due to physical problems (4 items), social functioning (2 items), mental health (5 items), bodily pain (2 items), vitality (4 items), and general health perceptions (5 items) (1). For each parameter, scores are coded, summed, and transformed to a scale from 0 to 100, with higher scores indicating better health.

A questionnaire which contained the SF-36 and a variety of questions on lifestyle, health service utilization, and self-reported physician-treated conditions was administered by health visitors to a random sample of adults from West Glamorgan, aged 20-89 years, in their own homes, drawn from the register of Family Health Services Authority.

Internal consistency, or the extent to which items on a scale are correlated with each other, was assessed by Cronbach's alpha, an inter-item correlation statistic with a range of 0-1, with higher values indicating that the scale taps an underlying single dimension (6).

Construct validity, the extent to which pre-defined hypotheses are supported was assessed by comparing responses between groups which would be expected to differ in health status: those with a long-standing limiting disability, or who had attended their general practitioner in the previous 2 weeks, or were hospital inpatients or outpatients in the previous 12 months.

The Kruskal-Wallis test was used to determine whether groups differed in scores on the eight parameters of the SF-36 and the [X.sup.2] test was used to assess differences in proportions.

Results

From the 1201 individuals selected 192 (16%) were found to have moved or died, and 827 (69%) completed questionnaires were obtained. Excluding those who had moved or died yielded a corrected response rate of 82%. Among the 827 respondents 216 (26%) were aged between 65 and 89 years with this group having a mean age of 73.9 (SD 6.0) years. Comparison with 1991 census data indicated that the response rate in those over 65 was marginally higher than those aged 20-64 years, with 26% of respondents being aged 65 and over compared with 22% of those sampled (7). For those aged 65 and over data on the SF-36 questionnaire were 98.8% complete.

Cronbach's alpha statistics for the eight variables of the SF-36 instrument for respondents aged 65 and over are given in Table I.

Table I. Cronbach's alpha statistic for variables of the SF-36 questionnaire

Variable                     No. of subjects     Alpha
Physical functioning         216                 0.94
Role limitations:
Physical                     215                 0.92
Emotional                    213                 0.94
Social functioning           214                 0.85
Mental health                212                 0.82
Vitality                     211                 0.84
Pain                         212                 0.89
General health perceptions   216                 0.83