Introduction: the Medical Outcomes Study short form 36 health survey (
SF-36) is being increasingly used and recommended as a suitable measur
e of subjective health status. However it is unlikely that any measure
will be appropriate for all groups. We wished to determine the suitab
ility of the SF-36 for assessing quality of life in older stroke patie
nts. Methods: a screening questionnaire was used to identify prevalent
cases of stroke from a random sample of 2000 subjects aged 45 years a
nd over. The SF-36 was included as part of a self-completion questionn
aire posted to each stroke patient. Data quality indicators were analy
sed. Results: we identified 104 cases of stroke and the response rate
for the SF-36 questionnaires sent was 83%. Completion rates for indivi
dual items ranged from 66 to 96%. All items in the role physical and r
ole emotional scales had completion rates < 75%. The percentage of sub
jects for whom an individual scale score could be computed ranged from
67 to 96%, being lowest for the role physical and role emotional scal
es. Floor effects were high (> 15%) for these two scales and for the s
ocial functioning and physical functioning scales. Ceiling effects wer
e substantial (> 15%) for the two role effect scales and for social fu
nctioning and bodily pain. Conclusions: this study has shown high resp
onse rates from older stroke patients to a postal questionnaire incorp
orating the SF-36. The poor completion rates and consequent inability
to compute scores for a large proportion of responders in certain scal
es raises concerns about the perceived relevance of these sections. Re
sults for the response effects suggest that, on its own, the instrumen
t is not suitable for assessing outcome. When data quality indicators
were examined, it appears that postal administration of the SF-36 is n
ot appropriate for assessing quality of life of older stroke patients.