Get access

Psychometric properties of the SF-36 in the early post-stroke phase

Authors

  • Suzanne Hagen BSc MSc CStat,

  • Carol Bugge MSc BN,

  • Helen Alexander BA BSc MSc PhD


Suzanne Hagen, Nursing Research Initiative for Scotland, Glasgow Caledonian University, Cowcaddens Road, Glasgow G4 0BA, UK.
E-mail s.hagen@gcal.ac.uk

Abstract

Background.  Patients who have had a stroke are a large group in whom long-term disability is common and therefore impaired general health is likely. The Short Form 36 (SF-36) is a popular measure of general health that has been used with this patient group, but not all aspects of its psychometric properties have been established for use in this context, and its use in the early post-stroke phase has been neglected.

Aims.  To examine the reliability, validity and sensitivity to change of the SF-36 (UK version I) in patients in the early post-stroke period.

Design.  A prospective, observational study of stroke outcomes was carried out.

Research methods.  From May 1996 to April 1997, patients who had had a stroke were identified by 24 general practices in Scotland and were recruited within 1 month of their stroke, whether in hospital or at home. Outcome measures including the SF-36 were administered at one, 3 and 6 months after onset.

Results.  The internal consistency of the eight subscales at all three time-points was good except for 1 month Vitality (α = 0·6824) and 3 month General Health (α = 0·6650), which were borderline in comparison with the criterion value of 0·7. Construct validity was adequate overall, although correlations between Role Physical and General Health and the Barthel Index and Canadian Neurological Scale were lower than hypothesized. Most SF-36 subscales were sensitive to change between 1 and 3 months post-stroke, but none detected change between 3 and 6 months.

Conclusions.  There were some practical problems in using the SF-36 in an acutely unwell stroke population. However, analysis of psychometric properties suggested that most of the subscales were adequately reliable and valid. Sensitivity to change was poorer in the later stages of the study.

Ancillary