Keywords

health services research/mt (methods), health status indicators, measurement error, quality of life, study design, survey methods

 

Authors

  1. Miller, Donald R. ScD
  2. Clark, Jack A. PhD
  3. Rogers, William H. PhD
  4. Skinner, Katherine M. PhD

Abstract

Patient-based assessments of functional status and well-being, such as the short form survey (SF-36) developed in the Medical Outcomes Study, are increasingly used to monitor patient health status and the effects of medical care. Although they have been used in many populations under a variety of circumstances, little is known about how environmental factors, such as place of administration, influence patients' responses. We administered the SF-36 as part of comprehensive quality-of-life assessments to 304 participants in the pilot phase of the Veterans Health Study, a large observational study of health outcomes in VA ambulatory care. SF-36 questionnaires were self-administered twice by patients, once at home and again as part of an interview in a clinic setting. For most participants, less than a week passed between the 2 administrations. Scores from these questionnaires were compared as a measure of reliability and of the possible influence of place of administration. Questionnaires self-administered at the time of the interview yielded mean scores that were significantly (P < .05) higher across all 8 SF-36 scales (physical function, role function with physical and emotional limitations, vitality, bodily pain, social function, mental health, general health perceptions) and both the physical and mental component summary scales. With scores scaled from 0 to 100, differences ranged from 2.1 (bodily pain) to 5.7 (role limitations due to emotional problems). Mean physical function was 56.8 at the time of the interview, and 52.4 at home. Higher scores from questionnaires administered at interview outnumbered lower scores by 3 to 2 for most scales. These differences remained even after restricting the sample to those with the highest cognitive function scores and the shortest interval between administrations. Because selection factors and order of administration could not be completely dismissed, a large number of other administrative, clinical, and sociodemographic factors were examined, which, however, failed to provide adequate explanation for these differences. Careful consideration should be given concerning the physical and social environment in the administration of health-related quality-of-life assessments. Findings from this study suggest that more favorable measures of self-reported functional status and well-being may be expected from clinic administrations of instruments.