OBJECTIVES. The present study assessed changes in the Medical Outcomes
Study 36-item short-form health survey (SF-36) during a Ill-month per
iod and examined the relation of those changes to selected baseline ch
aracteristics. METHODS. The study was a IF-month follow-up evaluation
of 786 disadvantaged adults aged 50 to 99 years old who had participat
ed in a randomized controlled clinical trial in the general medicine o
utpatient clinics of a major academic medical center. Descriptive and
psychometric analyses of changes in the SF-36 scale scores during a 12
-month period were performed, and two series of multivariable logistic
regressions of increases or decreases greater than one standard error
of measurement (SEM) versus stability on selected baseline characteri
stics were done. Measures were the eight SF-36 scales. RESULTS. Mean b
aseline scores on the SF-36 scales were substantially below age-specif
ic national norms. Problematic floor and/or ceiling effects were found
for the bodily pain, social function, role-physical, and role-emotion
al scales, consistent with age-specific national norms. Internal consi
stency was unacceptable for the general health perceptions scale, adeq
uate for the social function scale, and good for all the other SF-36 s
cales. Improvements greater than one standard error of measurement wer
e found for between one fifth and one third of the patients, and decli
nes greater than one standard error of measurement were found for betw
een one fifth and one third of the patients. Selected baseline charact
eristics generally were unrelated to either improvements or declines o
n the SF-36 scales. CONCLUSIONS. The SF-36 scales appear to be suffici
ently sensitive for measuring changes in health outcomes during a I-ye
ar period in older patients with debilitating disease. Little of the m
easured change, however, was predictable.