The aim of this study was to determine the reliability and validity of
relatives' assessment of patients' quality of life and to measure the
agreement between patients' and relatives' responses to the Short For
m 36 quality of life questionnaire, at discharge from and 6 months fol
lowing intensive care treatment. Ninety-nine patient-relative pairs we
re studied. Reliability was quantified by using measures of internal c
onsistency (Cronbach's alpha and correlation coefficients) and reliabi
lity coefficients. Relatives' responses met the required standards of
reliability and validity, but reliability was consistently weaker in t
he mental health dimension. Relatives' and patients' scores differed s
ignificantly in six dimensions at discharge; however, agreement betwee
n patients' and relatives' responses, as measured by the Kappa statist
ic, was fair, improved over 6 months, and was greatest in aspects conc
erning physical health. We conclude that relatives are able to give a
good proxy assessment of functional aspects of quality of life.