On statistical characteristics of quality of life assessment.

In recent years, the assessment of drug effects on quality of life (QOL), which is the ability of a person to function normally in society as perceived by the patient, has become a popular issue in clinical trials. Since the response may vary widely from patient to patient and from time to time, the questionnaire (or instrument) for assessing the QOL of patients is required to be sensitive and reliable for detection of meaningful differences. QOL measurements are known to be correlated over time and among QOL indices. In this paper, we study some statistical characteristics of a given QOL instrument, such as precision and power, using a time series model that accounts for the correlation structure of QOL measurements over time. Formulae for sample size determination for clinical trials with time-correlated outcome measurements are presented.