Selection of quality-of-life measures for a prevention trial: a psychometric analysis

Quality-of-life (QoL) outcomes have become increasingly important in the evaluation of health interventions. The objective of the present study was to determine which of three generic QoL instruments was most suitable for use in an 8-year nutritional primary prevention trial. We compared the Duke Health Profile, the Nottingham Health Profile, and the Short Form Health Survey Questionnaire (SF36). We conducted the comparison in two stages: (1) a statistical analysis of data from a pilot study (n = 963) comparing the psychometric properties of the three instruments; and (2) an assessment of the practicality of the tools. With regard to psychometric properties, convergent validity was comparable for the three scales, and the correlation with Global Health Assessment ranged from 0.24 to 0.72. Discriminant validity was best for the SF36, with a difference between scores in healthy subjects and those with chronic disease in the range of 4.4 to 15.8 (scores could range from 0 to 100, where 100 indicates perfect health). Reproducibility was good for all three instruments, with a test-retest intraclass correlation coefficient of over 0.60 for most dimensions. DHP and SF36 performed best in terms of responsiveness. We judged the practicality of the three tools as satisfactory. We chose the SF36 for its high responsiveness. We also selected the Duke Health Profile for its practicality and favorable psychometric properties.