Test-Retest Reliability of the Women's Health Initiative Physical Activity Questionnaire

Abstract

PURPOSE: Since few physical activity (PA) questionnaires were designed to measure the lifestyles and activities of women, we sought to examine the test-retest reliability of a physical activity questionnaire (PAQ) designed for women and to explore differences in reliability by race/ethnicity. METHODS: Participants (n=1092) were post-menopausal women aged 50–79 years, randomly selected from the baseline sample of participants in the Women's Health Initiative (WHI) Observational Study. Women were selected from all 40 WHI clinic centers, stratified by age and race/ethnicity (66% white, 13% African American, 14% Hispanic). The WHI PAQ collects usual frequency, duration and pace of recreational walking, frequency and duration of recreational activities or exercises (mild, moderate and vigorous), household, and yard activities. Summary variables were calculated by multiplying the frequency by duration and the metabolic equivalent (MET) level for that activity to obtain MET-hours per week. Approximately half of the women (n=569) repeated questions on recreational PA, the other half (n=523) repeated questions related to household and yard activities (mean 3 months apart). Test-retest reliability was assessed with weighted kappa and intraclass correlation coefficients (ICC). RESULTS: Overall, questions on recreational walking, moderate recreational PA, and vigorous recreational PA had higher test-retest reliability (weighted kappa range 0.50-0.60), than questions on mild recreational PA (weighted kappa range 0.35-0.50). The ICC for moderate to vigorous recreational PA was 0.71 (95% CI 0.67, 0.75) and total recreational PA was 0.75 (95%CI 0.71, 0.78). Substantial reliability was observed for the summary measures of yard activities (ICC 0.71; 95%CI 0.66, 0.75) and household activities (ICC 0.60; 95%CI 0.55, 0.66). No meaningful differences were observed by race/ethnicity for these measures. CONCLUSIONS: The WHI PAQ demonstrated moderate to substantial test-retest reliability in a diverse sample of post-menopausal women. An important next step to understanding the psychometric properties of this PAQ will be to assess validity.