Reliability and Validity of Self-reported Muscle-strengthening Exercise in the Cancer Prevention Study-3

Abstract

Purpose: This study examined the 1-yr reliability and construct validity of survey items relating to time spent on muscle-strengthening exercise (MSE) in a subset of a large prospective cohort.

Methods: Participants (n = 293 men, 433 women; age, 32-73 yr) were selected from the Cancer Prevention Study-3. Information was collected using a 1-yr presurvey and postsurvey and four 7-d diaries throughout the year. The presurvey and postsurveys collected time spent on MSE in two ways: one question captured MSE activities performed during a typical 24-h period (24-h survey), and another question captured leisure-time physical activities performed in hours per week and months per year (LTPA survey). Time spent on MSE using the LTPA survey was calculated for individual MSE items and summed for total MSE time. One-year reliability was assessed by comparing the responses between the presurvey and postsurvey using Spearman's correlation coefficients. Construct validity was assessed by computing Spearman's correlation coefficients between responses from the postsurvey items and the diary. Additional analyses were conducted to examine whether reliability or validity varied by sociodemographic factors.

Results: Reliability estimates for all MSE items were moderate (≥0.40) or strong (≥0.60) overall and across demographic strata. Reliability estimates were strongest for total MSE on the LTPA survey (Spearman ρ = 0.75; 95% confidence interval (CI), 0.71-0.78) compared with the 24-h survey (0.59; 95% CI, 0.54-0.64). In contrast, the validity estimates were similarly strong for the total MSE on the LTPA survey (Spearman ρ = 0.71; 95% CI, 0.67-0.75) and the 24-h survey (Spearman ρ = 0.68; 95% CI, 0.64-0.72).

Conclusions: The CPS-3 surveys have acceptable 1-yr reliability and validity for self-reported time spent on MSE. Reliability and validity estimates are acceptable across all sociodemographic subgroups.