Convergent Validity of Ecological Momentary Assessment to Assess Free-Living Sedentary Behavior and Physical Activity

Abstract

Self-report measures of physical activity (PA) can result in respondent over-reporting. Accelerometers provide direct estimates of movement but lack context. Ecological momentary assessment (EMA) may elicit PA estimates that are less prone to bias than traditional self-report measures, while providing context. PURPOSE: To examine the convergent validity of EMA-assessed PA compared to an accelerometer. METHODS: In 2014, the Pathways between Socioeconomic Status and Behavioral Cancer Risk Factors (PATHS) Study enrolled 238 free-living adults from Dallas, TX. Participants completed surveys of PA behaviors [International Physical Activity Questionnaire (IPAQ) and Behavioral Risk Factor Surveillance System (BRFSS)], and wore an accelerometer on their waist while concurrently completing daily EMAs for 7 days. Weekly summary estimates included: sedentary time, moderate-, vigorous-, and moderate- to vigorous intensity physical activity (MVPA). Spearman correlation coefficients and Lin’s concordance correlation coefficients (LCC) were used to examine the linear association and agreement for EMA and self-report measures compared to accelerometry. RESULTS: Participants were aged 43.3 (±13.1) years and predominantly Black (51.7%), overweight/obese (74.8%), and low income (52.4%). Based on accelerometer data, participants spent an average of 3,400.8 (±864.0) minutes per week sedentary; median (25th, 75th percentile) levels of MVPA was 121.5 (66.0, 225.0) minutes per week. The linear associations of EMA and self-reports with accelerometer estimates were statistically significant for sedentary time (EMA: ρ=0.16 [p=0.03]), moderate intensity PA (EMA: ρ=0.29 [p<0.01]; BRFSS: ρ=0.17 [p=0.02]; IPAQ: ρ=0.24 [p<0.01]) and MVPA (EMA: ρ=0.31 [p<0.01]; BRFSS: ρ=0.17 [p=0.02]; IPAQ: ρ=0.20 [p<0.01]). Only EMA estimates were statistically significant compared to accelerometer for agreement (moderate intensity PA: LCC=0.30 [95%CI: 0.18, 0.43]; vigorous intensity PA: LCC=0.06 [95%CI: 0.03, 0.10]; MVPA: LCC=0.28 [95%CI: 0.16, 0.41]). CONCLUSIONS: EMA showed better correlation and agreement to accelerometer estimates than traditional self-report methods. These findings suggest that EMA may be a practical alternative to assess PA in free-living settings. Funded by MRSGT-10-104-01-CPHPS (to DK)