Accuracy of physical activity assessment during pregnancy: an observational study

Prenatal physical activity may improve maternal and infant health and lower future disease risk for both mother and baby; however, very few physical activity assessment methods have been validated for use during pregnancy. The purpose of this study was to evaluate the accuracy of a subjective physical activity record (PAR) and an objective activity monitor, against a reference standard to quantify moderate and vigorous physical activity (MVPA) in pregnant women. The reference standard was based on participant interviews to determine if a woman was an exerciser and confirmed with information obtained from the PAR and a heart rate monitor.

Fifty-two pregnant women completed a physical activity record (PAR) and wore a SenseWear® Mini Armband (SWA) activity monitor over a 7-day period at 18 weeks gestation. Total minutes spent in MVPA were totaled from both modalities and evaluated against the reference standard using contingency analysis and Pearson's chi-square test to evaluate the number of women meeting minimum prenatal physical activity recommendations (at least 3, 30 minute sessions of exercise per week). Both modalities were also tested individually and collectively to assess their ability as indicators of activity using empirically determined cut-offs as indicated by receiver-operator characteristic curves. These experimentally-derived criteria were also tested with Pearson's chi-square test.

According to the reference standard, 13 of 52 participants (25%) met the criterion of 3, 30 minute sessions of volitional, moderate-intensity activity. When compared to the reference standard, both the PAR and SWA overestimated exercise status; 42 (81%) and 52 (100%) participants, respectively, achieved 90 minutes of MVPA (P < 0.0001 for both comparisons). Single-modality predictors of MVPA did not show a significant correlation. A composite predictor of MVPA offered the most favorable option for sensitivity and specificity (true positives, n = 8 and true negatives, n = 36) using cut-offs of 280 and 385 minutes/week for the PAR and SWA, respectively.

Compared to the reference standard, time spent in MVPA obtained from the PAR or SWA overestimated the prevalence of women meeting prenatal exercise recommendations. The most accurate predictor of women meeting current prenatal exercise guidelines was identified by using the PAR and SWA collectively.