Abstract / Summary
Abstract Background: Adherence to physical activity (PA) guidelines in pregnancy is low globally, and evidence in Chinese women is scarce, and available Chinese studies have focused predominantly on early pregnancy, leaving the correlates of guideline adherence in late pregnancy—when activity declines most steeply—poorly characterized. This study aimed to describe physical activity patterns among pregnant women in three Chinese provinces, examine centre-level variation in these patterns, and identify factors associated with meeting international PA recommendations in a sample predominantly in late pregnancy. Methods: A cross-sectional survey of 335 pregnant women in three provinces (94.6% in the third trimester) used validated Chinese PPAQ, EBBS and P-ESES scales. Compliance was defined as ≥150 min/week moderate-intensity sports/exercise activity (≥450 MET-min/week, ACOG-aligned). Multivariable logistic regression adjusted for nine parameters including maternal age, gestational age, education, employment and study centre; a pre-specified sensitivity analysis refitted the model using an all-domain adherence outcome (≥450 MET-minutes/week across all PPAQ domains). Results: Centres differed in sedentary time and light-to-moderate PA intensity (p<0.05) but not in overall compliance or in barriers, benefits and self-efficacy (all p>0.05). Univariable analyses linked barriers (OR=0.92), benefits (OR=1.03) and self-efficacy (OR=1.05) to PA adequacy, yet only perceived barriers remained significant in the adjusted model (aOR=0.93, p=0.004; Nagelkerke R²=0.085). Conclusions: Barriers were the only psychosocial factor independently associated with meeting PA recommendations, suggesting that perceived barriers may represent a candidate target for future intervention studies. In a pre-specified sensitivity analysis using the all-domain outcome (79.1% adherent), none of the psychosocial correlates remained associated with adherence, whereas both Zhejiang and Shaanxi showed lower adjusted odds than Guangdong, indicating that the psychosocial association was specific to the sports/exercise domain. Findings support transferability across settings rather than province-level generalization; the cross-sectional design limits causal inference.