Figure 4. Adjusted Associations Between Eliminating Outdoor Play and Adiposity Measures After Lockdown Among Those Who Reported Any Outdoor Play Before Lockdown.

Base models (A, point estimates in natural units; C, in ratios) were adjusted for baseline prelockdown measures and precision variables (age, sex, and days since end of lockdown) and interacted with cohort (P values for interaction = .01 for BMI and .03 for BMI z score). Final models (B, point estimates in natural units; D, in ratios) were further adjusted for confounders: ethnicity, maternal education, and prelockdown household income. Final models were fit for primary school–aged children (Growing Up in Singapore Towards Healthy Outcomes [GUSTO]) only because of null crude association in the preschool-aged children. Point estimates are provided in natural units and ratios of eliminated to not eliminated (due to log-transformation); 95% CIs and P values are reported.

^aP < .05.