Methods

We conducted a retrospective cohort study using Kaiser Permanente Southern California (KPSC) electronic health record data. Youth between 5 and 17 years with continuous health care coverage were included if they had an in-person visit with at least 1 BMI measure before the pandemic (March 2019-January 2020) and another BMI measure during the pandemic (March 2020-January 2021 with at least 1 BMI after June 16, 2020, ie, about 3 months into the pandemic). Youth with complex chronic conditions were excluded.²^,3 Race and ethnicity based on caregiver report or birth certificates were used to compare with the underlying population. Outcomes were the absolute distance of a youth’s BMI from the median BMI for sex and age,⁴ weight adjusted for height, and overweight or obesity (≥85th or ≥95th percentile of BMI for age, respectively).⁵^,6 We fit mixed-effect and Poisson regression models accounting for repeated measures within each individual, using an autoregressive correlation structure and maximum likelihood estimation of covariance parameters to assess each outcome. Similar to an interrupted time-series design, we included a binary indicator representing the periods before or during the pandemic plus a calendar month by period interaction term. We divided youths into 3 age strata (5.0-<12, 12-<16, 16-<18 years) based on age at the start of the pandemic.

Models were adjusted for sex, race and ethnicity, state-subsidized health insurance, neighborhood education, neighborhood income, and number of parks in the census tract. Mixed-effects models also included BMI-for-age class at baseline. All analyses were performed with α = .05 for 2-sided tests using SAS version 9.4 (SAS Institute Inc). The KPSC institutional review board approved the study and granted a waiver for informed consent.

Read more...