Assessing the construct and predictive validity of the early healthy lifestyles screening questionnaire

Background The purpose of this study was to assess the construct validity of the Early Healthy Lifestyles (EHL) behavioral screening questionnaire and to establish its predictive validity on change in physical growth in infants and children. Methods This retrospective, longitudinal cohort study used data collected alongside annual well-child visits (WCV) occurring at an integrated healthcare system (Pennsylvania, USA). The EHL launches in the electronic health record two weeks prior to each scheduled WCV for children aged 0–26 months. The EHL is available in English, requires 2–5 min to complete, and can be completed by parents in the patient portal prior to arrival, on a clinic iPad in the waiting room, or in the exam room via staff interview. Patients with EHL questionnaires collected between the ages of 1 to 2 years and between March 2016 and March 2020, were included in this analysis. Confirmatory factor analysis (CFA) was used to model the underlying latent structure of the questionnaire. Four different latent factors, diet, feeding, activity, and sleep, were used. To evaluate predictive validity, all questionnaires were aggregated into a single composite measure for each child. In this aggregation, any evidence of a specific behavior was coded to the same group. Lasso regression was used to identify which variables were related to the outcome of interest. Results Nine thousand six hundred eighty-four questionnaires were collected and analyzed for 5,444 children. Overall, the model had adequate fit (TLI: 0.913; CFI: 0.935; RMSEA: 0.029), and, in all cases, estimated coefficients were significant. Any reported consumption of juice (coefficient: 0.13; 95% CI: 0.04, 0.22; p-value: 0.006), parent use of a screen during play (coefficient: 0.14; 95% CI: 0.04, 0.25; p-value: 0.009), and parent concern that the child eats too much (coefficient: 0.20; 95% CI: 0.00, 0.39; p-value: 0.049) were associated with increases in BMI z-score from 1-year to 5-years. Conclusions The EHL is one of few brief behavioral questionnaires that is both integrated into routine preventive care in a healthcare system and measures multiple lifestyle domains. This study demonstrates the content validity of the EHL and its predictive validity in assessing change in BMI z-score from 1 to 5 years of age.