Factors Associated With Changes in Fruit Intake During Young Adulthood: A Classification and Regression Tree Analysis of Longitudinal Data

Abstract

Objective: To examine whether distinct participant groupings for changes in fruit intake (FI) levels between ages 23 and 31 years are identifiable based on both time-varying and time-invariant sociodemographic and behavioral variables.

Methods: Data were derived from the National Longitudinal Survey of Youth-1997, US. Change in FI frequency constituted the dependent variable. For 21 variables, changes and averages in 2007-2011 were calculated. Classification and regression tree analysis was conducted using Generalized, Unbiased, Interaction Detection, and Estimation software.

Results: Analysis isolated 5 variables (changes in smoking, drinking alcohol, and television viewing, plus 5-year mean of income-to-poverty ratio and computer use) and associated cutoff values to identify 7 groups of participants with differing degrees of FI change.

Conclusions and implications: Multiple groupings existed within upper social strata; a majority maintained healthy behaviors whereas some adopted substance use stress-coping mechanisms. Some low-income individuals demonstrated a capacity to adopt healthy behaviors. Dietary interventions could identify behavioral clustering, with emphasis on drinking, smoking, and screen time.

Keywords: alcohol; classification and regression tree; fruit intake; health-related behaviors; screen time; smoking.