Modeling multiple risks during infancy to predict quality of the caregiving environment: contributions of a...

Area of Science:

• Child Development
• Maternal Health
• Statistical Modeling

Background:

• Understanding infant risk factors is crucial for predicting caregiving quality.
• Existing methods may not fully capture the complexity of multiple risk interactions.
• Research in low-income, non-urban settings is vital for equitable child development insights.

Purpose of the Study:

• To compare variable-centered and person-centered methods for modeling infant risk factors.
• To predict caregiving environment quality at six months using early childhood risk indicators.
• To explore these relationships in a population from low-income, non-urban communities.

Main Methods:

• Compared bivariate, multiple regression, cumulative risk index, and latent class analysis (LCA).
• Utilized nine risk factors (demographics, maternal psychosocial) assessed at two months.
• Analyzed data from 1047 infants in Pennsylvania and North Carolina.

Main Results:

• Latent class analysis (LCA) provided a more intuitive summary of risk organization than other methods.
• Identified five distinct risk classes: married low-risk, married low-income, cohabiting multiproblem, single low-income, and single low-income/education.
• LCA revealed variations in caregiving quality prediction across race and site based on family configurations.

Conclusions:

• Person-centered models, like LCA, offer valuable insights into the complex interplay of multiple risk factors.
• LCA enhances understanding of how different combinations of risks impact infant caregiving environments.
• Findings underscore the utility of LCA for nuanced prediction in diverse populations.