Multilevel assessment of prenatal engagement in home visiting

Low engagement in prenatal home visiting may limit programme effectiveness to improve birth outcomes. Multiple factors may influence engagement.

A retrospective cohort study of first-time mothers enrolled in home visiting prenatally in southwest Ohio from 2007 to 2010. The primary outcome was enrolment by 20weeks' gestation; a secondary outcome included home visit frequency. Two multilevel assessments were conducted using random intercept multilevel modelling; maternal covariates were nested first within the home visiting agency and then within the ZIP code. In the first model, variations attributable to individual agency and agency volume were assessed. In the second model nested within the ZIP code, violence rates by ZIP code and interaction terms between violence rates and maternal factors were evaluated.

Of 837 women, 25.3% enrolled ≤20 weeks and 7.4% enrolled early and received ≥75% of expected visits. The first model demonstrated a significant variation in early enrolment based on clustering by agency (p<0.001), however, agency volume was not a significant predictor. In the second model, violence rate was not associated with early enrolment (AOR 0.92, p=0.08), but an interaction term with maternal race was significant (p=0.02). The effect of increasing community violence disproportionately affected early enrolment among white women (AOR 0.80, p=0.005) compared with black women (AOR 0.95, p=0.30). In both the random intercept multilevel models, teenagers demonstrated a decreased likelihood of enrolling early (AOR 0.58, p=0.046 and AOR 0.49, p=0.004).

Prenatal home visiting engagement is related to maternal, agency and community factors, presenting multiple opportunities to optimise programme implementation.