![[ Visit Client Website ]](images/banner.jpg)
As in other areas, Multnomah County Health Department’s (MCHD’s) home visitation (HV) program employs nurses and community health workers to provide education and facilitate use of health and social services for pregnant women at higher-risk for adverse pregnancy outcomes. The purpose of this analysis is to conduct a quantitative evaluation of HV’s effectiveness on pregnancy outcomes.
METHODS:
HV participation data were linked to 2008-2009 birth records for singletons using deterministic and probabilistic strategies. Propensity scores (PS) were generated using the following predictors of HV participation: maternal age, race, parity, education, insurance status, WIC participation, medical risk, smoking and their significant interactions. PS-Matching was performed with a Greedy algorithm, overall and separately for the following subgroups: teens, blacks/African-Americans, and first-time mothers. Generalized-estimating-equation modeling, accounting for the matching, was used to generate relative risks (RR) and 95% confidence intervals (95% CI) to compare the following outcomes for HV participants and non-participants: receipt of adequate prenatal care (PNC), small-for-gestational age, preterm birth, and Neonatal Intensive Care Unit admission. We used tests-for-trend to evaluate whether outcomes improved with increasing numbers of visits among HV participants.
RESULTS:
Of the 19,455 births in the sample, 1,743 had mothers who received HV and 97.1% of them were successfully one-to-one matched on PS to non-HV participants, resulting in 1,693 matched pairs with similar characteristics. In unadjusted analyses with all 19,455 births, HV was significantly associated with worse outcomes. With PS-matched analyses, HV was marginally significantly associated with adequate PNC (RR= 1.04, 95% CI= 0.97-1.10, p<0.10) among all women and among teens, blacks/African-Americans, and first-time mothers. For HV participants, having more visits was associated with an increased likelihood of adequate PNC (p<0.0001). HV had no effect on other outcomes studied.
CONCLUSIONS:
HV was associated with adequate PNC, but not improved birth outcomes. HV may occur too late to improve birth outcomes. However, it may affect important unstudied outcomes, such as parent-child attachment and mental health. PS-matching can be useful in generating comparable groups and reducing selection bias when evaluating public health impacts of MCH programs and policies at the local level.