Use of routine program data & evaluation data to maximize the impact of prevention of mother to child transmission of HIV (PMTCT) programs in Nigeria & Malawi
Radin, Anna K
MetadataShow full item record
Background Great progress has been made in preventing mother to child transmission of HIV (PMTCT), however programmatic and operational challenges remain. Utilizing routinely collected PMTCT program data to refine epidemiologic estimates and describe facility-level factors that influence maternal and infant outcomes is critical to addressing these challenges. Objectives In this dissertation, we (1) compared HIV positivity rates from PMTCT data to HIV prevalence estimates from antenatal care sentinel surveillance (ANC-SS) in Nigeria (2014-2015); (2) estimated local government area (LGA) HIV prevalence in Nigeria using PMTCT data (2015); and (3) determined the association between facility-level factors and maternal and infant HIV outcomes in Malawi’s national PMTCT program (2015-2016). Methods We used exploratory analyses and linear regression to quantify the difference between site- and state-level HIV prevalence estimates using PMTCT testing data (n=2.2 million) compared to ANC-SS data (n=36,431). We used Empirical Bayes approaches and multilevel modeling to generate LGA HIV prevalence estimates using PMTCT data. Finally, we used multilevel logistic regression to measure the association between facility level factors and individual-level maternal and infant outcomes using enrollment data from the National Evaluation of Malawi’s PMTCT Program (NEMAPP) (n=3,489 mother baby pairs, 54 facilities), facility staffing data, and PMTCT program data. Results ANC-SS-based HIV prevalence estimates at the facility and state levels differ significantly from estimates generated using routine PMTCT program data in Nigeria. Many states (62%) in Nigeria have statistically significant within-state variation in LGA-level HIV prevalence. In Malawi, mothers who sought care at public facilities (vs faith-based) were significantly more likely to have infants testing positive for HIV at 6 weeks postpartum and were less likely to be on ART during pregnancy. Mothers seeking care at sites with a high provider-to-patient ratio, a high proportion of patients newly diagnosed with HIV, and who traveled > 2 hours to reach the facility were significantly more likely to be on ART during pregnancy. Conclusions This dissertation demonstrates how routinely collected program data can be used to generate granular descriptive epidemiologic data, and to complement evaluation, survey, and surveillance data in multilevel analyses, allowing leaders to diagnose issues and intervene accordingly to improve PMTCT programs.