Trends and predictors of mother-to-child transmission of HIV in an era of protocol changes: Findings from two large health facilities in North East Nigeria

Background Research studies have demonstrated a reduction in the risk of mother-to-child transmission of HIV (MTCT) to less than 2%, or 5% in non-breastfeeding and breastfeeding populations, respectively, with antiretroviral interventions. However, the risk of MTCT in routine health-facility settings, where service delivery is usually sub-optimal needs monitoring. Method We conducted a retrospective review of data from 2008–2014, in two health facilities in Adamawa State, Nigeria. Descriptive statistics were used to estimate overall MTCT rate and MTCT rate by year, and period of prevention of mother-to-child transmission of HIV (PMTCT) protocol implementation. We conducted simple and multiple logistic-regression analyses, to identify predictors of MTCT. Results Data from 1,651 mother-to-infant pairs, with HIV deoxyribonucleic acid (DNA) polymerase-chain reaction (PCR) test results from 2008 (n = 49), 2009 (n = 246), 2010 (n = 280), 2011 (n = 335), 2012 (n = 290), 2013 (n = 225) and 2014 (n = 226) were analysed. The overall MTCT rate among HIV exposed infants (HEIs) was 9.7% (95% CI 8.3% – 11.1%) at a median age of 8 weeks (IQR = 6–20). The MTCT rate decreased from 14.3% (4.4%-24.2%) in 2008 to 4.9% (2.1%-7.7%) in 2014 (p = 0.016). The MTCT rate was the lowest (5.4% [3.7% – 7.0%]) when all pregnant women living with HIV received triple antiretroviral therapy, as treatment or prophylaxis (ARVT/P). Using the pooled data, we found that infant age, breastfeeding option, antiretroviral regimen and year were predictors of MTCT. The adjusted odds of MTCT were significantly higher, when neither mother nor HEI received ARVT/P (Adjusted odds ratio (AOR) 26.4 [14.0–49.8], and lower amongst infants born in 2012, compared with those born in 2008 (AOR 0.2 [0.0–1.0]). Conclusion The MTCT rate declined significantly between 2008 and 2014 in these two routine health-facility settings in Nigeria. Our study suggests that ARVT/P yields the lowest MTCT. Thus, efforts to scale up lifelong ARVT/P (Option B+) in Nigeria should be accelerated. We conducted a retrospective review of routine facility-held records; and analyzed routine individual-level patient data extracted from facility-based registers, from January 2008 to December 2014 (S1 File). All data were analyzed cross-sectionally. Records reviewed included those from early-infant diagnosis (EID)/Infant follow-up registers and HIV PCR request/result forms accessed from the PMTCT and medical record units of the hospitals. The review was conducted at Specialist Hospital (SH) Yola, a health facility providing predominantly secondary healthcare services in the Yola North Local Government Area (LGA), and Federal Medical Centre (FMC) Yola, a tertiary health facility in the Yola South LGA. The two health facilities commenced delivery of PMTCT services in September 2007 and January 2008, respectively. Other services rendered by both of these facilities included antenatal care (ANC), general out-patient and specialist services. The facilities render services to pregnant women and HIV-exposed infants (HEIs), in accordance with national cascades and protocols (S1 Fig and S1 Table). Using Lasec® DBS Collection kits with 5 spots, infant dried blood spot (DBS) samples were collected and couriered to the Federal Medical Centre, Jalingo. This centralized HIV deoxyribonucleic acid polymerase-chain reaction (DNA PCR) testing facility uses the Roche® brand of PCR machine; and is located approximately 167 kilometres away from Yola. We defined MTCT rate as the proportion of tested HIV-exposed infants (HEIs) who tested HIV positive. This formed the basis of our evaluation of PMTCT effectiveness as it allowed us to assess MTCT reduction in routine health-facility settings. For the overall MTCT rate, the MTCT rate by year, and the MTCT rate by PMTCT protocol period, the denominators were the total number of HEIs tested for HIV, with DNA PCR results: for the entire seven-year period, during a specific year, and during the specific protocol period; while the numerators were the number of HEIs that tested HIV positive: for the entire seven-year period, during a specific year, and during the specific period. We defined Period 1, as the period when only the 2007 National Guidelines were used (Jan 2008 –Jan 2010), and Period 2 as the period when only the 2010 National Guidelines were used (Jun 2012-Dec 2014). The transitional period refers to the period when both the 2007 and the 2010 Guidelines were used (Feb 2010 –May 2012). We reviewed regimens taken by HIV- positive pregnant women and HEIs, in order to confirm the duration of the transition period. From the pooled data, we estimated the transitional period to be two years and four months (Feb 2010 –May 2012). We defined exclusive breastfeeding as feeding an infant with breast milk only [25]. This excluded the use of formula feed, or any other liquids or solids [25]. The use of the prescribed medications and oral rehydration salt (ORS) for diarrhea was, allowed, as per WHO definitions [25]. We defined mixed feeding as feeding an infant with both breast milk and formula feed, or any other liquid, or solids [25]. Lastly, replacement feeding/not breastfed at all, refers to avoiding all breastmilk and feeding an infant with an appropriate replacement milk [25]. The data were analyzed by using STATA 14.2 [26]. Descriptive statistics (proportions) were used to describe the overall MTCT rate, MTCT rate by year, and the PMTCT protocol periods. Simple and multiple logistic regression analyses were conducted to establish the predictors of MTCT. Covariates were selected, based on the current literature on the risk factors for MTCT [19,21, 23, 24, 27–29]. Specifically, we included gender, infant age, breastfeeding option, ART/ARV prophylaxis receipt, hospital and year, in our logistics regression model. A p-value <0.05 was considered statistically significant. No adjustment was made in estimating the confidence intervals. All 158 infants, without first HIV DNA PCR test results, were excluded from all analyses. HIV-exposed infants excluded from the analyses differed significantly by age and feeding option, as well as by infant and maternal ARV use, compared with HEIs included in the analyses (S2 Table). For longitudinally linked results, if an infant tested HIV negative at first test and HIV positive at second testing, the HIV positive result was used in the analysis. Ethics approval for the study was obtained from the Senate Research Committee of the University of the Western Cape. Permission to access patient records was obtained from the research ethics committees of the two health facilities involved in this study, and from the Adamawa State Ministry of Health (SMoH). No individual patient-level consent was obtained; as no patient was interviewed. All data were de-identified: names of patients were not extracted. Individual patient-level consent was not required by the ethics review committees.