Background
Transmission rates of HIV from an infected mother to her child range from 15 to 45% in the absence of any intervention; however, this can be reduced to less than 5% with antiretroviral therapy (ART) [
1,
2]. Yearly, 1.3 million HIV infected women become pregnant worldwide, and an estimated 85% of them receive ART to prevent mother-to-child HIV transmission (PMTCT) [
3]. Yet this observed high coverage in ART has not translated to elimination of MTCT, which is estimated to account for 90% of new HIV infections in children under 15 [
4]. Although a reduction from 190,000 to 150,000 infections globally per year was documented between 2015 and 2020, this decrease is still far from the World Health Organization (WHO) target of zero new infant infections by 2030 [
5,
6], and varies widely across highly affected countries [
2].
Preventing vertical HIV transmission requires passing through a complex cascade that spans approximately two years and multiple biological periods for women-infant pairs; covers multiple services spanning antenatal care, institutional delivery, breastfeeding and pediatric support; and requires progression through each service stage to benefit from the linked steps across the PMTCT cascade [
7‐
9]. Given that preventing vertical transmission requires successful completion of each PMTCT step – where progression through HIV testing, ART initiation and continuity with high adherence, safe delivery, adequate infant feeding, and infant antiretroviral prophylaxis is conditional on completion of previous cascade steps – services must be provided in a high quality and timely manner [
10]. The complexity of the PMTCT cascade – including the fragmentation of services and actors and its conditional nature – impedes successful implementation of evidence-based PMTCT guidelines [
8].
The WHO recommends that infants exposed to HIV during the intra- and peripartum periods be tested at 4–6 weeks of age or at the earliest opportunity beyond that time to optimize treatment strategies for infected infants [
11,
12]. In 2018, however, only 59% of HIV-exposed infants globally received a virologic test by 2 months of age [
13].
The prevalence of HIV in Mozambique is one of the highest in the world, with 13.2% of the adult population (15–49 years old) infected with HIV [
14]. Estimates of progress toward Elimination of MTCT (EMTCT) as of 2020 from UNAIDS and the Mozambican Ministry of Health (MoH) indicate that coverage of early infant HIV diagnosis by 8 weeks was 83%, and the total MTCT rate across pregnancy, delivery and the breastfeeding period was 13% [
15,
16]. In 2020, the number of new infections averted due to PMTCT in Mozambique is estimated at 20,000[13,000 – 34,000], with a lower transmission rate in infants tested before 8 weeks of life (5% vs. 16%) [
16].
Health service delivery is greatly influenced by the robustness of health systems [
17]. By identifying modifiable health systems factors that impact quality of care and service uptake, targeted strategies can be designed to optimize the delivery of health interventions [
18]. To date the majority of research on determinants of PMTCT service performance (including early infant diagnosis) in sub-Saharan Africa relies on individual-level data to describe demographic predictors [
19‐
21]. Limited evidence exists on the influence of supply-side determinants of PMTCT performance. We aimed to address this gap by investigating the influence of facility-level determinants on the performance of the PMTCT cascade, with a focus on early HIV testing in exposed infants. The identification of modifiable factors is important to recommend adaptations in service delivery and improve health outcomes.
Discussion
National protocols for infant HIV diagnosis in Mozambique recommend that the first PCR test for HIV-exposed infants be conducted between 1–9 months of age, with subsequent rapid tests between 9–18 months for those with a negative PCR test through 9 months of age (with a second PCR test for confirmation of positive rapid tests) [
28]. Programmatically, the MoH established 8 weeks of life as the target age for early infant diagnosis [
28].
In exploratory analysis, without accounting for each facility’s contribution to the sample size, the overall proportion of tests conducted within 8 weeks, after 8 weeks, and proportion positive are 73.5, 26.5 and 6.5% respectively. However, to account for heterogeneity in sample size of the HFs and to be able to report confidence intervals, we estimated single and pooled proportions of the outcomes. Our exploration identified considerable heterogeneity in early PCR testing across the HFs included in this study, ranging from 15 to 81% of infants tested within 8 weeks of life and a pooled proportion of 48%. Comparatively, the MoH reported an average population coverage and Manica province coverage of PCR within 8 weeks of 66 and 70% respectively, in 2018 [
29]. Our lower proportion might be attributable to the small sample of facilities in our study, with stratified selection of the largest facilities in each district, rather than the largest in the province overall.
The pooled proportion of PCR tests conducted increased to 69% when we consider both testing before and after 8 weeks of life, implying that at least 21% of all PCR tests were delayed beyond 8 weeks. The proportion of tests conducted after 8 weeks is comparable with the national average of 23% in 2018 [
29]. The number of PCR test results in our sample exceeds the number of PCR tests performed (2,382 versus 2,366). A potential explanation for this difference is that some exposed infants are transferred to the CCR services with PCR tests already performed in other services (e.g., postpartum, or pediatric care) or other health facilities and then the results of these PCR tests are forwarded to the CCR services where the infant is being followed up.
We found an overall proportion of positive PCR diagnoses of 6%, though six facilities had greater than 10% positive PCR tests. It is important to note that some facilities with higher proportion of positive tests had a small number of total tests, suggesting that testing may have been prompted by symptomatic infection in some cases. Comparatively, the national average MTCT rate for 2018 was estimated at 9.9%; however, this rate is higher for infants tested after 8 weeks (16.9%) compared with those infants tested within 8 weeks (7.8%) [
29].
Our results suggest that the caregiver-to-population ratio (a measure of workforce availability), glove stockouts and distance to the reference laboratory are associated with both EID and ever having been tested by PCR. Catchment area size and the involvement of multiple NGOs in the facility were associated with EID only (testing by 8 weeks of life). As hypothesized, increased staffing was associated with better performance in testing, but not with HIV transmission, as we suspect that individual patient-level factors continue to drive differential retention in care and ARV adherence, which in turn impacts viral suppression and the likelihood of vertical transmission [
30,
31]. Counterintuitively, stockouts of gloves were associated with higher likelihood to perform PCR. A potential explanation for this is that having stockouts in the pharmacy does not necessarily mean that there are no gloves in specific services, as each service typically maintains their own small stock [as confirmed by our service readiness assessment (data not shown)]. In case of stockouts, health facility managers often prioritize gloves for priority and/or services that put health workers at increased risk, such as services with interactions with HIV-infected patients.
Larger facilities are frequently located close to reference laboratories, while small and rural facilities are required to send their blood samples for processing at distant laboratories [as confirmed by our service readiness assessment (data not shown)]. Our sample had more rural facilities than urban facilities (as is the case throughout Mozambique, where small, rural facilities greatly outnumber large, urban facilities), which may explain the positive association between PCR testing and distance to the reference laboratory. Larger catchment areas were found to be associated with EID [
30], which is expected as larger catchment is associated with having larger facilities to meet the population needs, closer reference laboratories, and potentially higher likelihood of finding suspected cases compared with facilities serving smaller catchment areas [as confirmed by our service readiness assessment (data not shown)].
Contrary to our hypothesis, the presence of multiple NGOs supporting the facility, compared with not having any, was negatively associated with EID. A potential explanation is that having multiple NGOs working in the same location might lead to poor task coordination. This might particularly impact smaller facilities with few human resources, in which an overload of administrative tasks related to filling forms and reports for different projects may reduce available time to focus on clinical duties.
Surprisingly, facility urbanicity was not statistically associated with any of our outcomes, likely due to the small number of urban facilities compared with rural facilities. However, our visual exploration in Fig.
2 suggests that – overall – urban facilities tend to perform better than rural facilities.
Some results presented here align with published literature. In 2018 the WHO estimated that globally, only 56% of all HIV-exposed infants had access to EID by the second month of age [
32], which is similar to our findings. A retrospective cohort study in a district hospital in Zambezia, Mozambique found that the median age for first PCR test for infants was 5 months (IQR, 2–7 months) and 16% among 105 infants tested positive [
33]. However, this study was conducted in 2007–2008, prior to the introduction of Option B + and significant investments to expand PCR capacity. In Ethiopia a cohort study including 266 exposed infants found that 41% had had a PCR test within 6 weeks of life, and 13.2% of infants tested positive for HIV [
34].
A systematic review that assessed accessibility of services for EID of HIV in sub-Saharan Africa, including Mozambique, found that stockout of supplies, weak infrastructures, inadequate human resources training and lack of sufficient reference laboratories with PCR capacity were ongoing challenges for delivering PMTCT [
35]. Our results support the findings of this review.
There were some limitations to our study that merit caution in interpreting our results. First, this study was not designed to attribute causality. Second, the results are from a facility-level analysis and do not include individual-level determinants of PMTCT performance and vertical HIV transmission. Lastly, these results are from a relatively small sample of facilities in one province, and caution is merited in generalizing the results. Despite these limitations, there are notable strengths of our study: We used multiple data sources, the readiness data was collected with a standard tool, testing data was abstracted directly from patient registers, the study covers all districts in one province including urban and rural facilities, and facility-level findings are useful to inform local health managers in decision making.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.