What will it take for the Global Plan priority countries in Sub-Saharan Africa to eliminate mother-to-child transmission of HIV?

BACKGROUND: The 2016 ‘Start Free, Stay Free, AIDS Free’ global agenda, builds on the 2011-2015 ‘Global Plan’. It prioritises 22 countries where 90% of the world’s HIV-positive pregnant women live and aims to eliminate vertical  transmission of HIV (EMTCT) and to keep mothers alive. By 2019, no Global Plan priority country had achieved EMTCT; however, 11 non-priority countries had. This paper synthesises the characteristics of the first four countries validated for EMTCT, and of the 21 Global Plan priority countries located in Sub-Saharan Africa (SSA). We consider what drives vertical transmission of HIV (MTCT) in the 21 SSA Global Plan priority countries. METHODS: A literature review, using PubMed, Science direct and the google search engine was conducted to obtain global and national-level information on current HIV-related context and health system characteristics of the first four EMTCT-validated countries and the 21 SSA Global Plan priority countries. Data representing only one clinic, hospital or region were excluded. Additionally, key global experts working on EMTCT were contacted to obtain clarification on published data. We applied three theories (the World Health Organisation’s building blocks to strengthen health systems, van Olmen’s Health System Dynamics framework and Baral’s socio-ecological model for HIV risk) to understand and explain the differences between EMTCT-validated and non-validated countries. Additionally, structural equation modelling (SEM) and linear regression were used to explain associations between infant HIV exposure, access to antiretroviral therapy and two outcomes: (i) percent MTCT and (iii) number of new paediatric HIV infections per 100 000 live births (paediatric HIV case rate). RESULTS: EMTCT-validated countries have lower HIV prevalence, less breastfeeding, fewer challenges around leadership, governance within the health sector or country, infrastructure and service delivery compared with Global Plan priority countries. Although by 2016 EMTCT-validated countries and Global Plan priority countries had adopted a public health approach to HIV prevention, recommending lifelong antiretroviral therapy (ART) for all HIV-positive pregnant and lactating women, EMCT-validated countries had also included contact tracing such as assisted partner notification, and had integrated maternal and child health (MCH) and sexual and reproductive health (SRH) services, with services for HIV infection, sexually transmitted infections, and viral hepatitis. Additionally, Global Plan priority countries have limited data on key SRH indicators such as unmet need for family planning, with variable coverage of antenatal care, HIV testing and triple antiretroviral therapy (ART) and very limited contact tracing. Structural equation modelling (SEM) and linear regression analysis demonstrated that ART access protects against percent MTCT (p<0.001); in simple linear regression it is 53% protective against percent MTCT. In contrast, SEM demonstrated that the case rate was driven by the number of HIV exposed infants (HEI) i.e. maternal HIV prevalence (p<0.001). In linear regression models, ART access alone explains only 17% of the case rate while HEI alone explains 81% of the case rate. In multiple regression, HEI and ART access accounts for 83% of the case rate, with HEI making the most contribution (coef. infant HIV exposure=82.8, 95% CI: 64.6, 101.1, p<0.001 vs coef. ART access=-3.0, 95% CI: -6.2, 0.3, p=0.074). CONCLUSION: Reducing infant HIV exposure, is critical to reducing the paediatric HIV case rate; increasing ART access is critical to reduce percent MTCT. Additionally, our study of four validated countries underscores the importance of contact tracing, strengthening programme monitoring, leadership and governance, as these are potentially-modifiable factors. Global and national data published in English were sought to synthesise HIV- and PMTCT-related characteristics of EMTCT-validated countries and the 21 SSA Global Plan priority countries. PubMed, Google scholar and Science Direct were used to search for relevant peer- reviewed articles in English, using the terms MTCT, EMTCT and PMTCT effectiveness. As this paper mainly focuses on current status and is not a review of progress over many years, once a document with updated information on a particular indicator or topic was obtained, no additional searches were conducted for prior documents on that indicator. Documents or papers that only focused on a sub-national level such as one clinic, hospital or region were excluded as information from subnational settings were not relevant for this synthesis. Additionally, key individuals participating in global think tanks and expert groups were identified and contacted for relevant global or country-level reports, data, fact sheets and press releases on EMTCT or measuring PMTCT effectiveness. Although the EMTCT criteria specify that all criteria should be met in at least one of the lowest sub-national levels, this synthesis is restricted to national level, given the dearth of reliable data at subnational levels in SSA settings [14, 15]. We applied three theories, the WHO’s six building blocks to strengthen health systems, van Olmen’s Health System Dynamics framework and Baral’s socio-ecological model for HIV risk to understand and explain the differences between EMTCT validated and non-validated countries. The WHO theory states that strengthening six health system building blocks, namely (i) leadership and governance, (ii) health care financing, (iii) the healthcare workforce, (iv) medical products and technologies, (v) information and research and (vi) service delivery improves access, coverage, quality and safety of interventions, resulting in improved outcomes including equity, responsiveness and efficiency [16]. The van Olmen Health Systems Dynamics framework recognizes the existence of an overarching context within which health systems function, as well as the importance of leadership and governance, service delivery, resources (infrastructure, human resources, finances and knowledge and information) and population characteristics on goals and health outcomes [17]. Baral’s social ecological model acknowledges rings of influence on HIV risk, beginning at the individual level, expanding to social and sexual networks, community, public policy and HIV epidemic stage [18]. We considered these three models because they each adopt a different approach, ranging from system-specific [16], system within a context [17] to individual within a system and context [18]. We integrated the information from these theories to compare and understand the EMTCT-validated and SSA Global Plan priority countries, and consider what it will take to achieve EMTCT in the latter. We used both linear regression and Structural Equation modelling (SEM) to estimate the contribution of infant HIV exposure (HEI) and ART access to percent MTCT (%MTCT) and the paediatric case rate (case-rate). The outcome case-rate was defined as the number of new paediatric HIV cases per 100 000 births, and the outcome %MTCT was percentage HIV transmission (numerator: number of new HIV infections amongst infants born to HIV infected mothers; denominator: number of HIV infected mothers), both assessed for 2017. The main exposure for both models was infant HIV exposure (HEI), defined as (number of HIV-positive pregnant women in 2017 *100)/total number of births in 2017. Due to limitations of sample size (number of countries), we reduced the number of variables in the models by creating an “ART-score”. This was defined as: ART-score = total access to treatment = ART access = % of HIV-positive on treatment (general population) + % HIV-suppressed (general population) + % of pregnant women on ART. Prior to deriving the additive score, we checked for internal consistency in the 3 ART-items (Cronbanch’s alpha=0.916) and also performed a confirmatory factor analysis to check how well the three items were loading on a single factor (root mean square error of approximation (RMSEA)<0.05, Tucker and Lewis Index (TLI)>0.95, comparative fit index (CFI)>0.95, coefficient of determination (CD)=0.94). We followed published rules of minimum levels of the fit indices, which specify minimum requirement for model acceptance for RMSEA values must be less than 0.06 and for CFIs and TLIs as 0.90 [19]. We also performed confirmatory factor analysis of the overall Structural Equation Model to check how well the model fitted the data (RSMEA<0.05, TLI, CFI>0.95, CD=0.88) and used the Sobel test to test for the significance of the mediating factor in both the structural equation model and the linear regression model [20]. For the linear regression model, we used the ‘product of coefficient’ method to test the mediation effects [20].