Benefits and Costs of Integrating Sexual-Reproductive Health and HIV Services in Kenya and Swaziland

Overview

The overarching aim of the Integra Initiative is to strengthen the evidence base on the impact of integrating family planning (FP), postnatal care (PNC) and HIV services in sub-Saharan Africa. Specifically, in the study the investigators aim to test the following hypotheses: the provision of integrated services, compared to separate services, will: 1. lead to increased uptake of a range of SRH services . 2. attract a greater number and diversity of clients. 3. lead to increased quality of a range of SRH services 4. lead to healthier sexual and reproductive behavior. 5. lead to reduced stigma at health facilities. 6. lead to the more efficient use of resources, with a lower unit cost of provision of key services. For the purposes of this study integration is defined as offering clients both HIV and postnatal care (PNC) or HIV and family planning (FP) services in the same visit. To better understand how services can be integrated in different countries this study focuses on two key models of integration in Kenya and Swaziland. * The first model focuses on integration of FP and HIV services (integrated FP model) and entails performing HIV testing, STI screening and management, cervical cancer screening, condom promotion within FP consultations, as well as active referral to antiretroviral (ART) units for HIV-positive clients. The FP model will be evaluated in Kenya only. * The second model focuses on integration of PNC and HIV services (integrated PNC model) and will be implemented in both Kenya and Swaziland. The model focuses on the provision of PNC services to mother and baby, FP services, repeat HIV testing for mother, HIV testing for infant and referral to HIV services for HIV positive mothers and infants, as well as referrals for clients requiring other additional services.

There are many well-established reasons that support the rationale for integrating or linking sexual and reproductive health (SRH) and HIV services in developing countries with generalized HIV epidemics - primarily in sub-Saharan Africa. Yet the evidence base for the impact of integrated service delivery on health outcomes and costs remains weak. Partly this is a result of methodological difficulties. There is an emerging body of literature addressing the challenges of using randomized controlled trials to assess the impact of public health interventions. Particularly in cases such as the Integra Initiative, where the causal chain (between intervention and outcome) is long, and where there are is a broad range of outcomes that need to be explored, and where there is already some a degree of integration occurring in some clinic settings, attempting to conduct a randomized controlled trial is not appropriate. Consistent with evaluation designs described by Habicht and colleagues, the Integra design includes evaluation of performance and impact to try to make two types of causal inference: adequacy and plausibility. Evaluation of adequacy will assess whether the expected changes in provision, service utilisation and cost-effectiveness have occurred in intervention facilities. Evaluation of impact will assess the plausibility that changes in service, health and behavioral outcomes are due to the Integra Initiative. The case for such plausibility will be built from the following strands of evidence: * Comparing findings in 'intervention' facilities with those in facilities chosen as 'comparison' sites prior to the evaluation * Exploring a dose-response relationship between the measured extent of integration and the study outcomes * Measuring changes in performance over time, to demonstrate a logical sequence between the intervention (integration) and outcomes. * Measuring change in each step of the logic model - a prerequisite for any attribution to the intervention * Triangulating findings from a mix of research methods to capture a range of perspectives and insights from different disciplines. The study will employ a controlled pre- and post-test quasi-experimental, or non-randomised, design and utilises multiple research methods (cohort study, community survey, clinic assessments, costing tools and qualitative interviews). Since the research is being conducted in real-life health delivery settings where programmatic contamination is possible due to ongoing health programme interventions over the study period, the control group will be referred to as a 'comparison group', for which outcomes will be compared over time up to two years after implementation.