Voluntary medical male circumcision: a framework analysis of policy and program implementation in eastern and southern Africa

Abstract

Background: Following confirmation of the effectiveness of voluntary medical male circumcision (VMMC) for HIV prevention, the World Health Organization and the Joint United Nations Programme on HIV/AIDS issued recommendations in 2007. Less than 5 y later, priority countries are at different stages of program scale-up. This paper analyzes the progress towards the scale-up of VMMC programs. It analyzes the adoption of VMMC as an additional HIV prevention strategy and explores the factors may have expedited or hindered the adoption of policies and initial program implementation in priority countries to date.

Methods and findings: VMMCs performed in priority countries between 2008 and 2010 were recorded and used to classify countries into five adopter categories according to the Diffusion of Innovations framework. The main predictors of VMMC program adoption were determined and factors influencing subsequent scale-up explored. By the end of 2010, over 550,000 VMMCs had been performed, representing approximately 3% of the target coverage level in priority countries. The "early adopter" countries developed national VMMC policies and initiated VMMC program implementation soon after the release of the WHO recommendations. However, based on modeling using the Decision Makers' Program Planning Tool (DMPPT), only Kenya appears to be on track towards achievement of the DMPPT-estimated 80% coverage goal by 2015, having already achieved 61.5% of the DMPPT target. None of the other countries appear to be on track to achieve their targets. Potential predicators of early adoption of male circumcision programs include having a VMMC focal person, establishing a national policy, having an operational strategy, and the establishment of a pilot program.

Conclusions: Early adoption of VMMC policies did not necessarily result in rapid program scale-up. A key lesson is the importance of not only being ready to adopt a new intervention but also ensuring that factors critical to supporting and accelerating scale-up are incorporated into the program. The most successful program had country ownership and sustained leadership to translate research into a national policy and program. Please see later in the article for the Editors' Summary.

Figure 1. Comparison of Diffusion of Innovation and ExpandNet frameworks.

A comparison of the DOI and ExpandNet frameworks is shown. Whereas the DOI describes the process through which innovations are adopted and diffused through the population, the ExpandNet framework specifically addresses the diffusion and scale-up of public health interventions. As highlighted in the figure, the ExpandNet framework reflects the earlier thinking of the DOI and incorporates the elements of context as environment, communication channels as scale-up strategy, and adoption by the population as adoption by the organization implementing the intervention.

Figure 2. Diffusion of Innovation adoption status.

The typical distribution of categories of adopters according to the DOI theory is shown in this figure. The very first adopters or innovators typically represent 2.5% of the population. The next group of adopters, called early adopters, represent about 13.5% of the population and are followed by the early majority, who represent approximately 34% of the total population. The next 34% of the population to adopt are the late majority; this group is followed by the remaining 16% of the population, who are the laggards, the last to adopt a new innovation.

Figure 3. Association of scale-up element scores and Diffusion of Innovation adoption status.

The correlation between each of the six elements of scale-up and DOI adoption status is shown in this figure. The scores obtained (ranging from 0 to 3) for each element by each country is shown (on the vertical axis) in relation to the adoption status (shown on the horizontal axis). Having a pilot program appears to be the strongest predictor of adoption status; this can be seen in the linear clustering of the countries. Conversely, having conducted a situational analysis appears to the least predictive of adoption status; the clustering of the countries is less linear and appears more random. BOT, Botswana; KEN, Kenya; LES, Lesotho; MAL, Malawi; MOZ, Mozambique; NAM, Namibia; RWA, Rwanda; SAF, South Africa; SWA, Swaziland; TAN, Tanzania; UGA, Uganda; ZAM, Zambia; ZIM, Zimbabwe.