Preventing HIV and HSV-2 through knowledge and attitudes: A replication study of a multi-component community-based intervention in Zimbabwe

INTRODUCTION: Approximately two-thirds of HIV-infected individuals reside in sub-Saharan Africa. The region accounts for 68% of the new HIV infections occurring worldwide with almost one-half of these infections being among young adults aged 12–24 years. Cowan and colleagues conducted a community-based, multi-component HIV intervention aimed at youth in rural Zimbabwe. Despite some changes in knowledge and attitudes, the community-based intervention did not affect the prevalence of HIV or HSV-2. We selected this frequently cited study for replication since it incorporates individual-, community-, and structural- level intervention components that are often considered in global HIV/AIDS prevention programs. Additionally, the intervention could be easily scaled-up, which is especially important in the context of limited resources. Although this study indicated no intervention effects in reducing HIV, the authors acknowledged some key methodological challenges. Our replication analysis provided important insights regarding the impact of these challenges to the interpretation of the results of this study. METHODS: Our replication study focused on replicating Cowan’s findings and assessing the robustness of Cowan’s results to alternative analytical models based on their study design. We determined how out-migration occurring during Cowan’s study may have affected the population characteristics, the intervention exposure level, and the study findings. While the original intervention targeted knowledge and attitudes as a mechanism to decrease HIV/HSV-2, the Cowan study evaluated the intervention effects on knowledge, attitudes, and prevalence of HIV or HSV-2 separately. To better identify the pathway describing the interrelationship among the intervention and knowledge, attitudes, and prevalence of HIV or HSV-2, we assessed whether increases in knowledge or attitudes were associated with decreased HIV or HSV-2 prevalence. RESULTS: We replicated the original findings with minor discrepancies during the pure replication. Our additional analyses revealed that the study population characteristics changed over time in ways that may have affected outcomes. These changes also affected the levels of intervention exposure, with 48.7% males and 75.5% females of the intervention group receiving low-level exposure. Both genders with higher level intervention exposure experienced higher increments in multiple knowledge, attitude, and sexual risk behavior outcomes. Unfortunately, these did not translate to a significant reduction in HIV or HSV-2 regardless of the level and combination of knowledge and attitude domains. However, males receiving high-level intervention exposure compared to control indicated significantly lower odds of having HIV or HSV-2 under a Bayesian modeling paradigm. CONCLUSIONS: Our findings suggest a more robust conclusion on the study intervention effects. Further study based on a design that more consistently maximizes the exposure level of the intervention is necessary and should ideally be an evaluated goal in similar studies. Evaluation of the intervention impact for key subgroups of the target population is important and would better advise the use and scale-up of the evaluated interventions in various contexts. Our observation of a consistent lack of relationship between knowledge/attitudes and HIV/HSV-2 suggests a need to explore and include relevant additional and or complementary interventions, e.g., promoting effective skills in reducing risky sexual behaviors and addressing cultural and structural bottlenecks that may reduce HIV/HSV-2 risk among youth.