Incorporating geographic context into intervention evaluation: cholera and malaria vaccine trials

Status: This talk is in preparation - details may change

This paper presents two case studies that show the value of adding a geographic component to vaccine trials for cholera and malaria. Generally we think of clinical trials as being well controlled and completely randomized experimental studies. However, rarely is the spatial distribution of trial subjects considered important to the randomization procedure. As a result, disease risk may not be homogeneously distributed across study subjects, and inference about the efficacy of a particular treatment regime may be confounded. Individuals are randomized according to treatment allocation, not infection risk. So, despite treatment being randomly and uniformly distributed across study subjects, the risk of infection for individuals randomly allocated will not be uniformly distributed, since these risk factors are not accounted for during the randomization procedure. The oral cholera vaccine trial was conducted in Matlab, Bangladesh. This double-blind trial measured the efficacy of two vaccines, the B subunit-killed whole cell and the killed whole cell only vaccines. Females aged 15 years and older and children aged 2–15 were the target groups in the trial. Three vaccine doses were given to 62,285 people in the target group in six-week intervals. The vaccine trial used a passive surveillance system to identify cholera cases in the study area during 3 years of follow-up. All households were mapped and information about the water and sanitation environment was collected through a health and demographic surveillance system. The spatial database was used to determine the proportion of people who were vaccinated by neighborhood. The malaria case study determines the effect of geographic context on RTS,S vaccine efficacy in Lilongwe, Malawi for a cohort of 1,623 infants in two age groups: 6-12 weeks at enrollment and 5-17 months at enrollment. The vaccine was administered as 3 doses with or without a booster. We mapped all households and administered surveys every 6 months to capture factors about the geographic context that could affect the vaccine performance, including household ecological characteristics, neighborhood ecology, bed net use, and travel history (location and duration). We hypothesized that RTS,S effectiveness would vary within the study area that it would be impacted by neighborhood-level differences including community bed net use, malaria transmission intensity, nearness and duration of standing water, and proportion of household members travelling to high malaria prevalence regions. In both trials we used multilevel regression analysis to investigate the influence of individual factors and neighborhood ecological factors on vaccine efficacy. Results show that an important geographic contextual variable that influences oral cholera vaccine effectiveness is proportion of neighborhood vaccination particularly in women 15 and older. RTS,S efficacy was modified by individual and neighborhood bed net use. This paper concludes that geographic context is essential for properly evaluating the effectiveness of some interventions.