50 per dose. In the original model we adjusted for a potential differential coverage among children likely to suffer rotavirus mortality [1]. For the current model we eliminated that assumption since we are explicitly modeling the co-distribution of risks and access. The distributional impact of vaccination in a given country was modeled by incorporating data on the disparities in vaccine coverage by wealth quintile at the national level and by estimating the distribution of rotavirus mortality risk by wealth quintile. Both of these were estimated using available data (2003 or later) from the most recent Demographic and Health Surveys of the 25 GAVI-eligible countries

[26]. Countries were selected based on the availability of data at the time of the analysis. Countries with earlier surveys were excluded given that disparities may change over time due to ongoing efforts to achieve universal coverage. Table 1 shows the countries

LEE011 datasheet find more and the year of the survey. For immunization coverage, DPT2 coverage was used as a proxy to estimate the distribution of rotavirus vaccination by quintile. No specific publications were identified with data on the distribution of rotavirus or diarrheal mortality by wealth quintile. As a result, we used alternative proxy measures to estimate the potential distribution of rotavirus mortality across wealth quintiles. We used three proxy measures: post-neonatal infant mortality, less than −2 standard deviations in weight for age Z-scores, and less than −3 standard deviations in weight for age Z-scores [26]. The first of these was expected to correlate with rotavirus

mortality risk as a proxy for health care access, while the latter two were expected to be proxies for physical susceptibility due to their demonstrated association with diarrheal mortality [27]. Post-neonatal infant mortality (between 1 and 11 months of age) was used since it closely corresponds with the primary ages of rotavirus mortality. However it is unclear whether other measures like 1–59 months mortality would be a more appropriate proxy. The rates of low weight for age and post-neonatal infant mortality by quintile were used to estimate the fraction of each outcome that would occur in a given quintile. For each of these proxies, of the quintile fraction was applied to the estimated national annual rotavirus deaths to estimate the rotavirus deaths for each quintile. Given the uncertainty as to which proxy would best estimate the distribution, the average of the estimated deaths based on the three proxies were averaged for each quintile, resulting in a single estimate of rotavirus mortality that would occur in each quintile. In addition, we also used each of the proxy measures to conduct a sensitivity analysis of the main outcomes. These are shown as a range in Table 4. Overall model parameters are shown in Table 2 and key inputs for the distributional analysis are shown in Table 3.