There was significant heterogeneity between the included studies; and based on the results of subgroup analyses, the authors deem this attributable in part to differences in study continent, adjustment for confounders and ascertainment of community-acquired pneumonia diagnosis (clinical diagnosis vs death records pneumonia diagnosis). They note that confounding as a result of the existence of other factors that were not usually adjusted for in the included studies (e.g. socioeconomic status, malnutrition) could not be explored.