In Reply Dr Raven highlights the importance of developing accurate population-based estimates of excess mortality among people with mental disorders. She raises several methodological issues that are relevant to developing estimates of disease burden in general and to understanding excess mortality in individuals with mental disorders in particular.
The letter regarding our article1 notes concerns about generalizing from meta-analytic findings of elevated relative risk for mortality to population-based estimates of excess mortality. This potential issue applies to all studies that seek to develop population estimates from observational data.2 While narrowing inclusion criteria may reduce heterogeneity, it may also introduce other biases into estimates. For instance, many community-based mortality studies do not include individuals with the most severe disorders or institutionalized and marginalized populations; thus, including only these studies could result in underestimation of excess mortality. To address the issue of sample heterogeneity, we chose to use broad inclusion criteria and then present aggregate estimates of mortality risk alongside estimates stratified by key parameters including psychiatric diagnosis. This is an established approach that has been used in public health studies that have used meta-analysis to estimate the impact of a wide range of risk factors on excess mortality.3
Walker ER, McGee RE, Druss BG. Analysis of Mortality in Mental Disorders—Reply. JAMA Psychiatry. 2015;72(11):1150-1151. doi:10.1001/jamapsychiatry.2015.1568