The EVIDEM framework and its usefulness for priority setting across a broad range of health interventions

Goetghebeur et al [1] propose that the EVIDEM can be used to compare various interventions across disease areas in order to prioritize interventions. They suggest that the EVIDEM can analyze single interventions, and the performance of competing interventions can subsequently be compared in a performance matrix. While we value the qualities of the EVIDEM because of its scope and breadth, we have doubts on the results consistency of the EVIDEM to compare competing interventions, particularly when setting priorities across broad healthcare service areas (e.g. in designing the national health benefit package) for two main reasons.First, the EVIDEM framework ignores the contextual nature of priority setting process by assuming a set of universal priority setting criteria [2]. In reality, the priority setting process is context specific and different sets of criteria lead priority setting of health interventions in different contexts. As the examples of studies in Nepal [3], Chile [4], and Ghana [5] show, the set of criteria identified for using in priority setting of health interventions were different between countries. Therefore, we suggest that the setting of prioritization criteria needs to be locally determined or verified, implying that the EVIDEM needs to be flexible to allow change/modification of the components to suit the local context.Secondly, the EVIDEM is vulnerable to interventions ranking inconsistency where performance evaluation of a broad range of competing interventions is mandated. For example, the EVIDEM framework requires different expert panels to assess the performance of every single intervention separately. This may lead to inconsistency of the results as different expert panels may have different considerations across the broad range of interventions to be assessed. As shown in the EVIDEM Turner Syndrome case study [1], the panel of experts estimated growth hormone intervention to achieve 41% of maximum value. However, in the absence of estab