Measuring implementation in global mental health: validation of a pragmatic implementation science measure in eastern Ukraine using an experimental vignette design

Abstract

Background

There is mounting evidence supporting the effectiveness of task-shifted mental health interventions in low- and middle-income countries (LMIC). However, there has been limited systematic scale-up or sustainability of these programs, indicating a need to study implementation. One barrier to progress is a lack of locally relevant and valid implementation measures. We adapted an existing brief dissemination and implementation (D&I) measure which includes scales for acceptability, appropriateness, feasibility and accessibility for local use and studied its validity and reliability among a sample of consumers in Ukraine.

Methods

Local qualitative data informed adaptation of the measure and development of vignettes to test the reliability and validity. Participants were veterans and internally displaced persons (IDPs) recruited as part of a separate validity study of adapted mental health instruments. We examined internal consistency reliability, test-retest reliability, and construct and criterion validity for each scale on the measure. We randomly assigned half the participants to respond to a vignette depicting existing local psychiatric services which we knew were not well regarded, while the other half was randomized to a vignette describing a potentially more well-implemented mental health service. Criterion validity was assessed by comparing scores on each scale by vignette and by overall summary ratings of the programs described in the vignettes.

Results

N = 169 participated in the qualitative study and N = 153 participated in the validity study. Qualitative findings suggested the addition of several items to the measure and indicated the importance of addressing professionalism/competency of providers in both the scales and the vignettes. Internal consistency reliabilities ranged from α = 0.85 for feasibility to α = 0.91 for appropriateness. Test-rest reliabilities were acceptable to good for all scales (rho: 0.61–0.79). All scales demonstrated substantial and significant differences in average scores by vignette assignment (ORs: 2.21–5.6) and overall ratings (ORs: 5.1–14.47), supporting criterion validity.

Conclusions

This study represents an innovative mixed-methods approach to testing an implementation science measure in contexts outside the United States. Results support the reliability and validity of most scales for consumers in Ukraine. Challenges included large amounts of missing data due to participants’ difficulties responding to questions about a hypothetical program.