Abstract
Valid mental and substance use disorders and suicide risk screening tools are needed for community case finding of individuals who may not otherwise seek care. We evaluated the Proxy Mental Wellness Tool-3 (mwTool-3-proxy) a three-item screener that asks about the mental health of another adult, against a diagnostic gold standard in Mozambique and South Africa. The mwTool-3-proxy adapts the three items of the Mental Wellness Tool-3, developed in Mozambique using Mini International Neuropsychiatric Interview diagnoses as the criterion standard, regression modeling and expert consultation to determine the best three items for identifying any mental disorder. The Mental Wellness Tool-3 has been validated in South Africa, Spain and the United States, and is being validated in three countries in the Asia-Pacific and Israel. Pairs of adults in South Africa and Mozambique at primary and tertiary healthcare facilities were separately screened with the mwTool-3-proxy and diagnosed using the Mini International Neuropsychiatric Interview. We calculated the sensitivities and specificities for predicting any mental and/or substance use disorder and suicide risk among the proxy individual. We performed additional analyses restricted to respondents who were relatives of one another and who lived in the same household. The prevalence of any Mini International Neuropsychiatric Interview-diagnosed disorder among the 229 pairs in both countries was 35.6% (38.5% in Mozambique; 32.9% in South Africa). The pooled sensitivity of the mwTool-3-proxy for identifying any disorder among the proxy individual was 73.01 (95%CI: 65.5-79.65) - 70.24 (95%CI: 59.27-79.73) in Mozambique and 80.00 (95%CI 69.17-88.35) in South Africa. The mwTool-3-proxy is a culturally-relevant, ultra-brief valid measure that can improve mental and substance use disorders and suicide risk case detection with strong sensitivity at the community and household level and offer a means to efficiently and feasibly collect clinical and population-level service needs data.