No abstract

No abstract

No abstract

We initially identified 33 EV indictors grouped in six categories: 1) geographic setting (eight, e.g., urban vs. rural); 2) study population (seven, e.g., gender); 3) implementation characteristics (six, e.g., adherence efforts); 4) institutional and legal context (three e.g., stigma); 5) ability to scale with quality (two e.g., implementation scale); and 6) HTC-specific indicator (seven, e.g., service delivery mode). After the first round of expert elicitation, we added one indicator and dropped one. Four (12%) indicators were excluded due to lack of variability (>90% studies reported the same characteristics) and four (12%) due to excessive (>70%) missing data, cumulatively comprising 27% of total weights. Seventeen of 25 (68%) remaining indicators comprised the top 80% of the total renormalized weights. The bottom five least weighted indicators were: 1) WHO region 0.4 %; 2) WHO subregion 0.8%; 3) country 0.8%; 4) national per capita government health spending 1.5%; and 5) country-level income 1.7%, and the top most weighted indicators: 1) target age group 6.4%; 2) service delivery mode 5.7%; 3) type of post-test counseling 5.5%; 4) stigma for intervention 5.5%; and 5) HIV epidemic type 5.4%. Summary/Conclusion: More attention should be given to EV for translation of evidence to real-world global health practice. Our study proposes a target-specific definition for EV: The likelihood that intervention effects observed in a set of studies will be replicated if implemented in a different target setting. Intervetion-specific indicators should be carefully explored for other EV tools. Validation of our tool is underway.