Development and Validation of a Sociodemographic and Behavioral Characteristics-Based Risk-Score Algorithm for Targeting HIV Testing Among Adults in Kenya

Abstract

To inform targeted HIV testing, we developed and externally validated a risk-score algorithm that incorporated behavioral characteristics. Outpatient data from five health facilities in western Kenya, comprising 19,458 adults ≥ 15 years tested for HIV from September 2017 to May 2018, were included in univariable and multivariable analyses used for algorithm development. Data for 11,330 adults attending one high-volume facility were used for validation. Using the final algorithm, patients were grouped into four risk-score categories: ≤ 9, 10–15, 16–29 and ≥ 30, with increasing HIV prevalence of 0.6% [95% confidence interval (CI) 0.46–0.75], 1.35% (95% CI 0.85–1.84), 2.65% (95% CI 1.8–3.51), and 15.15% (95% CI 9.03–21.27), respectively. The algorithm’s discrimination performance was modest, with an area under the receiver-operating-curve of 0.69 (95% CI 0.53–0.84). In settings where universal testing is not feasible, a risk-score algorithm can identify sub-populations with higher HIV-risk to be prioritized for HIV testing.

Screening Adults for HIV Testing in the Outpatient Department: An Assessment of Tool Performance in Malawi

“…Although standard of care screening had the highest sensitivity, it also had lower specificity than either full or reduced tools at their optimal cutoff scores. Similar to other literature, newly-diagnosed HIV infection was highly correlated with having suspected STI and ≥ 3 sexual partners in the past 12-months, [33][34][35], although adding these questions did not improve screening tool performance.…”

“…Therefore, in contexts with controlled or near-controlled epidemics, it may be particularly challenging to identify adult HIV testing screening tools with sufficiently high specificity to offer efficiencies over standard of care, and sufficiently high sensitivity to capture most infected individuals. A recent paper analyzed the performance of a similar adult outpatient screening tool in Kenya, and found that the optimal set of questions (which included both demographic characteristics and sexual risk behavior questions) would reduce the number of people requiring testing by 75%, but would miss approximately half of HIV-positive individuals [ 33 ]. Taken together with our findings from Malawi, it is evident that screening tools are not an optimal solution for HIV testing in outpatient departments.…”

Risk-Based Screening Tools to Optimise HIV Testing Services: a Systematic Review

“…Second, we found that in settings where patients may not be forthcoming about risk factors or where clinicians are not likely to ask, the implementation of risk-based tools prompted the offer of testing and improved HIV testing uptake [44,66,67]. Except for three papers [17,30,54], no other study discussed how privacy and confidentiality were maintained when administering the screening tool. Third, most screening tools were simple enough to allow their use by non-professional health workers, such as lay counsellors or self-assessments [39].…”

“…Currently, it is uncertain how widely used the tools are, whether tools are validated, which tools are used for what populations and how feasible and acceptable tools are to patients and providers. To date, results have varied; some programmatic implementation of screening tools suggests increased yield and positivity [6••, 13••, 14], while other reports raise concerns that these screening tools may mean people with undiagnosed HIV are not tested and missed due to limited criteria [15][16][17].…”

Development and Evaluation of a Digital HIV Risk Assessment Tool Incorporated Within an App-Based Self-Testing Program

“…The incorporation of behavioral risk factors, especially regarding sex and drug use, is common among existing risk staging tools 33 but less common is the use of contextual factors related to social determinants of health. In the sub-Saharan African context, we found 5 other published risk assessment tools for HIV that incorporated contextual factors: 2 were targeted to young women in South Africa, [34][35][36] and incorporated variables on financial dependence and school absence, and 3 were developed for use in Kenya with information on occupation, 37 Uganda with a variable for education, 38 and rural South Africa with variables for education, place of residence, and SES quintile. 39 In addition, studies using machine learning techniques previously identified education level, dwelling situation, and wealth as important social predictors of HIV in sub-Saharan Africa.…”