Implementation of risk triaging in primary healthcare facilities in Sub-Saharan Africa: A systematic review

BackgroundOne challenge facing treatment programs for HIV and other chronic conditions in sub-Saharan Africa (SSA) is how to target interventions to optimize retention in care and other outcomes. Most efforts to target interventions have identified predictive features among high risk patients after negative outcomes have already been observed. An alternative for identifying patients at high risk of negative outcomes is "risk triaging," or identifying vulnerable or higher risk patients before they experience an interruption in care or other negative outcome. We conducted a systematic review of the use of risk triaging tools at the primary healthcare (PHC) level in SSA. MethodsWe searched PubMed and other databases for publications after 1 January 2012 that reported development or implementation of risk triaging tools for PHC use in SSA. We extracted information on three outcomes: 1) characterization of the risk triaging tools; 2) tool performance metrics (sensitivity, specificity, positive and negative predictive value, area under the curve); and 3) health system effects (efficiency, acceptability, resource utilization, cost). We report outcomes for each eligible study and identify lessons for use of risk triaging. ResultsOf 1,876 articles identified, 28 were eligible for our review. Thirteen addressed HIV, 10 TB, 1 TB/HIV, and 4 other conditions. Approximately 60% used existing, retrospective data to identify important risk factors for an outcome and then construct a scoring system, but no implementation of these tools was reported. The remaining 40% designed a tool using existing data or experience and reported implementation results. More than half (16/28, 58%) of the tools achieved sensitivities >80%; specificity was much lower. Only one tool, the World Health Organizations 4-symptom screen for tuberculosis, had been scaled up widely. While most studies claimed that their tools could increase the efficiency of healthcare delivery, none of the studies provided examples of tangible health system impacts. ConclusionMost of the tools identified were at least somewhat successful in identifying potential risks but uptake by health systems has been minimal. Although well-designed risk triaging tools have the potential to improve health outcomes, implementation will require commitment at the policy, operational, and funding levels.