Abstract
Bacterial sepsis is an important cause of mortality in low- and middle-income countries, yet distinguishing patients with sepsis from those with other illnesses remains a challenge. Currently, management decisions are based on clinical assessment using algorithms such as Integrated Management of Adolescent and Adult Illness. Efforts to develop and evaluate point-of-care tests (POCTs) for sepsis to guide decisions on the use of antimicrobials are underway. To establish the minimum performance characteristics of such a test, we varied the characteristics of a hypothetical POCT for sepsis required for it to be cost-effective and applied a decision tree model to a population of febrile patients presenting at the district hospital level in a low-resource setting. We used a case fatality probability of 20% for appropriately treated sepsis and of 50% for inappropriately treated sepsis. On the basis of clinical assessment for sepsis with established sensitivity of 0.83 and specificity of 0.62, we found that a POCT for sepsis with a sensitivity of 0.83 and a specificity of 0.94 was cost-effective, resulting in parity in survival but costing $1.14 less per live saved. A POCT with accuracy equivalent to the best malaria rapid diagnostic test was cheaper and more effective than clinical assessment.