Diagnostic accuracy of a clinical score with a rapid diagnostic test for enteric fever in Nepal, Bangladesh and Cambodia

Background Distinguishing enteric (typhoid and paratyphoid) fever from other causes of fever is challenging. The diagnostic accuracy of Typhoid Rapid Diagnostic Tests (RDT) are suboptimal but might improve if combined with a clinical score. We aimed to develop a clinical score, incorporating a point of care C-reactive protein (CRP) and Typhoid RDT, that could be used for febrile patients attending a health facility with limited laboratory capability, to identify those suitable for further investigations such as a blood culture, an empirical antimicrobial choice that covers enteric fever, or to rule out enteric fever as a likely diagnosis. Methods Overall, 1168 Bangladeshi, Cambodian and Nepali children and adults attending hospital with fever were randomly split (80:20) into derivation and validation cohorts. There were 208/1168 (18%) with blood-culture-confirmed enteric fever. In the derivation cohort we studied the association of symptoms, signs, CRP, laboratory results, and the Test-it ™ typhoid IgM RDT with blood culture confirmed enteric fever using logistic regression and developed several clinical scores. We determined the diagnostic accuracy of the clinical scores in the validation cohort. Results The Typhoid RDT had a sensitivity of 63% and specificity 73% in the validation cohort. A basic clinical score including fever ≥ 3 days, headache, abdominal pain, diarrhoea, absence of cough, and CRP ≥ 10 mg/L had a sensitivity of 80% and specificity 56%. Adding the Typhoid RDT result to the clinical score resulted in a sensitivity of 87%, a specificity of 58%, a PPV of 34%, and a NPV of 95%. Conclusions The diagnostic accuracy of a Typhoid RDT was only marginally improved by combining with a clinical score. Enteric fever was very unlikely with a negative clinical score.