Diagnostic test accuracy of simplified algorithms for diagnosing acute rheumatic fever: a systematic review

Rheumatic heart disease, the long-term sequel to acute rheumatic fever, remains a prevalent public health problem in Africa and other low to middle-income regions of the world. Diagnosing acute rheumatic fever and using the modified Jones criteria in high-prevalence areas remains challenging.

We assessed the (i) diagnostic accuracy of simplified diagnostic algorithms among children, adolescents, and adults with suspected acute rheumatic fever, and (ii) the impact of different diagnostic criteria on the development of rheumatic heart disease (PROSPERO CRD42022344077). The MEDLINE, Embase, and Conference Proceedings Citation Index-Science were searched for relevant reports (date: 15th March 2025).

Here we identify 12,075 records, and three studies (four reports) meeting our eligibility criteria. Simplified diagnostic algorithms using only clinical data at community health centre-level (AUC 0.69, sensitivity 66% and specificity 68%), or adding 12-lead electrocardiogram and simple laboratory investigations at district-level facilities (AUC 0.76, sensitivity 77% and specificity 67%) perform worse than models including the full-set of laboratory investigations and echocardiography at National referral hospitals (AUC 0.91, sensitivity 84% & specificity 87%). Using modified Jones criteria without echocardiography results in an important loss of sensitivity (sensitivity 79%, specificity 100% & AUC 0.90). Progression to rheumatic heart disease is reported in 2.5-5% of children and young adults in high-prevalence areas who do not meet the full modified Jones criteria.

Simplification of the modified Jones criteria in areas without access to echocardiography and laboratory investigations may lead to underdiagnosis of acute rheumatic fever. Some patients who do not meet the modified Jones criteria for definite acute rheumatic fever diagnosis may still progress to develop rheumatic heart disease.