Etiological spectrum of persistent fever in the tropics and predictors of ubiquitous infections: a prospective four-country study with pooled analysis

Persistent fever, defined as fever lasting for 7 days or more at first medical evaluation, has been hardly investigated as a separate clinical entity in the tropics. This study aimed at exploring the frequencies and diagnostic predictors of the ubiquitous priority (i.e., severe and treatable) infections causing persistent fever in the tropics. In six different health settings across four countries in Africa and Asia (Sudan, Democratic Republic of Congo [DRC], Nepal, and Cambodia), consecutive patients aged 5 years or older with persistent fever were prospectively recruited from January 2013 to October 2014. Participants underwent a reference diagnostic workup targeting a pre-established list of 12 epidemiologically relevant priority infections (i.e., malaria, tuberculosis, HIV, enteric fever, leptospirosis, rickettsiosis, brucellosis, melioidosis, relapsing fever, visceral leishmaniasis, human African trypanosomiasis, amebic liver abscess). The likelihood ratios (LRs) of clinical and basic laboratory features were determined by pooling all cases of each identified ubiquitous infection (i.e., found in all countries). In addition, we assessed the diagnostic accuracy of five antibody-based rapid diagnostic tests (RDTs): Typhidot Rapid IgM, Test-it Typhoid IgM Lateral Flow Assay, and SD Bioline Salmonella typhi IgG/IgM for Salmonella Typhi infection, and Test-it Leptospira IgM Lateral Flow Assay and SD Bioline Leptospira IgG/IgM for leptospirosis. A total of 1922 patients (median age: 35 years; female: 51%) were enrolled (Sudan, n = 667; DRC, n = 300; Nepal, n = 577; Cambodia, n = 378). Ubiquitous priority infections were diagnosed in 452 (23.5%) participants and included malaria 8.0% (n = 154), tuberculosis 6.7% (n = 129), leptospirosis 4.0% (n = 77), rickettsiosis 2.3% (n = 44), enteric fever 1.8% (n = 34), and new HIV diagnosis 0.7% (n = 14). The other priority infections were limited to one or two countries. The only features with a positive LR ≥ 3 were diarrhea for enteric fever and elevated alanine aminotransferase level for enteric fever and rickettsiosis. Sensitivities ranged from 29 to 67% for the three RDTs targeting S. Typhi and were 9% and 16% for the two RDTs targeting leptospirosis. Specificities ranged from 86 to 99% for S. Typhi detecting RDTs and were 96% and 97% for leptospirosis RDTs. Leptospirosis, rickettsiosis, and enteric fever accounted each for a substantial proportion of the persistent fever caseload across all tropical areas, in addi