Evaluations of Diagnostic Tests for Undifferentiated Febrile Illness on the Thailand-Myanmar (Burma) Border

Watthanaworawit, Wanitda (2018). Evaluations of Diagnostic Tests for Undifferentiated Febrile Illness on the Thailand-Myanmar (Burma) Border. PhD thesis The Open University.

DOI: https://doi.org/10.21954/ou.ro.0000d17e

Abstract

Background
Dengue, leptospirosis and rickettsial infections are the most common non-malaria causes of acute undifferentiated febrile illness in Southeast Asia. Current diagnostic tests are inadequate for clinical use.

Methods
We conducted a two year prospective fever study in three outpatient clinics on the Thailand-Myanmar border, recruiting patients aged at least five years with acute undifferentiated febrile illness. The study was divided into two parts: non-malaria and malaria patients. Specimens were tested to determine the causes of fever and to evaluate the clinical diagnostic accuracy of the tests for early diagnosis of dengue, leptospirosis and rickettsial infections.

Results
A total of 1,029 febrile patients were recruited: 908 non-malaria and 121 malaria patients. A laboratory confirmed diagnosis was made in 34.5% of non-malaria patients, of which 15.9% were dengue, 6.0% leptospirosis, 6.0% murine typhus, and 3.2% scrub typhus. Co-infection was found in 1.7% with mostly leptospirosis and scrub typhus (1.5%). In malaria patients, co-infection was found in 8.3% in most cases with scrub typhus (7.4%).

An immunochromatographic test (ICT) (Non-structural protein 1 [NS1] and Immunoglobulin M [IgM]/Immunoglobulin G [IgG] detection) and real-time reverse transcriptase polymerase chain reaction (rRT-PCR) were evaluated for dengue diagnosis using acute plasma specimens. Sensitivities of the ICT and rRT-PCR were 86.1% (95% confidence interval [CI] 79.4-91.3) and 97.2% (95% CI 93.0-99.2), and specificities were 94.9% (95% CI 93.0-96.4) and 99.1% (95% CI 98.1-99.7), respectively compared against IgM/IgG enzyme-linked immunosorbent assay (ELISA). Combining rRT-PCR with ICT improved the sensitivity of the diagnostic process to 98.6% (95% CI 95.1-99.8).

The IgM ICT and 47kDa quantitative real-time PCR (qPCR) for scrub typhus and 17kDa qPCR for murine typhus were evaluated using acute plasma specimens for the IgM ICT and acute buffy coat specimens for the qPCRs. Sensitivities of the IgM ICT, 47kDa qPCR and 17kDa qPCR were 27.3% (95% CI 15.0-42.8), 22.7% (95% CI 11.5-37.8) and 31.5% (95% CI 19.5-45.6) respectively. Specificities of the IgM ICT, 47kDa qPCR and 17kDa qPCR were 93.5% (95% CI 91.5-95.1), 99.6% (95% CI 98.9-99.9) and 99.7% (95% CI 99.0-100), respectively compared to IgM ELISA/indirect immunofluorescence assay (IFA).

A 16S rRNA qPCR was validated and implemented for leptospirosis diagnosis. The analytical performance was good, it was found to be 100% specific and had a limit of detection of one copy per microliter (μl) of DNA template.

Sub-microscopic malaria infection detected using 18S rRNA qPCR was found in 17.5% of non-malaria patients. The geometric mean of parasitaemia was very low (281.1 parasites/μl [95% CI 172.5-458.2]). It was unlikely to be the cause of fever at this low level.

C-reactive protein (CRP) result was able to distinguish between dengue virus infection (9.0 milligrams per litre (mg/l) (Interquartile range [IQR] 7.9-18.0)) and bacterial infections (106.5 mg/l [IQR 40.8-166.5] for leptospirosis, 55.2 mg/l [IQR 40.7-114.0] for scrub typhus and 24.7 mg/l [IQR 15.0-48.6] for murine typhus, P<0.0001 for all).

Conclusion
The new generation ICT that included NS1 antigen detection was clinically useful and appropriate to implement in the field for early diagnosis of acute dengue infection. The rRT-PCR for dengue could replace the gold standard serology for early diagnosis using a single specimen. The IgM ICT, 47kDa and 17kDa qPCRs were inadequate for diagnosis of scrub typhus and murine typhus due to low clinical diagnostic sensitivity. For leptospirosis 16S rRNA qPCR, the excellent analytical performance warrants further investigation on its clinical usefulness. The CRP was found to be a useful tool to clinicians for determining whether a patient had a viral or bacterial infection. Early diagnosis of scrub typhus, murine typhus, and leptospirosis remains challenging. Development and clinical evaluation of improved diagnostic tests is urgently needed.