Low Diagnostic Performance of Thick Blood Smears of 50 µL in Comparison with Blood Direct Examination of 10 µL and Leukoconcentration Techniques among Loiasis-Suspected Patients with Low Microfilaremia in Gabon, Central Africa, using the STARD-BCLM Guidelines

Abstract

Background The aim of the study was to determine performance indicators of the Thick Blood Smear of 50 µL (TBS-50), following Standards for the Reporting of Diagnostic Accuracy Studies-Bayesian Latent Class Analysis (STARD-BLCA) guidelines. TBS-50 was compared to two common parasitological techniques – the blood direct examination of 10 µL and the leukoconcentration of 5 mL. Methods The study population was recruited among the patients of the Department of Parasitology-Mycology-Tropical Medicine of the Faculty of Medicine of the Université des Sciences de la Santé between July 2018 and July 2019. Age, sex, symptoms and eosinophilia variables were recorded from laboratory registers and medical files. The blood direct examination of 10 µL, TBS-50 and the leukoconcentration technique of 5 mL were performed for each patient. The Classica formula and BLCA were used to determine the diagnostic accuracies of the three techniques as well as the prevalence. Three models were built within the framework of BCLA: main model I and alternative models II and III for sensitivity analysis considering the leukoconcentration technique as the gold standard. Results In total, 191 patients consented to be included. The prevalence of Loa loa microfilaremia was 9.4% [95% CI: 5.7–14.5] (n = 18/191) with direct blood examination/TBS-50 and 12.6% [8.2–18.1] (n = 24/191) for leukoconcentration. The mean microfilaremia levels were comparable between blood direct examination (1005 [200-15000] mf/mL) and TBS-50 (815 [100-16720] mf/mL) (p = 0.97). Comparing TBS-50/direct blood examination with the leukoconcentration method, the sensitivity was 75.0% [53.3–90.2], the specificity was 100.0% [97.8–100.0], the positive predictive value was 100.0% [81.5–100.0], and the negative predictive value was 96.5% [92.6–98.7]. The prevalence estimates of microfilaremic loiasis were 9.7% [6.2–13.7], 10.2% [6.3–14.4] and 11.2% [6.7–16.4], respectively, in ascending order of the models. The outputs of main model I showed a sensitivity of 78.9% [65.3–90.3], a specificity of 100.0% [99.3–100.0], a positive predictive value of 99.1% [87.2–100.0] and a negative predictive value of 93.0% [87.3–97.7] for direct blood examination/TBS-50. Conclusions TBS-50/blood direct examination has a low sensitivity. In one in five cases, the result will be falsely declared negative using these two methods.