The article by Poposki et al. on visceral leishmaniasis as a cause of fever of unknown origin presents clinically valuable observations but exhibits substantial methodological limitations. Differential exclusion of undiagnosed cases introduces selection bias and precludes generalizable prevalence estimates. The analytical strategy lacks etiological coherence by comparing a specific diagnosis against heterogeneous conditions. Multiple univariate comparisons without adjustment for family-wise error rate and absence of multivariable analysis to control confounding limit causal inference. The cross-sectional assessment of clinical features after diagnostic confirmation precludes temporal inference. Future studies would benefit from nested case-control designs, transparent reporting of diagnostic pathways, and appropriate statistical adjustment.
Kostadinov K. Selection bias and inferential validity in fever of unknown origin studies. Comment on: Poposki et al. Visceral leishmaniasis as a leading cause of fever of unknown origin in immunocompetent adults. Folia Med (Plovdiv) 2026;68(3):е184403.
Dear Editor,
Comments on: Poposki K, Jakimovski D, Shopova Z, Osmani A, Trajkova I, Bosilkovski M. Visceral leishmaniasis as a leading cause of fever of unknown origin in immunocompetent adults: A prospective, observational, single-center study. Folia Med (Plovdiv) 2025;67(6):e167515.
The article by Poposki et al.[
Substantial selection bias is introduced by the exclusion of 36 patients (28% of eligible participants), 22 without established diagnoses, and 14 lost to follow-up. Patients in whom diagnosis remains elusive are likely to differ systematically from those with confirmed etiologies, yet this differential exclusion is not addressed analytically. The reported 17% prevalence of visceral leishmaniasis is conditional on diagnostic confirmation within a tertiary referral center and cannot be interpreted as representative of fever of unknown origin in the source population or community setting. The analytical strategy further conflates a specific diagnostic entity (visceral leishmaniasis) with a heterogeneous reference group encompassing infections, malignancies, inflammatory disorders, and miscellaneous conditions. This precludes meaningful inference about diagnostic discrimination, as the comparison group lacks etiological coherence and violates the assumption of exchangeability required for valid causal or predictive inference.[
Statistical methodology raises additional concerns. Multiple univariate comparisons were conducted across demographic, clinical, and laboratory variables without adjustment for familywise error rate, inflating the probability of Type I error substantially.[
Caution is warranted in interpretation regarding temporality and causal inference. Although framed in etiological terms, the design is fundamentally descriptive and cross-sectional with respect to clinical and laboratory comparisons. Temporality between exposure (visceral leishmaniasis) and clinical features is not established, as assessments occurred after diagnostic workup and potential treatment initiation. The observed associations, while consistent with established pathophysiology, do not distinguish whether hematologic or biochemical abnormalities preceded, accompanied, or followed diagnostic recognition. Additionally, the implicit survival requirement for diagnostic confirmation introduces potential collider stratification bias[
Future investigations would benefit from more rigorous comparative frameworks, such as nested case-control sampling within the fever of unknown origin cohort, propensity score methods to balance measured confounders, or inverse probability weighting to address selection bias from diagnostic incompleteness. Transparent reporting of diagnostic pathways, time-to-diagnosis distributions, and reasons for non-diagnosis would improve external validity and permit assessment of spectrum bias.[
The author declared that no clinical trials were used in the present study.
The author declared that no experiments on humans or human tissues were performed for the present study.
The author declared that no informed consent was obtained from the humans, donors or donors’ representatives participating in the study.
The author declared that no experiments on animals were performed for the present study.
The author declared that no commercially available immortalized human and animal cell lines were used in the present study.
The author has declared that no competing interests exists.
Artificial intelligence tools were used solely to support the preparation of this manuscript. Specifically, AI assisted with language refinement, grammatical editing, and improvement of clarity and academic tone in selected sections, as well as with structuring and organizing draft text. The authors critically reviewed, verified, and edited all AI-generated suggestions before including them. No AI tool was used for data analysis, statistical modeling, interpretation of results, or drawing scientific conclusions. The authors take full responsibility for the content, accuracy, originality, and ethical integrity of the manuscript.
No funding was reported.
The author solely contributed to this work.
Kostadin Kostadinov
All of the data that support the findings of this study are available in the main text.