Answers are grounded in this paper and cite their sources.
Dear Editor, We read with interest the retrospective cohort study by Bača et al. evaluating tuberculosis (TB) infection and disease among pediatric contacts in the Czech Republic.1 The authors report that chest computed tomography (CT) identified TB-consistent abnormalities in 13 of 45 children with a normal chest radiograph and was associated with an ∼50% increase in case detection. This conclusion may nevertheless be overstated because of selective verification and incorporation bias, an issue not included among the study’s stated limitations. CT was not applied uniformly across the cohort. Its use depended on age, epidemiological risk, immunological test results and radiograph findings, and the indication broadened during the study. Therefore, the 45 children with a normal radiograph who underwent CT represent an enriched subgroup rather than a random sample. The 29% CT yield cannot be interpreted as the incremental diagnostic accuracy of CT in pediatric contacts; it may reflect pre-test enrichment. A further concern is the reference standard. TB was defined by imaging findings consistent with TB or microbiological confirmation, together with a positive tuberculin skin test and/or interferon-gamma release assay. Thus, the CT result could contribute directly to the disease classification against which CT was judged. Such incorporation bias can inflate apparent diagnostic performance, particularly when radiological abnormalities are nonspecific.2,3 Microbiological specimens were obtained from 76 children and 10 were positive, but the authors do not clarify how many of the 13 CT-positive cases were microbiologically confirmed or adjudicated. Consequently, “additional detection” may represent additional radiological suspicion rather than confirmed TB. This distinction matters clinically. In children, CT exposes patients to ionizing radiation and may require sedation in younger children. A nonspecific abnormality may lead to repeat imaging, invasive sampling, prolonged follow-up, and, in a child with positive immunological testing, potentially unnecessary multidrug treatment. Conversely, a normal CT should not overrule microbiological investigation when the epidemiological probability of disease remains high. Pediatric CT findings require clinical, epidemiological, and microbiological integration rather than treatment as an independent diagnostic endpoint.4 A blinded independent re-reading of CT images, analyses restricted to prespecified CT indications, and an independent composite reference standard excluding CT findings would strengthen the inference. Reporting microbiological confirmation, subsequent clinical progression, treatment response, and inter-reader agreement in CT-positive children would also be important. Until these analyses are provided, CT should be described as a useful adjunct in selected high-risk contacts, not as a proven explanation for the increased TB detection in this cohort. Ercan Yılmaz (Conceptualization [equal], Data curation [equal], Formal Analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Visualization [equal], Writing – original draft [equal]), Nezihe Koker Özer (Conceptualization [equal], Data curation [equal], Formal Analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Visualization [equal], Writing – original draft [equal], Writing – review & editing [equal]), Erdem Topal (Conceptualization [equal], Data curation [equal], Formal Analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Visualization [equal], Writing – review & editing [equal]) The authors declare no conflict of interest. None. Research data are not shared. Not applicable. The authors used ChatGPT AI for language editing.
No takes yet. Share an insight, caveat, or question.
Yılmaz et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: