Commentary: Diagnostic accuracy of oral swab for detection of pulmonary tuberculosis: a systematic review and meta-analysis

A Commentary on
Diagnostic accuracy of oral swab for detection of pulmonary tuberculosis: a systematic review and meta-analysis

by Zhang, F., Wang, Y., Zhang, X., Liu, K., Shang, Y., Wang, W., Liu, Y., Li, L., and Pang, Y. (2024). Front. Med. 10:1278716. doi: 10.3389/fmed.2023.1278716

Introduction

Pulmonary tuberculosis caused by Mycobacterium tuberculosis remains a major public health concern, requiring accurate diagnosis for effective treatment, prevention, and transmission control, particularly in vulnerable populations. Among available diagnostic approaches, oral swab is regarded as a promising non-invasive and alternative test, especially in cases where sputum collection is difficult. Thus, many studies have explored its potential in managing this disease (1–3).

While analyzing the paper published by Zhang et al., Diagnostic Accuracy of Oral Swab for Detection of Pulmonary Tuberculosis: A Systematic Review and Meta-Analysis (https://doi.org/10.3389/fmed.2023.1278716), we found that the paper provides valuable insights on the topic; however, we identified an error that requires correction and felt it necessary to bring this to your attention.

To evaluate publication bias, we replicated the meta-analysis using R software (4), employing the metafor and meta packages. The analysis was conducted using the same dataset as Zhang et al. (1), which comprised 16 studies including both adult and paediatric populations. A key advantage of the metafor and meta packages is their capacity to provide separate assessments of publication bias for sensitivity and specificity. This stands in contrast to Deeks' test, which examines bias collectively through the diagnostic odds ratio (DOR).

Deeks' test, as described by Deeks, Macaskill, and Irwig (5), relies on a weighted linear regression of the logarithm of the odds ratio (logOR) on the inverse square root of the sample size (1/√n). When zero values are present in the 2 × 2 contingency table, the resulting estimates may become infinite, requiring a continuity correction by adding 0.5 to each cell (6). Within this framework, publication bias is indicated if P < 0.10.

This commentary therefore focuses on evaluating the diagnostic accuracy of the test, with particular attention to the influence of the identified publication bias.

Subsections relevant for the subject

In the Results section, page 4, subsection “Publication bias assessment”, the statement “The funnel plot clearly indicated the absence of significant publication bias in this meta-analysis (P = 0.99)” is inaccurate. Given the z-Stat4life community's interest in this topic, we replicated the meta-analysis using R Program (metafor and meta) (7, 8) and we found indications of the publication bias in the funnel plots and Egger test. For metafor, in adults, the P-values were 0.001 both for sensitivity (Se) and specificity (Sp) and for meta (Se: P = 0.007; Sp: P = 0.015), contrary to the paper's findings.

In the aggregate data, the results for the meta package showed a non-significant publication bias for sensitivity (Se: P = 0.054), while the metafor package yielded a similar outcome (Se: P = 0.068). These findings could be impacted by substantial residual heterogeneity (τ² ≈ 5.86). For Sp, however, a statistically significant publication bias was detected (P = 0.001), with comparatively lower residual heterogeneity (τ² ≈ 2.14) across both meta and metafor.

Deeks' test indicated P = 0.054, suggesting presence of publication bias (p < 0.10). This was visually apparent in the funnel plots and was further supported by Egger's test. Additionally, significant asymmetry was identified using Deeks' test in both the adult-specific dataset and the combined dataset (adults and children).

Discussion and final considerations

Publication bias affecting both Se and Sp in this meta-analysis may inflate the test's diagnostic accuracy, potentially misleading clinical decisions. Although the analyses consistently indicated the presence of publication bias in adults, the distinctive methodological characteristics of various approaches may result in divergent results depending on the software packages and analytical techniques used. For this reason, the use of complementary analytical strategies is strongly recommended to support a critical and comprehensive evaluation of the findings.

Given the importance of this topic, it is crucial to perform subgroup analyses while taking the following points into consideration:

• The selection of statistical software and modeling strategies can substantially affect the results, particularly in scenarios involving high heterogeneity or limited sample sizes;

• Ensuring the reproducibility and reliability of diagnostic meta-analyses requires not only the implementation of diverse methods for detecting publication bias but also a deep understanding of the underlying assumptions and limitations of each approach;

• Clear and transparent reporting of the analytical code, along with any modifications applied (such as continuity corrections), is vital for enabling reproducibility and facilitating cross-study comparisons;

• Comprehensive disclosure of all results, including non-significant or negative findings combined with a strong commitment to methodological rigor, is essential to reducing bias and strengthening the credibility of scientific evidence.

Therefore, attention to these issues and appropriate editorial actions are essential to maintain the journal's scientific quality and credibility.

Author contributions

NP: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. BC: Data curation, Formal analysis, Writing – review & editing. NM: Data curation, Formal analysis, Writing – review & editing. JV: Writing – review & editing. IA: Writing – review & editing. LG: Data curation, Supervision, Validation, Writing – original draft, Writing – review & editing.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. NP and LG gratefully acknowledge the partial financial support provided by the Portuguese Foundation for Science and Technology (FCT) through project UID/00006/2025, UIDB/00006/2020 (URL: https://doi.org/10.54499/UIDB/00006/2020) and doctoral grant UI/BD/154312/2022 (URL: https://doi.org/10.54499/UI/BD/154312/2022), within the Centre of Statistics and its Applications, University of Lisbon (CEAUL).

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Gen AI was used in the creation of this manuscript.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed.2025.1568093/full#supplementary-material

References

1. Zhang F, Wang Y, Zhang X, Liu K, Shang Y, Wang W, et al. Diagnostic accuracy of oral swab for detection of pulmonary tuberculosis: a systematic review and meta-analysis. Front Med. (2024) 10:1278716. doi: 10.3389/fmed.2023.1278716

PubMed Abstract | Crossref Full Text | Google Scholar

2. Jinfeng W, Zhen Y, Fuzhen Z, Cong Y. Meta-analysis of the diagnostic value of oral swab for pulmonary tuberculosis. Chinese J Antituberc. (2024) 46:815–22. Available online at: https://www.zgflzz.cn/EN/10.19982/j.issn.1000-6621.20240161

Google Scholar

3. Church EC, Steingart KR, Cangelosi GA, Ruhwald M, Kohli M, Shapiro AE. Oral swabs with a rapid molecular diagnostic test for pulmonary tuberculosis in adults and children: a systematic review. Lancet Glob Health. (2024) 12:e45–54. doi: 10.1016/S2214-109X(23)00469-2

PubMed Abstract | Crossref Full Text | Google Scholar

4. R Core Team (2024). R: A Language and Environment for Statistical Computing [Computer software manual]. Vienna, Austria. Available online at: https://www.r-project.org/

Google Scholar

5. Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol. (2005) 58:882–93. doi: 10.1016/j.jclinepi.2005.01.016

PubMed Abstract | Crossref Full Text | Google Scholar

6. Borenstein M, Hedges LV, Higgins JP, Rothstein HR. Introduction to Meta-Analysis. New York: John Wiley & Sons, Ltd. (2009). doi: 10.1002/9780470743386

Crossref Full Text | Google Scholar

7. Doebler P, Holling H. Meta-analysis of diagnostic accuracy with mada. Biom J. (2015) 57:751–65. doi: 10.1002/bimj.201400152

PubMed Abstract | Crossref Full Text | Google Scholar

8. Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw. (2010) 36:1–48. doi: 10.18637/jss.v036.i03

Crossref Full Text | Google Scholar