Efficacy of Punctal Occlusion Techniques in Tear Drainage Disorders: A Systematic Review and Meta-Analysis [Letter]

Dear editor

This letter concerns the systematic review and meta-analysis of punctal occlusion for tear drainage disorders by Qutub et al,¹ a PROSPERO-registered, PRISMA-reported synthesis. Its Conclusion section correctly states that comparative meta-analysis did not show statistically significant improvements “in tear production or film stability”, and its Discussion correctly warns that within-group gains are susceptible to placebo effects and regression to the mean. However, the figure captions, the abstract, and the Discussion do not report the review’s most informative finding, the four null primary pooled comparisons. This letter names six discrepancies and the correction that each discrepancy needs.

First, a production error at the center of the thesis: the caption to Figure 2 contradicts its own plot. The caption reports the pooled effect as “demonstrated a significant improvement in TFBUT”. However, the Figure 2 interval and the Results text agree on WMD 0.37 (95% CI −0.94 to 1.67; p = 0.58). A caption reporting the pooled estimate and its interval accurately, as the Results text already does, would remove the contradiction.

Second, an analytic error: the abstract reports a Schirmer p = 0.01 that does not correspond to any efficacy estimate anywhere in the paper. The abstract describes significant improvement on within-group analysis by follow-up period (p = 0.01). However, Figure 5 pools between-group comparisons within follow-up strata, with subgroup p values of 0.57, 0.05 and 0.02. Figure 5 prints 0.01 exactly once: the test for subgroup differences (Chi2 = 8.64, df = 2, P = 0.01, I2 = 76.9%), which asks whether effects differ across strata rather than whether the intervention worked. If the abstract took its p value from the subgroup-difference test, a heterogeneity test has been presented as an efficacy result. The review’s one significant comparative estimate, the six-month Schirmer subgroup (WMD 2.00; 95% CI 0.32 to 3.68), rests on one study (Packer et al), a limitation that belongs wherever the six-month Schirmer estimate is cited.

Third, an analytic error: the forest plots use a different unit of analysis from the Methods. The Methods state twice that the unit of analysis is eyes. However, Figures 2, 4, 6 and 7 use the Table 1 participant counts instead: Packer et al contribute 103 and 54 rather than 198 and 104, Sherwin et al 48 and 48 rather than 96 and 96, and Rabensteiner et al 15 and 15 rather than 30 and 30. The unit of analysis matters because the Results ask readers to read the pooled estimates cautiously over inter-eye correlation, and readers cannot tell whether the printed intervals account for inter-eye correlation until the authors state the unit of analysis.

Fourth, a wider production error: the Figure 4, 6 and 7 captions assert a benefit that their intervals do not support. Figure 4 “indicates an overall improvement in tear production” against a pooled 0.61 (95% CI −0.96 to 2.18; p = 0.45). Figure 6 “suggests a reduction in fluorescein staining scores in the intervention group, indicating improvement in ocular surface integrity” against a pooled −0.38 (95% CI −0.92 to 0.16; p = 0.17). Figure 7 “suggests a decrease in staining scores in the intervention group compared with the control group, indicating improvement in ocular surface damage” against a pooled −0.18 (95% CI −0.75 to 0.40; p = 0.55). All three intervals are compatible with a null effect. The overstated captions matter in a clinical setting because punctal occlusion is often the last step before permanent occlusion by thermal cautery in cases of severe refractory dry eye. Figures 3 and 5 carry purely descriptive captions, which read as uneven caption drafting rather than a deliberate reading of the plots.

Fifth, a reporting error: the Discussion prints a staining estimate that does not appear in any of the printed analyses. The Discussion describes corneal fluorescein staining as “markedly improved”, with a mean difference of about −0.47 at p < 0.00001 and reductions at both short- and long-term follow-up. However, the only comparative analysis of corneal fluorescein staining printed is Figure 6 (WMD −0.38; 95% CI −0.92 to 0.16; p = 0.17). Staining subgroup analysis by follow-up duration was not provided, and Table 4 prints each study’s own before-and-after values without producing a pooled estimate. The authors may identify the analysis behind −0.47 and, if it pools the Table 4 within-group data without a printed trace, publish it with its study set and heterogeneity statistics.

Sixth, another reporting error: the same Discussion paragraph describes 14 studies involving 1567 participants, while the abstract and the Results both report 1537. Table 1’s Total column sums to 1537, so the Discussion’s 1567 is wrong.

Finally, the abstract’s Conclusion retains the benefit language while omitting the null comparative findings two paragraphs above the abstract’s Conclusion, and does not match the Conclusion section, which opens by stating the null findings. This letter disputes neither the safety data, the within-group improvements, nor the symptomatic benefit the authors report, and captions written separately from analyses explain the caption errors without implying the analyses themselves are wrong. The review’s most informative finding is its own null result: punctal occlusion does not show a significant comparative advantage on any of the four outcomes, although individuals improve consistently in uncontrolled comparisons. Restating the captions, identifying the source of the p = 0.01 and −0.47 estimates, naming the unit of analysis, and conforming the abstract to the Conclusion section would give the paper one consistent message throughout.

Funding Statement

There is no funding to report.

AI Disclosure

The authors used Claude Opus 5.0 to assist in drafting and structuring the manuscript text. All scientific content, data, interpretation, and conclusions were provided, verified, and approved by the authors; the authors take full responsibility for final content.

Disclosure

The authors report no conflicts of interest in this communication.