Sensitivity of the Pooled Surgical Site Infection Estimate to Study Design in a Meta-analysis of Prophylactic Negative Pressure Wound Therapy in Spine Surgery

Dear Editor,

The systematic review and meta-analysis by Muthu et al ¹ provides an important and methodologically rigorous synthesis of the available evidence regarding prophylactic closed-incision negative pressure wound therapy (NPWT) in spine surgery. By incorporating 13 studies and applying random-effects modeling with Knapp–Hartung adjustment, the authors offer a clinically relevant quantitative assessment of surgical site infection (SSI) prevention in a field where randomized evidence remains limited. The work represents a meaningful contribution to perioperative spine care and informs ongoing discussions regarding selective versus routine NPWT utilization. One methodological consideration merits focused discussion: the influence of the historically controlled study by Imtiaz et al (2024), as included in the SSI meta-analysis. In the forest plot presented by Muthu et al, this study demonstrates a directionally discordant effect (log odds ratio 2.55; 95% CI 1.07–4.04), favoring the control group, and contributes approximately 9% statistical weight to the pooled estimate. ¹ Notably, it includes a markedly imbalanced cohort (50 NPWT patients vs 5630 historical controls) and employs a retrospective design using a historical comparator. Historically controlled studies are particularly susceptible to secular trend bias, changes in perioperative infection-prevention protocols, institutional practice evolution, and confounding by indication. Methodological guidance for synthesis of nonrandomized studies recommends careful handling of such designs and preferential consideration of adjusted estimates where available. ² Furthermore, ROBINS-I emphasizes that confounding is typically the dominant source of bias in nonrandomized intervention studies, and historical comparisons are especially vulnerable to bias due to unmeasured temporal changes. ³

The overall pooled SSI estimate reported by Muthu et al (OR 0.42; 95% CI 0.28–0.63) is accompanied by substantial heterogeneity (I² ≈ 78%). ¹ Because the Imtiaz study is both directionally opposite to the majority of included studies and statistically influential, its inclusion likely attenuates the apparent protective effect of NPWT and contributes materially to between-study heterogeneity. Exclusion of this study would therefore be expected to (1) shift the pooled odds ratio further below unity, indicating a stronger apparent protective effect of NPWT; (2) reduce heterogeneity by removing a discordant outlier; and (3) modestly alter confidence interval width due to reduced study count. Importantly, such exclusion would not materially affect pooled analyses of wound complications, reoperation, or length of stay, as this study does not contribute to those outcomes.

¹ A prespecified sensitivity analysis excluding historically controlled cohorts—or alternatively restricting the primary SSI synthesis to prospective or non-historical designs—would clarify the robustness of the central finding and strengthen causal interpretability. Given the clinical and economic implications of adopting prophylactic NPWT in spine surgery, such clarification would enhance confidence in translating these findings into practice.