Response to Letter: “Mesh Application in Large Hiatal Hernias

    loading  Checking for direct PDF access through Ovid


We appreciate the interest of Antoniou et al in our recent article titled “Laparoscopic Repair of very Large Hiatus Hernia with Sutures Versus Absorbable Mesh Versus Nonabsorbable Mesh: A Randomized Controlled Trial.” 1 They correctly highlight the limitations of the data we reported. In particular, they highlight the difference between clinical significance versus statistical significance. Before generalizing outcomes from any randomized trial to clinical practice, any differences identified must be determined to be not only statistically significant, but also clinically significant. In other words, the effect size or magnitude of difference must be big enough to be significant in the clinical context. Using P values alone to assess the significance of data is not appropriate, and we hope the readers of our article will also consider the clinical data, which were extensively described in our report. 1
We are acutely aware that the data presented in our randomized trial might be interpreted in different ways, perhaps influenced by bias and beliefs held by individual readers. Our study showed that some outcome differences reached statistical significance in secondary analyses. However, these differences were small and unlikely to be clinically significant. The primary outcome measures were not statistically different. All data were presented in the report and we hope that readers will carefully consider the outcomes and draw reasonable conclusions. Our study was powered a priori to identify a 25% reduction in the incidence of hiatus hernia (identified at x-ray or endoscopy) between either of the mesh groups versus the suture repair group. We are confident that all readers will agree that a reduction of this magnitude is clinically significant. A secondary power calculation showed that when the 2 mesh groups were combined, then for comparison of mesh versus sutured repair the study was powered to detect a 13% difference. This is still likely to be of clinical importance. The actual outcomes that were identified to be different at P < 0.05 were all small, and we do agree with Antoniou et al that these differences are unlikely to be clinically significant.
However, we disagree with the implication that the trend differences between the Timesh repair versus sutured repair groups can be claimed to be clinically significant. Although a trend is evident in the data presented, this trend never reached statistical significance, so it is not reasonable to claim this is a clinically significant difference. As pointed out in our article, it is possible that these differences might become significant in a larger trial, but equally they might not, so we encourage more randomized trials, which address this question. It should be noted, however, that when setting up our trial, the power calculations and recruitment targets were based on outcomes reported from previous trials, 2 and our trial actually enrolled more patients than any previously reported randomized trial, 2,3 including the trial reported from Antoniou et al's group in 2005. 4
We agree with Antoniou et al that longer follow-up is required, and individuals recruited are currently completing endoscopy and barium meal x-ray radiology at 3–4 years follow-up. These data will be analyzed and reported when the trial will be completed. Currently, the only randomized trial reporting long-term follow-up failed to show any difference for mesh versus no mesh at median 5 years follow-up, 5 similar to the results reported in our article. It is now 10 years since Antoniou et al's group described 12 months follow-up in their earlier trial, 4 and we would be very interested to see their long-term follow-up data.
    loading  Loading Related Articles