Abstract
W
First, the authors primarily expressed disapproval of our results for permanent recurrent laryngeal nerve (RLN) injury (favoring traditional clamp-and-tie hemostasis when compared with surgical devices), and did not identify reservations regarding other endpoints, suggesting that there may be an undue focus on an individual result rather than an objective concern with our methodology.
Second, many device analyses suffer from an underdetection and underreporting of RLN injury because of inadequate assessment and underpowering for this endpoint. We only included studies that assessed RLN function according to the gold-standard methodology of pre- and postoperative laryngoscopy, and included the two studies that were adequately powered for a comparative assessment.
Third, the authors also inaccurately state that Starion is “an electrocautery device” and that one of the studies we used was “not randomized.” Starion utilizes thermal fusion technology to seal and divide vessels simultaneously in “the absence of electrical transit through tissues” (3), and the study to which they refer is classified as a randomized controlled trial (4).
Fourth, in this case we felt that conventional meta-analysis offered a restricted view as there are several hemostatic devices and outcomes to consider (5). Limiting our study to a standard meta-analysis in such a case where multiple comparisons are required may have led to selection bias. As an example, there are 14 pair-wise comparisons for the Harmonic scalpel versus clamp-and-tie in the literature but only 3 for the Harmonic scalpel versus Ligasure with regard to permanent RLN paralysis. The disproportionate representation of direct comparisons (known as “co-occurrence”) has previously been reported and can bias results through device comparisons against inferior “straw men” (agents perceived as easy to beat) while avoiding comparisons against other effective treatments (5).
Fifth, even when conventional meta-analysis is applied, the Harmonic scalpel is associated with an increased risk of permanent RLN injury over clamp-and-tie hemostasis. From the direct comparisons in our article, there is an odds ratio of 1.74 and a statistical trend (p=0.08) favoring clamp-and-tie hemostasis over the Harmonic scalpel.
Sixth, from a technical perspective, dissection near any nerve is safer in the absence of heat. Lateral thermal spread is a well-recognized problem associated with ultrasonic coagulation and bipolar systems that can place the RLN at risk. The clamp-and-tie technique is not associated with this problem as it involves no heat production.
Seventh, while we note the comment that “not all devices are similar in a given treatment arm,” this is a recognized limitation of studies where there is rapid evolution of surgical technology. We accounted for this factor in our methodology by adjusting the results by year. Similarly, to answer the authors' comments about heterogeneity, we limited the heterogeneity between studies by conducting a metaregression analysis that adjusted for publication year (technological evolution of devices) and Jadad score (quality of randomized controlled trials).
Network meta-analysis is a valuable evidence synthesis technique developed precisely to overcome the limitations associated with conventional meta-analysis. By integrating all the data from both direct and indirect comparisons, network meta-analysis provides an unbiased, wider view of the evidence landscape (5). As a result, it can prevent the assessment of evidence through a “keyhole” and can offer better visualization of the “evidence horizon” (6).
Our network meta-analysis represents the largest meta-analysis to date on hemostatic devices in thyroid surgery. It offers preliminary and robust evidence to guide clinical decisions and policy makers to adopt safer thyroid operations and can also contribute to the future development of national or international guidelines regarding the application of hemostatic devices in thyroid surgery.