Meta-analysis can be used for both qualitative and quantitative evaluation of existing literature by comparing and integrating the results of different studies and taking into account variations in characteristics that could influence the overall estimate of the outcome of interest. Although meta-analysis has been traditionally applied and was mostly confined to randomized controlled trials (RCTs), meta-analytical techniques using nonrandomized controlled trials (NRCTs) might be a good method for use in some clinical settings in which either the number or the sample size of the RCTs is insufficient[18, 19]. To our knowledge, this is the first comprehensive meta-analysis comparing RT versus ET.
RT is often perceived as being more time-consuming, because of the additional set-up time required. Operating times depend mainly on the experience and skill of the surgeon. In this meta-analysis, we found that there was no significant difference in operating time between RT and ET. This may be attributable to the shortened learning curve with RT, as it has been suggested that robotic systems make the technique easier to learn, even by relatively inexperienced endoscopic surgeons. With increasing experience, set-up time gradually decreased, and the actual time may be shorter in RT. There was no significant difference in conversion rates between RT and ET.
Although RT offers a number of advantages over ET, including improvements in manual dexterity, ergonomics, and visualization, the results of the present meta-analysis suggest that there is no additional clinical benefit for RT over ET. The disadvantages of RT are a higher rate of complications and a greater amount of drainage fluid. It has been suggested that the characteristics of RT might reduce complications because, using the Da Vinci Surgical System, robotic arms are used for retraction and dissection, and their use has been found to reduce unnecessary procedures and to minimize iatrogenic tissue injury during retraction. Consequently, our result is difficult to explain, and more studies are needed before such a conclusion can be drawn. There was no difference in post-operative hospital stay between the two groups, implying that the time required for patients to resume daily activities might be similar between RT and ET.
Oncologic outcomes after thyroid cancer surgery are affected by the extent of lymph-node dissection and the completeness of thyroidectomy[21, 22]. Some studies have concluded that more lymph nodes are harvested via RT compared with ET, and that the robotic method may improve the long-term prognosis in patients who undergo surgery for thyroid cancer[13, 14]. In this analysis, we found no significant differences between RT and ET in the number of lymph nodes harvested; however, long-term follow-up evaluation is necessary to evaluate the exact oncologic outcomes of RT for thyroid cancer.
In the studied articles we found significant heterogeneity in operating time and number of lymph nodes harvested, which may be explained by the differences in personnel skills, extension of lymph-node dissection, and period of the learning curve. Because of this heterogeneity, we used a random-effects model in this meta-analysis.
There are several limitations to this meta-analyis, and consequently, the results should be interpreted with caution. First, the data came from NRCTs, and the overall level of clinical evidence was low. It has been reported that NRCTs might either overestimate or underestimate the magnitude of the measured effect in an intervention study, regardless of quality scores. However, Abrahama et al. found that meta-analyses carried out on well-designed NRCTs of surgical procedures were probably as accurate as those carried out on RCTs, and all six studies included in this study were NRCTs. Second, there was heterogeneity between the two groups because it was impossible to match the patient characteristics across all of the studies. We applied a random-effects model to take variation between studies into consideration, and we believe that the heterogeneity would have had very limited influence. Finally, it is possible that investigative groups might be more likely to report positive results, and that studies with significant outcomes are more likely to be published, therefore, potential publication bias might be present in our analysis.