Conversion rates in robotic thyroid surgery: A systematic review and meta‐analysis

Abstract Objective To define the conversion risk to open procedure during robot‐assisted thyroid surgery (RATS) identifying potential specific subclasses of procedures or accesses at higher conversion risk. Methods In a PRISMA‐compliant framework, all original prospective studies providing RATS conversion rates from multiple databases were pooled in a random‐effects meta‐analysis. Conversion rates were compared between different typologies of thyroid surgery and robotic access. Results 13 studies were deemed eligible. Four conversions from two studies were reported out of 398 procedures. No significant heterogeneity was observed (Cochran's Q p = 0.932; I2 = 0%). The pooled conversion rate was 1% (95% confidence interval, 0.1%–2%). The ANOVA‐Q test failed to show significant differences when comparing type of thyroid surgery or robotic access (respectively p = 0.766 and p = 0.457). Conclusion While the conversion rate appears consistently low across studies, prospective data collection and systematic reporting of procedural complications are required for framing high‐risk procedures and accesses.

required, among other causes, for excessive bleeding, previously undetected neoplastic infiltration or unexpected disease extension, or technical issues. [5][6][7] Despite the potential need for conversion being recognized in most case series, conversion rates are reported inconsistently in the literature and span from large case series with no conversion 8 to significantly preliminary smaller series with rates higher than 15%. 9 To the authors' knowledge, no study has systematically explored the risk of conversion in RATS or addressed whether different RATS procedures (e.g. Total thyroidectomy ,TT, hemithyroidectomy (HT), or radicalisation thyroidectomy) or approaches (e.g. transoral, facelift, gasless transaxillary, retroauricolar, robotic-assisted breast-axillo insufflation thyroidectomy) hold significantly different conversion rates. Defining such risk of conversion appears pivotal, as the need for a neck incision for controlling the surgical field, despite not hindering the procedure outcomes, nullifies the major advantage of RATS, that is, the scarless or near-scarless approach. 10 This systematic review and meta-analysis aims at delineating the risk of conversion in RATS and assessing whether specific subclasses of procedures or accesses should be regarded at higher risk.

| METHODS
This review was registered in the International Prospective Register of Systematic Reviews under the number CRD42021277928.

| Search strategy
A systematic review and meta-analysis was conducted between 12 September 2021, and 20 January 2022, according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses reporting guidelines. 11 We completed systematic electronic searches for studies written in English, Italian, German, French, or Spanish published until the search date that reported original data obtained from humans and focussed entirely or partly on RATS in humans.
On 23 September 2021, we searched MEDLINE, Embase, Web of Science, Scopus, Cochrane Library, and http://ClinicalTrials.gov databases using wide search strategies for thyroid-, thyroid surgery-, and robot-related terms. The detailed search strategy with the number of unique items retrieved from each database is available in Table 1.
We included any study dealing with RATS in humans. We excluded cadaver studies, meta-analyses, systematic and narrative reviews, and case reports, though references from review articles were hand-checked for additional potentially relevant studies. No minimum study population was required. We included only prospective studies that explicitly reported conversion rates (even if nil) and specified the robotic technique of choice and the type of thyroid surgeries that had been performed.
Abstracts and full texts were reviewed in duplicate by different authors (B.M. and L.N.). To maximise the rate of inclusivity in the early stages of the review, at the abstract stage, we included all studies deemed eligible by at least one rater. Then, during the fulltext review stage, disagreements were resolved by consensus between raters.

| Search results
Among the 1356 unique research items initially identified, a total of 184 articles were selected to undergo full-text evaluation. Ultimately, 13 studies published between 2010 and 2019 were retained for further analysis (see Figure 1). Table 2 reports the characteristics and demographics of the included studies. 11 articles were prospective cohort studies, [15][16][17][18][19][20][21][22][23][24][25] one was a prospective case-control study 26 and one was a randomized controlled trial, RCT. 27 Four studies were performed in Europe all occurring during TGA total thyroidectomies. One conversion was due to excessive bleeding in a female patient, another one was due to unexpected high glandular volume (multinodular goitre) and two were due to previously undetected significant tumour extension (specifically, a papillary cancer invading the cricothyroid area and a follicular carcinoma with cranial extension). All converted procedures were completed via a midline neck incision and without any further complication. Data on procedures, approaches, and conversion rates are reported in Table 3. Final histology reports for included patients are reported in Table 4.

| DISCUSSION
To the authors' knowledge, this is the first systematic review to address specifically the risk of RATS conversion into open-neck procedures. Despite the small effect size, we found that the conversion rate is consistent between different studies and with the 1% pooled rate emerging from our meta-analysis. This issue, intrinsic to all minimally invasive procedures irrespective of the surgical site, has not been addressed by already published review works, either because they were focussed on other safety features 28 or because they simply explored the differences between robotic and open procedures, the latter being unaffected per se by conversions. 29,30 Analogously, the same risk has not been specifically assessed also for endoscopic thyroidectomies procedures, with review works and meta-analytic comparison focussing again on different aspects of the surgical procedures and other patient outcomes. 31,32 The 1% conversion rate stemmed from four different events during TGA total thyroidectomies reported in two studies from different groups. 15  Abbreviations: BMI, body mass index.; F, fair; G, good; lung, and blood institute study quality assessment tools; NHI-SQAT, national heart; OCEBM, Oxford centre for evidence-based medicine; PCCS, prospective case-control study; PCS, prospective cohort study; RATS, robot-assisted thyroid surgery; RCT, randomized controlled trial; SD, standard deviation.  Even if we take into account the experience of high-volume tertiary centres or the pilot experience on specific accesses or high-risk patient classes, the mere existence of such a considerable gap in reported conversion rates claims the possibility of a reporting bias. On the basis of this potential bias, this meta-analysis was based only on prospective studies, thus allowing for a higher level of evidence and also for avoiding duplicated results that may be generated by partially overlapping case series presented in different articles or multi-centre studies.
On the other hand, the small number of prospective studies available in the literature and their relatively small sample sizes prevented us from drawing conclusions on the potential differential risk between different types of RATS procedures or accesses.
Although TT might indeed hold a higher conversion risk than HT, being thyroidectomy and TGA respectively the most common procedure and access in this meta-analysis, it's not surprising that all conversion cases belong to these groups. Analogously, the sample size is too small to draw any reliable comparison with the conversion rate for endoscopic procedures and the allocation bias to RATS versus endoscopy-assisted procedures or open procedures might be considerable, as the single RCT included demonstrates. Furthermore, the relatively small size of included studies determines a more considerable publication bias. Nevertheless, these biases were considered too low to hinder the overall value of our conclusions.
Another limitation of this meta-analysis stems from the heterogeneity of RATS eligibility criteria in terms of nodule/thyroid volume and the inconstant reporting of BMI criteria, which do not allow to draw any conclusion on which thyroid-and patient-specific characteristics determine a higher risk of conversion. Last, databases searches didn't locate any eligible stud reporting conversion rates for transoral robotic thyroidectomy (TORT), so this approach was not covered in our analysis or included in any subgroup evaluation.
Nevertheless, a recent wide retrospective multicentric review suggested TORT has similar conversion rates as those emerging from our systematic review, around 0.7%. 36

| CONCLUSION
By providing a novel insight on the conversion risk in RATS, this meta-analysis calls nevertheless for greater attention to this often neglected surgical adverse event. Only routine implementation of a F I G U R E 2 The pooled conversion rate in patients undergoing robot-assisted thyroid surgery (RATS). Effects and summaries were calculated using a random-effect model weighted by the study population common and unbiased reporting system into prospective multicentric studies might allow for better defining patient groups, RATS procedures, and accesses at higher conversion risk, and providing a sounder risk assessment. Irrespective of future studies, the conversion risk should be adequately discussed with patients in everyday practice, given its impact on the secondary outcomes of RATS. Even if aggregate conversion rates are low, an average 1% risk of neck scar is worth exploring with patients who mostly see RATS as the scarless answer to their thyroid disease.