Background
Since the introduction of robot-assisted laparoscopic surgery, it has transformed numerous operations across specialties. In fact, roughly 3000 robotic units have been introduced into the United States of America (USA) over the last decade [
1]. The platform offers advantages such as three-dimensional optics, elimination of tremors, and improved range of motion [
2]. Concerns about proper implementation, utilization, and lengthier operating room (OR) time have been raised. Despite rapid expansion, there remain questions about whether the use of the robot translates to an improvement in clinical outcomes or improves the efficiency of surgery.
In general surgery, the number of robot-assisted procedures has grown. Between 2012 and 2018, a large statewide collaborative described a more-than-10-fold increase in robot-assisted general surgery cases [
3]. Over one million cholecystectomies are performed annually in the USA, and it was one of the first robot-assisted general surgery procedures attempted. Regardless of early adoption, the optimal technique to implement when using the robot has not yet been established [
4] (such as multi-port or single-port cholecystectomy [
2,
5]), and there have been few reviews comparing these procedures [
6‐
9]. Those that do exist lack evaluation of pertinent clinical outcomes (i.e., specific postoperative complications) [
6,
7,
9], fail to consistently address all types of robot-assisted and laparoscopic comparisons (multi-port versus single-port) [
7,
8], and have reached disagreeing conclusions regarding outcomes such as operative times [
6‐
9] or incisional hernia rates [
6,
8,
9]. Updating the literature to compare surgical techniques in a systematic way provides surgeons with data to guide clinical practice and potentially improve outcomes and efficiency.
This systematic review analyzes the clinical effectiveness of robot-assisted surgery compared with the laparoscopic approach for cholecystectomy for benign gallbladder disease.
Methods
This review is part of a larger review commissioned by the Department of Veterans Affairs (VA) on the clinical outcomes and cost-effectiveness of robot-assisted procedures for general surgery. The review process was supported by a Technical Expert Panel (TEP) consisting of general surgeons, who specialize in robot-assisted surgery and are policymakers from across the country. This systematic review is reported using PRISMA standards, and the protocol for the larger review was registered in PROSPERO: CRD42020156945.
Literature search
All searches included PubMed, Embase, and Cochrane (all databases) from January 2010 to March 2020. The search used a broad set of common terms relating to “robotic surgical procedures” or “robotic-assisted,” “cost-effectiveness,” and “cholecystectomy.” We excluded studies published prior to 2010 since robot-assisted procedures were not widely being performed, and earlier studies likely captured surgeons within their “learning curve.” This decision was supported by our TEP (see Supplemental Data Content
1 for complete search strategy).
Study selection and data collection
All stages of the review were completed by two independent team members, and disagreements were reconciled through a discussion. Studies were included at either the abstract or the full-text level if they (1) studied patients undergoing elective cholecystectomy for non-cancer indications, (2) included one group of patients treated with a robot-assisted technique, (3) had a comparison to patients treated with a laparoscopic approach, and (4) measured intraoperative, perioperative, or postoperative outcomes. Both randomized controlled trials (RCTs) and observational studies were included. Abstracts were included in the review and underwent the same quality assessment and duplication exclusion as full texts. All exclusion criteria are included in our literature flow (Supplemental Data Content
1).
Data extraction was completed in duplicate. All discrepancies were resolved with full group discussion. We abstracted data on the following: study design, patient characteristics, sample size, intraoperative outcomes, postoperative outcomes, long-term functional outcomes, duration of follow-up, and data needed for the Cochrane Risk of Bias tool or Cochrane Risk of Bias In Non-randomized Studies – of Interventions (ROBINS-I) [
10,
11]. Data are reported as differences between the robot-assisted and laparoscopic groups using summary statistics (means, medians, or proportions as appropriate) in the results and figures.
Risk of bias and certainty of evidence
RCTs were assessed for quality (risk of bias) with the Cochrane Risk of Bias tool [
10]. We used the ROBINS-I [
11] for observational studies. We also used the criteria of the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) working group to assess the overall certainty of the evidence [
12]. Each outcome was measured on consistency, directness, and precision with an overall certainty of evidence of high, moderate, low, or very low.
Statistical analysis
Due to the heterogeneity in clinical outcomes of both the RCTs and the observational studies, we did not conduct a meta-analysis. The data synthesis is narrative. We presented the data by grouping studies based on the number of surgical access ports used. This grouping is important given that clinical outcomes (i.e., incisional hernia rates or operative times) may differ based on the number of ports, and interpretation of the data must include this context. The three comparison groups were as follows: robot-assisted multi-port compared with laparoscopic multi-port, robot-assisted single-port compared with laparoscopic multi-port, and robot-assisted compared with laparoscopic (unknown port number). Statistical analysis was done using R v4.0.2.
Conclusion
Our review found that OR time is longer for robot-assisted cholecystectomy as compared with laparoscopic cholecystectomy. There was no evidence of differences in intraoperative complications or conversion rates between surgical approaches. LOS, readmissions, and SSIs also had no evidence to support differences between techniques. Pain was examined, but the methods used within the studies were too heterogeneous to make conclusions regarding this outcome. Finally, rates of incisional hernias may be different when comparing approaches with different numbers of ports; however, when accounting for the use of the same number of ports, there was no evidence of a difference in outcomes.
Our search yielded four RCTs, four propensity-matched studies, and 36 observational studies, and thus, this review on robot-assisted cholecystectomy is the largest to date (see Supplemental Data Content
7) [
6‐
9]. There were several limitations to the prior reviews. The first was that the most recent review published included 26 studies published up to 2017. Second, the prior reviews made inconsistent conclusions regarding OR time with two identifying no differences [
7,
8] and two identifying longer time needed for robot-assisted cases [
6,
9]. Finally, three of the four previously published reviews grouped all postoperative complications together [
6,
7,
9], making it difficult to reach conclusions regarding complication severity. Our work updates and expands on the prior reviews performed and is the most comprehensive thus far with 44 studies, including those published up to 2020. Our review identified a lack of data needed to examine the differences in operative technique. Only two of the four RCTs were designed to study patient clinical outcomes as their primary outcome of interest, while the other two were primarily examining surgeon-related outcomes [
27,
32]. The majority of studies were observational, with concerns for selection bias regarding which technique may be preferentially utilized for certain patients. While 19 of 44 studies compared techniques with the same number of ports, the majority compared single-port techniques with multi-port techniques or did not specify. In the studies where these data were reported, we consistently found differences in OR time between techniques.
Although the OR time appears longer for robot-assisted cholecystectomy, with the largest median difference in time (found in the RCTs) at 38 min, such a difference may represent a variety of modifiable factors such as surgeon learning curve, OR staff efficiency, and case selection [
32]. Differences in outcomes between techniques must be considered within the context of the OR staff learning curve. The surgeon learning curve is a well-characterized concept that has been applied to robot-assisted surgery. We attempted to control for this variable by only including studies after 2010; however, it is possible that some of the data remain influenced by this factor. Indeed, while 90% of the reviewed studies acknowledged the possibility of a learning curve, only five reported data and assessment on these trends [
14,
19,
26,
50,
55]. The learning curve also applies to the OR staff’s setup and takedown of the robot unit and to the flow of the operation. Robotic instrument exchanges by inexperienced staff can compromise the efficiency of the surgery and may contribute to longer OR times reported. This is juxtaposed onto the familiar passing of instruments and exchanges occurring in conventional laparoscopic cholecystectomy. Given the nuanced nature of the factors contributing to OR time, it is challenging to conclusively state that the increased time seen in the robot-assisted technique is due to the lack of experience. Furthermore, while interpreted as a negative outcome, longer OR times may be encouraging safer practices, particularly for more advanced gallbladder pathology.
Our review had several limitations. The first is that the studies only examined surgery for benign, elective gallbladder disease—a process that is associated with low complication rates overall. Thus, differences in technique are unlikely to greatly affect these outcomes. As the use of the robot-assisted technique continues to expand, it is increasingly being applied in non-benign and non-elective settings for complex gallbladder disease and cases. Given the differences in complexity for such indications, the results from this review may not be generalizable to these populations. Second, we were unable to test for publication bias and cannot make any conclusions about its possible existence. Third, the quality of studies and heterogeneity in outcome measurement limited our conclusions. Fourth, this review did not address the differences in cost which may represent an important difference between techniques.
In summary, OR time was found to be significantly longer for the robot-assisted technique, although a variety of factors may explain such differences. The other clinical outcomes did not differ between techniques for benign, elective gallbladder disease. Understanding the differences in outcomes for robot-assisted surgery is critical as the use of this technology is being introduced across surgical disciplines and will increasingly be used to address more challenging pathology. Future work should focus on RCTs or propensity-matched studies that include clinical endpoints as primary outcomes (i.e., operative time, pain, or incisional hernias, measured in a standard fashion) and make appropriate comparisons when examining the number and type of ports used. Given the expanding use of robot-assisted cholecystectomy, these studies should also consider and control for other indications (i.e., acute cholecystitis, malignant disease). An analysis of the costs of the robot-assisted technique relative to the potential benefits including an analysis of how the robot may improve or add new challenges for surgeon ergonomics will also contribute to the data used when considering the use of one platform over the other. Robotic technology will become more ubiquitous, thus understanding its impact remains of paramount importance in the quality control and implementation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.