An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Ho C, Tsakonas E, Tran K, et al. Robot-Assisted Surgery Compared with Open Surgery and Laparoscopic Surgery: Clinical Effectiveness and Economic Analyses [Internet]. Ottawa (ON): Canadian Agency for Drugs and Technologies in Health; 2011 Sep. (CADTH Technology Report, No. 137.)
Over the last decade, there has been a rapid uptake of robot-assisted laparoscopic surgery. Standard laparoscopic approaches to surgical procedures have improved patient care in some fields, such as cholecystectomy. For more complex operations, such as radical prostatectomy, a laparoscopic approach is associated with a long learning curve and is technically challenging for the surgeon.194 Robot-assisted surgery has been reported to provide benefits to the patient and surgeon. This health technology assessment reviews the published literature on four types of robot-assisted surgery: radical prostatectomy, nephrectomy, hysterectomy, and cardiac cases. Many other robot-assisted surgeries have been reported, but we have limited the scope to these surgeries, because they encompass the most commonly performed procedures.
The clinical review of this technology assessment included 51 studies for the indication of prostatectomy,29–79 26 for hysterectomy,80–105 10 for nephrectomy,106–115 and eight for cardiac surgery.116–123 All studies used prospective or retrospective observational designs. Based on the interpretation of primary estimates from meta-analysis, the following observations were made: robot-assisted surgery was shown to reduce the length of hospital stay compared with open prostatectomy, laparoscopic prostatectomy, open hysterectomy, laparoscopic hysterectomy, and laparoscopic partial nephrectomy; blood loss and transfusion rates were reduced with robot-assisted surgery, compared with open prostatectomy, laparoscopic prostatectomy, and open hysterectomy; robot-assisted surgery reduced positive margin rate compared with open prostatectomy in pT2 patients, and reduced postoperative complication rates compared with open hysterectomy and laparoscopic hysterectomy; and robot-assisted surgery increased operative time compared with open prostatectomy and open hysterectomy, and reduced operative time compared with laparoscopic prostatectomy. All these differences, which were identified in the clinical review, were statistically significant. Findings on robot-assisted cardiac surgery are scarce, but seem to favour robot-assisted surgery in terms of length of hospital stay. These observations were drawn from primary analyses of all data and include statistically significant findings. None of the evidence is derived from findings in gold standard randomized controlled trials (RCTs). Instead, it is drawn from a collection of observational studies of prospective and retrospective designs. RCTs conducted to verify these findings are warranted. Second, a persistent presence of statistically significant heterogeneity was associated with many meta-analyses in this review and did not appear to be associated with study quality or study design, and analyses based on other criteria, such as surgeon expertise, were not feasible; thus, residual confounding is a limiting factor. Furthermore, given the controversies in the meta-analysis of observational data and synthesis in the presence of unexplained heterogeneity, interpretations of pooled evidence need to be made carefully. In addition to pooled estimates, summaries of reported directions of intervention effectiveness and the associated levels of statistical significance were thus also provided in this report. Lastly, because there is likely to be uncertainty about the clinical relevance of differences between surgical approaches that were observed for clinical outcomes, such as differences in length of hospital stay and extent of blood loss, this aspect needs to be considered during decision-making.
In prostatectomy, the reduction in positive surgical margin rates will likely result in better cancer control outcomes and reduced secondary interventions for prostate cancer recurrence. Although these data are unavailable for RARP, the positive surgical margin rate can be extrapolated from open surgical data, because a positive surgical margin is a pathological measure and would be standardized. The shorter operating time for RARP compared with laparoscopic surgery can have an impact on surgical waiting lists. For example, if a surgeon can perform two RARPs compared with one laparoscopic surgery for an assigned operating day, the wait times will decline. Alternatively, if a surgeon can perform three open prostatectomies in the same time, then the wait lists may be adversely affected, lengthening the time a patient is on the wait list. The effect on surgical wait times has not been reported in this context.195 The comparison of postoperative complications reveals no advantage to one surgical approach. Heterogeneity among the studies and the reporting techniques also makes it difficult to draw conclusions.
Initiating a surgical robotics program has been associated with a learning curve. The initial experience worldwide involved the transition from an open approach or a laparoscopic approach to RARP.196–198 With RARP, several learning curve estimates have been published, ranging from a few cases to several hundred.40,197,199–202 One difficulty in interpreting the literature on surgical learning curves is the definition of a learning curve. Proficiency in RARP can be measured using different variables, including operative time, blood loss, complications, length of hospital stay, positive surgical margins, cancer control, and surgeon comfort. While these are individually important, the learning curve for each outcome measure can differ.40 There is no standard definition of the learning curve that is accepted in the surgical literature.198
A variety of ways to reduce the learning curve have been promoted for RARP or LRP, including mentoring of the novice surgeon by an experienced surgeon, mini-fellowship training, formal full fellowships, graduated responsibility during the procedures for trainees, and robotic team training.203–206 The literature is limited regarding the demonstrable benefits of these interventions and approaches.
The concern about the learning curve includes complications that result from surgeon inexperience with the technique. Several authors have made recommendations about case selection during the learning curve, based on experience. These recommendations include selecting patients with prostate gland volume less than 60 cm3,207 lower BMI,208 and less extensive disease.209 Complications during the early Canadian experience have been documented195 and these complications may counter any benefits provided by RARP in patient recovery, quality of life, and overall health. An organized, cautious approach to the implementation of surgical robotics programs in Canada must be considered. Because the outcomes of radical prostatectomy are related to surgeon experience,210,211 the use of robots at regional or tertiary care hospitals in a “centre of excellence” is a potential model to be considered.
More partial nephrectomy for small renal masses are being performed because of the increasing discovery of incidental masses with the use of cross-sectional imaging. Partial nephrectomy is typically performed for small kidney tumours that are presumed to be renal cell carcinoma, with the goals of complete extrication of the tumour and maximal preservation of kidney function (nephron sparing). The operation is technically challenging, and increasingly, laparoscopic and robot-assisted approaches have been reported. The review of the literature did not identify any adequate comparative studies for OPN and RAPN. Few studies compare LPN with RAPN. A shorter hospital stay was observed for RAPN, but the data are otherwise inconclusive. This is likely a factor of the recent introduction of RAPN worldwide. The first reported series was published in 2005,212 and the earliest paper suitable for this analysis was published in 2008.106 Other considerations regarding RAPN need to be acknowledged, but do not appear in this HTA because of a lack of suitable data. First, RARP facilitates the ability of surgeons to perform more complex surgeries, compared with LPN. Thus, patients who may have needed an OPN or a radical nephrectomy are having successful RAPN. There are insufficient data to address this argument. Second, an aspect of nephron-sparing surgery is the warm ischemic time (WIT) that is a result of clamping the renal blood vessels to allow the surgeon to resect the mass. With longer WIT, the risk of renal injury increases, with a resultant loss of kidney function. Several reports suggest RARP shortens the WIT compared with LPN, but a statement about the impact on renal function cannot be made.
Limited data showed that robot-assisted hysterectomy shortened the length of hospital stay, and reduced blood loss and transfusion rates and postoperative complications compared with open surgery and laparoscopic surgery, but it took longer to perform than open surgery. Although robot-assisted cardiac surgery seems to provide for a shorter length of stay compared with non–robot-assisted surgery, the paucity of the data and the heterogeneity among trials precluded any conclusion.
In the economic review, there were 30 economic evaluations58,86,96,102,115,119,123,129–151 of robotic surgery in the four indications: 15 in prostatectomy, four in cardiac surgery, two in nephrectomy, eight in hysterectomy, and one in multiple indications (prostatectomy, cardiac surgery, nephrectomy). There was variation between studies regarding their conclusions about the costs and cost-effectiveness of robotic surgery; however, there was also variation between studies in the estimation and inclusion of costs. The estimation of QALYs in three cost-utility studies in radical prostatectomy was unclear. Most studies had limitations in the reporting of methods and results, and the relevance of most studies to a Canadian setting was also limited. One Canadian analysis in hysterectomy suggests that robotic surgery may be less costly than open surgery if the robot is used for five surgeries per week.
Because of the frequency with which this procedure is performed in Canada, radical prostatectomy was chosen as the indication for the economic evaluation. A cost-minimization analysis was conducted because an impact of robotic radical prostatectomy on major outcomes was not found in the clinical review. Robotic radical prostatectomy had shorter lengths of stay than open prostatectomy and laparoscopic radical prostatectomy, thus reducing hospitalization costs; however, the estimated per-patient costs of the robotic technology were large, leading to higher net incremental total costs of robotic radical prostatectomy, compared with open (incremental costs $3,860 per patient) surgery and laparoscopic (incremental costs $4,625) surgery. Other factors affecting incremental costs were the useful life of the equipment, specialist fees, currency exchange rates, changes in recurring costs, and annual caseload. The probabilistic sensitivity analysis suggests that RARP is more expensive than ORP and LRP in approximately 75% of cases, and that cost-saving situations with robotic surgery would largely be due to a variation in hospitalization costs.
The population impact analysis suggests 4,030 patients could undergo robotic surgery with a da Vinci robot in Canada annually, if the number of centres operating a robot expands from 11 to 31 (assuming similar institutional characteristics and average caseloads to those using a robot now). Consideration of large non-teaching general hospitals or hospitals with smaller capacity would expand the number of potential robotic centres to 85, and the annual patient population to 11,050. Considering the reduced hospitalization costs that result from decreased lengths of stay in each of the four indications, the net institutional costs for operating a robotics program for seven years is estimated to be $2.9 million, assuming an average robotics case and an annual caseload of 130 patients per year. When considering indication-specific programs, cardiac surgery is estimated to be the least costly, with a net program cost of $0.9 million over seven years, and prostatectomy the most expensive, with a net program cost of $3.5 million over seven years.
The limitation of the clinical review of this report is a lack of prospective RCTs of robot-assisted compared with laparoscopic or open surgical approaches.213 This analysis is based on mostly single-institution observational studies, which means that the level of evidence is not as robust as that of RCT data. More comparative studies assessing postoperative outcomes, such as sexual function and continence, are needed. Many outcomes showed heterogeneity across trials, but no apparent potential causes of heterogeneity — including trial quality, trial design, sample size, definition of outcomes, and surgeons’ experience — adequately explained these differences. Reporting of the potential covariates, such as surgeon expertise, was not provided, or was provided in formats that precluded categorization of many of the studies with outcome data available, and thus the potential for sensitivity analyses was limited. Enhanced reporting of future studies with such information is needed; even in studies where data were provided, a lack of sufficient detail about factors such as surgeon expertise may result in the presence of residual confounding. For localized prostate cancer, no RCT has been published, and there are several potential reasons. In general, localized prostate cancer has a long natural history; thus, even with surgical intervention, survival is measured 10 years to 20 years later. As a result, no studies exist. The outcomes that are analyzed here reflect short-term variables that have been reported. Until long-term data become available, no further conclusions can be drawn beyond those outlined. Another reason for the lack of RCT data is the fact that surgeons go through a learning curve when a new technology is introduced into the operating room. Few surgeons, if any, are considered to be experts at open prostatectomy, laparoscopic prostatectomy, and robot-assisted prostatectomy. Thus, any comparative study would include the surgeon as a variable. This is a potential source of bias for an RCT.
As new technologies are introduced, results involving small numbers of patients, technical modifications, and learning curves are more likely to be accepted for publication in the medical literature. Many of the studies that provided the basis for this analysis represent early experiences with robot-assisted surgery and are being compared with open surgical techniques with which the surgeons have experience using. Some papers cited here compare surgical outcomes between RARP and open surgery featuring small numbers of patients during the learning curve for the surgeons.29,61,72 A review on prostatectomy found that there was no evidence of publication bias by Begg’s test or Egger’s test.214
In Canada, most radical prostatectomies are performed via an open surgical approach. Thus, any advantages for robot-assisted surgery are weighed against open surgery outcomes and cost. For this HTA, the clinical data analyzed are not from Canadian centres and, as a result, potential sources of bias must be acknowledged (publication bias and patient selection bias).
The systematic review for the economic assessment was conducted in a rigorous manner. Most of the data used in the economic evaluation and the health services impact analyses were obtained from Canadian sources. Current data on the use of robotic equipment at all Canadian centres were made available. Analyses were provided in a disaggregated manner throughout the report, to allow for further assessment of the results. Sensitivity analyses were conducted throughout.
There were limitations in the estimation of the cost of training in the economic evaluation. Intuitive Surgical, Inc., requires surgeons who are training in robotic surgery with the da Vinci Surgical System to undergo its initial training program, and these costs were included in the economic evaluation. Their overall impact in the analysis was small. There are no similar requirements for laparoscopic surgery. Robotic surgery and laparoscopic surgery are associated with learning curves that require additional training and mentorship, and these costs are difficult to estimate and could not be captured in the analysis.
Lengths of stay and their between-group differences were estimated from meta-analyses of international studies, under the assumption that marginal differences in length of stay would reflect what might be seen in Canada. At the time of the analysis of the data for this report, CIHI did not yet have reliable data on lengths of stay for the robotically performed procedures that are considered in this report. These data will likely become more reliable in the future, as more robotic surgeries are performed, more current data become available, and estimation methods are refined.
Hospitalization cost estimates derived from CIHI data would necessarily include the cost of disposable surgical equipment. Because the classification of robotic surgeries in CIHI’s Discharge Abstract Database is recent, identification and costing methods for robotic surgeries is incomplete, and it is unclear whether the cost of robotic disposables has been included in the hospitalization costs. The costs of disposables for open and laparoscopic surgeries are likely to be included, but because of the level at which these costs are allocated in the CIHI method, if any of these costs are included, they are likely allocated uniformly across all surgical approaches. This implies that all our current estimates of hospital costs, regardless of surgical approach, may include an averaged allocation of the cost of surgical disposables, and all hospitalization costs would therefore be inflated by this average amount. If robotic disposables are included in this amount, they likely do not contribute a large relative weight, because few robotic surgeries are performed. Accounting for the cost of disposables separately in the economic analyses implies some double counting of these costs, but the fact that all hospitalization costs are inflated by the same amount led to the decision to assess them separately in the base case analysis. A sensitivity analysis that removed these costs from the cost-minimization analysis in prostatectomy showed that they had little impact in the open surgery comparison, and some impact in the laparoscopic surgery comparison; however, they did not affect the conclusions. In the budget impact analysis, these costs are presented separately, to allow for calculation with and without their consideration. In the population impact analysis, the number of hospital beds was used as a characteristic to identify institutions that are likely to adopt this technology. Surgical volume may have been a better indicator, but these data were unavailable.
Finally, there may be benefits of robotic surgery that are difficult to evaluate and that were not included in the economic assessment, such as the ergonomics of robotic surgery and the potential impact on surgeon fatigue and performance.
The primary economic evaluation applied the clinical results on robotic surgery in radical prostatectomy to a Canadian health care setting. The methods that were used to conduct the analysis were valid, and the patient populations to which the results apply appear to be representative of the types of cases seen in Canadian settings. Because national hospitalization data on robotic surgery are still being developed, it is difficult to assess how the lengths of stay reported in the clinical section of this report compare with those of actual Canadian surgical cases. The health care service use and costs used in the economic evaluation and budget impact analysis came mainly from Canadian sources.
RCTs are needed for the evaluation of clinical outcomes in all surgical procedures. There are limited data on outcomes from the Canadian centres using the robot are available. The decision to conduct a cost-minimization analysis was based on the absence of evidence for between-group differences in major outcomes. General QOL data in prostatectomy (and for the other indications) for the selected surgical comparisons were limited, and more research in this area may be useful. Longer-term data on patient outcomes in robotic surgery are also needed.
Except where otherwise noted, this work is distributed under the terms of a Creative Commons Attribution-NonCommercial- NoDerivatives 4.0 International licence (CC BY-NC-ND), a copy of which is available at http://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
