To date several authors have published cost analyses demonstrating that robotic hysterectomy for benign and oncologic indications is more costly compared to the laparoscopic approach.[1-4] Wright et al. published the most robust cost model to date comparing the cost of hysterectomy according to approach: laparoscopic versus robotic. Reviewing 264 758 hysterectomies for benign indication at 411 hospitals, the authors demonstrated that patients undergoing robotic hysterectomy experienced similar morbidity compared to patients undergoing laparoscopic hysterectomy. The robotic approach was associated with $2189 in additional cost compared to laparoscopy. Despite being more expensive than laparoscopy, other studies have found robotics to be less expensive and more effective than laparotomy. Wright's study is landmark given its size, the incorporation of fixed costs of acquiring and maintaining the robot, and the use of propensity score matching to address treatment allocation bias inherent to retrospective, non-randomized cohort studies.
An absence of randomized data
The use of RCT where clinical equipoise exists is central to the modern tenet of evidence-based medicine. There is paucity in the quantity and quality of RCT in surgery. Occasionally, pressure, either extrinsic or intrinsic, to introduce technology overcomes the restraint and collective effort required to perform RCT and technologies or novel surgical techniques are deployed without evidence proving their benefit. The intra-aortic balloon pump for the treatment of cardiogenic shock is one such technology that was deployed without rigorous RCT given widely shared beliefs in logic behind its theoretical benefit. After years of clinical use and entrenched provider bias preventing post-market RCT, Thiel et al. published a landmark RCT demonstrating that the use of the intra-aortic balloon pump is not superior to supportive care among intensive care unit patients with cardiogenic shock.
Although the stakes are not nearly as high, robotic surgery was introduced into the surgical arena with a similar lack of pre-clinical randomized trials with early adopters convinced of its theoretical benefit. With nearly 10 years of clinical experience, the likelihood of meaningful RCT comparing the robotic approach to laparoscopy and laparotomy diminishes as providers develop practice bias. While retrospective studies are helpful in comparing robotics to other modalities, the very same biases, which become impediments to RCT, complicate the interpretation of retrospective studies.
Provider bias regarding the allocation of patients to the robotic versus laparoscopic versus open approach introduces the potential for confounding, which would otherwise be avoided through RCT. The remedy for confounding in retrospective studies includes matched cohort analysis and propensity score matched analysis. However, once matched, the criteria used for matching can no longer be examined for association with study outcomes. For example, once matched for body mass index (BMI), BMI should no longer be evaluated as a predictor of outcome in retrospective studies regarding robotics. While stratified analyses provide a potential solution to this problem, prospective randomized controlled trials powered to allow sub-group analysis remain the gold standard.
Then again, a meticulous look at surgical research and the process of introducing medical devices to the US market provides compelling evidence why RCT may have limited utility in this arena. As elegantly stated by McCulloch et al., operations are complex procedures. Quality and results are affected by many factors, including but not limited to learning curve, variations in technique, blinding and quality control.
The concept of the learning curve was first introduced in behavioral psychology, and subsequently applied in the aircraft industry. A learning curve describes the increase in learning coincident with experience or repetition. In surgery, every new surgical technique, instrumentation or modification to an existing technique has a learning curve. Unlike drug trials, this is a unique variable in surgical RCT, especially minimally invasive surgery, and is difficult to control between treatment arms. During the learning curve, errors and adverse outcomes are more likely, regardless of pre-clinical training. During that learning curve, modifications, and variations of a technique or operation with a novel instrument has the potential to influence surgical outcomes and success. Solutions for the above problems unique to surgical RCT include the use of appropriate statistical techniques, and performance of surgical phase 2 trials prior to RCT in order to elucidate learning curve analysis of participating surgeons, suitable end-points for subsequent RCT and quality issues, among others.
The reasons for the lack of RCT in surgical research are far more complex, and perhaps out of the scope of this review. However, it is important to mention the lack of funding for surgical research, and the role of government in such endeavors. The United States Congress in 1997 amended the US Food and Drug Administration (FDA) Act on medical devices and radiation-emitting products in an effort to streamline the process of bringing safe and effective drugs, medical devices, and other therapies to the US market. Paradoxically, the current process is gradual and relies heavily presumably on expert committee members calling into question the weight of evidence versus expert opinion in the approval process.
A closer look at this process reveals a lack of transparency and a complex method that is daunting for any surgeon-investigator in this field of research. Offices of medical product regulation within the FDA are further divided into the Center for Biologics Evaluation and Research (CBER), the Center for Drug Evaluation and Research (CDER), and the Center for Devices and Radiological Health (CDRH). In contrast, in Europe, to legally place a medical device on the European market, medical devices must meet the ‘Medical Device Directives’, and devices or products conforming with such directives can obtain a ‘CE mark’. Clearly in the European Union the process is simplified providing physician-scientists an edge on medical/surgical devices research. Of note, an FDA or CE mark-approved medical device rarely requires RCT; moreover, it is not a ‘safety mark’ rather it means compliance with government regulatory process by the industry prior to market. Albeit a complex process, it is intended to verify that the device complies with safety, health, and environmental protection requirements: a process not suitable for RCT.