Last week, I wrote about factors that lead to the premature adoption of surgical technologies and procedures, the “bandwagon” or “fad” effect among surgeons, if you will. By “premature,” I am referring to widespread adoption “in the trenches,” so to speak, of a procedure before good quality evidence from science and clinical trials show it to be superior in some way to previously used procedures, either in terms of efficacy, cost, time to recover, or other measurable parameters. As I pointed out before, laparoscopic cholecystectomy definitely fell into that category. The popularity of the procedure spread like wildfire in the early 1990s before there was good quality data supporting its superiority to the “old-fashioned” gold standard procedure of open cholecystectomy. Another example, although not nearly as dramatic because the number of patients for whom the procedure would be appropriate is much smaller, is transanal endoscopic microsurgery. However, the difficulties in practicing science- and evidence-based medicine don’t just include fads and bandwagon effects. The example of laparoscopic cholecystectomy notwithstanding (which was largely driven by marketing and patient demand), surgical culture tends to be deeply conservative in that surgeons can be very reluctant to change practice even there is very strong evidence from clinical trials saying that they should.
Recently, a rather revealing study was published in Cancer1 about just such an example in urology, specifically in the treatment of kidney cancer. Before I get to the study, a little background is first required. The previous gold standard treatment for renal cell carcinoma confined to the kidney was open radical nephrectomy. By “open” I mean no laparoscopy and a surgical approach that uses traditional large incisions. “Radical” nephrectomy is different from a “simple” nephrectomy in that more than just the kidney is removed; the ipsilateral adrenal gland is also removed, and a complete regional lymph node dissection is performed as well. Over the last decade or two, newer procedures have shown their efficacy. Radical nephrectomies can be performed using the laparoscope with both equivalent oncologic outcomes in terms of recurrence, disease-free survival, and overall survival, with attendant advantages of laparoscopy, such as less pain, faster return of bowel function, and fewer days in the hospital. Another example is nephron-sparing nephrectomy for smaller tumors. All “nephron-sparing” means is that a total nephrectomy is not carried out. Rather, only the part of the kidney with the tumor is removed, leaving the rest of the kidney. The advantages include better preservation of long term renal function and reducing overtreatment of patients with indolent tumors, and a number of studies have also demonstrated oncologic outcomes equivalent to those due to radical nephrectomy.
The question this study, Diffusion of surgical innovation among patients with kidney cancer, addresses is how many patients are benefitting from this newer surgical technology. Basically, what the researchers, led by Dr. David Miller at UCLA, did was to examine data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program (commonly known as the SEER database) and the Centers for Medicare and Medicaid Services. This database is a repository of cancer incidence, treatment, and mortality. The methodology was rather complex, but involved using the database to identify a population-based cohort of patients 66 and older from 1997 to 2002 with kidney cancer. These cases were then linked to Medicare claims information, a linkage that was achieved for more than 90% of the cohort. A preliminary cohort of 6,515 Medicare patients with who were diagnosed between 1997 and 2002 with localized or regional, nonurothelial kidney cancer. Out of this group, 1,026 patients were excluded for whom claims denoting surgical treatment for kidney cancer could not be found, and three patients with bilateral cancer were also excluded. The final cohort included 5,483 cases (84.2% of the preliminary cohort). The investigators then examined how many of these patients received laparoscopic nephrectomy or partial nephrectomy. Patient characteristics and comorbidities, the most important of which was tumor size (which determined which tumors might be appropriate for partial nephrectomy), were correlated with treatment choice and surgeon-specific factors. From a bit of rather complex statistical modeling, an estimate was made for how much of the variation in surgical procedure was due to surgeon factors rather than patient factors.
What they found suggests that for many patients with kidney cancer, what surgery is recommended depends more on their surgeon’s practice style than on what the characteristics of the patient’s disease would suggest as the most appropriate procedure. First, they did note that during the time period of the study the proportion of patients receiving partial nephrectomy increased significantly from 7.1% to 14.6% (P<0.01). The increase was even more substantial in the subset of patients with tumors 4 cm or less (8.9% to 23.5%, P<0.01), which is the subset of patients for whom partial nephrectomy is most likely to be surgically appropriate. Also in patients with smaller tumors, the proportion undergoing laparoscopic radical nephrectomy increased from 1.2% to 20.3% (P<0.01). Even for those with tumors larger than 4 cm, it rose from 0.8% to 16.2% (P<0.01). On the surface, this would appear to indicate a fairly rapid change in surgical practice, as these increases occurred during only six years. So how do the authors come to the conclusion that many patients who should get these newer procedures don't?
Their conclusions come from an analysis of the variance in rates of use of these procedures. Basically, the investigators modeled the factors that could account for variance in the usage of these two procedures. After looking at all the patient-attributable sources of variance and other sources of variance that could be identified, they derived a "left-over" surgeon-derived variance. For partial nephrectomy, these unmeasured surgeon factors contributed to 17.5% of the variance in surgical choice, the second largest contribution to the variance in surgical choice. Tumor size contributed a larger proportion at 19.6%. For laparoscopic radical nephrectomy, unmeasured surgeon factors contributed 37.5% of the variance, which far larger than any other factor. The conclusion is that surgeon practice factors often contribute more to the choice of surgical procedure for kidney cancer than tumor and patient characteristics. In other words, if this study is to be believed, habit trumps science- and evidence-based medicine more often than we would like to admit. As the authors conclude:
Specifically, despite their potential advantages relative to open radical nephrectomy, partial nephrectomy and laparoscopy are used relatively infrequently in this population; moreover, much of the variance in their use is attributable to surgeon-specific factors rather than patient- or tumor-specific factors. Thus, for many older patients with kidney cancer, the surgery provided may depend more on their surgeon’s practice style than on the characteristics of the patient and his or her disease. Consequently, the timely dismantling of residual barriers to surgeons’ adoption of partial nephrectomy and laparoscopy is an important step toward improving the quality of care provided to patients with kidney cancer.
This study, of course, is not without limitations, and I’m not sure that its results still hold true in 2008. The main reason is that the data examined comes from, at the most recent, six years ago. These were the early years after urologists’ general acceptance of partial nephrectomy and laparoscopy. Indeed, the very rapid rate of increase in the usage of these procedures over the six year period examined is quite consistent with what happens in the early years of adoption of a new procedure. Against this interpretation, the authors cite data suggesting that the rates of these procedures stabilized in 2003 and 2004, the most recent year for which good data is available. Another weakness in the study is that it was a study of only older patients over 65. Perhaps the most difficult to deal with weakness of the study is that the investigators couldn’t identify every single factor that might have influenced a surgeon’s decision for one procedure over another. It is thus an assumption, albeit a reasonable one, that all or most of the excess variance is related to surgeon factors rather than patient factors.
Even though this study has a number of limitations, let’s for the moment assume that its results are accurate (at least for 2002) and that there is indeed a disconnect between what science-based medicine suggests as the surgical procedure for patients with kidney cancer that provides the best balance between efficacy and risk and what many patients are getting and that this disconnect is due to surgeon factors. Let’s assume for the moment that, for whatever reason, urologists are slow to adopt these procedures. Why would this be? Why, for example, would laparoscopic cholecystectomy in general surgery or laparosocopic prostate surgery in urology have caught on so quickly, while laparoscopic nephrectomy or nephron-sparing nephrectomy apparently are not diffusing into the urologic community as quickly as the clinical evidence would suggest that they should? That is the topic of an accompanying editorial by Drs. Lee Richstone and Louis R. Kavoussi of the North Shore-Long Island Jewish Health system2. After pointing out that previous studies have shown that the use of these techniques is, unsurprisingly, associated primarily with teaching hospitals, high volume hospitals, and urban centers, they then emphasize just how striking the underutilization of these techniques are:
The under-use of NSS and laparoscopy reported by Miller et al.8 indeed is striking. In their report of 5483 patients who underwent surgery for renal cancer between 1997 and 2002, only 11.1% of patients underwent NSS, and 11% of radical nephrectomies were performed laparoscopically. Even when the indications for NSS and LRN were strongest (tumors 4 cm and >4 cm, respectively), the use of NSS was only 23.5% and the use of LRN was only 16.2% in 2002. These are incriminating numbers. The finding that these practice patterns reflect surgeon/hospital-based variance is supported strongly by our own data. At our institution, 49% of all patients who underwent surgery for RCC underwent an NSS approach, including 85% of patients with tumors 4 cm. Moreover, 87% of these partial nephrectomies were performed laparoscopically. When a higher stage tumor mandated radical nephrectomy, 71% of those procedures were performed laparoscopically (unpublished data). It appears that, de facto, a 2-tiered system of urologic care is developing.
They also noted that robotic prostatectomy has rapidly increased in usage, despite the fact that it is a technically demanding procedure with a steep learning curve, and this was a procedure that was introduced into surgical practice a decade after nephron-sparing surgery and laparoscopic nephrectomy. Indeed, it is estimated that in 2007 the majority of prostatectomies were performed robotically in the U.S.. The reluctance to acquire new skills is clearly not the stumbling block. Nor does the diffusion rate of these new techniques appear to be proportional to their objective benefits, validated in multiple clinical trials. Indeed, the benefits of laparoscopic nephrectomy over open nephrectomy in terms of convalescence and decreased pain are huge, whereas the only validated benefits of robotic versus open prostatectomy are reduced intraoperative blood loss and (possibly) a quicker return of urinary continence. Otherwise, convalescence and morbidity appear to be fairly similar. Nor is the reluctance to adopt laparoscopic nephrectomy and partial nephrectomy due to lingering concerns over whether they are adequate cancer operations, either. It is well-accepted by now that they are.
So what is the reason? One possibility is that it’s because the incidence of renal cell carcinoma is nowhere near as high as it is for prostate cancer, less than one quarter. Case volume is important for the acquisition of new skills. It is thus speculated that surgeons won’t take the effort to learn a new procedure if they do not see a lot of patients needing that procedure. Certainly, that is a plausible reason why laparosocopic cholecystecomy spread so rapidly to basically all general surgeons doing gallbladder surgery within a few years. Gallbladder surgery is “bread and butter” surgery to general surgeons. It’s a very common operation. Moreover, there was a great deal of hype and advertising, making patients highly aware of the operation. Surgeons who could not do laparosocopic cholecystecomy were finding that they were no longer getting patient referrals for one of their most commonly performed operations. They had to learn the procedure fast or risk losing their practices. If it’s true, however, that it’s the difference in incidence that drives the difference in diffusion of these surgical therapies, an implication of this is that low volume surgeons are now disproportionately referring their kidney cancer patients to high volume centers rather than learning the new operation.
One possible explanation for why this might be is that there is a huge marketing push for robotic prostatectomy. It’s an advertising angle that many hospitals use to attract patients. There is also a major industry marketing effort to promote the technology, and it’s working. Patients are seeking out centers that have this technology. Despite the expense and the dubious proposition of whether the capital expenditure for the robotic system is justifiable on strictly a fiscal basis, surgical robotic systems are proliferating. As was the case 15 years ago for general surgeons and laparosocopic cholecystectomy, if a urologist wants to be seen as “modern” and up on all the latest techniques, he or she has to learn robotic prostatectomy, or a practice has to have a surgeon who does the procedure to whom his or her partners can refer such patients. There is no equivalent marketing push for laparoscopic nephrectomy (laparoscopy is, after all, old news compared to robotic surgery) or nephron-sparing nephrectomy. Alternatively, the increasingly poor reimbursement climate may be contributing:
Conversely, urologists are likely to refer patients for robotic prostatectomy more often than for NSS or laparoscopy. This is multifactorial. For the urologist without robotic training, deciding to perform an open radical prostatectomy rather than to refer a patient for a robotic procedure is a significant, and not necessarily profitable, commitment. Consultations regarding prostate cancer decision-making are lengthy and complex. The operation itself is time consuming and is accompanied by significant complication rates and postoperative concerns, such as incontinence and erectile dysfunction. Because the reimbursement rates per hour for surgical procedures have dropped 28.5% in less than a decade, whereas office-based evaluation and management billing has increased inversely, a disincentive to perform complex, time-consuming operations has developed that may have led to a disproportionate rate of referral for prostatectomy. In contrast, for the surgeon without the skills to perform NSS or a laparoscopic nephrectomy, the less complicated, open radical nephrectomy still can be offered. The procedure is a straightforward, less time-consuming operation with which all urologists are familiar. There are fewer postoperative concerns and morbidity to contend with. In addition, despite the cited benefits with respect to long-term renal functional outcomes, many urologists still question the necessity of nephron-sparing surgery in the presence of a normal contralateral kidney. In addition, neither the patients nor the industry drive the treatment to the extent that they do for prostate cancer surgery.
My purpose in writing this is most definitely not to cast aspersions on urologists. Rather, it was to point out that there are many barriers to practicing science- and evidence-based medicine and surgery in our present medical system that have nothing to do with science. These barriers include a perverse reimbursement system, industry-driven marketing that influences patient demand and surgeon practice, the difficulties inherent in acquiring new skills (a problem that is much more pronounced in surgery and specialties that are highly procedure-oriented), patient expectations and demands, as well as the general human conservatism that makes it difficult to persuade surgeons to change want to change practice from what they have been trained to do since the beginning of their surgical careers, or, as I like to call it, the “we’ve always done it this way” bias. Overcoming these barriers is critical to assuring that the best possible care is available for patients in as many regions as possible.
Of course, determining when a procedure has passed the threshold of scientific and clinical evidence to warrant diffusing out “into the trenches” is a question that is at least as complex and difficult as getting the benefits of clinically validated new procedures out to practicing surgeons, but that may have to be a topic for another week.
- Miller, D.C., Saigal, C.S., Banerjee, M., Hanley, J., Litwin, M.S. (2008). Diffusion of surgical innovation among patients with kidney cancer. Cancer, 112(8), 1708-1717. DOI: 10.1002/cncr.23372
- Richstone, L., Kavoussi, L.R. (2008). Barriers to the diffusion of advanced surgical techniques. Cancer, 112(8), 1646-1649. DOI: 10.1002/cncr.23369