It is getting so hard to accrue to randomized trials in urologic surgery that we have pretty much stopped doing them altogether. This is perhaps most apparent if one visits surgery sessions at the AUA Annual Meeting. The most typical presentation describes how a surgeon changed technique, say from using cautery to an athermal approach, and then compares results before and after the change. Alternatively, researchers may report the results of surgeons who happen to do an operation one way vs those who use an alternative approach. Randomized trials are not unknown, but they are generally so small as to virtually guarantee nonsignificant results. The first 3 trials I saw at AUA2023 had sample sizes of 120, 40, and 146, and none had significant results. The third trial had such low statistical power, the chance of a significant result was the same for an effective as for an ineffective treatment.

The advantages of randomized trials over observational studies do not warrant repeating here. What does need reemphasizing is that, if observational studies are unreliable, and if the evidence base of much urological surgery comprises observational studies, then the evidence base of much urological surgery is unreliable.

I can’t say that I blame urologists for the current dire state of affairs. The problem is that the sort of randomized trial methodologies that have become common in recent years are no longer appropriate for urological research. Trials have become extremely complex, with dozens of eligibility criteria, multiple end points, and an overwhelming regulatory burden. As such, the cost per patient is very high, $25,000 or more. Moreover, it is increasingly hard to get patients to consent to trials, at least in part because they are so complex. Patient consent forms are now often 20 pages or longer, despite clear data that the longer the consent form, the less patients understand.¹ Patients who don’t understand trials don’t go on them.

These problems are redoubled because we expect only small differences between alternative surgical approaches and therefore require very large trials. If a urological procedure has, say, a 25% adverse event rate, and we think a surgical modification would reduce risk to 20%, the sample size would be 2,000. I doubt that any funding body would be interested in contributing the $50 million required for a trial of, say, fascial sparing vs conventional radical prostatectomy, and it would be hard to find 2,000 patients willing to take part.

At Memorial Sloan Kettering Cancer Center (MSKCC), we take 2 approaches to allow randomized trials in urology and other surgical areas. The first is dramatic trial simplification. Instead of 50, 70, or even 100 eligibility criteria,² we generally have no more than 3 or 4. Indeed in some trials we have only a single criterion based on the uncertainty principle²: if the doctor is uncertain which of 2 procedures is best, then the patient is eligible to be randomized between them. We also reduce the number of research tests, end points, visits, and questionnaires, generally to zero, using instead data taken in routine care. The principle is, “If a researcher wants to know it, a doctor should want to know it.” As a researcher, I need to record whether a particular patient on a trial has recurred. But the patient’s urologist will also want to know about the recurrence in order to consider salvage treatment. We have described this approach as a “clinically integrated randomized trial”³ because the clinical experience of the patient and provider are almost identical, irrespective of whether the patient is on or off study.

Once we have simplified trials, we then take a patient-centric approach to lower barriers to accrual. This is partly achieved directly by trial simplification: we can tell patients that, if they agree to take part, they won’t have to do any additional tests, procedures, appointments, or questionnaires. But we add 2 other tweaks to randomization. The first is to randomize the doctor, not the patient. The doctor will use one approach for a period of, say, 3 months, then switch for a few months, and then switch back. We have found that patients feel very comfortable hearing: “I will evaluate you and decide what is in your best interests. Only if I’m really not sure, I’m truly 50:50 on the best approach, will I do what the computer recommends.” This is in some distinction to “I’m going to flip a coin to decide what treatment to give you.” The second alternative to traditional randomization is called “2-stage consent.” Just as it sounds, this breaks up consent into 2 stages. In the first stage, patients are told that the hospital is conducting research, asked for consent about data collection and then told that, if they agree, their name will be “put in a hat” and they may be randomly selected to hear about a novel treatment approach. Patients who consent are then randomized. Those randomized to the control group are not contacted further; those randomized to the experimental arm then undergo a second consent where they are told about the new treatment and asked if they would like to try it (and if not, due to the intent-to-treat principle, they are analyzed in the experimental arm irrespective of their answer). We have demonstrated that 2-stage consent is highly acceptable to both patients and doctors, maintains excellent understanding of consent and has a low refusal at second-stage consent.⁴ One doctor told us, “I’m never going back to traditional consent, ever.”

Using these methodological approaches, MSKCC has completed 7 single-center randomized trials, with several more underway and accruing rapidly. Only 1 trial was externally funded. Total accrual is approaching 10,000, including several large trials in radical prostatectomy^5,6 and nephrectomy.⁷

There is little doubt that the US clinical trial system is shuddering to a halt, especially for surgical trials. At MSKCC we have shown that simple modifications to the traditional approach are possible and can dramatically improve clinical trials.