CHALLENGES IN SURGICAL INTERVENTION

” The randomized clinical trial is the epidemiologic centerpiece of clinical epidemiology”

- Rothman KJ, epidemiologist, 1945-

Or, depending on how the inclusion runs; “An experiment which any damn fool can design and frequently does” – Anonymous.

This section is dedicated to the many methodological and practical challenges that arose whilst conducting the randomized clinical trial described in Appendix A.

The strengths with a trial that consists of a prospective, multi-center, large sample size design with several surgeons involved and a random allocation with double-blinding will undoubtably be major and the bias will be minimized. I will leave out these recognizable epidemiological strengths, and instead I would like to emphasize some more “hidden” methodological obstacles that has appeared to me during the conduction of the SUMMER trial, that I could not fully expand in the study protocol.

Surgical intervention studies traditionally follow several years of surgical clinical experience. This is very different from, for example, drug trials. It may be considered strange and unscientific to practice a surgical method without having the underlying proven evidence for it. Like for example in paper IV. But there may also be a valuable point to it.

A respected professor of surgery (urology) once asked us younger unexperienced researchers,

“Can we separate the handicraft from the craftsman?”

As a surgeon with a surgical intervention involving a surgical procedure to carry out, he got my attention. Indeed, surgery is a complex intervention. I believe that observational non-randomized studies can have a struggle to cope with all the hidden confounders that can led to bias.

However, there is a challenge in evaluating innovation in surgery even by randomized controlled trials.

The purpose of the SUMMER trial is to assess if the onlay mesh is better than just suturing the defect, in respect to recurrence. To separate the therapeutic effect of the mesh from the surgeons’ skills can be difficult. But there are some ways to get around this. Stratification of centers (a sort of indirect control for surgeons’ or a surgical unit’s skills) and involving several surgeons of different seniority are adopted in the SUMMER trial. Additionally, this RCT is being conducted at a perfect time, in my opinion. It is not done in an early stage of the development of its methods. Surgeons has practiced the onlay-mesh method repair for several years and the comparator is well established. Consequently, the learning curve of the technique should not be a concerning bias in the trial. Also importantly, it is still not late in time for the trial. A preserved balance of interest to be treated by both methods between the groups is still considered to be high. The latter of course greatly facilitates the surgeon’s interest in including patients in the trial and the study participants interest in participating.

Trial recruitment is another obstacle. Patients should theoretically be unlikely to be reluctant to participate in the SUMMER trial, since the trial involves surgical procedures that are fairly

similar. In contrast, more substantially different surgical procedures can be more problematic for the recruitment. Again, the similarity in surgical procedures within the allocations also reduces the imbalance of varied surgical expertise and minimizes bias.

“Surgical randomization” by itself can be another difficulty. A surgeon’s uncertainty of what treatment the patient will receive and its subsequent effect is usually not welcomed in the

operating room. Undeniably, as surgeons we want to be in control of what is best for the patient, after we have controlled for the tissue and the anatomical conditions of the procedure. A

randomization to an allocation prior to surgery can therefore be an advantage in these cases. The disadvantage is that it can lead to an increased risk of excluding trial participants during the procedure, due to surgical issues. The gold standard analysis of an intention to treat protocol can therefore be disturbed. The SUMMER trial has a randomization close to time to intervention with an assigned allocation in the operating room following the control of the hernia defect. A crossover of procedure after randomization is completely excluded and this maneuver will favor the principle of preserving the intention to treat analysis. Although, one cannot still avoid thinking that there may be a risk that surgeons decide to exclude the advanced cases from the trial before randomization, as they think that a mesh should be applied in those cases. For example; an umbilical hernia with a very thin aponeurosis with a 2 cm defect. This can lead to a reduction in the chance of detecting differences in recurrence rates between the groups.

Standardization is commonly considered a positive aspect and is the cornerstone of randomized controlled trials. However, a surgical procedure as part of an intervention has several interacting components that can challenge units’ ability to differentiate and standardize aspects such as procedures before surgery, during surgery and the staff around the trial participant after surgery.

I believe we have managed this in a balanced way for the SUMMER Trial. However, we have been very strict with general anesthesia and usage of the same mesh for all the trial subjects in the study protocol. The downside will be that some potential trial participants may not be included in the trial.

Blinding of the outcome investigator and the trial participants are crucial in the SUMMER trial to reduce bias, but can be a barrier for recruitment of patients. The uncontroversial harm-to-benefit profile of the surgical procedures in the trial can compensate for this. However, the benign small umbilical hernia repairs contribute to patients not being prone to return for the long-term follow-ups. It will be a challenge to achieve compliance for this, both from the trial subjects and

from the involved surgeons to track the patients. The trial design could have been modified better for the follow-up controls in order to reduce drop-outs.