Editorial Comment for George et al.

“If you torture the data long enough, it will confess.”

— Ronald H. Coase

This is a commonly repeated quote, usually altered to some extent. I have always preferred the adaptation “If you torture your data long enough, it will tell you what you want.” This article by George and associates is a good case in point.

While there is no doubt that contemporary surgeons are—out of unfortunate necessity—becoming more facile with surgery in obese patients, this article provides an overly optimistic picture of the ease and safety of laparoscopic partial nephrectomy (LPN) in the obese compared to nonobese patient. Part of this derives from the authors' choice to use the WHO body mass index (BMI) groupings of <25 (normal weight), 25 to 30 (overweight), and >30 (obese). I understand that they did this to align their findings with the numerous publications using this same categorization, but for the surgeon, these groupings are not clinically useful. I think that most of us would agree that operating on someone with a BMI of 29.9 is not much different than operating on someone with a BMI of 24.9. This means that the distinction between the normal weight and overweight groups is not useful to the surgeon. As such, at the very least the authors should have looked at BMI groups of <30 vs >30.

Moreover (and this is where the quote above comes in), statistical analysis of three groups can obfuscate differences that would be apparent if only two groups are used, owing to nonregular trends in the data. For example, in table 5, the p-value for the three-group comparison of overall complications was not significant (p-value for the chi-square test is 0.11 by my calculation): 11/74 (15%) for BMI <25, 26/180 (14%) for BMI 25 to 30, and 42/181 (23%) for BMI >30. Collapsing the two lower BMI groups, however, reveals the statistically (and clinically) significant difference. Of 254 patients with BMI <30, 37 (15%) had a complication, compared to 42 of 181 patients (23%) with BMI >30. The p-value for this comparison is 0.0235 by Fisher's exact test. To be fair, however, most of the differences are driven by postoperative hematuria/bleeding requiring angioembolization (4 of 254 patients [1.6%] with BMI <30 compared to 11 of 181 patients [6.1%] with BMI >30, Fisher's exact test p-value=0.0151), which the authors did identify as being associated with BMI (p=0.033). It seems to me that a 6.1% rate of hemorrhage requiring angioembolization is a very important finding that merits strong consideration!

I certainly agree that LPN should be offered to obese patients, even to the super morbidly obese. LPN, when technically possible, is preferred to open surgery especially since open surgery is probably more impacted by obesity than is laparoscopic surgery, for both the patient and the surgeon. At our institution, like that of the authors, we offer LPN based on tumor characteristics rather than patient body habitus. However, the obese patient should be counseled that their body habitus increases the risk of complications. The authors conclude “LPN demonstrates equivalent perioperative outcomes in normal, overweight, and obese patients.” In fact, their data in this article—when tortured appropriately—tell us otherwise.