CCBYNC Open access
Research

Comparison of postoperative outcomes among patients treated by male and female surgeons: a population based matched cohort study

BMJ 2017; 359 doi: https://doi.org/10.1136/bmj.j4366 (Published 10 October 2017) Cite this as: BMJ 2017;359:j4366

Re: Comparison of postoperative outcomes among patients treated by male and female surgeons: a population based matched cohort study

I would like to preface with the statement that I do not, in any way, want to undermine the results of this study or the important societal dialogue that is sure to follow. Thank you to the authors for your interesting analysis on an interesting topic, with very thought-provoking results. I believe that their ultimate conclusion may very well still be valid and certainly warrants further exploration.

I am, however, very curious about their analysis with respect to year of surgery. This seems the be the most significant potential confounder that was unfortunately not included in their matching process.

The authors reports that there was no significant difference between the two groups in year of surgery, presumably on the basis of a standardized difference of 0.05. This seems like an inappropriate use of this statistical test resulting in an inaccurate conclusion. Simply reviewing Table 5 visually (or better yet, graphing the distribution of the two groups by year) makes it readily apparent that the variables are not normally distributed (as assumed in the standardized difference calculation), but rather inversely linearly related. The male surgeons are disproportionately represented in the early years and the female surgeons in the later years.

The authors also report that the year of surgery was not associated with the primary outcome, referring to Supplementary Table 5. While the second column would support this assertion, there are glaring discrepancies in columns 3-6. The absence of difference in the composite outcome masks very compelling decreases in death and length of stay and increasing complication rate with increasing year of surgery. While the claim that year of surgery was not statistically associated with the primary outcome may still hold true, I would suggest that matching by year of surgery is absolutely necessary (at least on an approximate basis) as there is good evidence that procedures in 2007 and 2015 are not readily comparable.

Competing interests: No competing interests

11 October 2017
Michael S Jansz
Resident
UBC
Vancouver, BC