On my way to my 85-year-old dad who suffered from paralytic ileus after laparotomic hemicolectomy, I heard the following news: “Patients from female surgeons have less long-term postoperative complications (including death)“ (see: Wallis et al., JAMA Surg.). Is that what the study really revealed?

This was a population-based retrospective cohort study conducted in adults undergoing one of 25 common elective or emergent surgeries in Canada between 2007 and 2019. An adverse postoperative event was defined as the composite of death, readmission, or complication assessed at 90 days and one year following surgery, among patients operated by either female or male surgeons.

Among 1,165,711 included patients, 13% were treated by a female surgeon. Adverse postoperative outcomes occurred in 14.3% at 90 days and in 25.0% at one year following surgery.

Multivariable-adjusted rates of the composite end point were higher among patients treated by male than female surgeons at both 90 days (13.9% vs 12.5%; adjusted odds ratio [AOR], 1.08; 95%CI, 1.03-1.13) and one year (25.0% vs 20.7%; AOR, 1.06; 95%CI, 1.01-1.12). Similar patterns were observed for mortality at 90 days (0.8%vs 0.5%; AOR 1.25; 95%CI, 1.12-1.39) and one year (2.4%vs 1.6%; AOR, 1.24; 95%CI, 1.13-1.36).

Now what I don’t understand: Female and male surgeons differed considerably with respect to age, years in practice, and annual number of procedures done (all higher for males). Moreover, male surgeons did – proportionally – more cardiothoracic-, neuro-, orthopedic, urological, and vascular surgery, and female surgeons did more general, obs & gynaecological and plastic surgery. And, patients operated by female surgeons were 8 years younger and were more frequently also female (80% versus 59% for male surgeons). So, comparing male to female surgeons is as comparing apples to pears. Apart from a measure of comorbidity, no other patient information was available.

With all these differences between male and female surgeons and their patients, I would have expected that statistical adjustment would considerably change crude estimates. But that is not the case. Crude estimates (in supplement) were nearly identical to adjusted estimates: incidences for composite endpoint 14.4% vs 12.6% at day 90 and 25.8% vs 19.6% at 1 year. This is not mentioned or addressed in the discussion. What are your thoughts?

My dad recovered quickly. His surgeon was female.