I must admit that my initial reaction to the now famous study by Ashish Jha and colleagues—showing that female internists achieve slightly better 30-day inpatient mortality rates than male internists—was one of annoyance. “Here we go again,” I thought. “Data mining at the service of political correctness.” And I was pleased to read David Shaywitz reply to the study with a piece in Forbes aptly titled “When Science Confirms Your Cherished Beliefs—Worry.”
That said, I must give credit to the study authors for generating a lot of interesting discussion and for stimulating Saurabh Jha to write his magnificent commentary “Homme Fatale.”
Now, Saurabh makes a lot of excellent points, and many are erudite and highly amusing. But I must disagree with his main thesis that the study is essentially pointless.
Saurabh argues correctly that finding that male physician mortality rates are 0.4% worse than female is bound to be misinterpreted as a “lazy generalization.” Differences in the mean value of bell-shaped curves obfuscate the complex reality of the individual data points that compose those curves.
I could not agree more with him on that. We should not be quick to conclude, as much of the mainstream press has, that the study is “Evidence of the Superiority of Female Doctors.” It is not the case, as Saurabh points out that any given female doctor, if given 100 patients to treat, will see 89.8% survive, whereas anyone of her male counterparts will only muster an 89.4% survival rate. In that sense, gender is not a drug, and being male is not necessarily a weaker sex, medically-speaking.
Nonetheless, we shouldn’t dismiss the study result offhand as insignificantly informative, simply because it represents how women internists fare on average compared to men. Besides, the effect was present in the same direction (i.e., mortality rates were better for female internists than males) across a number of disease states and subgroups of patients, and this is indicative that something is going on.
Furthermore, the significance of a finding is not strictly correlated to its magnitude. That the effect of gender was small does not mean it is not revealing of some deeper truth. (Remember that rust is only slightly heavier than iron. If scientists had ignored that fact, we would still be talking of phlogiston instead of oxygen).
What, then, can we infer from the study finding that can elevate it beyond the level of a mere curiosity?
Here I must simply follow the main clue that the study authors give. That clue, however, may lead to a very unpopular conclusion, and perhaps that is why the authors have been so careful not to emphasize it.
We should first recall that the data was mined from the Medicare inpatient databases. The practice setting, therefore, is one that is tightly regulated and subject to numerous government directives, “quality measures,” and pay-for-performance incentives.
In the introduction section of the paper, the authors provide several lines of prior evidence that male and female physicians practice differently. Of those, only one pertain to inpatient practice: females, the authors tell us, follow clinical guidelines more closely than males.
Now, it is possible that the study indicates that adherence to clinical guidelines provides better outcomes (that was not the point of the paper, and cannot be concluded from the data). But the authors fail to note that, for a number of years, the medical literature has referred to guideline-based therapy as “cookbook medicine.”
Could it be, then, that the appropriate interpretation of the study by Jha et al. should be that women are better cooks than men? If it is so, I leave it to David Shaywitz to tell us whether that would put it in the category of “science confirming of cherished beliefs.”