“Comparison of Hospital Mortality and Readmission Rates for Medicare Patients Treated by Male vs Female Physicians.”
Tsugawa Y, Jena AB, Figueroa JF, et al.
JAMA Intern Med. 2017 Feb 1;177(2):206-213 [Full text]
When this work was published in 2017 it became something of a sensation. Altimetric listed it as the #3 article of 2017 and it has been referenced in >300 news stories. But the work did not occur in a vacuum and was based on previous work noting differences in practice by gender. The authors note:
“Literature has shown that female physicians may be more likely to adhere to clinical guidelines,1–3 provide preventive care more often,4–11 use more patient-centered communication,12–15 perform as well or better on standardized examinations,16 and provide more psychosocial counseling to their patients than do their male peers.14“
It was within this context that the authors attempted to quantify the outcome difference by gender of the treating physician. While we will see this required a number of assumptions to design such a study, the authors were deliberate in all attempting to control for numerous factors and reported the “same” results in multiple ways (by varying the factors adjusted for). In the end the results were impactful not just for the results observed but that the implications speak to our entire health system.
This was an observational cross-sectional study that included a random sample (20%) of Medicare beneficiaries aged 65 years or older hospitalized with acute medical conditions between January 1, 2011 and December 31, 2014. The entire hospitalization was assigned to a single internist based on the most Medicare Part B spending during the hospitalization. If a patient was transferred between hospitals this was merged and considered a single hospitalization. If the patient left against medical advice or if it was an elective hospitalization, these were excluded.
The main outcomes of interest were 30 day mortality and 30 day readmission. If a patient was admitted at the end of the review period (12/2014) they were excluded from 30 d mortality analysis. Similarly if discharged at the end of the review period (12/2014) they were excluded from 30 d readmit analysis.
The same outcomes were looked at in a subset of common conditions (sepsis, pneumonia, CHF, COPD, UTI, acute renal failure, arrhythmia, and GIB).
Interestingly, there were overall differences in gender in distribution of type and location of hospital employed. Females were more likely to work in
- large (7460 [41.9%] vs 13 628 [35.7%])
- nonprofit (13,947 [78.2%] vs 28,850 [75.6%])
- major teaching hospitals (5168 [29.0%] vs 8061 [21.1%])
- in the Northeast region (4746 [26.8%] vs 8574 [22.7%])
Additionally, though the patient characteristics were similar, female physicians more often were the identified physician for female patients compared to males (258,091 [62.1%] vs 722,038 [60.2%]).
In regards to the main outcomes (30 d mortality, 30 d readmission), the results were adjusted for contributing factors in three ways: Based on 1) Patient characteristics 2) Hospital Indicators (comparing physicians w/in same hospital 3) Physician specific characteristics (physician age, medical school attended, allopathic vs osteopathic training). Results were reported for the three adjustments and overall results reflect adjustments for ALL of these variables together.
The 30 day mortality rate of patients cared for by female physicians was lower, 11.07% vs. 11.49% (adjusted risk difference, –0.43%; 95% CI, –0.57% to –0.28%; P<0.001), with a NNT 233.
The 30 day readmission rate was 15.02% vs. 15.57% (adjusted risk difference, –0.55%; 95% confidence interval, –0.71% to 0.39%; P<0.001) with a NNT 182.
The subset of common acute conditions had many with a significant 30d mortality difference or 30d readmission difference. Two conditions, sepsis and pneumonia were significant for BOTH mortality and readmission.
Finally, the patients were divided by illness severity and the gender specific differences persisted among all groups except for the 2nd least sick (30d mortality) or least sick (30d readmission, see Figure).
At first glance the NsNT of 233 and 182 for 30d mortality and 30d readmission, respectively, may seem modest. This is likely through our lens of a typical “journal club” study with hundreds to thousands of patients, expecting an intervention to have a certain impact to act upon (i.e. NNT in the 10s not 100s). However, this is a national sample and just among the 1.5 million evaluated, the impact would reflect 6000+ hospitalizations with a potential mortality impact and 8000+ readmissions potentially prevented. The authors extrapolate further:
“…given that there are more than 10 million Medicare hospitalizations due to medical conditions in the United States annually and assuming that the association between sex and mortality is causal, we estimate that approximately 32,000 fewer patients would die if male physicians could achieve the same outcomes as female physicians every year.”
While there are certainly issues to consider, such as the setting or system in which the female physician works (recall there were gender differences here), the authors attempted to adjust for these. Furthermore, we know the care for a patient no longer truly falls to one internist or physician. It is a team, multidisciplinary effort. Some have posited that perhaps this outcome reflects team leadership. Unfortunately, the reasons for the difference cannot be proven with this study but are speculated based on the existing literature on gender differences. These would be clear starting points with which systems can intervene to realize practice improvements and reward physicians fulfilling these metrics, with females perhaps having a head start.