Respect and supreme condolences to those who lost their lives, especially during the early days. We primary care docs seem to have died at higher rates than others - don’t forget that proper PPE was not available, and often discouraged by administration beyond surgical masks... if you could even find them. Leadership threatened me when I refused to wear anything less than an N95 in my 100 sq ft examining rooms, knowing full well that this shit was airborne from the start. Many nurses and others were fired when they defiantly wore respirators to protect themselves above the egregious lowest common denominator.
I assume this trend held through the first year or two. Asymptomatic spread has always been the most fearsome obstacle, and an under protected primary care work force was a sitting duck.
“A new study, led by researchers in the Perelman School of Medicine at the University of Pennsylvania, assessed 1,004 reported cases of COVID-19 deaths among healthcare workers globally as of May 13, 2020, and found that more than half of these reported deaths were among physicians. Notably, however, family physicians appeared to be affected much more often than “frontline,” hospital-based physicians. This report was published in the Journal of the American Board of Family Medicine.”
Here's my skeptic (with a lower case "s" [1]) take on this paper:
1) Their counterfactual model should have been them modeling 2012-2016, then predicting deaths for March 2017 through December 2018. Could then evaluate the model's accuracy of the predicted deaths. That's how I would have tested a model some consultant brings to me, see how it compares to historical records.
2) The confidence intervals are very high (-3 to 25; -8 to 56; 27 to 136, etc) would like to see more discussion on this.
3) The inclusion of non-active, elderly physicians, while important, skews the data heavily. Despite comprising only 22% of the total physician population studied here, they accounted for 58% of excess deaths.
4) The exclusion of young physicians (under 45, hooray I am still "young") because there were not enough deaths to track *could* lean towards the hypothesis that Covid 19 isn't especially dangerous to young people, as even those constantly encountering Covid didn't have an excess death signal.
5) Why is there no excess mortality after April 2021 but there continued to be excess deaths in the vaccinated general population? I don't feel the authors properly thought this out, as there are enough counterfactuals to falsify their hypothesis that this was due to the vaccines.
6) Would like to see them investigate cause of death. Here, they appear to attribute the excess deaths entirely to Covid, yet during this study period there was an increase of 20% for accidental cause of deaths in the general population. Granted accidents only represent ~15% of all-cause mortality in this age group, but physicians have a higher substance abuse rate than the general population, and while the authors attribute the correlation of deaths to the fact it was pre-vaccine era, it also correlates to the most stressful period of the pandemic.
7) Why did they predict the active physicians would have half the mortality rate of non-active (1.2% vs 2.3%) in the 75-85 age cohort? - This will come up below in my thoughts on the discussion.
8) Really wish they broke this out by jurisdiction. Would be interesting to know if physicians in NYC experienced a 700% increase in deaths in NYC April of 2020. Is this data anywhere?
"Across age groups, physicians had substantially lower excess mortality than the general population; however, active physicians had lower excess mortality than nonactive physicians despite their higher risk of contracting SARS-CoV-2 infection. The findings suggest that personal protective equipment use, vaccine requirements, infection prevention protocols, adequate staffing, and other workplace-based protective measures were effective in preventing excess mortality."
Several points in this do not make sense based on their data or seem easily falsifiable:
1) The claim "active physicians had lower excess mortality than nonactive physicians" is only true in absolute terms, but this is an instance where relative increases provide more insight. Active physicians providing direct care saw an 18.8% increase in all-cause mortality compared to non-active +17%.
Either way this could be an example of HARKING, hypothesizing after the results are known, because you could still frame this to support the same hypothesis by reframing the expected impact of NPI (i.e., "Look there was only an absolute increase in the relative increase between cohorts of 1% and it should be higher" - that may be true)
2) How can they credit PPE, infection prevention and other workplace based protective measures on lower excess mortality rates when there isn't a noteworthy difference between active and nonactive physicians (I don't find a 1% absolute difference on the relative increase compelling)?
3) How beneficial is it to compare physicians to the general population when physicians lack the #1 and #2 confounders for Covid mortality outside of age, specifically physicians have much lower BMI than the general population and are in the 1% to 2% income earners. Obesity and income (especially medical access) are highly correlated to Covid outcome so I find it strange the authors wouldn't account for those differences and instead credit NPI.
4) Want to stress the comparison of Active to Inactive physicians ignores that 92% of inactive physicians were over the age of 65 compared to 28% of active physicians.
5) Without Covid they predicted 2.3% of the over 75 population would die in the inactive cohort compared to 1.2% in the active over 75 group. This indicates confounders present in the inactive population not addressed in the study. These are pretty obvious confounders too.
What would be fascinating, was if there was a difference in mortality between the types of "active" physicians. In the notes it appears an ER physician on the front lines, an Orthopedic Surgeon who likely doesn't encounter many Covid patients directly (you aren't going to get your knee replaced if you have an active Covid infection), and a Psychiatrist who likely worked remote via Zoom all count as "active", yet they clearly have different exposure risks. I assume they could have further segmented the population to answer more questions.
Thanks for sharing! Not being cynical, I read everything through the lens of skepticism.
Great article ! Now get in touch with Bill Maher who said otherwise on last weeks show . Makes me irate .
JAMA link is broken. Link below is working right now:
https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2800889
It was last Friday so February 3rd
I’ll look for it .
He commented on the Covid deaths with his panel .
Respect and supreme condolences to those who lost their lives, especially during the early days. We primary care docs seem to have died at higher rates than others - don’t forget that proper PPE was not available, and often discouraged by administration beyond surgical masks... if you could even find them. Leadership threatened me when I refused to wear anything less than an N95 in my 100 sq ft examining rooms, knowing full well that this shit was airborne from the start. Many nurses and others were fired when they defiantly wore respirators to protect themselves above the egregious lowest common denominator.
Here’s the early study from UPENN
https://www.pennmedicine.org/news/news-releases/2020/july/among-healthcare-workers-family-primary-care-doctors-most-at-risk-of-dying-from-covid19
I assume this trend held through the first year or two. Asymptomatic spread has always been the most fearsome obstacle, and an under protected primary care work force was a sitting duck.
“A new study, led by researchers in the Perelman School of Medicine at the University of Pennsylvania, assessed 1,004 reported cases of COVID-19 deaths among healthcare workers globally as of May 13, 2020, and found that more than half of these reported deaths were among physicians. Notably, however, family physicians appeared to be affected much more often than “frontline,” hospital-based physicians. This report was published in the Journal of the American Board of Family Medicine.”
This is just tragic.
Here's my skeptic (with a lower case "s" [1]) take on this paper:
1) Their counterfactual model should have been them modeling 2012-2016, then predicting deaths for March 2017 through December 2018. Could then evaluate the model's accuracy of the predicted deaths. That's how I would have tested a model some consultant brings to me, see how it compares to historical records.
2) The confidence intervals are very high (-3 to 25; -8 to 56; 27 to 136, etc) would like to see more discussion on this.
3) The inclusion of non-active, elderly physicians, while important, skews the data heavily. Despite comprising only 22% of the total physician population studied here, they accounted for 58% of excess deaths.
4) The exclusion of young physicians (under 45, hooray I am still "young") because there were not enough deaths to track *could* lean towards the hypothesis that Covid 19 isn't especially dangerous to young people, as even those constantly encountering Covid didn't have an excess death signal.
5) Why is there no excess mortality after April 2021 but there continued to be excess deaths in the vaccinated general population? I don't feel the authors properly thought this out, as there are enough counterfactuals to falsify their hypothesis that this was due to the vaccines.
6) Would like to see them investigate cause of death. Here, they appear to attribute the excess deaths entirely to Covid, yet during this study period there was an increase of 20% for accidental cause of deaths in the general population. Granted accidents only represent ~15% of all-cause mortality in this age group, but physicians have a higher substance abuse rate than the general population, and while the authors attribute the correlation of deaths to the fact it was pre-vaccine era, it also correlates to the most stressful period of the pandemic.
7) Why did they predict the active physicians would have half the mortality rate of non-active (1.2% vs 2.3%) in the 75-85 age cohort? - This will come up below in my thoughts on the discussion.
8) Really wish they broke this out by jurisdiction. Would be interesting to know if physicians in NYC experienced a 700% increase in deaths in NYC April of 2020. Is this data anywhere?
9) “Reproducible code is available at: http://github.com/mkiang/excess_physician_mortality. We note that due to our data use agreement, we cannot share data.”
Bummer
Thoughts on discussion. From the article:
"Across age groups, physicians had substantially lower excess mortality than the general population; however, active physicians had lower excess mortality than nonactive physicians despite their higher risk of contracting SARS-CoV-2 infection. The findings suggest that personal protective equipment use, vaccine requirements, infection prevention protocols, adequate staffing, and other workplace-based protective measures were effective in preventing excess mortality."
Several points in this do not make sense based on their data or seem easily falsifiable:
1) The claim "active physicians had lower excess mortality than nonactive physicians" is only true in absolute terms, but this is an instance where relative increases provide more insight. Active physicians providing direct care saw an 18.8% increase in all-cause mortality compared to non-active +17%.
Either way this could be an example of HARKING, hypothesizing after the results are known, because you could still frame this to support the same hypothesis by reframing the expected impact of NPI (i.e., "Look there was only an absolute increase in the relative increase between cohorts of 1% and it should be higher" - that may be true)
2) How can they credit PPE, infection prevention and other workplace based protective measures on lower excess mortality rates when there isn't a noteworthy difference between active and nonactive physicians (I don't find a 1% absolute difference on the relative increase compelling)?
3) How beneficial is it to compare physicians to the general population when physicians lack the #1 and #2 confounders for Covid mortality outside of age, specifically physicians have much lower BMI than the general population and are in the 1% to 2% income earners. Obesity and income (especially medical access) are highly correlated to Covid outcome so I find it strange the authors wouldn't account for those differences and instead credit NPI.
4) Want to stress the comparison of Active to Inactive physicians ignores that 92% of inactive physicians were over the age of 65 compared to 28% of active physicians.
5) Without Covid they predicted 2.3% of the over 75 population would die in the inactive cohort compared to 1.2% in the active over 75 group. This indicates confounders present in the inactive population not addressed in the study. These are pretty obvious confounders too.
What would be fascinating, was if there was a difference in mortality between the types of "active" physicians. In the notes it appears an ER physician on the front lines, an Orthopedic Surgeon who likely doesn't encounter many Covid patients directly (you aren't going to get your knee replaced if you have an active Covid infection), and a Psychiatrist who likely worked remote via Zoom all count as "active", yet they clearly have different exposure risks. I assume they could have further segmented the population to answer more questions.
Thanks for sharing! Not being cynical, I read everything through the lens of skepticism.
______________
[1] https://blogs.scientificamerican.com/cross-check/dear-skeptics-bash-homeopathy-and-bigfoot-less-mammograms-and-war-more/