Wiener, Renda Soylemez, Lisa M Schwartz, and Steven Woloshin. 2011. “Time trends in pulmonary embolism in the United States: evidence of overdiagnosis..” Archives of internal medicine 171 (9) (May 9): 831–837. PMID 21555660
Found this via a chap who commented on Weingart’s now famous PE episode.
I think this paper raises a lot of good points surrounding the “epidemic” of overdiagnosis of PE. The VTE literature is vast and complex yet some of the most basic questions lack substantive evidence.
The evolution of our approach to VTE has dramatically altered what a diagnosis of VTE actually means.
This paper is one of the first I’ve seen that tries to examine whether or not it’s fair to say that we are overdiagnosing PE. It is of course indirect evidence. Direct evidence would be a comparative trial of treatment vs no treatment.
- they used the Nationwide Inpatient Sample to get the data. This is a huge dataset that is all ICD-9 codes from hospital discharges. ICD-9 codes have their problems but in this question they’re less relevant
- they compared the number of diagnoses of PE with the recorded mortality from PE. If all the PEs being diagnosed were of equal lethality then finding more of them and treating them should result in a lower mortality rate
- below is a graph of 3 potential outcomes
- the numbers of patients involved are in the hundred of thousands
- I think the little graph tells the story pretty well:
- there was a big increase in the number of PEs but no change in the mortality rate implying that we were just finding lots of PEs that don’t seem to kill people
- they tease this out with a bit more stats in the paper if you’re interested