Journal Club: Death certificates for diabetes

This week we will investigate how frequently death certificates capture diabetes prevalence:
McEwen et al, Temporal Trends in Recording of Diabetes on Death Certificates
Cheng et al, Sensitivity and Specificity of Death Certificates for Diabetes: As Good as it Gets?

Results (spoiler alert): McEwen et al – Diabetes was recorded on 41% of death certificates and as the underlying cause of death for 13% of decedents with diabetes. Cheng et al – Among 1641 men and 1568 women, 378 decedents had a history of diabetes; 168 of whom had diabetes listed anywhere on their death certificates.

1 Comment

Filed under global health

One response to “Journal Club: Death certificates for diabetes

  1. kjforeman

    Oh man, I have wasted so many hours tracking down discrepancies in diabetes death reporting in the US. So I thought I’d chime in with some more details I didn’t see in the above abstracts.

    There’s of course the issues around the switch from ICD9 to ICD10 in the late 1990s. But that’s to be expected, since all causes of death sort of jump around when there’s a coding shift.

    The more surprising jump is 1989. In 1989, even though the total number of death certificates with diabetes mentioned anywhere remained roughly the same, the proportion of those where diabetes was identified as the underlying cause increased by about 20%:
    https://www.dropbox.com/s/0ewrhnyoxbot0yk/diabetesIcd9.pdf

    At first I thought perhaps ACME (the software used to transcode death certificates, which applies some rules to try to pick an “appropriate” underlying cause) had undergone changes in 1989, but that turns out not to be the case. Rather, it was the standard US death certificate itself that changed.

    An additional line was added to the death certificate in 1989, which resulted in changes to how doctors were placing and ordering diabetes and cardiovascular disease. Instead of putting them on the same line, they were now free to have a separate line for diabetes, oftentimes incorrectly putting diabetes above CVD (which ACME then assigns diabetes as the underlying cause for):
    https://www.dropbox.com/s/zy3blutkmbqjagl/diabetesByLinesFromBottom.pdf

    Lu, Anderson, and Kawachi have more information on incorrect causal sequences here: http://aje.oxfordjournals.org/content/171/10/1069.short