After the attacks

Last week, a newly published article about trends in US heart disease mortality (King et al. 2025) got some attention in social media. The article describes changed mortality patterns during the period 1970–2022: at the beginning of the period, heart disease mortality was dominated by ischemic heart disease (IHD), above all acute myocardial infarction (AMI). During the half-century, there has been a large decrease in overall heart disease mortality, but the decrease has been largest for AMI, so that the mortality pattern now is dominated by other types of heart disease, e.g. chronic IHD, hypertensive heart disease, and functional heart conditions, e.g. heart failure and arrhythmia. It is important to note that we are talking about underlying causes of death, which can be added because exactly one is registered for each death, and not cases where e.g. heart failure is reported as a complication of AMI.

I have written in my earlier Swedish blogs about similar patterns for Sweden, e.g. on 2 December 2011. With my morr R package, cause of death patterns over time can be plotted using the data available via WHO (2025). Unfortunately, these data cannot distinguish between AMI and other kinds of IHD for ICD versions before ICD-9, which the first countries, e.g. US, started using in 1979, and Sweden in 1987. That distinction is not possible even with recent data based on ICD-10 for some countries using the abridged List 1, e.g. Russia. Made with the functions from the morr package, fig. 1 and fig. 2 show sex-specific cause of death patterns for all ages for the US and Sweden.¹

The graphs show that the share of all deaths with heart disease as underlying cause has decreased since the 1980s, both in Sweden and the US, and that this largely has been driven by a large decrease in AMI. Non-AMI heart disease has remained relatively constant, especially in men, with some smaller variations, as with covid-19 entering the scene as a competing cause in 2020. Within that group, there has also been some drift from non-AMI IHD towards other heart disease (e.g. heart failure) and hypertensive disease. It seems rather clear that much of this is due to changed practices in certification, coding and selecting underlying cause, e.g. the shift to ICD-9, which was accompanied by an increase in other heart diseases, both in the US and Sweden. The remarkably low share of deaths due to hypertensive disease in Sweden in the early 1980s points to a strong tendency to attribute cardiac deaths, where hypertension may have been involved, to IHD.

It is important to note that there will soon not be much room left for further lowering the share of all deaths due to heart disease by decreasing AMI mortality, because the share of heart disease deaths due to AMI will be so small that further decrease will not make much difference. And the decreasing share of AMI deaths may also have consequences for the validity of using models based on older data for population-level attribution of shares of IHD, or heart disease, mortality to risk factors like cholesterol, which may largely influence mortality through their effects on conditions closely related to AMI (Menotti and Puddu 2019).

For the overall decrease in the share of deaths due to circulatory disease, as defined in ICD-8 to ICD-10, on which I made a quiz, as I wrote about in my 9 December 2024 post, the decrease in cerebrovascular disease has also been significant. However, I have plotted that group next to vascular dementia and other dementia-related categories (including e.g. Alzheimer disease and unspecified dementia). If one looks at these as a group, the changes have not been that dramatic. And that would probably have been the default way at looking at them, had cerebrovascular disease not been moved from the chapter for neurological disease to circulatory disease in ICD-8. With ICD-11, the category will be moved back, and it may again seem natural to look at age-related neurological disease as a spectrum, from more acute strokes over vascular dementia to mixed, Alzheimer or unspecified dementia, similar to how the remains of the circulatory chapter group AMI with chronic IHD, as well as functional and hypertensive heart disease. One may also note that the group of arterial disease except for IHD and stroke has decreased in parallel with the increase in dementia. This group includes conditions like generalized atherosclerois (ICD-10 I70.9), which may be regarded as garbage codes, when used for underlying causes of death (Naghavi et al. 2010), and may in the past often have been used in patients with dementia without other obvious causes of death.

As noted in my December post, there are still countries with more than 50 percent of deaths due to circulatory disease, and many of these are former Eastern Bloc countries. However, we should not, in that context, regard these as simply present-day versions of Northern European and Anglosphere high-AMI countries during late 20th Century. As I noted in my 19 July 2022 post, their life expectancy may be as high as some Western countries like US in recent years, at least among women. And the pattern within the circulatory categories may also be very different from the older Western pattern. As an example, fig. 3, shows the cause distribution for Lithuania, for the same categories as in fig. 1 and fig. 2.

As can be seen, Lithuania has a low proportion of deaths due to dementia, and a high proportion due to circulatory disease, like the US and Sweden 50 years ago. But there are also striking differences, with the IHD category being larger among women than among men in Lithuania, and AMI only consisting a small fraction of IHD, regardless of time period. This is in line with the findings by Timonin et al. (2021), discussing differences in ascribing IHD deaths to AMI between Norway and Russia (using detailed Russian data not available via WHO (2025), as noted above). It remains to be seen whether future development, e.g. the adoption of ICD-11, will lead to greater convergence between Western and Eastern countries regarding these patterns.

References

---

1. The figures may be reproduced in R by cloning the blog repository and running 2025-07-04-attack.r in the subdirectory postdata/2025-07-04-attack. The causes are listed in the capat vector in that file, using the cause definitions with ICD codes from the morr configuration file. For Sweden, the AMI data for the ICD period (1969–86) have been entered manually from PDF reports available via SCB (2019), and may thus contain transcription errors. To use these with morr (only for all ages), you have to copy the files ami-all-4290.csv and ihdnoami-all-4290.csv from postdata/2025-07-04-attack/data to the inst/extdata directory for morr.↩︎

Webbplatsen drivs av Karl Pettersson och har genererats med Hakyll. Källkod. Om fel uppstår vid besök, kan uppgifter om besökares IP-adress och tid för felet lagras i interna loggar och användas vid felsökning.