August 1, 2022

expert reaction to study looking at certain medications and risk of non-fatal heat-related heart attacks

A study published in Nature Cardiovascular Research at non-fatal heart attacks triggered by heat exposure and medication intake.

Prof Stephen Evans, Professor of Pharmacoepidemiology, London School of Hygiene & Tropical Medicine, said:

“This is an interesting research letter.  There have been many studies of heat-related effects on mortality and heart attack rates and it is well-known that extreme outside temperatures affect these rates (both hot and cold).

“The findings that particular drugs were associated with higher heat-related risks is of interest, but the authors are right to be cautious in the interpretation of their findings.  They note that despite the careful design, it is possible that users of anti-platelets and beta-blockers, both types of drugs given to people who have had or are at increased risk of a heart attack (MI) are the most affected by heat.  While the association was not seen with other drugs there is uncertainty in the results, and it is possible that the users of those drugs were more vulnerable to heat-related effects rather than that the drugs themselves causing the effect.

“In addition to this note of caution there are some other limitations not noted by the authors.  They say, “Information on medication intake was not available for most patients who experienced a fatal MI”, and consequently they only study non-fatal heart attacks.  This leads to notable uncertainty, since it is possible that those who suffer a heart attack and die within 28 days are protected by these drugs and so the survivors are disproportionally represented in those that they study.  This might not be so and the only way to know is to study all heart attacks, both fatal and non-fatal.

“The authors analyse the data, quite reasonably, using odds ratios (a relative measure) but for public health importance the absolute effects are more important.  Absolute rates of heart attacks increase notably with age and the effects in younger people are based on very small numbers.  The data on drugs was obtained by questioning the patients which can have some uncertainty in its validity.  The fairly large (25%) proportion of the non-fatal heart attacks who had missing data on medication also increases the uncertainty.

“This is an interesting hypothesis-generating study, and merits further studies of all heart attacks with ascertainment of medication through prescription data, possibly supplemented by questionnaires.

“It is MOST important that patients do not stop or change their medicines taken for prevention of heart attacks based on this study.  The risks associated with stopping drugs may be very much greater than continuing them, even in a heatwave.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This study does certainly provide some evidence that people who take certain types of cardiovascular medication may be more likely to have a non-fatal heart attack in warm weather (compared to more average weather), than is the case for people who do not take those medications.  Like any finding based on statistical analysis and modelling, there’s some uncertainty about that evidence.  But, much more importantly, that’s a correlation, and it doesn’t show that the difference in risk is actually caused by the medication.  It might be, or it might not be.  The researchers (and the press release) make this explicitly clear – that’s why they call for more research.  (Yes, researchers do very often call for more research – but in this case I don’t think anyone can clearly say what should be done about the findings unless there has been more research first.)

“There’s been good evidence from previous research, including research by the same research team using the same database and the same statistical methods [see reference number 1 in the research paper], that heart attacks are more likely in weather that is colder than average or hotter than average.  Even in those findings, though, there remains a bit of doubt as to exactly how the increased risk arises.  It could well be caused directly by the effects of hot (or cold) temperatures on the body, but as well as that, people do behave differently in more extreme weather conditions than when the temperature is average, and perhaps some aspects of what they do differently is part of the cause of the increased heart attack risk.

“This new research is looking at a more subtle question – whether the correlation between hot weather and heart attack risk is different in people who take, or don’t take, various medications that are prescribed for cardiovascular disease (heart attacks, strokes, etc.).  The researchers report that they did find a difference in risk for people who take anti-platelet medications or beta-receptor blockers, with greater increase in heart attack risk in hot weather for those who take them than for those who don’t.

“The big problem here is that these medications aren’t prescribed for no reason.  People who take them are likely to be, on average at least, in a different state of health than are people who don’t take them – perhaps those who take them are, on average, sicker in some way than those who don’t, or just have a different state of health even if you wouldn’t call them sicker.  And maybe that’s why high temperatures have a bigger effect in increasing their heart attack risk.  In other words, the risk difference might be caused by their underlying health, and not by the effects of the medication at all.  (In the jargon, this is called ‘confounding by indication’ – ‘confounding’ because some factor other than the one of main interest may be causing the observed results, and ‘indication’ meaning the reason why the person has been prescribed the medication.)

“The researchers are perfectly open that this is a possibility, and it’s mentioned in the press release too.  In the research paper, the researchers give several reasons why the effect they found might not all arise in this way, but also several reason why it might arise this way.  You just can’t tell the position from these findings – and that’s why it makes good sense for the researchers to ask for more research in different large data sources to disentangle what’s going on.  Using different data sources could also help with the issue that these finding are based only on data from a particular area of Germany – things might well be different elsewhere, where prescribing decisions, people’s own choices about what to do in different weather conditions, and other aspects may be different from in the Augsburg area.

“Note that the main comparisons of risk in the paper are between temperatures of 7.5 degrees (when heart attack risk was lowest) and 24.2 degrees.  A temperature of 24 degrees does not sound very hot, compared to daily maxima of over 30 degrees, or even 40 degrees, that we’ve seen recently.  But the temperatures in the research paper are daily averages, not maximum temperatures, and a daily average of 24 degrees probably means it was pretty warm at night as well as during the day.

“I’d usually try to give some figures for the absolute risks of heart attacks from a study looking at risk changes.  The research gives only relative risk measures – that is, how different the increase in heart attack risk in warm weather is for people who take the medications, in comparison to people who don’t take them.  But I can’t give absolute figures because of the way the research was done.  The researchers used what’s called a case-crossover research design, which is an ingenious method that (in this research) takes all the people who had a non-fatal heart attack, directly compares the weather when they had the heart attack with the weather on other days when they didn’t have a heart attack (in this case, on the same days of the week in the same month), and then puts together those individual comparisons to give an overall measure.  This means that the findings are based on people being compared with themselves, so avoiding having to deal with some differences between individuals who had or didn’t have heart attacks.  But it also means that there aren’t direct absolute measures of heart attack risk, that I could report.

“Finally, it’s worth asking what might be done about these findings, if further research does confirm the risk differences and indicate that they aren’t just caused by underlying disease.  All the researchers say is that this could “help clinicians, patients and public health officials develop targeted strategies to reduce the burden on cardiovascular disease under climate change.”  The concern about climate change is, of course, because hotter weather might well lead to more heart attacks, and if something can be done about prescribing patterns (for instance) to reduce this, that would be good.  But there can’t be any suggestion yet about what these targeted strategies might be, in detail.

“I’m no clinician, but I’d urge anyone who may be taking these medications and is concerned about the findings to talk to their doctor or other health care provider before changing what they do in terms of taking the medication.”

Further information

A more detailed issue, perhaps, arises in comparing these new findings with the findings from the same group [in reference 1 in their paper] indicating that heart attack risk is increased, on average, in weather than is hotter or colder than average. In their previous work, they used data from all patients who had a heart attack, including fatal heart attacks, and the statistical methods (case-crossover research) were similar to those used on the new research.  But the previous research found that the minimum risk of heart attacks was at a daily average temperature of 18.4 degrees, a lot warmer than the minimum risk temperature of 7.5 degrees in the new research.  I’m sure that difference is because the earlier research based the minimum risk temperature on a much larger number of heart attack cases, with different characteristics because the earlier research included all heart attacks (including fatal ones), over a longer period (1987-2014 rather than 2001-2014 in the new research), and over the whole year (instead of only May to September as in the new research).  But that does raise a concern, perhaps, about the fact that the new research is based only on non-fatal attacks.  The previous research does give some separate figures for 2001-2014 (for the whole of the year), and there were about 8,200 non-fatal heart attacks and about 5,300 fatal ones.  The researchers explain in their new paper that they could not include fatal heart attacks because they did not have adequate information on the medication that the patients were using, but that does leave out substantial numbers of heart attacks.  But perhaps, if they could have looked at fatal heart attacks too, the patterns in risk in relation to temperature and medicine would have been different in some important way from what they found with the non-fatal attacks.  We can’t know if they would have been, because there are no data – but maybe the further research on these questions could look at that aspect too.

‘Triggering of myocardial infarction by heat exposure is modified by medication intake’ by Kai Chen et al. was published in Nature Cardiovascular Research at 16:00 UK time on Monday 1 August 2022.

DOI: 10.1038/s44161-022-00102-z

Declared interests

Prof Stephen Evans: “No conflicts of interest.”

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  My quote above is in my capacity as an independent professional statistician.”