July 19, 2022

expert reaction to study looking at cardiovascular disease and diabetes risk after Covid

A study published in PLOS Medicine looks at cardiometabolic outcomes up to 12 months after COVID-19 infection.

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

“This is a good study in many ways.  It’s based on data from a large database of records from over 1,300 GP practices in England, and the results are based on data from over 850,000 patients registered at those practices.  While there’s always a possibility that the patients at those practices aren’t typical of all patients in England, the main findings come from comparing all the people on the database with a diagnosis of Covid-19 with a sample of other, control, patients who, individually, matched the Covid cases in terms of year of birth, sex, and the practice where they were registered.  The statistical methods and analysis generally seem very appropriate to me.

“However, there are some inevitable difficulties in interpreting the findings.  Generally, these issues arise from the way the study was conducted, and particularly from the fact that it’s observational.  The researchers did not intervene in what the patients or their doctors did – in fact they had no opportunity to do so, even if they had wanted to, because the study is entirely based on patient records, analysed after the events in the records occurred.  The researchers found that patients, who had had Covid, had on average an increased risk of cardiovascular disease (heart disease, strokes, blood clots and so on), and also an increased risk of diabetes, in a period of time after they were diagnosed with Covid, compared to the control patients who had not had Covid.  This could be because Covid itself causes the increased risk, but there are other possibilities.

“That’s because there are other differences between the Covid-19 cases and the controls, apart from one group having been diagnosed with Covid and the other not.  Though the cases and the controls were matched in terms of some factors, it wouldn’t have been possible to match them on everything that is potentially relevant.  The Covid cases, for example, had a higher tendency to be overweight than the controls, and had on average more pre-existing health conditions.  So maybe the observed difference in risk is due, in part or entirely, to these other differences between the groups, rather than the Covid itself.

“The researchers, as you’d expect, were perfectly aware of this issue.  It’s possible to make statistical adjustments to allow for other differences between groups, and the researchers made these adjustments for several such factors, including age and sex (where it still makes statistical sense to adjust, despite the use of these two factors for matching), and also ethnicity, body mass index, blood pressure, and a measure of pre-existing conditions.  This helps, but it can’t get rid entirely of the questions about what causes what.  There may be other relevant factors that weren’t adjusted for – for instance, the researchers mention that they had no data on deprivation so did not adjust for that.

“Also, the question of what causes what is complicated by the impossibility of knowing exactly what order things occurred in.  The researchers explain that their findings could be explained in other ways.  Possibly the Covid caused increases in the risk of cardiovascular disease and diabetes, or possibly people who already had undiagnosed cardiovascular disease or diabetes had the course of those diseases affected by Covid (so that the diseases were more likely to be diagnosed in people who had Covid, even though they existed already).  Or perhaps undiagnosed disease affected in some way the chances that people would be infected with Covid.  Another possibility, also mentioned by the researchers, is that people who had had Covid had more contact with their GPs, because of the Covid, and therefore might be more likely to have a diagnosis of diabetes afterwards, even though the disease (rather than its diagnosis) might be no more common in the cases than the controls.  (Type 2 diabetes can often be present without having been diagnosed.)  Overall, then, though it’s quite possible and plausible that the results come from Covid tending to increase the risk of cardiovascular disease and diabetes, we can’t be certain that this is what’s going on.

“This is of a matter of increased risk of some cardiovascular diseases and of diabetes, averaged across all the people that were studied, in those who had a Covid diagnosis compared to comparable people who had not had a Covid diagnosis.  It’s an increased risk across the population – no study like this could tell you which specific individuals are at risk and which are not.  The researchers do report that increased risk was greater in people with higher body mass index, or who had more pre-existing conditions, or whose ethnicity was categorised as ‘Asian’ (by which they mean that their ethnic origins were in the south Asian sub-continent, not the whole of Asia).  But they do not give clear results on how much higher the risk is for people in these categories, and anyway we’ve got to remember that nothing here can be taken to be a question of cause and effect.

“It’s perhaps reassuring that the increased risks of cardiovascular disease and of diabetes do fall off fairly quickly after the diagnosis of Covid.  That’s particularly true for cardiovascular disease, though the increased risk has largely or possibly entirely gone by about three months after the initial Covid diagnosis.  The increased risk of diabetes falls off more slowly with time, but it does fall.  An interesting feature of the findings is that there seems to be an increased risk of both cardiovascular disease and of diabetes for a short time before the Covid diagnosis, though the increase after the diagnosis is considerably greater.  The researchers point this out, but do not really explain how the increased risk before the Covid diagnosis arises.  I could speculate that it might be because a new case of cardiovascular disease or diabetes might possibly increase the chance of contracting Covid soon after, or perhaps it arises only because of anomalies in when diagnoses of any of the diseases (CVD, diabetes or Covid) get onto the database – but really I don’t know.

“The increases in the risk of both cardiovascular disease and diabetes, in the first four weeks after a Covid diagnosis, are really quite large.  In 100,000 patients who had Covid, roughly 300 had a new diagnosis of cardiovascular disease in the first four weeks after their Covid diagnosis.  In 100,000 comparable control patients who did not have Covid, the number of new cardiovascular disease diagnoses in the corresponding period of time was below 30.  The difference is not so marked for diabetes.  In 100,000 patients who had Covid, just short of 100 had a new diabetes diagnosis in the first four weeks after the Covid diagnosis, but for 100,000 controls, the corresponding number was slightly under 40.  But all these numbers do at least make it clear that most Covid patients don’t actually become a new case of diabetes or cardiovascular disease.”

Prof Amitava Banerjee, Professor of Clinical Data Science and Honorary Consultant Cardiologist, Institute of Health Informatics, UCL, said:

“In terms of novelty, there have been analyses showing increased risk of both increased cardiovascular disease and increased cardiometabolic (diabetes and CVD) risk following COVID.  There is a group which is publishing a lot in this space from Washington University St Louis (led by Ziyad Al-Aly): https://www.nature.com/articles/s41591-022-01689-3 (1 year CVD outcomes); https://pubmed.ncbi.nlm.nih.gov/35325624/ (1 year diabetes outcomes).

“This large, well-conducted electronic health record study supports findings from other research in the UK and other countries showing increased risk of cardiovascular disease and diabetes post-COVID.  These new analyses suggest slightly increased risk in the whole population in the first few weeks after COVID, but higher risk in certain people, such as those with pre-existing heart disease or diabetes.  The underlying mechanism of the increased risk is unclear and more research is needed.  As well as the direct effects of COVID-19, the diagnosis and management of heart disease, diabetes and other underlying conditions has been adversely affected by the pandemic, albeit indirectly.  Yet again, the science supports avoiding SARS-CoV-2 infection at individual and population level, wherever possible.”

‘Cardiometabolic outcomes up to 12 months after COVID-19 infection. A matched cohort study in the UK’ by Emma Rezel-Potts et al. was published in PLOS Medicine at 19:00 UK time on Tuesday 19 July 2022.

DOI: 10.1371/journal.pmed.1004052

Declared interests

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.”

Prof Amitava Banerjee: “AB is supported by research funding from the National Institute for Health Research, British Medical Association, AstraZeneca, and UK Research and Innovation; and is trustee of the South Asian Health Foundation and of LongCOVIDSoS.  He is chief investigator of the NIHR-funded STIMULATE-ICP study in individuals with Long COVID.”