July 7, 2020

expert reaction to study looking at statins and all-cause and cardiovascular mortality in US veterans 75 years and older

Research, published in JAMA, looked at statins and all-cause and cardiovascular mortality in US veterans 75 years and older.

Prof Peter Sever, Professor of Clinical Pharmacology & Therapeutics, Imperial College London, said:

“There is conflicting evidence that the benefits of statins in reducing cardiovascular events in the elderly is similar to the benefits observed in younger age groups.  The best evidence to date is derived from a meta-analysis of randomised controlled trials, on account of the fact that in individual studies the number of patients in the older age groups (>70-75 years) has been too few to allow robust assessment of efficacy of statins.  The Cholesterol Treatment Trialists Meta-analysis (Lancet 2019; 393: 407-15) showed that benefits of statins  on major vascular events and vascular death were generally consistent throughout the age range.  However, when these analyses were separated into those with and without prior vascular disease, although for secondary prevention the risk reductions in vascular events with statins were similar in both young and old, in primary prevention there was a diminution in these benefits in those over the age of 70 years.  There was no excess of statin related adverse effects in any age group.

“It is against this background that Orkaby and colleagues report data from an observational Veterans Health Administration Study carried out in over 300,000 patients, average age of 81 years, who were followed up for approximately seven years.  The patients, who were mainly white males, were new statin users with no prior history of heart disease or stroke.  This study included patients with multiple comorbidities – the type of patients who were unlikely to have been recruited into clinical trials of statins.  The outcomes included a remarkable 25% reduction in all cause mortality and a 20% reduction in cardiovascular death.  These benefits are approximately three times greater than those reported in the clinical trials of statins.

“There are always problems with the interpretation of observational data due to confounding and other influences, nevertheless, these data are impressive and reassuring and suggest that statins are effective and safe in elderly people in whom the incident rates of cardiovascular disease are very high, and in whom the benefits of statins are potentially very great.  Whether the magnitude of benefits reported in this study will be confirmed in future studies will await the outcome of two ongoing clinical trials of statins which are currently underway in the elderly population.”

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

“Given the very considerable amount of research, and of media discussion over many years, about the use of statins, it might seem surprising that there is not much hard evidence on health outcomes if statins are prescribed to people of 75 years and older for what’s called primary prevention of diseases of the heart and circulation.  (That means prevention of these diseases in people who have not previously had them, in contrast to preventing re-occurrence or worsening of the disease in people who have already had them.)  One reason is that the best evidence on the effectiveness of statins comes from clinical trials, and the clinical trials of statins have not involved many patients aged 75 and over.  And often such trials have excluded people with other health problems such as cancers or dementia, which are more likely to occur in people over 75.  This new study attempts to fill this evidence gap, and it does provide useful new evidence. But because the way it had to be carried out, the results have to be interpreted carefully.

“The new study is not a randomised clinical trial.  (Two clinical trials that will provide further data are ongoing, but their results will not be available for several years.)  It is observational – that is, the researchers used records of statin prescriptions and various health outcomes in a large group (almost 330,000) of people aged 75 and over.  They looked at whether death rates from all causes, death rates from disease of the heart and circulation (cardiovascular disease), and various rates of occurrence of the diseases that statins might be most likely to reduce (heart and circulatory diseases involving atherosclerosis, the build-up of fatty material inside arteries that can lead to heart attacks and strokes, which are abbreviated as ASCVD, for atherosclerotic cardiovascular disease), whether or not they were fatal, to see if these differed between people taking statins and people not taking statins.  They found quite strong evidence that starting on statins is associated with reductions in the rates of deaths from all causes, and from cardiovascular disease, and also with reductions in the rate of occurrence of ASCVD.  And the reductions they found are quite substantial.  For instance, out of every 1000 people studied who were not taking statins, nearly 100 would die on average each year.  (That’s quite a high death rate, but remember that they were all aged at least 75, that the average age when they started in the study was 81, and that they were followed up on average for about 7 years.)  The researchers estimated that taking statins would, if the association really is causal, reduce the number of deaths in these 1000 people by, on average, about 20 a year.

“There are a lot of points to take into account when interpreting these result, though.  First, they apply only to people agreed 75 and over, who were free of ASCVD when they began to be studied, and who had not been on statins before they began to be studied.  So they don’t give any direct information about effects of statins in people who already have (or previously had) ASCVD, or in people who started taking statins before they were 75 – the question of interest was whether it might be worth starting patients (who are free of ASCVD so far) on statins for the first time at ages of 75 of more.  Then, the people studied all came from the US Veterans Health Administration service, which generally provides healthcare to people who served in the US armed forces.  For many reasons they are unlikely to be typical even of the US population as a whole.  That shows up in the data – for instance, only 3% of the study participants were women, and only 9% were non-white.  Maybe the results would have been different in the US elderly population as a whole, let alone in other populations.

“Because the new study is observational, it shares the issues faced by all observational studies.  People who get a new statin prescription are bound to differ in several ways from people who do not get one.  If nothing else, there would in general be some reason why the statins were prescribed.  Data from the study indicate that statin-users were more likely than non-users to have certain diseases or conditions – particularly, high levels of lipids (including cholesterol) in the blood, which isn’t surprising given that a major reason for statin prescription is to control lipid levels.  Statin-users were also more likely than non-users to be taking a range of other medications, including drugs to reduce blood pressure.  So any difference in death rates or ASCVD rates between statin users and non-users might be wholly or partly caused by these other differences, and not the statins.  The researchers made adjustments to their results to try to allow for these other differences as far as they could, using a method called propensity score overlap weighting.  That method is relatively new, though advantages have been claimed for it compared to some other methods.  But the researchers themselves point out that, despite its use (and also the use of an approach called new-user design), there may be differences between statin users and non-users that could not be adjusted for, perhaps because no data were available to adjust for them.  Though I think the researchers used appropriate statistical methods, they could not entirely avoid these limitations stemming from the fact that the study is observational.

“Because of this, and because the group that they studied may not be typical of the whole US population or other populations, I support their recommendation that further research is done.  This further research could also look at potential adverse effects of statin treatment.  How common such adverse effects might be has been a matter of controversy, with many researchers taking the view that the adverse effects are relatively rare, but some others disagreeing, particularly when statins are used for primary prevention as in this study.  But the new study did not assess the prevalence of adverse effects.”

Prof Kausik Ray, Professor of Public Heath, Imperial College London, said:

“In the context of primary prevention statins are recommended for use in those with an estimated 10 year risk above a certain threshold.  Since age is the strongest predictor of 10 year risk, older patients simply by virtue of their age are likely to cross this threshold on that basis.  This is particularly a challenge in an ageing population where there are little randomised trial data for what should be done in the setting of primary prevention.  Older patients often have comorbidities which excludes them from randomised trials and more likely to be frail and intolerant of medications.  These patients have shorter life expectancy and are at risk of deaths from both cardiovascular and non-cardiovascular causes such as cancer or dementia which will not be modifiable by lipid lowering treatments.

“It is against this background that this large retrospective cohort study using the VA health system database in the USA was conducted.  Approximately 326 000 people were studied including those with significant comorbidities.  The average age was 81 years, the study population was 97% men and 90% white.  The association between new statin prescriptions versus no statins were compared on all-cause mortality, cardiovascular mortality and fatal and non-fatal cardiovascular events.  The authors used propensity matching to reduce the likelihood of biases and statistical adjustment to account for confounding.  During 6.8 years > 200 000 died ~2/3, of whom ~ 56 000 died from cardiovascular causes.

“During the study, in those people taking statins deaths from any cause were 25% lower, from cardiovascular causes were 20% lower, and cardiovascular events were 8% lower, mostly driven by revascularisation procedures being 11% lower; but there were no significant associations with reduced risk of non-fatal heart attacks or non-fatal strokes.  There were no subgroups where the beneficial associations with mortality were significantly attenuated including those with a prior history of dementia or arthritis.

“There are some anomalies with trials.  Firstly the benefit on mortality is 2.5 times greater than what the trials would suggest.  One quarter of deaths were from cardiovascular causes, and statins were associated with a 20% lowering of cardiovascular deaths but a 25% lower risk of deaths overall, suggesting in this study they were associated with a lower risk of non-cardiovascular deaths.  This is biologically unlikely and at variance with trial data.

“Furthermore the associated benefits with stroke and heart attacks are more modest than in trials.  The absolute number of deaths per 1000 potentially prevented was 19 but for cardiovascular deaths was a more modest 3.  One would expect a larger effect size on cardiovascular death with lipid lowering and if cardiovascular deaths accounted for the vast majority of deaths then one would expect a large impact on deaths from any cause.  A large proportion of the 200 000 deaths would be from unknown causes (found dead/ died in sleep) and may be cardiovascular in origin to explain these differences.

“Whilst these data do not provide evidence that starting statins are beneficial in the elderly, which requires the results of two large ongoing trials, these data at least suggest that for those elderly patients in whom statins are initiated, it should continue.  There was no excess risk of deaths from any cause and indeed potential benefits, including those with dementia or arthritis at baseline.  A lower association with revascularization (coronary artery bypass operations, angioplasties and stenting) has large health economic benefits as these procedures are expensive, require extensive hospital stay and have a higher risk of complications among the elderly.  Apart from the observational data there is little data on non-whites and women.  These data however set the scene for the results of prospective randomised trials to be completed.”

Dr Riyaz Patel, Associate Professor of Cardiology, UCL, said:

“This is a well conducted study from the US, exploring whether statin use among older people over the age of 75, without previous heart attacks or strokes, is associated with lower death rates.

“This topic has been controversial with several mixed results in the past, showing either no benefit or little benefit of statins in this age group.  This in turn has led to uncertainty whether statins should be started or even stopped in older people.

“The researchers used sophisticated analyses to address the question.  Importantly they did not exclude too many people so the studied population would resemble a more real world group.  However this is still observational data and complex biases around the factors that dictate prescribing of statins to certain individuals and not to others, could play a part despite the statistical attempts to account for this.  Also the study only included men so we can’t draw firm conclusions beyond this group.  As the authors state, further work is needed and hopefully more concrete answers will come from clinical trials.

“Nonetheless, this is a sound study and the findings are consistent with what we understand about risk and the nature of the benefit of statins in all age groups.”

Prof Stephen MacMahon, Principal Director of The George Institute for Global Health, Professor of Medicine at the University of Oxford, and Professor of Cardiovascular Medicine at UNSW Sydney, said:

“There are many weaknesses in this study that make its conclusions unreliable.  It is not a randomised trial, which is the gold standard used by all drug regulators and most clinical researchers to assess benefits and risks of drug treatments.

“In this case, the results themselves raise serious questions about the reliability of the study.  For example, while there is a lower risk of death among those taking statins, the greatest reduction is in deaths from diseases not known to be associated with cholesterol.  Additionally, there is no reduction in those conditions known to be particularly sensitive to treatment with statins, such as myocardial infection and ischaemic stroke.

“These inconsistencies are almost certainly due to biases that are impossible to avoid, even with the most sophisticated statistical analyses.  In many ways, the study results illustrate precisely why randomised trials are essential and why other approaches frequently produce misleading results.”

‘Association of statin use with all-cause and cardiovascular mortality in US veterans 75 years and older’ by Ariela R. Orkaby et al. will be published in JAMA at 16:00 UK time on Tuesday 7 July 2020.

DOI: 10.1001/jama.2020.7848

Declared interests

Prof Peter Sever: “Research grants and consultancy payments: Pfizer, Amgen.”

Prof Kevin McConway: “Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.”

Prof Kausik Ray: “Disclosures include consulting for companies who have developed lipid lowering therapies.”

Dr Riyaz Patel: “Received honoraria from Amgen and Sanofi, companies that make new non -statin drugs to control cholesterol.”

Prof Stephen MacMahon: “No conflicts of interest.”