Author + information
- aRobertson Centre for Biostatistics and Clinical Trials, University of Glasgow, Glasgow, Scotland, United Kingdom
- bNational Heart and Lung Institute, Imperial College London, London, United Kingdom
- ↵∗Address for correspondence:
Prof. John G.F. Cleland, Robertson Institute of Biostatistics and Clinical Trials Unit, University of Glasgow, University Avenue, Glasgow, Scotland G12 8QQ, United Kingdom.
There are no randomized, placebo-controlled trials, contemporary or historical, showing that aspirin given for longer than 28 days, at any dose, reduces cardiovascular mortality after a myocardial infarction (3). Some large, long-term trials did show a reduction in nonfatal myocardial infarction, but others showed a significant excess. The results of published meta-analyses have been driven by small, outrageously positive trials, suggesting publication bias (4). Aspirin is a fine example of the mess created by people jumping to premature conclusions based on wishful thinking and flawed data. Prescribing aspirin, long-term, has become at least a habit and perhaps an addiction; rehabilitation of offenders is likely to be a long and difficult process with many lapses.
How do we get out of this mess? Ideally, further randomized, placebo-controlled trials should be conducted. This approach requires that patients and physicians agree that it is foolish to continue to ignore the possibility that aspirin is a waste of time or worse. An environment conducive to fresh trials will only be possible if the intellectual equipoise that uncertainty provides is restored. Being certain and being wrong is a difficult position to extricate oneself from. An apparently safe but useless treatment is still harmful because it displaces other useful interventions and may give a false sense of security.
There are many ways to create a climate of constructive uncertainty. Review of the results of the original trials of chronic aspirin therapy conducted 50 years ago, upon which the antiplatelet secondary prevention meta-analysis depends, might well be enough to change majority opinion (3). Randomized trials do not suggest that withdrawing aspirin from patients with stable coronary artery disease is associated with risk and might be beneficial. The theory that vascular occlusion is always primarily thrombotic is challenged by the evidence that hemorrhage into plaque may often be the primary event (5). Substantial, placebo-controlled trials of aspirin conducted for primary prevention have proved neutral (or worse) thus far, making it feasible to conduct trials in higher risk groups (4). Concerns exist that aspirin may accelerate declines in cognitive dysfunction as well as hearing and visual acuity (4), which might account for why some aspirin-evangelists are unable to follow the scientific arguments proposed by those who take a more critical view of the evidence. Aspirin also increases the risk of hemorrhage, likely contributing to the iron-deficiency anemia that is now highly prevalent in older people with coronary disease.
Subgroup analysis of the landmark trials that form the evidence base for guideline-recommended therapy for heart failure also suggests that aspirin may detract from the benefits of angiotensin-converting enzyme inhibitors, mineralocorticoid antagonists, and possibly even beta-blockers (6). Importantly, international guidelines have not reached a consensus on recommending aspirin for patients with heart failure, even if they have coronary artery disease. Another approach that might add to uncertainty is to investigate the effects of aspirin at a population level, using administrative data, as done by Madelaire et al. (7) in this issue of JACC: Heart Failure.
Before considering this evidence, it is worth reflecting on the goals of treating heart failure and many other diseases (Figure 1). Fundamentally, these goals are to restore or maintain well-being, prevent or correct disability, restore or maintain independence, and to prolong life for as long as the patient wishes. If treatments also relieve or prevent headaches, toothaches, chest pain, troponin leaks, or episodic paresis that is good, but such benefits should not be used as surrogate measures of efficacy when important outcomes can actually be measured. For instance, of patients who have had a myocardial infarction and subsequently experience a stroke, approximately one-third will die within 1 month, one-third will recover with little or no long-term disability, and only one-third will survive with persisting major disability. It is the consequences that such events have on well-being, disability, independent living, and prognosis that are of fundamental importance, not the event itself. An intervention that reduces supposedly serious events but does not subsequently improve well-being, disability, or mortality should be treated with suspicion (4). For instance, flecainide is highly effective at reducing nonfatal myocardial infarction but only because it increases prehospital mortality. Use of nonsteroidal anti-inflammatory drugs, of which aspirin is one, may be associated with an increase in sudden death (8). Several long-term trials of aspirin have suggested a reduction in nonfatal myocardial infarction, but a reduction in non-fatal events has not translated into a reduction in cardiovascular death (4).
So what did Madelaire et al. (7) show in their article? In Denmark, administrative data from a number of sources can be linked to de-identified individual personal records to investigate the incidence, prevalence, and natural history of disease and their associations with prescribed therapies, such as aspirin. Madelaire et al. (7) identified 59,838 patients admitted with a new diagnosis of heart failure (approximately 1,800 patients per million population per year) between 2007 and 2012; in-patient mortality was about 10%, and approximately 65% of survivors were not in sinus rhythm or were anticoagulated at discharge. Very similar data have been reported for the United Kingdom (9). Within 90 days, of 24,936 patients discharged in sinus rhythm and not on anticoagulant agents, 2,616 (10.5%) died and 4,627 (18.6%) were re-hospitalized for heart failure. Only 12,277 patients (about 21%) from the original cohort were suitable for further analysis. These patients had a median age of 76 years (interquartile range [IQR]: 65 to 84 years), and 51% were women. This information provides a very useful epidemiological snapshot relevant to an ongoing study such as COMMANDER HF (A Randomized, Double-blind, Event-driven, Multicenter Study Comparing the Efficacy and Safety of Rivaroxaban with Placebo for Reducing the Risk of Death, Myocardial Infarction or Stroke in Subjects with Heart Failure and Significant Coronary Artery Disease Following an Episode of Decompensated Heart Failure) comparing rivaroxaban versus placebo for patients with ischemic heart disease and a reduced left ventricular ejection fraction recently hospitalized for worsening heart failure (10).
Variables used by Madelaire et al. (7) for propensity matching included age, sex, year of study, heart failure severity, several comorbidities, and concomitant medical treatment (7). Unfortunately, there was no record of left ventricular ejection fraction. Only a minority of patients were reported to have ischemic heart disease (28%), and <10% had a history of myocardial infarction. These findings are fairly consistent with epidemiological studies in primary care, although not secondary care, and may reflect under-reporting of ischemic heart disease (9).
Altogether, 3,840 matched pairs of patients (13% of the initial population) taking or not taking aspirin were identified (7). Clearly, this total is a large number of patients compared with most randomized trials, but it is still not a randomized trial. Over a median follow-up of 1.7 years, in the matched population, 37% of patients died, approximately 4% had a stroke, and 6% were admitted for bleeding, with similar rates for aspirin and placebo. These outcomes suggest that aspirin might be useless but not harmful. However, the rates of (mainly non–ST-segment elevation) myocardial infarction (4.8% vs. 3.6%; p = 0.009) and hospitalization for heart failure (31% vs. 29%; p = 0.03) were higher with aspirin. The increased risk of heart failure hospitalization is consistent with many, although not all, previous analyses and, if true, equates to a number needed to harm of just 48. The effect on myocardial infarction could just reflect residual confounding but certainly does not provide evidence of benefit from aspirin. Further analyses, including subgroups and taking into account patients initiated on aspirin during follow-up, did not provide convincing evidence of substantial benefit or harm.
The COMPASS (Cardiovascular Outcomes for People using Anticoagulation Strategies) trial compared aspirin 100 mg/d alone versus rivaroxaban 5 mg twice daily versus a combination of aspirin 100 mg/d plus rivaroxaban 2.5 mg twice daily in patients with coronary or peripheral vascular disease, approximately 20% of whom had heart failure (11). Unfortunately, the study was stopped early due to a small but highly significant advantage to the combination compared with aspirin alone. This early cessation tends to overestimate the benefits of the combination but, more importantly, underestimates that of rivaroxaban monotherapy. Careful review of the data suggested that rivaroxaban monotherapy might have had the same benefit as the combination had follow-up been extended by an additional 6 months. This extended assessment would have been a huge step forward in the management of aspirin addiction but is now another lost opportunity.
Madelaire et al. (7) have shown that most patients discharged from the hospital in sinus rhythm are not prescribed aspirin, including one-third or more of those with ischemic heart disease. Therefore, it seems that a randomized controlled trial is feasible to ensure that aspirin is either given or withheld appropriately based on evidence. The odd thing is, whenever you open a trial for enrollment, suitable patients are often in short supply. Perhaps the best treatment for patients is to administer a clinical trial?
↵∗ Editorials published in JACC: Heart Failure reflect the views of the authors and do not necessarily represent the views of JACC: Heart Failure or the American College of Cardiology.
Dr. Cleland has received research funding and honoraria for advisory boards and speaking from Bayer.
- 2018 American College of Cardiology Foundation
- ↵Tolstoy K. Leo Tolstoy quotes. Available at: https://www.goodreads.com/quotes/106513-wrong-does-not-cease-to-be-wrong-because-the-majority. Accessed December 5, 2017.
- ↵Brainyquotes. Robert Burns' quotes. Available at: https://www.brainyquote.com/quotes/robert_burns_182938. Accessed December 5, 2017.
- Cleland J.G.
- Madelaire C.,
- Gislason G.,
- Kristensen S.L.,
- et al.
- Sondergaard K.B.,
- Weeke P.,
- Wissenberg M.,
- et al.
- ↵National Heart Failure Audit. April 2015–March 2016. Available at: http://www.ucl.ac.uk/nicor/audits/heartfailure/documents/annualreports/annual-report-2015-6-v8.pdf. Accessed December 27, 2017.
- Zannad F.,
- Greenberg B.,
- Cleland J.G.,
- et al.
- Eikelboom J.W.,
- Connolly S.J.,
- Bosch J.,
- et al.