The ASPirin in Reducing Events in the Elderly trial (ASPREE) contributed important knowledge about primary cardiovascular disease (CVD) prevention among healthy older adults. The finding that daily low-dose aspirin (LDA) does not statistically prevent disability or CVD among adults aged over 70 years when compared with placebo, but does significantly increase risk of haemorrhage, immediately influenced clinical practice guidelines. In this article, we discuss nuances of the trial that may impact the extrapolation of the ASPREE trial results to the everyday individual clinical care of older adults.
For UK healthcare professionals only
The ASPirin in Reducing Events in the Elderly trial (ASPREE), published in 2018, was a landmark randomised-controlled trial (RCT) that contributed important knowledge about primary cardiovascular disease (CVD) prevention among healthy older adults.1 ASPREE found that daily low-dose aspirin (LDA) does not statistically prevent disability or CVD among adults aged over 70 years when compared with placebo, but does significantly increase risk of haemorrhage; findings that immediately influenced clinical practice guidelines.2
When used as a case study of large RCTs, ASPREE provides further, more existential, lessons for both researchers and physicians in the conduct and interpretation of large RCTs and the complexity of assimilating trial data into clinical practice guidelines. In this perspective, we discuss nuances of the trial that may impact the extrapolation of the ASPREE trial results to the everyday individual clinical care of older adults, particularly those who are concurrently taking non-steroidal anti-inflammatory drugs (NSAIDs) and proton-pump inhibitors (PPIs).
The results of ASPREE were seemingly discordant with the hitherto large body of primary prevention LDA trials conducted in non-elderly study populations. Specifically, most prior trials of LDA in high-risk primary prevention found reductions in non-fatal CVD (particularly myocardial infarction [MI]) with trends towards lower all-cause mortality,3 whereas ASPREE found no effect on MI and a trend towards higher mortality.1
Why is this? The ASPREE authors propose that it may be related to the fact that their patients were not required to have cardiovascular risk factors, or be above a certain threshold of estimated cardiovascular risk. Therefore, despite their older age, ASPREE participants may not have had the same increased cardiovascular risk as the patient populations in other primary prevention aspirin trials.1 Other explanations must include the consideration that elderly adults are fundamentally different from non-elderly adults with respect to drug therapy. Similarly, is it that the primary prevention population’s background propensity to CVD has changed in the modern era, reducing LDA’s meaningful contribution? Or is ASPREE a triumph of the large RCT to dispel misconceptions about innately ineffective therapies? While all of these explanations are possible, there are a few nuances to this trial worth considering when applying the results from ASPREE to individual older adults being treated in the clinic.
Duty of care to patients versus clinical trial imperatives
Chronic LDA increases the risk of upper gastrointestinal (UGI) haemorrhage, and this risk is increased by various factors including age over 65 years (by definition, all ASPREE patients) and concomitant use of NSAIDs. Conversely, prophylactic use of PPIs has been shown to significantly reduce bleeding risk.4
The recently published COMPASS (Cardiovascular Outcomes for People Using Anticoagulation Strategies) trial found that prophylactic PPI use did not significantly reduce the composite of clinically significant UGI bleeding events in patients taking low-dose anticoagulation and LDA.5 The point estimate, however, was in favour of PPI use. It should also be noted that COMPASS included patients that had already been on LDA long-term, and those who were already on a PPI were excluded from this study. Interestingly, in a subgroup analysis of aspirin-only patients (i.e. those randomised to aspirin and placebo), there was a statistically significant reduction in UGI events in the factorial arm of this trial who were also randomised to PPI. These points suggest that a high-bleeding risk subgroup on LDA (like the elderly patients in ASPREE) would probably still benefit from PPI consideration.
Thus, COMPASS appears to provide some support for current guidelines, which recommend that an informed discussion on the prophylactic use of PPIs should be standard-of-care for patients over 65 on LDA and who have at least one other bleeding risk factor.4 However, for the same patient not on LDA, a PPI would be unnecessary as a purely prophylactic intervention; i.e. the use or non-use of LDA would make the difference between the recommendation (or not) for prophylactic PPIs in many ASPREE-like patients. Here, the blinding process that is utilised to prevent bias within an RCT was in direct conflict with critical, actionable clinical information.
ASPREE raises ethical and medico-legal conundrums for primary-care physicians involved with clinical research. Out of the trial setting, if there was no discussion of the recommendations from expert societies about prophylactic PPIs, and an elderly at-risk patient commenced on LDA suffered a significant UGI haemorrhage, would it be considered that the doctor practised best care? Because a patient is in a trial and the doctor is blinded to actionable information, does that obviate the doctor’s duty-of-care to provide best practice?
The ASPREE protocol did not prevent the use of PPIs. But, because of blinding, ASPREE did not allow aspirin-arm patients to be started on prophylactic PPIs, who might otherwise qualify.
To maintain reproducibility, when counselling patients on risks and benefits of LDA, should ASPREE only be considered an evidence base for patients who do not tell their doctor whether they are or are not taking LDA? Since gastrointestinal bleeding was one of the main adverse outcomes of ASPREE, can ASPREE be considered quotable evidence for contemporary practice where at-risk older patients started on LDA are typically offered prophylactic PPIs?
Pharmacokinetics, pharmacodynamics, interactions and pre-trial review of trial protocols
The half-life of LDA is very brief, but its clinical effect is prolonged due to the irreversible inhibition of the platelet cyclo-oxygenase (COX)-1 enzyme. An increasingly recognised drug interaction occurs when the presence of another NSAID (such as ibuprofen) at the COX-1 enzyme prevents LDA’s brief window of opportunity to bind and irreversibly deactivate platelet COX-1; before LDA is rapidly eliminated. The Food and Drug Administration (FDA) state that “Timing of dosing of ibuprofen and low-dose aspirin is important for preserving the cardioprotective effect of aspirin”.6
The ASPREE trial protocol recognised this: “if they take NSAIDs they will be advised to take them at least half an hour after taking trial medication to minimise interaction.” However, this protocol recommendation is only appropriate for immediate-release aspirin, whereas ASPREE used a slow-release enteric-coated formulation of aspirin, which likely meant that no medication administration regimen would have prevented this NSAID interaction. For example, according to the FDA, an “ibuprofen regimen that does not attenuate the antiplatelet effects of enteric coated aspirin has not yet been identified”.6
Approximately 14% of ASPREE participants, enrolled in Australia and America were regularly taking conventional NSAIDs. This observation appears to be consistent with the British population, where around 14,000 patients have been estimated to be taking an NSAID and anticoagulation concurrently without gastroprotection.7 If the aspirin–NSAID interaction was considered significant enough to include measures to avoid it in the trial protocol (measures that unfortunately may have been futile given the use of enteric-coated aspirin in ASPREE), then it appears difficult to justify including data from the 14% of patients concurrently on NSAIDs in the primary outcome analysis. While a subgroup analysis of ASPREE did not report effect modification for LDA based on NSAID use, this analysis was not prespecified and underpowered, so the results are inconclusive. As such, concomitant NSAID use (and timing) presents just one of, potentially, many drug interactions that could limit the net benefits of primary prevention of LDA in older adults, reminding us that close evaluation for drug interactions is always advised for any older adults taking aspirin.8
Translating findings from trial populations to individual older patients outside of the trial
In ASPREE, the reported compliance with aspirin was 62%. Considering the 14% of patients concurrently taking NSAIDs, the number of patients effectively taking aspirin was potentially as low as 48% (because in those 14% of patients the NSAIDs may have prevented the extended-release formulation of aspirin from being able to deactivate the platelet COX-1 enzyme). It is difficult to ask any drug or intervention to perform when it doesn’t even get on the playing field. However, despite as few as 50% of patients in the LDA group effectively taking the medicine, there was a trend towards reduced cardiovascular events (hazard ratio 0.95; 95% confidence interval [CI] 0.83 to 1.08).1 Noncompliance related to polypharmacy is a major problem in older populations that is often not planned for in sample size estimates.
Furthermore, the follow-up time averaged only 4.7 years. So, in a potentially underpowered study there was a mechanistically consistent trend towards cardiovascular benefit. What should one tell a patient, especially a 70-year-old, hoping to live more than 4.7 years, who is motivated to take the drug?
Finally, it is impossible to counsel accurately regarding the main downside to LDA, as an unknown number of bleeding complications could/should have been prevented by otherwise standard-of-care prophylactic PPIs.
How should physicians deal with these issues related to the applicability of large homogenised trials to individual patient circumstances; before potentially rejecting beneficial therapies for their older patients? An understanding of systems complexity underpins a final lesson of ASPREE, as related to evidence-based guidelines. Unfortunately, while large RCTs are necessary and a fundamental step in the practice of evidence-based medicine, they often lack the fidelity to give useful information at the margin of clinical care for individual patients, particularly in the elderly. Although guidelines attempt to prioritise RCT evidence when providing recommendations for clinicians seeing individual patients, often RCTs are unable to provide clinical answers to individual patient circumstances, due to the very nature of clinical trial results – which are expressed as the average result from among the whole sample of included individuals participating in the trial. Thus, sometimes evidence outside of clinical trials needs to be drawn upon to formulate guideline recommendations. This might also help to explain the results from a JAMA study, which found that “among recommendations in major cardiovascular society guidelines, only a small percentage were supported by evidence from multiple RCTs or a single, large RCT. This pattern does not appear to have meaningfully improved from 2008 to 2018”.9
- Individual patient circumstances need to be considered when basing clinical decisions on data from large randomised-controlled trials. This is the case for aspirin and primary prevention in the elderly
- Sometimes the use of actionable clinical information might be in conflict with blinding in clinical trials. The use of this information to prevent side effects may change the risk/benefit balance
- Pre-trial review of trial protocols may allow optimisation of drug formulations
Conflicts of interest
MK is on the medical advisory board for Cannvalate, a medical cannabis company. JWM and JN have no conflicts to report.
We thank Professor Neville David Yeomans (The University of Melbourne) for his conceptual guidance and editing.
1. McNeil JJ, Wolfe R, Woods RL et al. Effect of aspirin on cardiovascular events and bleeding in the healthy elderly. N Engl J Med 2018;379:1509–18. https://doi.org/10.1056/NEJMoa1805819
2. Arnett DK, Blumenthal RS, Albert MA et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease. Circulation 2019;140:e596–e646. https://doi.org/10.1161/CIR.0000000000000678
3. Raber I, McCarthy CP, Vaduganathan M et al. The rise and fall of aspirin in the primary prevention of cardiovascular disease. Lancet 2019;393:2155–67. https://doi.org/10.1016/S0140-6736(19)30541-0
4. Bhatt DL, Scheiman J, Abraham NS et al. ACCF/ACG/AHA 2008 expert consensus document on reducing the gastrointestinal risks of antiplatelet therapy and NSAID use: a report of the American College of Cardiology Foundation Task Force on Clinical Expert Consensus Documents. J Am Coll Cardiol 2008;52:1502–17. https://doi.org/10.1016/j.jacc.2008.08.002
5. Moayyedi P, Eikelboom JW, Bosch J et al. Safety of proton pump inhibitors based on a large, multi-year, randomized trial of patients receiving rivaroxaban or aspirin. Gastroenterology 2019;157:682.e2–691.e2. https://doi.org/10.1053/j.gastro.2019.05.056
6. Kelty J. Regulatory history of the interaction between aspirin and other nonprescription NSAIDs. US Food and Drug Administration, Division of Nonprescription Drug Products, Office of Drug Evaluation. Published 2016. Available at: https://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/ArthritisAdvisoryCommittee/UCM607522.pdf [accessed May 2019].
7. NHS Business Service Authority. Medication safety dashboard. 2018. Available at: https://apps.nhsbsa.nhs.uk/MOD/MedicationSafety/atlas.html [accessed 30 October 2019].
8. Hamilton H, Gallagher P, Ryan C, Byrne S, O’Mahony D. Potentially inappropriate medications defined by STOPP criteria and the risk of adverse drug events in older hospitalized patients. Arch Intern Med 2011;171:1013–19. https://doi.org/10.1001/archinternmed.2011.215
9. Fanaroff AC, Califf RM, Windecker S, Smith SC Jr., Lopes RD. Levels of evidence supporting American College of Cardiology/American Heart Association and European Society of Cardiology guidelines, 2008–2018. JAMA 2019;321:1069–80. https://doi.org/10.1001/jama.2019.1122