Aspirin for Prevention of Cardiovascular Events in a General Population Screened for a Low Ankle Brachial Index:  A Randomized Controlled Trial FREE

Quiz Ref IDThe ankle brachial index (ABI), which is the ratio of systolic pressure at the ankle to the arm, is used in the diagnosis of peripheral artery disease affecting the legs. Also, a low ABI is associated with concomitant coronary and cerebrovascular disease¹ and, in healthy individuals, with an increased risk of future vascular events, independently of cardiovascular risk factors.² In a meta-analysis of 16 cohort studies of healthy individuals, the 10-year risk of a major coronary event in men with an ABI of 0.90 or less was 27% compared with 9% in those with an ABI in the normal range of 1.11 to 1.40.² Respective figures for women were 19% and 6%.

Because the ABI is a simple, inexpensive, noninvasive test, it could be used in population screening programs to identify a novel subgroup potentially amenable to preventive treatments. This is of particular importance given that current primary prevention strategies in the general population are of limited benefit.^3- 4 Guideline development groups have suggested that ABI screening be considered in primary care or in the community, especially among certain high-risk groups,^5- 6 whereas others do not recommend screening.⁷

Although an ABI might better define risk, a major argument against screening is lack of evidence for an effective intervention for those with a low ABI.⁷Quiz Ref IDAntiplatelet drugs, including aspirin, are effective in the secondary prevention of events of patients with known vascular disease,⁸ but aspirin may have only a modest effect as a primary prevention for healthy individuals.⁹ We aimed to determine whether screening the general population for a low ABI could identify a higher-risk group who might derive substantial benefit from aspirin therapy.

The Aspirin for Asymptomatic Atherosclerosis trial, conducted from April 1998 to October 2008, was a pragmatic intention-to-treat, double-blind, randomized controlled trial of once daily low-dose aspirin (100 mg) vs placebo involving individuals free of clinical cardiovascular disease and with a low ABI.

Ethical approval was granted by the UK National Health Service research ethics committees in Lanarkshire, Greater Glasgow, and Lothian health boards. All participants gave written informed consent.

The study population comprised men and women aged 50 to 75 years at baseline with no history of vascular disease. Between April 1998 and December 2001 volunteers were recruited from Lanarkshire, Glasgow, and Edinburgh in central Scotland by mailing individuals on Community Health Indexes that contain the age and sex of patients listed on the registers of general practitioners, 83% of whom agreed that their patients could be invited for screening. The study was advertised in newspapers and on fliers posted in general practice offices and announced that only those who had not had a heart attack or stroke or who were not taking aspirin or warfarin could participate in the screening. Those wishing to take part were invited to a local center for ABI screening administered by specially trained nurses.

At screening clinics, participants lay rested supinely for 5 minutes and right and left brachial, posterior tibial, and dorsalis pedis systolic pressures were measured using a sphygmomanometer and Doppler probe. The ABI was calculated as ratio of the lowest ankle pressure (lower of posterior tibial and dorsalis pedis and of left and right) to the higher pressure of either arm. Those with an ABI of 0.95 or lower were entered into the trial unless they had a history of myocardial infarction, stroke, angina, or peripheral artery disease; currently used aspirin, other antiplatelet or anticoagulant agents; had severe indigestion; had chronic liver or kidney disease; were receiving chemotherapy; had contraindications to aspirin; and had an abnormally high or low hematocrit value (measured after the screening). Baseline assessment for trial participants included medical history, medication, smoking status, blood pressure, serum total cholesterol levels, and socioeconomic status.¹⁰ Quality control was maintained by holding regular training days for staff members.

Participants were randomized to receive 100 mg of enteric-coated aspirin daily or placebo. Consecutive participant study numbers were assigned to aspirin or placebo with permuted blocks of size 8, which varied randomly. A staff member not involved in the study produced the computer-generated randomization list. Staff at a single pharmacy prepared the bottles containing aspirin or placebo and labeled them with study numbers. The trial recruits were allocated the next available study number and given the appropriately labeled bottle containing a year's supply of tablets. Randomization lists in sealed envelopes were available only to the pharmacy, an independent member of staff responsible for unblinding due to clinical necessity, and an independent statistician providing reports to the data monitoring committee.

Participants were followed up after 3 months, 1 year, and 5 years at special clinics and annually by telephone. They received a midyear letter enquiring generally about problems and an end-of-year newsletter. Participants kept a diary of adherence. A specially trained nurse assessed self-reported adherence annually and mailed a new supply of tablets. If participants stopped the drug for a nonmedical reason, they were encouraged to restart. If they started long-term antiplatelet medication due to a vascular event or for another reason, the study drug was discontinued. Participants were followed up until the end of the trial, irrespective of their experiencing an end point or adverse event.

Ascertainment of possible events was sought annually from participant follow-up, a study reply card attached to general practitioner notes, flagging for death at the NHS Central Registry, and linkage to databases of deaths and hospital discharges at NHS National Services Scotland. Confirmation of events was sought by review of hospital, general practitioner records, or both independently by 2 members of the outcome events committee with adjudication by the full committee in cases of disagreement. Vascular events were classified using modified American Heart Association criteria¹¹ (eAppendix). Confirmation of an adverse event required diagnosis by a physician and adherence to predefined criteria.

The primary end point was a composite of initial (earliest) fatal or nonfatal coronary event or stroke or revascularization. Two secondary end points were (1) all initial vascular events, defined as a composite of a primary end point event or angina, intermittent claudication or transient ischemic attack; and (2) all-cause mortality.

The trial was powered to detect a 25% proportional risk reduction in major vascular events. The predicted risk reduction was based on evidence that (1) event rates in asymptomatic participants with a low ABI were similar to those with symptomatic peripheral artery disease,^12- 14 suggesting that the risk reduction might be comparable with patients who have clinical disease (approximately 25%¹⁵) and (2) in stable angina, unlike acute coronary syndromes with thrombosis complicating atherosclerotic plaque, the risk reduction could reach 33%.¹⁵ Previous trials had included effects of nonadherence to treatment and of concomitant aspirin use. Adjustment was therefore not made for these effects. The power calculation was based on comparison of proportions whereas the planned survival analysis would be more sensitive. A sample size of 3350 gave 80% power at 5% significance (2 sided) to detect a reduction in event rate from 12% to 9%. The 12% rate in the placebo group was based on data from the Edinburgh Artery Study, a cohort with a similar population to our trial.¹³ The sample size required ascertainment of 350 events (200 placebo; 150 aspirin).

During the trial, the event rate was 60% of that predicted, consistent with a decrease in the number of events in Scotland over the same period. Therefore, extended follow-up was planned for another 4.5 years. Subsequently, on advice from the data monitoring committee, the trial was stopped 14 months early due to the improbability of finding a difference in the primary end point by the end date and an increase in major bleeding (P < .05) in the aspirin group. Despite stopping the trial early, the required number of events was achieved.

Study data were collected on standard forms, checked for completeness, and double keyed into an Access database. Analyses were by intention to treat, with 2-tailed tests of significance, and were performed using SAS/STAT software, version 9.1 SAS System for Windows (SAS Institute Inc, Cary, North Carolina) and SPSS for Windows release 14.0.0 (SPSS, Chicago, Illinois).

Kaplan-Meier plots of time from enrollment to composite primary and secondary end point events for the 2 treatment groups were produced. Incidence rates per 1000 person-years (95% confidence intervals [CIs]) were based on numbers of events and survival analysis follow-up times. Treatment efficacy was assessed using univariate Cox proportional hazards regression model (only treatment in the model).

Robustness of the result for the primary end point was assessed in a multivariate Cox proportional hazards regression model controlling for baseline characteristics of age, ABI, cholesterol levels, and systolic pressure as continuous variables, with sex, smoking status (current/previous/never), and quintile of socioeconomic status¹⁰ as categorical variables. All covariates were complete except for cholesterol, for which 20 missing values were imputed using linear regression.

Additional analyses were carried out on predefined subgroups across risk strata: male vs female and above or below the median age of 62 years; and post hoc for ABI cut points of 0.80, 0.85, 0.90; and excluding diabetes. Cox proportional hazards regression was used. A similar comparison was made between participants taking or not taking treatment as previously defined¹⁶ at 5 years of follow-up.

Invitations to ABI screening were sent to 165 795 people aged 50 to 75 years of whom 28 980 were screened (response rate 15.9% men, 18.9% women). Those screened had similar distributions of age, sex, and socioeconomic status to the target population.¹⁷ Of those screened, 4914 (17%) had a low ABI (≤0.95), 641 of whom met the exclusion criteria and 923 declined to participate in the trial. The remaining 3350 were randomly assigned to the aspirin (n = 1675) or placebo (n = 1675) groups (Figure 1). Participants were followed up for a mean (SD) of 8.2 (1.6) years. Ten participants (0.3%) were censored because they either emigrated or could not be contacted.

Baseline characteristics were similar in the aspirin and placebo groups (Table 1): 72% of participants were women; 27% were in the most deprived socioeconomic category; mean (SD) ABI was 0.86 (0.09); mean (SD) serum cholesterol 238.5 (41.9) mg/dL (to convert cholesterol from mg/dL to mmol/L, multiply by 0.0259); 33% were current smokers; and 3% reported that they had diabetes mellitus. Lipid-lowering therapy use (including statins) was reported by 4% at baseline, increasing to 25% at 5 years (26.0% aspirin group; 23.9% placebo group). Physician- or self-prescribed aspirin at 5 years was reported by 16.4% in the aspirin group and by 15.0% in the placebo group, including those who had had a cardiovascular event.

A primary end point event occurred in 357 participants (13.5 per 1000 person-years, 95% CI, 12.2-15.0). Figure 2 shows that no statistically significant difference was found in event rates over time between the groups (aspirin, 13.7; 95% CI, 11.8-15.9 vs placebo, 13.3; 95% CI, 11.4-15.4, events per 1000 person-years; HR, 1.03; 95% CI, 0.84-1.27). No statistically significant differences were found in individual primary end point events between the groups (Table 2).

The secondary end point of all vascular events occurred in 578 participants (22.8 per 1000 person-years, 95% CI, 21.0-24.8). The occurrence of these events did not differ over time between groups (aspirin, 22.8; 95% CI, 20.2-25.6 vs placebo, 22.9; 95% CI, 20.3-25.7 events per 1000 person-years; HR, 1.00; 95% CI, 0.85-1.17; eFigure). Table 2 shows that there was no evidence suggesting an effect of aspirin on angina, intermittent claudication, or transient ischemic attack.

All-cause mortality did not differ significantly between the aspirin group, 176 deaths (12.8; 95% CI, 11.0-14.8 per 1000 person-years) and the placebo group, 186 deaths (13.5; 95% CI, 11.6-15.6 per 1000 person-years; HR, 0.95; 95% CI, 0.77-1.16).

Adjustment for baseline characteristics did not affect the HR for the primary end point (HR, 1.00; 95% CI, 0.81-1.23). Comparisons of the primary end point by sex, age, and ABI cutpoints are shown in Table 3. Event rates were higher in older men and in those with a lower ABI cut point, but no significant differences were found in the HRs. Excluding participants with diabetes, the event rate per 1000 person-years in the aspirin group was 13.7 (95% CI, 11.7-15.9) and in the placebo group was 12.8 (95% CI, 10.9-14.9; HR, 1.07; 95% CI, 0.87-1.33).

Participants adhered to the study medication for 60% of person-years of follow-up (up to death, vascular or adverse event, or end of trial). More than 85% of the participants took the medication for at least 6 months. The effect of aspirin vs placebo on the primary end point was not significantly different (P = .31) between those who were still taking their medication at 5 years (aspirin, 10.9; 95% CI, 8.6-13.6 vs placebo, 11.9; 95% CI, 9.5-14.6 events per 1000 person-years; HR, 0.92; 95% CI, 0.68-1.24) and those not taking their medication (aspirin, 17.2; 95% CI, 14.0-20.8 vs placebo, 15.1; 95% CI, 12.2-18.6 events per 1000 person-years; HR, 1.14; 95% CI, 0.86-1.51). At the 5-year follow-up, those taking their medication did so for 88% of person-years throughout the entire trial, whereas those classified as not taking their medication did so for 24% of person-years.

Table 4 shows that 34 participants (2.0%) in the aspirin group had an initial event of a major hemorrhage compared with 20 (1.2%) in the placebo group (aspirin, 2.5; 95% CI; 1.7-3.5 vs placebo, 1.5; 95% CI, 0.9-2.3 per 1000 person-years; HR, 1.71; 95% CI, 0.99-2.97). Of these events, 11 in the aspirin group and 7 in the placebo group were intracranial including 3 fatal subarachnoid or subdural hemorrhages in the aspirin group compared with none in the placebo group. Differences in total number of adverse events between groups were similar but less marked than for initial events.

To our knowledge, this trial is the first to report on the effectiveness of aspirin in reducing major cardiovascular and cerebrovascular events in individuals from the general population who were free of clinical cardiovascular disease but at higher risk as identified by ABI screening. Quiz Ref IDWe did not observe a reduction in major vascular events or in the secondary vascular end point, which also included angina, intermittent claudication, and transient ischemic attack.

A number of factors may have contributed to these findings. Approximately, 70% of trial participants were women because men responded to the screening invitation at a slightly lower rate (following self-exclusion for clinical cardiovascular disease) and because of a lower population distribution of the ABI in women than in men^17- 18 leading to higher levels of trial eligibility among women. However, both the relative risk of a cardiovascular event for a given level of ABI² and the proportional risk reduction for vascular events in primary prevention trials have been shown to be comparable in men and women,⁹ suggesting that sex imbalance in this trial was unlikely to have affected greatly the results.

The rate of major cardiovascular and cerebrovascular events was lower than expected, consistent with recent decreases in event rates in Scotland as a whole. However, given that the overall 10-year event rate (equivalent to 12.7%) was within the accepted medium cardiovascular risk category (10%-20%), this was unlikely to explain fully the lack of an effect of aspirin. Also, the inclusion of participants with diabetes mellitus could have influenced our results (aspirin has been shown to have no effect on vascular events in asymptomatic patients with diabetes,^19- 21 and diabetes may inhibit the antiplatelet effect of aspirin²²), but only 3% of trial participants reported having diabetes at baseline and analysis excluding these individuals did not meaningfully change the results.

A likely major effect on our findings was the limited power of the study to detect a small effect of aspirin. A meta-analysis of secondary prevention trials of aspirin use among patients with symptomatic peripheral artery disease found a 25% proportional risk reduction in cardiovascular events, although this result was not statistically significant.²³ In the recent individual participant data meta-analysis by the Anti-Thrombotic Trialists' Collaboration,⁹ aspirin led to a smaller proportional risk reduction of 19% in serious vascular events in the secondary prevention trials and only a 12% reduction in the primary prevention trials.^24- 29 A formal comparison however found no significant difference in effectiveness between the primary and secondary prevention trials (test for heterogeneity P ≥ .1).⁹ In light of these findings, our initial estimate of a 25% reduction in event rate in this trial, based on evidence at that time, was likely to have been an overestimate. Statistical power was greater for the secondary end point analysis. However, lack of a statistically significant effect of aspirin for this outcome could in part be explained if aspirin has a lesser effect on chronic atherosclerotic narrowing (typical of angina and intermittent claudication) than on acute thrombotic or embolic events.

Although we found no statistically significant effect of aspirin on major events, the HRs and 95% CIs did not rule out the possibility of a risk reduction of up to 16% (or an increased risk of up to 27%). However, extrapolating from the numbers screened for participation in our trial, a risk reduction of this order means that between 500 and 600 people from the general population would need to be screened and prescribed aspirin to prevent a single major cardiovascular event over an 8-year period. It is highly questionable whether the additional resources required to screen and treat such a large number of people to prevent only a small number of events would be justified or indeed whether a larger trial to try and demonstrate such a small effect should be considered. In addition, any effect of aspirin on cardiovascular events needs to be balanced against the potential for harm. Quiz Ref IDAlthough numbers were small, the trial results suggested an increased incidence of major hemorrhage and gastrointestinal ulcer, although not severe anemia, in the aspirin group, and more participants in the aspirin group than in the placebo group had fatal intracranial adverse events. These findings are compatible with previous primary prevention trials in which a collective 50% increase was found in major gastrointestinal and other extracranial bleeds.⁹ Such adverse effects are of particular concern in the context of screening the healthy general population and when the absolute effects of aspirin in reducing major vascular events may be small.

In our trial, study medication was taken for 60% of trial person-years, with more than 85% of participants taking aspirin for 6 months or more. Similar findings were reported in a comparable study of antithrombotic agents involving high-risk men recruited from UK general practices.²⁶ Also, in a trial of aspirin on Italian general practitioner attendees with one or more cardiovascular risk factor, nearly 20% stopped medication within a year.²⁸ Although no significant interaction of the effect of aspirin on the primary end point was found between those taking or not taking the treatment medication, this result was based on small numbers, and stopping medication would undoubtedly have influenced the trial results. However, given that the trial population was typical of the local population and was followed up closely, it is doubtful that the proportion receiving treatment could be improved if an ABI screening was ever implemented in the community. The likely effect of aspirin in a general population screened for ABI must therefore be taken to include the effect of moderate adherence to treatment, as did the results of this trial. The reasons for stopping medication in the trial were much as expected, including taking aspirin for noncardiovascular reasons, known adverse effects of aspirin, other illnesses unrelated to aspirin, and no longer wanting to take the study medication.¹⁶ These factors may have been compounded by a lack of understanding of the ABI as a measure of risk.³⁰

An ABI of 0.90 or less is often used as the cut point for identifying individuals at high risk of peripheral atherosclerosis. However, the risk of future vascular events is a reverse J-shaped curve with a level of 1.1 to 1.4 being considered normal.² In ABI screening the cut point selected is a trade-off between level of risk and yield of individuals who might potentially benefit from treatment. We considered that a cut point of 0.95 was preferred given an estimated absolute increase in yield of 6% of the screened population. This choice of cut point could theoretically have affected the result of the trial. However, the major event rate was not very much higher among those with an ABI of 0.90 or less (15.6 per 1000 person-years, 95% CI, 13.8-17.6) than among those with an ABI of 0.95 or less (13.5 per 1000 person-years, 95% CI, 12.2-15.0). Also, no substantial differences in the effects of aspirin on the primary end point were found for different ABI cut points, although numbers were small for such an analysis.

Quiz Ref IDAlthough this trial was not of screening per se, the results indicate that using the ABI in the community to screen individuals free of cardiovascular disease for an ABI of 0.95 or less is unlikely to be beneficial if aspirin is the intervention of choice. However, given the increased level of risk among those with a low ABI, the use of alternative therapies, such as statins or more potent antiplatelet agents without attendant hemorrhagic risks may usefully be considered.^31- 33 In addition, given that the trial was a pragmatic trial in the context of ABI screening in the general population, aspirin might still have a net beneficial effect on patients with a low ABI, elevated risk factors, and a greater incentive to continue taking medication. Further trials of the management of patients with a low ABI identified in clinical practice as part of cardiovascular risk assessment programs would be justified.