Chronic kidney disease (CKD) is common in the general community. Population-based studies have shown that 10% to 15% of the adult population has CKD (1- 3), and this proportion is increasing (4). Most people with CKD have relatively mild disease and are unlikely to ever require dialysis or a kidney transplant; however, even early CKD confers an increased risk of cardiovascular events and death (5- 11). Recent work has demonstrated that lipid lowering (12) and blood pressure lowering (13- 14) are both effective in reducing the risk of cardiovascular events in people with early CKD; however, less is known about the balance of risks and benefits associated with other potential preventative therapies.

In trials of secondary prevention of cardiovascular disease, treatment with aspirin clearly delivers a net benefit of harm reduction over harm caused. The Antithrombotic Trialists' Collaborative Group meta-analysis of individual participant data confirmed that aspirin reduces the yearly risk of major cardiovascular outcomes (strokes and coronary events) by about 15 events per 1,000 patient-years (15). However, the same analysis demonstrated that, overall, among participants with no history of previous cardiovascular events (primary prevention), the absolute benefit of aspirin in preventing 0.6 events per 1,000 patient-years is comparable to the number of major gastrointestinal and extracranial bleeds caused (0.3 events per 1,000 patient-years). Analyses of the effect of aspirin in a number of defined subgroups failed to identify a primary prevention patient group that benefited from aspirin.

Patients with CKD have high cardiovascular risk; therefore, the absolute benefits of aspirin might be greater for them than for people with normal kidney function. However, patients with CKD have abnormal platelet function, leaving them at potentially increased hemorrhagic risk when treated with anticoagulants, including antiplatelet agents (16). Substantial uncertainty persists regarding the balance between the risks and benefits associated with antiplatelet agents for patients with CKD. Consistent with this uncertainty, patients with CKD (17) and end-stage kidney disease (18) have been shown to be less likely to be prescribed aspirin after an acute myocardial infarction.

The HOT (Hypertension Optimal Treatment) trial, one of the largest individual primary prevention trials, randomly allocated 18,790 participants 50 to 80 years of age with elevated diastolic blood pressure to aspirin or matching placebo for an average of 3.8 years (19). It was a primary prevention study with <2% of participants having a prior history of myocardial infarction. In the HOT study overall, a significant 15% reduction in major cardiovascular events (1.6 events per 1,000 patient-years) was observed, but this needs to be weighed against a significant 80% increase in the risk of major nonfatal bleeding (1.4 events per 1,000 patient-years). Analyses demonstrated there was no interaction between the blood pressure–lowering effect and aspirin effects (19- 20). Subsequent analyses of the HOT study population used serum creatinine thresholds to explore how impairment of renal function influences the effect of aspirin (5- 6). A trend to increased benefit from aspirin was demonstrated for patients with an elevated serum creatinine. However, the balance of benefits and harms associated with aspirin usage in CKD have not been previously reported, nor has the level of renal function below which benefits may overcome harms been established.

Since these subgroup analyses, estimated glomerular filtration rate (eGFR) using the Modification of Diet in Renal Disease equation (21) has become standard in the staging of CKD (22). This analysis, therefore, investigates whether the balance of benefits and harms of aspirin therapy in HOT study participants is influenced by kidney function evaluated continuously and categorically by CKD stage based on eGFR levels.

The HOT study design has been described in detail elsewhere (19,23). In brief, 18,790 participants age 50 to 80 years (mean 61.3 years) from 26 countries in Europe, North and South America, and Asia, and with a diastolic blood pressure between 100 and 115 mm Hg, were randomly assigned to 2 interventions in a factorial design: aspirin 75 mg daily (n = 9,399) or matching placebo (n = 9,391), and 1 of 3 diastolic blood pressure targets (≤90, ≤85, or ≤80 mm Hg). The blood pressure targets were randomly assigned in an open-label fashion. There was no exclusion on the basis of renal function. The conduct of the study was overseen by a steering committee and approved by national and local ethics committees and regulatory bodies at all participating centers. An independent safety committee regularly reviewed safety data. All patients provided written informed consent.

The current analysis included 18,597 participants assigned to aspirin or placebo for whom baseline serum creatinine values were available. Analyses of the change in renal function were performed on participants for whom serum creatinine values were also available at study end. Glomerular filtration rate was estimated using the 4-variable Modification of Diet in Renal Disease equation (21) and categorized using Kidney Disease Outcomes Quality Initiative (KDOQI) stages (24).

The primary end point of this study was a composite of major cardiovascular events consisting of myocardial infarction, stroke, and death due to cardiovascular disease. Secondary end points included myocardial infarction (nonfatal myocardial infarction and death due to coronary heart disease [including sudden death]), stroke (fatal and nonfatal stroke), cardiovascular mortality, total mortality, death due to kidney failure, and change in eGFR. Cardiovascular and mortality events were reviewed and validated by an independent clinical event committee. Secondary end points also included investigator-reported major hemorrhage (fatal, life-threatening, disabling, or requiring hospital admission) and minor hemorrhage (all other reported bleeding events). The first event of each relevant outcome type was included for analysis.

Participants were recruited from October 1992 until April 1994, with follow-up concluding in August 1997, resulting in an average follow-up period of 3.8 years (range 3.3 to 4.9 years) (19).

The risk estimates for each outcome associated with eGFR at baseline were estimated using a Cox proportional hazards regression model after adjusting for potentially confounding baseline covariates including age, sex, systolic blood pressure, history of diabetes mellitus, history of cardiovascular disease, total cholesterol, body mass index, and smoking. The variances of each risk estimate were calculated using the floating absolute risk method (25- 26). The regression lines for the risk estimates according to eGFR at baseline were fitted using linear regression analysis with inverse variance weighting (27).

The effect of randomization to aspirin was assessed according to baseline eGFR categories of ≥60, 45 to 59 (KDOQI CKD stage 3a), and <45 ml/min/1.73 m² (KDOQI stage 3b, 4, and 5). The hazard ratio (HR) and 95% confidence interval (CI) associated with active treatment for each end point was estimated using a univariate Cox proportional hazards regression model stratified by eGFR levels at baseline. The presence of heterogeneity in the treatment effect across eGFR categories was assessed by adding an interaction term to the relevant Cox model.

To assess whether there was any threshold level of eGFR, below which the effect of treatment changed, the risk estimate was investigated by fitting univariate Cox proportional hazards model below an eGFR threshold, which was progressively changed using 3 ml/min/1.73 m² increments.

The absolute effect of randomization to aspirin was calculated as the number of people in whom events were prevented or caused per 1,000 patients treated for 3.8 years for the overall study population and for categories of kidney function.

The study population, consisting of 18,597 of 18,790 (99.0%) randomized patients with serum creatinine data available, had a median eGFR of 73 ml/min/1.73 m² (interquartile range 63 to 84 ml/min/1.73 m²). Of these, 14,978 (80.5%) had an eGFR ≥60 ml/min/1.73 m², 3,083 (16.6%) had an eGFR of 45 to 59 ml/min/1.73 m², and 536 (2.9%) an eGFR of <45 ml/min/1.73 m² (Table 1). Only 9 patients (0.05%) had an eGFR of <15 ml/min/1.73 m².

A total of 671 people experienced at least 1 major cardiovascular event. Strokes were experienced by 289 participants, and myocardial infarctions were experienced by 349. There were 582 deaths from any cause, including 268 deaths from cardiovascular causes and 8 from renal failure (3 in the aspirin group and 5 in the placebo group). There were 15 fatal bleeding events (7 in the aspirin group and 8 in the placebo group), 187 nonfatal major bleeding events, and 208 minor bleeding events.

Patients with lower eGFR levels experienced greater rates of cardiovascular events, bleeding events, and death: major cardiovascular event rates were 32.2, 46.4, and 80.2 per 1,000 patients with eGFR ≥60, 45 to 59, and <45 ml/min/1.73 m², respectively; myocardial infarction rates were 17.2, 22.7, and 39.2 per 1,000 patients; stroke rates were 13.3, 23.4, and 31.7 per 1,000 patients; cardiovascular mortality rates were 12.6, 18.5, and 42.9 per 1,000 patients; major bleeding event rates were 10.8, 7.8, and 28.0 per 1,000 patients; and total mortality rates were 29.1, 32.8, and 84.0 per 1,000 patients. Event rates were increased by 70% to 100% for every halving of eGFR: major cardiovascular events (HR: 1.84, 95% CI: 1.37 to 2.46), myocardial infarctions (HR: 1.84, 95% CI: 1.09 to 3.11), stroke (HR: 1.87, 95% CI: 1.28 to 2.73), cardiovascular mortality (HR: 1.99, 95% CI: 1.32 to 2.99), bleeding events (HR: 1.77, 95% CI: 1.09 to 2.86), and total mortality (HR: 1.69, 95% CI: 0.82 to 3.49) (Figure 1).

As previously reported (19), aspirin significantly reduced the risk of major cardiovascular events for the overall study population for whom creatinine measurements were available (event rates for active and placebo groups were 3.32% and 3.90%, respectively; HR: 0.85, 95% CI: 0.73 to 0.98). The benefit provided by aspirin was significantly greater for subjects with low eGFR: risk reductions of 9% (HR: 0.91, 95% CI: 0.76 to 1.09), 15% (HR: 0.85, 95% CI: 0.61 to 1.17), and 66% (HR: 0.34, 95% CI: 0.17 to 0.67) were observed for patients with eGFR ≥60, 45 to 60, and <45 ml/min/1.73 m², respectively (p for interaction = 0.03) (Figure 1). Event rates for subjects treated with aspirin fell from 3.38% to 3.10% for those with eGFR ≥60 ml/min/1.73 m², from 5.01% to 4.26% for those with eGFR 45 to 60 ml/min/1.73 m², and from 11.76% to 4.17% for those with eGFR <45 ml/min/1.73 m² over the mean 3.8 follow-up years in the study. There was no interaction between assignment to different diastolic blood pressure targets and assignment to aspirin (all p values for interaction > 0.2; data not shown).

The protection afforded by aspirin for myocardial infarction increased as kidney function declined (Figure 2), although the interaction was of borderline statistical significance (p = 0.08). Subjects with the highest eGFR (≥60 ml/min/1.73 m²) had a borderline significant risk reduction of 22% (HR: 0.78, 95% CI: 0.61 to 1.00), subjects with an eGFR of 45 to 59 ml/min/1.73 m² had a risk reduction of 36% (HR: 0.64, 95% CI: 0.39 to 1.03), and subjects with eGFR <45 ml/min/1.73 m² had a risk reduction of 69% (HR: 0.31, 95% CI: 0.11 to 0.85). Event rates for subjects treated with aspirin fell from 1.93% to 1.52% for those with eGFR ≥60 ml/min/1.73 m², from 2.76% to 1.77% for those with eGFR 45 to 60 ml/min/1.73 m², and from 5.88% to 1.89% for those with eGFR <45 ml/min/1.73 m² over the mean 3.8 follow-up years in the study.

There was no significant benefit in the total study population from aspirin therapy for total mortality, cardiovascular mortality, or stroke (Figure 2). However, aspirin did confer significant protection for subjects with an eGFR <45 ml/min/1.73 m² for whom total mortality was reduced by roughly one-half (event rates for active and placebo groups 11.03% and 5.68%, respectively; HR: 0.51, 95% CI: 0.27 to 0.94), cardiovascular mortality by nearly two-thirds (event rates for active and placebo groups 6.25% and 2.27%, respectively; HR: 0.36, 95% CI: 0.14 to 0.90), and stroke by nearly four-fifths (event rates for active and placebo groups 5.15% and 1.14%, respectively; HR: 0.21, 95% CI: 0.06 to 0.75). The benefit for cardiovascular mortality and total mortality was significantly greater for patients with reduced kidney function than for subjects with normal kidney function (p for both interactions = 0.04) and was nearly significantly greater for stroke (p for interaction = 0.06).

We performed sensitivity analyses to identify any eGFR threshold level below which the benefit associated with aspirin therapy changed and to confirm that our analyses did not appear vulnerable to adjustments in the cut-off between the eGFR categories. The benefit from aspirin therapy progressively, but not linearly, increased (declining risk ratio among subjects randomly assigned to aspirin therapy) as eGFR declined for all end points (Figure 3). However, the risk reduction of aspirin therapy for cardiovascular mortality, total mortality, and stroke became large and significant when baseline eGFR was <45 ml/min/1.73 m².

In the overall study population, aspirin increased the risk of major bleeding by 61% (HR: 1.61, 95% CI: 1.21 to 2.14) (Figure 4). The risk of major bleeding associated with aspirin was nonsignificantly greater with categories of lower eGFR (HR: 2.81, 95% CI: 0.90 to 8.84 for eGFR <45 ml/min/1.73 m²; HR: 1.70, 95% CI: 0.74 to 3.88 for eGFR 45 to 59 ml/min/1.73 m²; and HR: 1.52, 95% CI: 1.11 to 2.08 for eGFR ≥60 ml/min/1.73 m²; p trend = 0.30). There were 15 fatal bleeds in the study, 7 among subjects assigned aspirin therapy and 8 among subjects assigned placebo. All fatal bleeds were in subjects with an eGFR ≥60 ml/min/1.73 m². There was a trend toward an increased risk of any bleeding as categories of eGFR declined, although the absolute numbers of events in subjects with reduced kidney function were few (p for interaction = 0.08). There was no interaction between assignment to different diastolic blood pressure targets and assignment to aspirin (all p values for interaction > 0.2; data not shown).

Overall, 6 major cardiovascular events will be prevented for every 1,000 participants treated for 3.8 years, while there will be 6 major bleeds and 6 minor bleeds (Table 2). Both the benefits and risks increase as kidney function declines, with the net benefit appearing to increase. For every 1,000 people with eGFR <45 ml/min/1.73 m² treated with aspirin for 3.8 years, 76 people will avoid a major cardiovascular event, and 40 myocardial infarctions, 40 strokes, 40 cardiovascular deaths, and 54 all-cause deaths will be prevented. Conversely, 27 major bleeding episodes and 12 minor bleeding episodes would be caused by aspirin therapy for persons with an eGFR <45 ml/min/1.73 m².

Aspirin therapy did not affect renal function in the overall study population nor within any eGFR category (Table 3).

Safe treatments to reduce the high risk of cardiovascular disease for patients with CKD are urgently required. In this analysis, we confirm that aspirin therapy prevented significantly more cardiovascular events, cardiovascular deaths, and all-cause deaths in patients with CKD than in subjects with normal kidney function. Aspirin therapy had no detrimental effect on renal function. Although the absolute risk of bleeding was greater for subjects with CKD, the overall cardiovascular benefits appear to outweigh bleeding risks. These results suggest that primary prevention with aspirin reduces the burden of cardiovascular disease and has an overall net benefit among high-risk patients with CKD.

Our analysis confirms previous findings in the HOT study population and in other studies that cardiovascular risk increases with declining CKD stage (5- 7,28- 29). Overall, the HOT study participants were at low risk, with a 5-year rate of myocardial infarction of 2.9% and of major cardiovascular events of 5.1%, observed in the control arm. However, the HOT study participants with stage 3b CKD randomly assigned to placebo had a 5-year myocardial infarction rate of 7.7%, with a 5-year major cardiovascular event rate of 15.5%. These high event rates underscore the need for a clear understanding of the risks and benefits of potentially effective preventative therapies.

Aspirin appears to produce greater absolute benefits for patients with CKD. The explanation for this greater benefit lies partly in the high baseline risk of these patients, translating a similar proportional benefit into a greater absolute benefit. In addition, the current analysis demonstrates greater proportional benefits with progressively lower eGFR. The explanation for this difference in proportional benefit is unclear. Patients with advanced CKD are known to have abnormal platelet function and evidence of a predisposition to both thrombotic and bleeding events. Patients with all stages of CKD (30- 33) have higher rates of thromboembolism than the general population. Conversely, patients with kidney disease appear to have an increased bleeding risk, with evidence in advanced kidney disease of decreased platelet aggregation and of a range of platelet abnormalities (34- 37). Observational and randomized clinical studies in dialysis patients sometimes (38), but not always, report increased bleeding rates with antiplatelet therapy (39- 41), although some of these studies excluded those at high bleeding risk.

There are few other randomized trials describing the effects of antiplatelet therapy involving substantial numbers of subjects with CKD. The Antithrombotic Trialists' Collaborative Group reported the results of an individual patient data meta-analysis of antiplatelet therapy that included 99 cardiovascular events in 2,632 hemodialysis patients. They found a 41% odds reduction (SE 16%) in the risk of cardiovascular events among hemodialysis patients (40), compared with a 22% odds reduction (SE 2%) seen in the overall study population, although the difference was not statistically significant. The efficacy and harm of aspirin and other antiplatelet agents may not be homogeneous. Post-hoc analyses of the effect of additional clopidogrel over standard therapy according to GFR categories has shown a benefit consistent with that in the population with normal renal function in some (42) but not all (29) studies. For patients with diabetes and albuminuria, the addition of clopidogrel to aspirin was associated with increased mortality not seen in nonalbuminuric diabetic patients or nondiabetic patients in a post-hoc analysis (43).

This analysis has some important limitations. Only 2.9% of the study population had an eGFR <45 ml/min/1.73 m², limiting our power to estimate bleeding risk in this group. The small number of participants with CKD stage 4 and above (98 had an eGFR <30 ml/min/1.73 m²) mean the findings cannot be extrapolated to patients with severe CKD or end-stage kidney disease. In addition, the HOT study, as in many randomized trials of aspirin administration, did not report bleeding episodes with the same precision as cardiovascular outcomes, and bleeding episodes were not validated by an expert committee. Furthermore, this is a post-hoc analysis of a trial that was not designed (or powered) to examine the effects of aspirin according to categories of kidney function. Finally, the participants in this trial were at increased cardiovascular risk due to the blood pressure–based entry criteria, meaning extrapolation to persons with normal blood pressure levels is not possible.

Our results indicate that CKD predicts increased cardiovascular risk and greater net benefit with aspirin therapy. This finding reinforces calls to consider CKD when making decisions regarding treatment to mitigate cardiovascular risk. These results suggest that aspirin might be used more widely as primary prevention for high-risk patients with CKD.

The authors gratefully acknowledge the kind assistance of Ingrid Warnold, PhD, AstraZeneca, Mölndal, Sweden.