One-year Results of a Factorial Randomized Trial of Aspirin versus Placebo and Clonidine versus Placebo in Patients Having Noncardiac Surgery

Intraoperative mortality is now rare,^1,2  but postoperative mortality and morbidity remain common.^3,4  Major bleeding and myocardial infarction are leading attributable causes of mortality in the 30 days after noncardiac surgery.⁵ 

Currently, there is no known safe way to prevent postoperative myocardial infarction.^6,7  β-Blockers prevent myocardial infarction after noncardiac surgery but increase the risk of stroke and death.⁶  Avoiding nitrous oxide is ineffective.⁷  Both clonidine and aspirin were attractive candidates to reduce perioperative myocardial infarction because they either reduce sympathetic nervous system activation and lower heart rate, or block platelet activation and clot formation. However, the large randomized Perioperative Ischemic Evaluation-2 (POISE-2) trial failed to confirm these hypotheses, as neither aspirin⁹  nor clonidine⁸  prevented myocardial infarction or death within 30 days of randomization. We did, however, demonstrate that a subgroup of patients, those with previous percutaneous coronary interventions, benefit from perioperative aspirin through a reduction in nonfatal myocardial infarction: 30-day hazard ratio for death or myocardial infarction 0.50 [95% CI, 0.26 to 0.95].⁸ 

Both aspirin and clonidine caused adverse effects. Clonidine provoked clinically important bradycardia and hypotension,⁹  whereas aspirin increased the risk of major bleeding.¹⁰  Both hypotension^11,12  and bleeding^13,14  are related to the risk of long-term complications, but it remains unclear whether perioperative treatment with either aspirin or clonidine causes long-term harm or benefit. We therefore evaluated the effect of aspirin or placebo and clonidine on the 1-yr risk of a composite outcome of death or nonfatal myocardial infarction in POISE-2 patients. Secondary and tertiary outcomes that were evaluated at 30 days were also considered at 1 yr.

Thirty-day pain was not assessed in the trial’s analyses of 30-day postoperative outcomes. Therefore, assessing 1-yr postoperative pain adds a new longer-term outcome that was not assessed in the short-term in the POISE-2 analyses of 30-day outcomes. Chronic postoperative pain after surgery is defined by pain that persists longer than natural healing and is not associated with other apparent causes. Typically, incision pain lasting more than 3 to 6 months is considered persistent. Chronic postoperative pain reduces patient function, and associated long-term opioid use can lead to abuse and addiction. It is therefore reasonable to consider preventive strategies. Aspirin is an effective analgesic and anti-inflammatory drug used for acute pain. Clonidine has been used in chronic pain and acute pain with varying degree of success.^15,16  Aspirin has yet to be tested for prevention of persistent pain, and there are only a few inconsistent studies of clonidine. We therefore also evaluated chronic pain 1 yr after surgery.

The POISE-2 Trial was an international, randomized, controlled trial with a 2-by-2 factorial design to assess the individual efficacy and safety of both clonidine and aspirin among patients having noncardiac surgery. We previously published detailed information about ethics approvals, registration, trial objectives, and methods.¹⁷  In brief, we recruited 10,010 patients from 135 hospitals in 23 countries between July 2010 and December 2013. Eligible patients were at least 45 yr of age, having inpatient noncardiac surgery, and had either established vascular disease or multiple risk factors. The main exclusion criteria were systolic blood pressure less than 105 mm Hg, heart rate less than 55 beats per minute, consumption of aspirin within 72 h before surgery, or having a recent episode of major bleeding (Supplemental Digital Content, table 1, https://links.lww.com/ALN/C227).

Patients were randomized through a computerized web randomization system using block randomization (n = 4) stratified by center and whether they were taking aspirin chronically (i.e., the continuation stratum if they were taking aspirin chronically or the initiation stratum if they were not taking aspirin chronically). Using a factorial approach, patients were assigned in a 1:1:1:1 fashion to clonidine/aspirin, clonidine/aspirin placebo, clonidine placebo/aspirin, or clonidine placebo/ aspirin placebo. Patients, health care providers, data collectors, and outcome adjudicators were blinded to treatment allocation.

Centers were encouraged to instruct their patients to hold usual antihypertensive medications on the morning of surgery. Study personnel reviewed participants’ vital signs in the presurgical area, and in consultation with the attending anesthesiologist determined which antihypertensive medications should be given, and in what doses. Patients were given 200 mg of aspirin or aspirin placebo just before surgery and took 100 mg daily for 30 days if they were in the aspirin initiation stratum, or 7 days if they were in the aspirin continuation stratum, with their regular aspirin thereafter. Two to four hours before surgery, patients fulfilling hemodynamic requirements also received 0.2 mg of oral clonidine or placebo and had a transdermal clonidine patch (which releases 0.2 mg per day and has physiologic effects within 24 h)¹⁸  or placebo patch applied to their upper arm or chest until 72 h after surgery.

All sites were trained in person or by video conference before enrollment started. Training included detailed definitions of outcomes, and case-report forms included specific definitions for each field. Major outcomes were centrally adjudicated to improve consistency.

Blood pressure and heart rate were measured 1 h after starting study drugs and every 4 h for the first 96 h after surgery. If a patient developed clinically important hypotension or bradycardia that did not respond to initial treatment (e.g., fluid bolus), then study personnel encouraged clinicians to remove the patient’s clonidine study patch. All decisions were left to the attending physicians, including the discontinuation of study drugs.

Blood was drawn for a troponin measurement 6 to 12 h after surgery and daily for the next 3 days. Electrocardiography was performed when an elevated cardiac biomarker was detected. Research personnel followed patients until 1 yr after randomization and submitted the case report forms and supporting event documentation directly to the web-based data management system. The primary, secondary, and tertiary outcomes were documented 1 yr after randomization. The primary outcome was a composite of death or nonfatal myocardial infarction, with details previously reported.^9,10  Outcomes are defined in Supplemental Digital Content, table 2 (https://links.lww.com/ALN/C227). Adjudicators evaluated key outcomes, and their decisions were used in the statistical analyses.

A preplanned secondary outcome was persistent incisional pain. Patients were asked whether they experienced any pain at their surgical site. Those who did were further asked to complete a Modified Brief Pain Inventory¹⁹  and Neuropathic Pain Inventory short form.²⁰ 

A statistical analysis plan was finalized and date-stamped in the investigators’ files before accessing data. The plan defined primary, secondary, and tertiary outcomes (Supplemental Digital Content, https://links.lww.com/ALN/C228). Statistical analyses were performed using SAS version 9.4 according to a prespecified analysis plan. Patients were analyzed according to the intention-to-treat principle, and those lost to follow-up were censored on the last day that outcome status was known. Outcomes were analyzed using Cox proportional hazards models, stratified by the alternative factorial allocation and whether participants took aspirin chronically.

For the primary outcome, we performed prespecified subgroup analyses based on neuraxial anesthesia, vascular surgery, vascular disease, and the number of revised cardiac risk index criteria.²¹  We also did a subgroup analysis based on the aspirin stratum within the aspirin factorial. In the statistical analysis plan, we prespecified the expected direction of effects in the subgroups. We performed one post hoc subgroup analysis.

When POISE-2 was designed, we did not plan a percutaneous coronary intervention subgroup analysis because we did not anticipate that physicians would enroll patients with a history of percutaneous coronary intervention. However, given that investigators randomized 470 patients with previous percutaneous coronary intervention, and given the ongoing uncertainty regarding the effects of antiplatelet therapy in these patients, we also undertook a percutaneous coronary intervention subgroup analysis for the aspirin factorial. We used Cox proportional hazards models that incorporated tests of interaction for all subgroup analyses. A P value less than 0.05 was prespecified to indicate statistical significance.

The Population Health Research Institute was the Sponsor and Coordinating Center for POISE-2 and was responsible for the randomization scheme, database, data validation, analyses, and trial center coordination. No POISE-2 funding source had a role in the trial design, conduct, data collection, analyses, or manuscript preparation or review. The Operations Committee designed the trial, prespecified the statistical analysis plan, and vouched for the data and analyses.

A total of 10,010 patients were randomized. Overall, more than 80% of the patients took at least 80% of the doses of the aspirin study drug, more than 97% of patients took the clonidine study drug before surgery, and more than 90% of patients kept the transdermal study patch on for at least 80% of the targeted duration of application.^9,10  Supplemental Digital Content figure 1 (https://links.lww.com/ALN/C229) and Supplemental Digital Content figure 2 (https://links.lww.com/ALN/C230) show the trial diagram for each treatment; only 47 (0.5%) patients did not complete the 1-yr follow-up and their data were included until the time of censoring. Baseline patient characteristics were similar in each treatment group, as shown in table 1. The mean age of patients was 69 yr, and 53% were male.

Aspirin had no effect on the primary outcome at 1 yr: hazard ratio = 1.00 (95% CI, 0.89–1.12; fig. 1). Aspirin had no effect on the incidence of new or recurrent cancer or chronic incisional pain. Aspirin had no effect on secondary or tertiary outcomes, as shown in table 2.

Clonidine had no significant effect on the primary outcome at 1 yr: hazard ratio = 1.07 (95% CI, 0.96–1.20, P = 0.218; fig. 2). Clonidine had no effect on the incidence of new or recurrent cancer or chronic incisional pain. Clonidine significantly increased the risk of congestive heart failure: 1.8 vs. 1.1%, hazard ratio = 1.54 (95% CI, 1.10–2.15, P = 0.012) but none of the other secondary or tertiary outcomes (table 3).

Figures 3 and 4 present subgroup analyses for the 1-yr primary outcome. Three of the prespecified subgroup effects were nominally significant; however, none was in the direction prespecified in the analysis plan. Specifically, aspirin reduced risk in patients who had a previous percutaneous coronary intervention, clonidine worsened risk in patients who did not have a neuraxial block, and clonidine worsened risk in patients with Revised Cardiac Risk scores of 3. The previously reported reduction in the 30-day risk of death and nonfatal myocardial infarction in the subgroup of patients who previously had percutaneous coronary intervention persisted at 1 yr: hazard ratio = 0.58 (95% CI, 0.35–0.95) in those with percutaneous coronary intervention versus hazard ratio = 1.03 (95% CI, 0.91–1.16) in those without previous percutaneous coronary intervention; P for interaction 0.033 (fig. 5 and Supplemental Digital Content, table 3, https://links.lww.com/ALN/C227). Reduced risk in the percutaneous coronary intervention subgroup was mainly attributable to a reduction in nonfatal myocardial infarction rather than death.

POISE-2, a large international trial, showed that neither aspirin nor clonidine significantly reduced death or nonfatal myocardial infarction at 30 days.^9,10  The current long-term follow-up of POISE-2 confirmed the absence of benefit or harm of either drug in the overall population in terms of death, nonfatal myocardial infarction, and stroke up to 1 yr after surgery. The 1-yr data also show that neither aspirin nor clonidine has an effect on noncardiovascular outcomes including new-onset cancer, cancer recurrence, or persistent incisional pain. Perioperative aspirin reduced the incidence of composite of death and nonfatal myocardial infarction at 1 yr in patients with previous percutaneous coronary intervention.

It was plausible that clonidine and aspirin would cause long-term harm given that they induced adverse effects around the time of surgery. Specifically, at 30 days, aspirin significantly increased the risk of major bleeding compared with placebo (230 patients [4.6%] vs. 188 patients [3.8%]; hazard ratio = 1.23; 95% CI, 1.01, to 1.49; P = 0.04), whereas clonidine increased the risk of hypotension (2385 patients [47.6%] vs.1854 patients [37.1%]; hazard ratio = 1.32; 95% CI, 1.24 to 1.40; P < 0.001) and nonfatal arrest (0.3% [16 patients] vs. 0.1% [5 patients]; hazard ratio = 3.20; 95% CI, 1.17 to 8.73; P = 0.02).^9,10  Bleeding and hypotension are both associated with organ injury and death.^11–14  Similar rates of adverse outcome events in the aspirin, clonidine, and placebo arms of the trial suggest that the hazards induced by aspirin and clonidine in the immediate perioperative period did not translate into a measurable long-term harm, with the possible exception of congestive heart failure from clonidine.

The primary 30-day results clearly showed that, in the overall population, aspirin provides no benefit but increases bleeding during the perioperative period.¹⁰  But patients who previously had percutaneous coronary interventions are a special population. Patients who received a bare-metal stent within 6 weeks or placement of a drug-eluting stent within 1 yr were excluded from POISE-2, but 470 patients who had a remote history of percutaneous coronary intervention participated in the trial. Among those patients, we previously reported that aspirin reduced the risk of the composite of death and nonfatal myocardial infarction at 30 days.²²  We now demonstrate that the benefit of aspirin in percutaneous coronary intervention patients persists at 1 yr. These data suggest that patients with a history of percutaneous coronary intervention should continue aspirin around the time of noncardiac surgery except, presumably, in patients with special risk of bleeding or in whom the consequences might be especially severe.

There was also a significant interaction for the percutaneous coronary intervention subgroup on myocardial infarction. However, this analysis was not specified a priori because when POISE-2 was designed we did not anticipate that physicians would randomize patients with a history of percutaneous coronary intervention. An additional concern is that the total number of myocardial infarctions in the percutaneous coronary intervention subgroup was fewer than 70. Large effects (such as the relative reduction of approximately 50% seen here) with small numbers of events typically prove to be overestimates, and often by large amounts. Overall, the credibility of the percutaneous coronary intervention subgroup effects in POISE-2 is only moderate.

A limitation of our analysis is that subgroup effects are often spurious. Indeed, we report two other a priori specified subgroup effects in which the interaction P values reach conventional levels of statistical significance: neuraxial blockade versus no neuraxial blockade and Revised Cardiac Risk grade 3 versus other grades. It nonetheless seems likely that both of these other apparent subgroup effects are spurious. Previous work suggests that neuraxial anesthesia is not protective for cardiovascular outcomes.²³ 

None of the defined secondary outcomes of POISE-2 differed significantly between groups at 1 yr. The tertiary outcome of congestive heart failure was increased at the 1-yr follow-up in patients randomized to clonidine (hazard ratio = 1.54; 95% CI, 1.10 to 2.15, P = 0.012). Although congestive heart failure was a tertiary outcome in this study, and the observed association between clonidine and long-term congestive heart failure may be a false positive association, it is not impossible that the increased risk of nonfatal cardiac arrest associated with clonidine could play a role. Were that the case though, we would have expected a stronger signal at 30 days and not at 1 yr of follow-up. Further investigation may be warranted to determine plausible biologic mechanisms for this putative association.

Our large study shows that neither clonidine nor aspirin ameliorates development of persistent incisional pain. Furthermore, the incidence of chronic postsurgical pain (approximately 5%) was in our patients considerably lower than reported in other postsurgical cohort studies^15,24 —possibly because of variations in pain assessment measures, but more likely because we included a broad range of surgeries rather than selecting operations that are particularly susceptible to persistent pain.

Neither aspirin nor clonidine had an effect on cancer occurrence or recurrence. Previous studies have shown that aspirin may prevent incident cancer during long-term follow-up, but the beneficial effect of aspirin is observed only after two decades of follow-up.²⁵ 

In conclusion, POISE-2 randomized 10,010 noncardiac surgical patients and did not show a significant effect of either aspirin or clonidine on prespecified outcomes, including death, cardiovascular complications, cancer, and chronic incisional pain at 1 yr of follow-up. Perioperative aspirin in patients with previous percutaneous coronary intervention reduced the composite of death and nonfatal myocardial infarction at 1 yr of follow-up, supporting the continued use of perioperative aspirin in this subgroup.

POISE-2 was funded through grants from the Canadian Institutes of Health Research (Ottawa, Canada), the Commonwealth Government of Australia’s National Health and Medical Research Council (Canberra, Australia), the Spanish Ministry of Health and Social Policy (Madrid, Spain), and Boehringer Ingelheim (Ingelheim am Rhein, Germany) who also provided the clonidine study drug. Bayer Pharma AG (Berlin, Germany) provided the aspirin study drug. The funding agencies had no role in study conduct, data analysis, or manuscript preparation. None of the authors has a personal financial interest in this work.

Dr. Kurz reports a financial relationship with Canadian Institute of Health Research (Ottawa, Canada). Dr. Meyhoff reports that in his role as Head of Research, his department received direct and indirect research funding from Boehringer Ingelheim (Ingelheim am Rhein, Germany), Ferring Pharmaceuticals (Saint-Prex, Switzerland), and Merck, Sharp and Dohme (Kenilworth, New Jersey), all outside submitted work. Dr. Xavier reports support from Population Health Research Institute, Hamilton Health Sciences (Hamilton, Ontario, Canada), and financial relationships with AstraZeneca (Cambridge, United Kingdom), Bristol Myers Squibb (New York, New York), Cadila Pharma India (Gujarat, India), and Sanofi (Paris, France). Dr. Amir reports support from Population Health Research Institute. Mr. Torres also reports support from Population Health Research Institute and acting as a speaker for 3M (Maplewood, Minnesota), and Pfizer (New York, New York). Dr. Wang reports support from Hamilton Health Sciences Corporation through Population Health Research Institute. Dr. Paniagua reports an ongoing financial relationship with CSL Behring (King of Prussia, Pennsylvania). Dr. Devereaux reports financial relationships with Abbott Diagnostics (Abbott Park, Illinois), Phillips Healthcare (Amsterdam, The Netherlands), Roche Diagnostics (Rotkreuz, Switzerland), and Siemens (Munich, Germany). The remaining authors declare no competing interests.