Adverse Events After Stopping Clopidogrel in Post–Acute Coronary Syndrome Patients
Insights From a Large Integrated Healthcare Delivery System
Background— A prior study from the Veterans Health Administration found a clustering of cardiovascular events after clopidogrel cessation. We sought to confirm and expand these findings.
Methods and Results— This was a retrospective cohort study of 2017 patients with acute coronary syndrome discharged on clopidogrel from an integrated health care delivery system. Rates of all-cause mortality or acute myocardial infarction (MI) within 1 year after stopping clopidogrel were assessed among patients who did not have an event before stopping clopidogrel. Death/MI occurred in 4.3% (n=71) of patients. The rates of death/MI were 3.07, 1.62, 0.70, and 0.95 per 10 000 patient-days for the time intervals of 0 to 90, 91 to 180, 181 to 270, and 271 to 360 days after stopping clopidogrel. In multivariable analysis, the 0- to 90-day interval after stopping clopidogrel was associated with higher risk of death/MI (incidence rate ratio, 2.74; 95% confidence interval, 1.69 to 4.44) compared with 91- to 360-day interval. There was a similar trend of increased events after stopping clopidogrel for various subgroups (women versus men, medical therapy versus percutaneous coronary intervention, stent type, and ≥6 months or <6 months of clopidogrel treatment). Among patients taking clopidogrel but stopping ACE inhibitor medications, the event rates were similar in the 0- to 90-day versus the 91- to 360-day interval (2.67 versus 2.91 per 10 000 patient-days; P=0.91).
Conclusions— We observed a clustering of adverse events in the 0 to 90 days after stopping clopidogrel. This clustering of events was not present among patients stopping ACE inhibitors. These findings are consistent with a possible rebound platelet hyper-reactivity after stopping clopidogrel and additional platelet studies are needed to confirm this effect.
Received July 1, 2009; accepted February 12, 2010.
In vitro platelet studies have demonstrated increased platelet reactivity after antiplatelet medication withdrawal.1–3 Among patients taking clopidogrel and aspirin for 1 year after drug-eluting stent placement, there were significant increases in proinflammatory and prothrombotic markers after clopidogrel cessation.1 This may be due to increased production of reticulated platelets, which have a higher thrombotic potential.4–7 However, there has been only limited clinical evidence to suggest that this increased platelet activity in vitro translates into adverse clinical thrombembolic outcomes.
Two prior studies suggest that the in vitro platelet findings of increased activation may translate into adverse clinical events after clopidogrel cessation. In the STRATEGY randomized clinical trial, there was an early (ie, <30 days) clustering of adverse outcomes after stopping thienopryidine therapy among patients with acute myocardial infarction (MI) receiving a drug-eluting stent (DES) or a bare-metal stent (BMS).8 Furthermore, a cohort study of acute coronary syndrome (ACS) patients from the VA demonstrated a nearly 2-fold increase in the risk of death or acute MI in the initial 90-day period after clopidogrel cessation.9 However, it is unknown if the observations of a clustering of adverse events after clopidogrel cessation from these studies can be generalized to broader ACS patient populations.
Accordingly, the objective of this study was to assess the risk of adverse outcomes after stopping clopidogrel among patients enrolled in a large integrated health care delivery system. Specifically, we assessed the incidence of death or MI in 90-day intervals after stopping clopidogrel among patients who were discharged on clopidogrel after ACS hospitalization. Furthermore, we expanded on the prior VA study by (1) assessing the clustering of adverse events in broader patient subgroups, specifically, women versus men, patients treated medically versus percutaneous coronary intervention (PCI) during index ACS hospitalization, patients receiving drug-eluting versus bare-metal stents, and different clopidogrel treatment durations (≥6 months or <6 months); (2) describing the prevalence of bleeding events around the time of stopping clopidogrel; and (3) evaluating whether the clustering of adverse events was specific to clopidogrel or a general effect of stopping medications.
WHAT IS KNOWN
We sought to confirm and expand on the findings of a prior study that demonstrated a clustering of cardiovascular events after clopidogrel cessation in a cohort of 1656 patients from a large integrated health care delivery system.
We found that there was a 2-fold increase in the risk of death or myocardial infarction in the 0 to 90 days after clopidogrel cessation compared with later time intervals (ie, 91 to 360 days) and compared with patients remaining on clopidogrel therapy.
WHAT THE STUDY ADDS
This study adds to the literature by (1) confirming the observations from the prior study; (2) demonstrating that the early clustering of adverse outcomes was consistent for multiple patient subgroups (women versus men, percutaneous coronary intervention versus medical therapy without stents, drug-eluting stents versus bare metal stents, and duration of clopidogrel treatment before cessation, ≥6 months versus <6 months); (3) highlighting the incidence of bleeding events (≈7%) around the time of clopidogrel cessation; and (4) showing that the clustering of adverse events was specific to clopidogrel discontinuation.
These findings support that the mechanism for these events may be due to increased platelet prothrombotic activity after clopidogrel withdrawal rather than a stent-specific mechanism and suggest the need for prospective platelet studies to verify this mechanism.
If confirmed, strategies will need to be tested of how best to attenuate this clustering of events after clopidogrel cessation.
Kaiser Permanente of Colorado (KPCO) is a nonprofit integrated health care delivery system that provides medical services to more than 430 000 members in the Denver, Colorado, metropolitan area. We conducted a retrospective cohort study of patients admitted with acute MI or unstable angina based on International Classification of Diseases 9th Revision Clinical Modification principal discharge diagnosis codes 410.xx and 411.xx. Baseline patient demographics, procedures, and comorbidities were derived from automated KPCO inpatient and ambulatory databases, which were defined by relevant ICD-9 (International Classification of Diseases, 9th Revision), CPT (Current Procedural Terminology), and/or DRG (Diagnosis-Related Group System) codes as previously described.10
All patients with acute MI or unstable angina admitted between January 1, 2002, and December 31, 2006, filled a prescription for clopidogrel within 30 days of hospital discharge, and remained event-free during clopidogrel treatment were included. We excluded patients who had an adverse event while on clopidogrel therapy because these events were not related to a potential rebound phenomenon since clopidogrel therapy had not been stopped prior to the observed event.
Clopidogrel Use and Clopidogrel Cessation
Clopidogrel use was assessed using the KPCO pharmacy, data which records the date dispensed and the number of days supplied for each dispensed medication. Clopidogrel was considered available and taken if there was a filled prescription for the medication that covered the date of follow-up based on the dispense date and number of days supplied.9 In the primary analysis, we allowed a 14-day gap between consecutive prescription refills before a patient was considered to have discontinued the medication. In secondary analysis, we decreased the gap to 7 days between prescription refills to categorize a patient as discontinuing a medication. The findings were consistent with the primary results and are not further reported.
The primary outcome was the combined end point of all-cause mortality or MI hospitalization within 1 year after cessation of clopidogrel therapy. Data on mortality were derived from the KPCO automated databases and validated by comparison with internal KPCO data sources and state death certificates. Hospitalizations for acute MI were based on primary ICD-9 discharge diagnosis codes 410.XX. Hospitalizations outside the HMO were also captured through the administrative claims data. Vital status information was available after hospital discharge on patients through December 31, 2007. Patients who lost KPCO insurance during follow-up were censored at the point that they lost coverage, which comprised 3.5% of the cohort.
The primary objective of the analysis was to assess the incidence and timing of adverse events after stopping clopidogrel among ACS patients. First, unadjusted incidence rates for all-cause mortality or MI were calculated for each 90-day interval after stopping clopidogrel (eg, 0 to 90, 91 to 180, 181 to 270, and 271 to 360 days) with 95% Fisher confidence intervals for person-time rates. Second, to assess the association between time interval after stopping clopidogrel and risk of adverse events, we used Poisson regression to calculate incidence rate ratios, adjusting for all variables including duration of clopidogrel treatment (Table 1). In these models, the primary independent variable of interest was the risk of adverse events in the first 0- to 90-day interval after stopping clopidogrel compared with the 91- to 360-day interval. We also tested first-order interaction terms for time interval after clopidogrel cessation and prespecified subgroups of interest (ie, sex, treatment during ACS hospitalization, type of stent, and duration of clopidogrel treatment).
We then performed a series of additional sensitivity analyses. First, we assessed the incidence of the death/MI outcome stratified by sex. Second, we evaluated the incidence of events according to therapy received for ACS, medical therapy or PCI during the index hospitalization. Third, among patients undergoing PCI, we assessed outcomes according to whether patients received a BMS or a DES stent during the index procedure. Fourth, we evaluated the incidence of events according to duration of clopidogrel treatment, ≥6 months or <6 months. Fifth, we compared the rate of adverse events in the first 0- to 90-day interval after stopping clopidogrel with the 91- to 180-day interval consistent with a prior study.9 Sixth, we shortened the initial time interval of interest after stopping clopidogrel to 45 days and compared the risk of adverse events in the 0- to 45-day period after stopping clopidogrel versus the 46- to 360-day period after stopping clopidogrel. Seventh, to account for the usual decline in event rates over time after ACS hospital discharge, we compared the risk of adverse events between patients who continued to take clopidogrel and patients who stopped clopidogrel using Cox proportional hazards models. We included a time varying categorical variable to indicate clopidogrel use as (1) still taking; (2) stopped taking clopidogrel and within 90 days of stopping; and (3) stopped taking clopidogrel and within 91 to 360 days of stopping. We then compared the rate of adverse events in the 0- to 90-day and 91- to 360-day intervals after stopping clopidogrel with patients still taking clopidogrel, which provided risk estimates for persons at comparable time points after hospital discharge. Seventh, because bleeding events can lead to clopidogrel discontinuation, we excluded patients with any bleeding event based on inpatient or outpatient diagnoses within 90 days of the calculated clopidogrel stop date and evaluated the risk of adverse events in the 0 to 90 days after stopping clopidogrel among patients without a bleeding event.
Next, among patients who were taking clopidogrel, we assessed whether there was a clustering of adverse events after stopping ACE inhibitors, for which there have not been prior concerns of a potential “rebound” effect. Among patients who stopped ACE inhibitor medications and did not stop clopidogrel concurrently (ie, within 6 months of stopping ACE inhibitors) (n=425), we evaluated rates of death/MI in 90-day intervals after ACE inhibitor cessation, similar to the primary analysis with clopidogrel. If there was no increase in adverse events in the initial 0- to 90-day interval after stopping ACE inhibitors compared with subsequent time intervals, the findings would further support that the clustering of adverse events after stopping clopidogrel was specific to clopidogrel rather than a general effect of discontinuing medications. Finally, due to concerns of model overfitting, we derived a propensity score for the probability of being in the 0- to 90-day versus the 91- to 360-day exposure group using multivariable logistic regression. Then, we constructed a multivariable Poisson regression model including the propensity score and the 0- to 90-day variable (versus 91- to 360-day) to assess the association with death/MI.
The study was approved by the Kaiser Permanente Colorado Institutional Review Board. Analyses were performed using the SAS statistical package version 9.1 (SAS Institute, Cary, NC).
Of 2108 patients hospitalized for ACS during the study period, 2017 (95.7%) filled their clopidogrel through the KPCO outpatient pharmacy and refill prescription data were available on all of these patients (Figure 1). Of these patients, 141 patients had death or MI while still taking clopidogrel and 217 patients were still taking clopidogrel at the end of the enrollment period or at the end of follow-up. This resulted in a final analytic cohort of 1656 patients who stopped clopidogrel and were event-free before stopping the drug.
In the study cohort, the average age was 65.1±12.6 years, and almost one third were female (30.9%) (Table 1). Approximately one fifth of the patients had a history of diabetes (20.7%) and a very small proportion of patients had evidence of prior clopidogrel use (2.1%). The median duration of clopidogrel therapy after hospital discharge was 90 days (interquartile range, 31 to 183 days).
Mean follow-up after stopping clopidogrel was 259 days and median was 365 days. All-cause mortality or MI occurred in 4.3% (n=71) of patients, with 2.3% of the events occurring in the 0- to 90-day interval, 1.4% in the 91- to 180-day interval, and the rest of the events in the latter time intervals. The incidence rate of death/MI per 10 000 patient-days during each 90-day interval after stopping clopidogrel was 3.07 (95% confidence interval [CI], 2.17 to 4.21), 1.62 (95% CI, 0.94 to 2.59), 0.70 (95% CI, 0.28 to 1.45), and 0.95 (95% CI, 0.43 to 1.79) for 0 to 90, 91 to 180, 181 to 270, and 271 to 360 days, respectively (P<0.001 for comparison between the 4 time intervals) (Figure 2). In multivariable analysis, including adjustment for total duration of clopidogrel treatment, the 0- to 90-day interval after stopping clopidogrel was associated with significantly increased risk of death/MI compared with the 91- to 360-day interval (incidence rate ratio [IRR], 2.74; 95% CI, 1.69 to 4.44).
The findings of an increased risk of adverse events associated with the 0- to 90-day interval were consistent for the various patient subgroups evaluated: women versus men, PCI versus medical therapy without stents, DES versus BMS, and duration of clopidogrel treatment before cessation (≥6 months versus <6 months) (Table 2). The risk of death/MI remained consistent in the comparison between the 0- to 90-day interval after stopping versus the 91- to 180-day interval (IRR, 2.04; 95% CI, 1.13 to 3.71) and in the comparison between the 0- to 45-day interval after stopping versus 46- to 360-day interval (IRR, 3.61; 95% CI, 2.23 to 5.86). In models accounting for time since hospital discharge and compared with patients remaining on clopidogrel therapy, the 0- to 90-day interval after stopping clopidogrel remained associated with adverse events (hazard ratio, 1.75; 95% CI, 1.16 to 2.62). However, there was no association between the 91- to 360-day interval after stopping clopidogrel and adverse outcomes (hazard ratio, 0.89; 95% CI, 0.56 to 1.43) compared with patients remaining on clopidogrel.
There were 115 (6.9%) bleeding events within the 90 days before the calculated clopidogrel stop date. Of these events, 57 (3.4%) resulted in a hospitalization or emergency department visit and 41 (2.7%) were related to a gastrointestinal bleed. In analyses excluding these patients, the 0- to 90-day interval after stopping clopidogrel remained associated with increased risk of death/MI (IRR, 2.65; 95% CI, 1.56 to 4.52) compared with the 91- to 360-day interval.
Among patients who stopped ACE inhibitors and did not also stop clopidogrel therapy concurrently, the event rates after stopping were similar in the first 0- to 90-day (2.67; 95% CI, 0.98 to 5.82; per 10 000 patient-days) compared with the 91- to 360-day interval (2.91; 95% CI, 0.60 to 8.50; per 10 000 patient-days) (P=0.91). Finally, the predicted risk of death/MI in the 0- to 90-day interval after stopping clopidogrel versus the 91 to 360 day interval remained consistent when adjusting for the propensity score variable instead of the individual covariates (IRR, 2.74; 95% CI, 1.70 to 4.43).
The objective of the present study was to assess whether there was a temporal relationship between stopping clopidogrel and death or MI among survivors of an ACS hospitalization. We found that there was a 2-fold increase in the risk of death or MI in the 0- to 90-day interval after clopidogrel cessation compared with later time intervals and compared with patients remaining on clopidogrel therapy. This study adds to the literature in several ways by (1) confirming the observations from the prior VA study; (2) demonstrating that the early clustering of adverse outcomes was consistent for multiple patient subgroups; (3) highlighting the incidence of bleeding events around the time of clopidogrel cessation; and (4) showing that the clustering of adverse events was specific to clopidogrel discontinuation. These findings support that the mechanism for these events may be due to increased platelet prothrombotic activity after clopidogrel withdrawal rather than a stent-specific mechanism. Additional prospective platelet studies are needed to verify this mechanism, and, if confirmed, strategies must be tested of how best to attenuate this clustering of events.
Several platelet studies have found an increase in platelet reactivity after antiplatelet withdrawal that may explain the clustering of death/MI events demonstrated in our study. Angiolillo et al1 found higher platelet aggregation and serum C-reactive protein levels after clopidogrel cessation among patients taking clopidogrel plus aspirin for 1 year after DES placement. After antiplatelet withdrawal, there may be increased platelet production by megakaryocytes, and these newly formed platelets may have more thrombotic potential that could potentially contribute to the clustering of clinical events seen in our study.4–7 In addition, high platelet reactivity after stent placement has been associated with increased risk of future events.12–16 The composite of these platelet studies provide the mechanistic support for the clinical observations in our study of an increased short-term risk of adverse events after clopidogrel withdrawal.
The findings of this study suggest a 2-fold risk of adverse events in the 0- to 90-day interval after stopping clopidogrel compared with subsequent time intervals. The magnitude of increased relative risk that we observed is consistent with a prior published study in a male veteran population, although the absolute risk was much smaller in the current study.9 This may be related to differences in the patient populations or clinical management between the 2 studies. The PCI patients who comprise the majority of the patients in our analysis were less likely to have prior MI, heart failure, chronic obstructive pulmonary disease, and to have prior clopidogrel use compared with the veteran cohort. Although the absolute rate of events was small, the number of events attributable to this potential rebound phenomenon is not trivial, given how frequently patients are discharged on clopidogrel therapy after ACS hospitalization (>800 000 admissions in 2005) and how commonly clopidogrel is prescribed in general.17,18 Given the growing evidence of a clustering of adverse events following clopidogrel cessation and because clopidogrel is currently recommended for a defined treatment period, studies are needed to test whether strategies to minimize the increase in platelet reactivity after the end of clopidogrel therapy can reduce adverse events during this initial post-treatment period.
There are important considerations in interpreting the results of this study. Ascertainment of clopidogrel use was based on automated pharmacy dispensing data. However, we had detailed pharmacy utilization data in terms of when the medication was dispensed and how pills were supplied, and pharmacy dispensing data are a validated measure of medication taking and is strongly correlated with a broad range of patient outcomes.10,19 We also do not know the specific reasons for stopping clopidogrel, but potential reasons include the end of the prescribed course—the expected reason for most of the patients—or the occurrence of complications from the medication, such as bleeding. In sensitivity analysis, excluding patients with a bleeding event did not change our primary findings of an increased risk of adverse events in the 0- to 90-day interval after stopping clopidogrel. The cohort included insured patients from a single integrated health care system, so our results may not be completely generalizable to other populations and practice settings. Further, we were unable to perform a subgroup analysis based on ACS presentation type due to the small number of patients presenting with unstable angina (2.5%) in the cohort and the inability to distinguish between ST-segment elevation MI versus non– ST-segment–elevation MI from administrative data; however, we would not expect the early clustering of adverse events after stopping clopidogrel to be different according to type of initial ACS presentation, based on the hypothesized mechanism of increased platelet reactivity after clopidogrel cessation. Finally, as with any retrospective cohort study design, we cannot confirm causality or completely exclude residual confounding, but this study confirms previous findings of an association between clopidogrel cessation and downstream increase in adverse outcomes. In addition, the event rates were similar in the 0- to 90-day versus 91- to 360-day intervals among patients taking clopidogrel but stopping ACE inhibitors, suggesting that the clustering of adverse events associated with clopidogrel is not a general effect of medication discontinuation but specific to clopidogrel. Further, there is an ongoing prospective study, PRACTICE I, which will measure platelet activity levels after clopidogrel cessation among patients treated for 6 versus 12 months after DES implantation.20 This study can potentially provide additional prospective evidence of increased platelet activity after stopping clopidogrel.
In conclusion, we found a 2-fold increase in the risk of death or MI in the 0- to 90-day interval after stopping clopidogrel compared with later follow-up intervals. This risk was consistent for patients receiving BMS or DES during index hospitalization or medical therapy without stents. In addition, there was a clustering of adverse events after stopping clopidogrel but not after stopping ACE inhibitor medications, suggesting that the clustering of events is not a general effect of stopping medications and consistent with the mechanistic concerns of hyper-platelet reactivity after clopidogrel cessation. These results confirm and extend previous findings and suggest an urgent need to develop strategies to attenuate this effect after the end of a prescribed clopidogrel treatment course.
Sources of Funding
This project was funded under Contract No. 290-05-0033 from the Agency for Healthcare Research and Quality, US Department of Health and Human Services, as part of the Developing Evidence to Inform Decisions about Effectiveness (DEcIDE) program. The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the US Department of Health and Human Services.
Drs Magid and Go are supported by the NHLBI funded Cardiovascular Research Network (U19 HL091179). Dr Go receives funding for the AHA Pharmaceutical Roundtable Outcomes Research Center (AHA #0875162N) for relevant work. Dr Ho is supported by a VA Research and Development Career Development Award (05-026-2).
Guest Editor for this article was Brahmajee K. Nallamothu, MD.
Angiolillo DJ, Fernandez-Ortiz A, Bernardo E, Ramirez C, Sabate M, Jimenez-Quevedo P, Hernãndez R, Moreno R, Escaned J, Alfonso F, Banuelos C, Costa MA, Bass TA, Macaya C. Clopidogrel withdrawal is associated with proinflammatory and prothrombotic effects in patients with diabetes and coronary artery disease. Diabetes. 2006; 55: 780–784.
Valgimigli M, Campo G, Arcozzi C, Malagutti P, Carletti R, Ferrari F, Barbieri D, Parrinello G, Percoco G, Ferrari R. Two-year clinical follow-up after sirolimus-eluting versus bare-metal stent implantation assisted by systematic glycoprotein IIb/IIIa inhibitor infusion in patients with myocardial infarction: results from the STRATEGY study. J Am Coll Cardiol. 2007; 50: 138–145.
Weiss NS. Estimating the impact of the discontinuation of medical interventions on health outcomes. Am J Epidemiol. 2009; 169: 653–656.
Cuisset T, Frere C, Quilici J, Morange PE, Nait-Saidi L, Mielot C, Bali L, Lambert M, Alessi MC, Bonnet JL. High post-treatment platelet reactivity is associated with a high incidence of myonecrosis after stenting for non-ST elevation acute coronary syndromes. Thromb Haemost. 2007; 97: 282–287.
Bliden KP, DiChiara J, Tantry US, Bassi AK, Chaganti SK, Gurbel PA. Increased risk in patients with high platelet aggregation receiving chronic clopidogrel therapy undergoing percutaneous coronary intervention: is the current antiplatelet therapy adequate? J Am Coll Cardiol. 2007; 49: 657–666.
Cuisset T, Frere C, Quilici J, Barbou F, Morange PE, Hovasse T, Bonnet JL, Alessi MC. High post-treatment platelet reactivity identified low-responders to dual antiplatelet therapy at increased risk of recurrent cardiovascular events after stenting for acute coronary syndrome. J Thromb Haemost. 2006; 4: 542–549.
Rosamond W, Flegal K, Friday G, Furie K, Go A, Greenlund K, Haase N, Ho M, Howard V, Kissela B, Kittner S, Lloyd-Jones D, McDermott M, Meigs J, Moy C, Nichol G, O'Donnell CJ, Roger V, Rumsfeld J, Sorlie P, Steinberger J, Thom T, Wasserthiel-Smoller S, Hong Y. American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics: 2007 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 2007; 115: e69–e171.
RxList. Available at: http://www.rxlist.com/script/main/hp.asp. Accessed June 30, 2009.
US National Institutes of Health Clinical Trials website. Available at: http://www.clinicaltrials.gov/ct2/show/NCT00684359?term=clopidogrel+rebound&rank=1. Accessed on October 28, 2008.