Drug-Drug Interactions Between Antithrombotic Medications & GI Bleeding (PDF)

Research Drug–drug interactions between antithrombotic medications and the risk of gastrointestinal bleeding Joseph A. Delaney MSc, Lucie Opatrny MD MSc, James M. Brophy MD PhD, Samy Suissa PhD @...

Research Drug–drug interactions between antithrombotic medications and the risk of gastrointestinal bleeding Joseph A. Delaney MSc, Lucie Opatrny MD MSc, James M. Brophy MD PhD, Samy Suissa PhD @ See related articles pages 357 and 369 Abstract Background: Anticoagulants and antiplatelet drugs (e.g., warfarin, clopidogrel and acetylsalicylic acid) are key thera- A ntithrombotic drugs are used for the prevention and treatment of cardiovascular disorders.1,2 However, co-prescribing these drugs or prescribing them with others such as nonsteroidal anti-inflammatory drugs can create important drug–drug interactions that can lead to an peutic agents in the treatment of cardiovascular diseases. increased risk of gastrointestinal bleeding.3–6 This increased However, drug–drug interactions may lead to a greatly in- risk may be much greater than the product of the risks asso- creased risk of gastrointestinal bleeding when these drugs ciated with each drug. We conducted this study to assess are combined. We assessed whether antithrombotic drug whether an increased risk of gastrointestinal bleeding due combinations increased the risk of such bleeding in a gen- to drug–drug interactions between antithrombotic medica- eral practice population. tions existed in a general practice population. Methods: We conducted a population-based, retrospective case–control study using records in the United Kingdom Methods General Practice Research Database from 2000 through 2005. Cases were identified as patients over 18 years of age Study design with a first-ever diagnosis of gastrointestinal bleeding. They We conducted a population-based, retrospective case–control were matched with controls by physician practice, patient study using records in the United Kingdom (UK) General age and index date (date of diagnosis of bleeding). All eligi- Practice Research Database from Jan. 1, 2000, through Dec. ble patients had to have at least 3 years of follow-up data in 31, 2005. This is a well-validated database of a network of the database. Drug exposure was considered to be any pre- more than 400 general practices in the United Kingdom6–8 scription issued in the 90 days before the index date. that has been widely used for pharmacoepidemiology re- Results: There were 4028 cases with a diagnosis of gastro- search, including studies of gastrointestinal bleeding.9,10 The intestinal bleeding and 40 171 matched controls. The pre- electronic medical records in the UK General Practice Re- scribing of acetylsalicylic acid with either clopidogrel search Database include all important medical events and all (adjusted rate ratio [RR] 3.90, 95% confidence interval [CI] prescriptions, since the general practitioner is the centre of 2.78–5.47) or warfarin (adjusted RR 6.48, 95% CI 4.25– health care in the United Kingdom. As a result, the database 9.87) was associated with a greater risk of gastrointestinal is a reliable source of information to study drug effects in a bleeding than that observed with each drug alone. The clinical setting.6–8 same was true when a nonsteroidal anti-inflammatory drug We defined a case as any patient 18 years or older whose was combined with either clopidogrel (adjusted RR 2.93, record in the database contained a first-ever entry of a com- 95% CI 1.74–4.93) or warfarin (RR 4.60, 95% CI 2.77–7.64). puterized code for gastrointestinal bleeding. The date of diag- Interpretation: Drug combinations involving antiplatelets nosis was taken as the index date for the case. Using inci- and anticoagulants are associated with a high risk of dence-density sampling, we selected up to 10 controls for gastrointestinal bleeding beyond that associated with each every case in the database matched by practice, patient age drug used alone. Physicians should be aware of these risks (± 2 years) and index date. To permit a full assessment of pa- to better assess their patients’ therapeutic risk–benefit tient comorbidity and lifestyle information, all patients had to profiles. have medical records with at least 3 years of data recorded be- DOI:10.1503/cmaj.070186 fore the index date. Une version française de ce résumé est disponible à l’adresse www.cmaj.ca/cgi/content/full/177/4/347/DC1 From the Division of Clinical Epidemiology (Delaney, Opatrny, Brophy, CMAJ 2007;177(4):347-51 Suissa), the Department of Epidemiology, Biostatistics and Occupational Health (Delaney, Brophy, Suissa), and the Division of Internal Medicine (Opatrny), McGill University Health Centre, Montréal, Que. CMAJ August 14, 2007 177(4) 347 © 2007 Canadian Medical Association or its licensors Research We obtained ethics approval for this study from the Scien- Table 1: Characteristics of patients with upper gastrointestinal bleeding (cases) and matched controls tific and Ethical Advisory Group of the UK General Practice Research Database and the McGill University Health Centre Group; no. (%) of patients* Research Ethics Board. Cases Controls Characteristic n = 4 028 n = 40 171 Outcome measures The primary exposure of interest was the co-prescription of Age, yr warfarin or clopidogrel with acetylsalicylic acid or a non- Mean (SD) 69.3 (17.6) 69.1 (17.7) acetylsalicylic-acid nonsteroidal anti-inflammatory drug. Range 18–104 18–105 Non-acetylsalicylic-acid nonsteroidal anti-inflammatory Male sex 2 171 (53.9) 17 237 (42.9) drugs were defined according to the British National Formu- Female sex 1 857 (46.1) 22 934 (57.1) lary (www.bnf.org), with the vast majority of prescriptions by Body mass index, kg/m2 general practitioners being for naproxen, diclofenac and < 18 105 (2.6) 690 (1.7) ibuprofen. (The full list of agents considered is aclofenac, 18–29.9 2 289 (56.8) 23 636 (58.8) 30–39.9 514 (12.8) 4 780 (11.9) dexketoprofen, diclofenac, diflunisal, etodolac, fenoprofen, ≥ 40 56 (1.4) 399 (1.0) ibuprofen, indomethacin, ketoprofen, ketorolac, mefenamic Not recorded 1 064 (26.4) 10 666 (26.6) acid, meloxicam, nabumetone, naproxen, piroxicam, sulin- Blood pressure dac, tenoxicam and tiaprofenic acid). The cyclooxygenase-2 High 959 (23.8) 8 848 (22.0) selective inhibitors rofecoxib and celecoxib are defined sepa- Borderline 978 (24.3) 8 264 (20.6) rately from older traditional nonsteroidal anti-inflammatory Normal 741 (18.4) 5 518 (13.7) drugs. Current drug use was defined as a prescription in the No reading in past year 1 350 (33.5) 17 541 (43.7) 90 days before the index date; this definition was selected to Smoking status minimize the misclassification of exposure, since it is diffi- Smoker 1 797 (44.6) 13 780 (34.3) cult to estimate durations of warfarin prescriptions, especially Nonsmoker 1 763 (43.8) 20 702 (51.5) because doses of warfarin may change during the course of a Not recorded 468 (11.6) 5 689 (14.2) prescription according to changes in the patient’s interna- Heavy alcohol use 395 (9.8) 791 (2.0) tional normalized ratio. Comorbid condition† Acid reflux disease 431 (10.7) 3 321 (8.3) Statistical analysis Peptic ulcer 76 (1.9) 403 (1.0) We used conditional logistic regression analysis to compute Helicobacter pylori infection 56 (1.4) 228 (0.6) odds ratios as an estimate of rate ratios of gastrointestinal Pulmonary embolism 89 (2.2) 410 (1.0) Deep-vein thrombosis 139 (3.5) 907 (2.3) bleeding associated with drug exposure. The odds ratio is a Myocardial infarction 358 (8.9) 2 014 (5.0) valid estimate of the rate ratio because we used incidence- Angina 672 (16.7) 4 477 (11.1) density sampling to select the matched controls.11,12 We esti- Stroke 329 (8.2) 1 489 (3.7) mated both the main effects and the statistical (multiplica- Atrial fibrillation 536 (13.3) 3 362 (8.4) tive) interactions using the same statistical model. The drug– Congestive heart failure 472 (11.7) 2 290 (5.7) drug interaction term shows the excess risk beyond what Rheumatoid arthritis 101 (2.5) 616 (1.5) would have been predicted from the combination of the indi- Other arthritis 1 252 (31.1) 10 841 (27.0) vidual effects of each drug. The adjusted effect of the drug Diabetes 512 (12.7) 3 204 (8.0) combination is the total risk of the drug combination, includ- Cancer 143 (3.6) 852 (2.1) Dementia 171 (4.2) 1 029 (2.6) ing the effect of each agent individually as well as any excess Liver failure 89 (2.2) 62 (0.2) risk due to the combination of the drugs. Renal failure 125 (3.1) 490 (1.2) We considered as covariates a past history (indicated by Chronic obstructive the presence in the patient’s medical record of at least 1 med- pulmonary disease 354 (8.8) 1 875 (4.7) ical code entered before the index date) of the following con- Drug use other than NSAID ditions: gastroesophageal reflux, peptic ulcer disease, a posi- or antithrombotic‡ tive test result for Helicobacter pylori, hypertension, liver Antibiotic 1 009 (25.0) 5 990 (14.9) failure, renal failure, arthritis, diabetes, cancer (any type), Antidepressant 632 (15.7) 3 702 (9.2) chronic obstructive pulmonary disease and dementia (any Corticosteroid 599 (14.9) 4 729 (11.8) Diuretic 1 370 (34.0) 10 348 (25.8) type). We also considered as covariates the possible indica- H2 antagonist 268 (6.7) 1 287 (3.2) tions for warfarin use, including atrial fibrillation, pulmonary Heparin 4 (0.1) 7 (0.02) embolism, deep-vein thrombosis, congestive heart failure, Paracetamol 1 336 (33.2) 7 934 (19.8) myocardial infarction, angina and stroke. Proton pump inhibitor 930 (23.1) 3 985 (9.9) We also compared demographic characteristics of the Note: SD = standard deviation, NSAID = nonsteroidal anti-inflammatory drug. cases and controls, including age, sex, smoking status, body *Unless stated otherwise. mass index and history of heavy alcohol use (as indicated by †Previous history of condition entered in database medical record before the database medical codes). A body mass index of less than index date (date of first-ever diagnosis of gastrointestinal bleeding). ‡Any prescription issued in the 90 days before the index date. 18 kg/m2 was considered to indicate underweight, of more 348 CMAJ August 14, 2007 177(4)  Research than 30 kg/m2 but less than 40 kg/m2 to indicate obesity, and The individual and combined effects of the study drugs are of 40 kg/m2 or higher to indicate morbid obesity. A positive shown in Table 2. The top section of the table describes the history of smoking (current or past) was grouped together as risk of gastrointestinal bleeding among patients prescribed a a single smoking variable given the cross-sectional nature of single antithrombotic agent. The lower section of the table smoking data in the database. describes the much higher risk among patients prescribed All statistical analyses were adjusted for potential con- combinations of these drugs. founders and markers of health status, as measured by pre- In particular, the combined prescription of acetylsalicylic scriptions in the 90 days before the index date or a diagnosis acid with either clopidogrel (adjusted RR 3.90, 95% CI 2.78– code for comorbid conditions entered in the database any 5.47) or warfarin (adjusted RR 6.48, 95% CI 4.25–9.87) was as- time before the index date, as well as age, sex, smoking status sociated with a greater risk of gastrointestinal bleeding than and body mass index. We used indicator variables for missing that observed with each drug alone. For example, a prescription demographic or lifestyle data to indicate the presence of a of acetylsalicylic acid alone was associated with an increased missing variable. risk of bleeding (adjusted RR 1.39, 95% CI 1.26–1.53), as was a prescription of warfarin alone (adjusted RR 1.94, 95% CI 1.61– Results 2.34); however, the effect of combining these 2 drugs, as shown above, yielded a significant interaction term (RR 2.23, 95% CI The characteristics of the cases and controls are described 1.46–3.41; Table 2). This interaction term shows the additional in Table 1. Known risk factors for gastrointestinal bleeding risk of gastrointestinal bleeding from a drug–drug interaction were found to be important predictors of increased risk, between warfarin and acetylsalicylic acid beyond the risks of even in the multivariable analysis. These factors were: male each agent. The adjusted effect (total risk) of this drug combina- sex (adjusted rate ratio [RR] 1.50, 95% confidence interval tion was high (RR 6.48, 95% CI 4.25–9.87). [CI] 1.40–1.62), heavy alcohol use (adjusted RR 4.00, 95% Similar effects were seen among patients prescribed any CI 3.45–4.63), smoking (adjusted RR 1.23, 95% CI 1.15– nonsteroidal anti-inflammatory drug (either a conventional 1.34), acetaminophen (paracetamol) use (adjusted RR 1.47, one or a cyclooxygenase-2 selective inhibitor) with either 95% CI 1.35–1.60) and liver failure (adjusted RR 7.00, 95% clopidogrel (adjusted RR 2.93, 95% CI 1.74–4.93) or warfarin CI 4.78–10.27). (adjusted RR 4.60, 95% CI 2.77–7.64). Table 2: Individual and combined effects of the study drugs on the risk of gastrointestinal bleeding Rate ratio (95% confidence interval) No. (%) of cases No. (%) of controls Drug use n = 4 028 n = 40 171 Crude Adjusted* Individual None† 2 124 (52.7) 28 264 (70.4) 1.00† 1.00† Warfarin 281 (7.0) 1 130 (2.8) 2.64 (2.31–3.03) 1.94 (1.61–2.34) Clopidogrel 160 (4.0) 532 (1.3) 3.16 (2.63–3.79) 1.67 (1.27–2.20) ASA 1 122 (27.9) 7 350 (18.3) 1.85 (1.71–2.00) 1.39 (1.26–1.53) NSAID‡ 678 (16.8) 3 707 (9.2) 2.02 (1.84–2.21) 1.78 (1.61–1.97) COX-2 inhibitor 129 (3.2) 630 (1.6) 2.12 (1.74–2.58) 1.64 (1.31–2.06) Adjusted Adjusted effect Combination interaction term§ for drug combination¶ None† 2 124 (52.7) 28 264 (70.4) 1.00† 1.00† Warfarin + ASA 48 (1.2) 82 (0.2) 2.23 (1.46–3.41) 6.48 (4.25–9.87) Warfarin + NSAID‡ 30 (0.7) 53 (0.1) 1.33 (0.78–2.25) 4.79 (2.79–8.21) Warfarin + COX-2 inhibitor 6 (0.2) 9 (0.0) 1.37 (0.44–4.30) 4.62 (1.48–14.43) Clopidogrel + ASA 73 (1.8) 133 (0.3) 1.75 (1.17–2.64) 3.90 (2.78–5.47) Clopidogrel + NSAID‡ 22 (0.6) 43 (0.1) 1.04 (0.56–1.93) 2.90 (1.58–5.35) Clopidogrel + COX-2 inhibitor 9 (0.2) 19 (0.1) 0.98 (0.40–2.44) 2.60 (1.09–6.23) Note: ASA = acetylsalicylic acid, NSAID = nonsteroidal anti-inflammatory drug, COX-2 = cyclooxygenase-2. *Adjusted for potential confounders and markers of health status, as measured by prescriptions in the 90 days before the index date or a diagnosis code for comorbid conditions entered in the database record any time before the index date, as well as age, sex, smoking status and body mass index. †Patients who were exposed to none of the study drugs; these patients constituted the reference group. ‡This class of drugs includes aclofenac, dexketoprofen, diclofenac, diflunisal, etodolac, fenoprofen, ibuprofen, indomethacin, ketoprofen, ketorolac, mefenamic acid, meloxicam, nabumetone, naproxen, piroxicam, sulindac, tenoxicam and tiaprofenic acid. §Estimated additional risk of exposure to the combination of the 2 drugs beyond the risk associated with exposure to each of the drugs individually (the risks of the individual drugs appear in the top half of the table). ¶Estimated total risk of gastrointestinal bleeding for a patient who is prescribed the 2 drugs simultaneously. CMAJ August 14, 2007 177(4) 349 Research Increased risk of patients with an increased risk of gastrointestinal bleeding. gastrointestinal bleeding In addition, our study design does not allow us to make infer- ences about the appropriateness of the prescriptions. Finally, Clopidogrel + NSAID we lacked the statistical power to look at the risk of gastroin- Clopidogrel + ASA testinal bleeding associated with a 3-way interaction of all antithrombotics. Warfarin + NSAID The low numbers for many of the drug exposures in our Warfarin + ASA study suggest that additional evidence should be gathered from other population databases before definitive conclu- NSAID sions can be reached. This may be especially true for cyclo- ASA oxygenase-2 selective inhibitors, which have known cardiac Clopidogrel and renal risks17,18 even though the risk of gastrointestinal bleeding associated with them is lower than that associated Warfarin with conventional nonsteroidal anti-inflammatory drugs.19 In our results, the risks associated with the cyclooxygenase-2 1 10 selective inhibitors and conventional nonsteroidal anti- Odds ratio (log) inflammatory drugs appeared to be comparable. Our results indicate that physicians need to be aware Figure 1: Risk of gastrointestinal bleeding among patients in and weigh the potential risk of gastrointestinal bleeding the United Kingdom General Practice Research Database who were prescribed acetylsalicylic acid (ASA), clopidogrel, warfarin due to drug–drug interactions with antithrombotic agents or any type of non-ASA nonsteroidal anti-inflammatory drug against the known therapeutic benefits 13 of these drug (NSAID), either alone or in combination. combinations. This article has been peer reviewed. A forest plot of the risks of gastrointestinal bleeding asso- Competing interests: None declared for Joseph Delaney, Lucie Opatrny or ciated with the different drugs, alone and in combination, ap- James Brophy. Samy Suissa has received consultancy fees from Sanofi- pears in Figure 1. Aventis for Lantus and leflunomide but not for clopidogrel, which is studied in this paper. Interpretation Contributors: All of the authors contributed to the conception and design of the study as well as the acquisition and interpretation of data. Joseph Delaney did the statistical analysis and drafted the manuscript. All of the authors were We found an increased risk of gastrointestinal bleeding as- involved in revising the article for important intellectual content and ap- sociated with drug–drug interactions among patients pre- proved the final version to be published. scribed antithrombotic agents. The increased risk observed Acknowledgements: This study was funded by the Canadian Institutes of when acetylsalicylic acid was combined with other anti- Health Research (CIHR) and the Canadian Foundation for Innovation. Samy thrombotic agents was similar to that seen in other studies.5 Suissa is the recipient of a Distinguished Scientist Award from CIHR. James However, our estimates were higher than those derived from Brophy and Lucie Opatrny are supported scholars by le Fonds de la recherche en santé du Québec. the meta-analyses of clinical trials,1,13 possibly because the monitoring of patients was less strict than that inside the clinical trial environment.14,15 Indeed, our study was popula- REFERENCES tion based, so our statistical inferences should hold in actual 1. Dentali F, Douketis JD, Lim W, et al. Combined aspirin–oral anticoagulant ther- clinical settings across a broad sampling of the United King- apy compared with oral anticoagulant therapy alone among patients at risk for cardiovascular disease: a meta-analysis of randomized trials. Arch Intern Med dom population. 2007;167:117-24. Our study does have limitations owing to its being based 2. Scarvelis D, Wells PS. Diagnosis and treatment of deep-vein thrombosis. CMAJ 2006;175:1087-92. on physician records. Although diagnoses and prescriptions 3. Buresly K, Eisenberg MJ, Zhang X, et al. Bleeding complications associated with in the UK General Practice Research Database are quite spe- combinations of aspirin, thienopyridine derivatives, and warfarin in elderly pa- tients following acute myocardial infarction. Arch Intern Med 2005;165:784-9. cific, since they are part of patient care management, the 4. Jonsson AK, Spigset O, Jacobsson I, et al. Cerebral haemorrhage induced by war- database may lack sensitivity because some events or pre- farin-the influence of drug–drug interactions. Pharmacoepidemiol Drug Saf 2007; 16:309-15. scriptions could have been missed,6–8 especially with drugs 5. Hallas J, Dall M, Andries A, et al. Use of single and combined antithrombotic ther- available over the counter. Our choice of the 90-day exposure apy and risk of serious upper gastrointestinal bleeding: population based case– period assures that any bias is directed toward the null, since control study. BMJ 2006;333:726. 6. Walley T, Mantgani A. The UK General Practice Research Database. Lancet 1997; any past user we may have misclassified as a current user 350:1097-9. should have a risk profile that is more similar to the risk 7. Jick SS, Kaye JA, Vasilakis-Scaramozza C, et al. Validity of the general practice re- search database. Pharmacotherapy 2003;23:686-9. profile of someone not prescribed antithrombotics than to 8. Lawrenson R, Williams T, Farmer R. Clinical information for research; the use of the risk profile of someone with such a drug exposure. There general practice databases. J Public Health Med 1999;21:299-304. 9. Gasse C, Hollowell J, Meier CR, et al. Drug interactions and risk of acute bleeding could also have been some degree of confounding by drug in- leading to hospitalisation or death in patients with chronic atrial fibrillation dication16 owing to the choice of nonsteroidal anti-inflamma- treated with warfarin. Thromb Haemost 2005;94:537-43. 10. Hollowell J, Ruigomez A, Johansson S, et al. The incidence of bleeding complica- tory drug prescribed, since general practitioners may selec- tions associated with warfarin treatment in general practice in the United King- tively prescribe a cyclooxygenase-2 selective inhibitor to dom. Br J Gen Pract 2003;53:312-4. 350 CMAJ August 14, 2007 177(4)  Research 11. Clayton D, Hills M. Statistical models in epidemiology. Oxford (UK): Oxford Uni- 17. Schneider V, Levesque LE, Zhang B, et al. Association of selective and conventional versity Press; 1993. nonsteroidal antiinflammatory drugs with acute renal failure: a population-based, 12. Suissa S. Novel approaches to pharmacoepidemiology study design and statistical nested case–control analysis. Am J Epidemiol 2006;164:881-9. analysis. In: Strom BL, editor. Pharmacoepidemiology. Sussex (UK): John Wiley & 18. Levesque LE, Brophy JM, Zhang B. Time variations in the risk of myocardial infarc- Sons; 2000. p. 785-805. tion among elderly users of COX-2 inhibitors. CMAJ 2006;174:1563-9. 13. Rothberg MB, Celestin C, Fiore LD, et al. Warfarin plus aspirin after myocardial in- 19. Bombardier C, Laine L, Reicin A, et al.; VIGOR Study Group. Comparison of upper farction or the acute coronary syndrome: meta-analysis with estimates of risk and gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid benefit. Ann Intern Med 2005;143:241-50. arthritis. VIGOR Study Group. N Engl J Med 2000;343:1520-8. 14. Meade TW, Roderick PJ, Brennan PJ, et al. Extra-cranial bleeding and other symp- toms due to low dose aspirin and low intensity oral anticoagulation. Thromb Haemost 1992;68:1-6. 15. Sarawate C, Sikirica MV, Willey VJ, et al. Monitoring anticoagulation in atrial fibril- Correspondence to: Dr. Samy Suissa, Division of Clinical lation. J Thromb Thrombolysis 2006;21:191-8. Epidemiology, Ross 4.29, Royal Victoria Hospital, 16. MacDonald TM, Morant SV, Goldstein JL, et al. Channelling bias and the incidence of gastrointestinal haemorrhage in users of meloxicam, coxibs, and older, non- 687 Pine Ave. W, Montréal QC H3A 1A1; fax 514 843-1493; specific non-steroidal anti-inflammatory drugs. Gut 2003;52:1265-70. [email protected] CMAJ August 14, 2007 177(4) 351