ORIGINAL INVESTIGATION Cyclooxygenase Selectivity of Nonsteroidal Anti-inflammatory Drugs and Risk of Stroke Mendel D. M. Haag, PharmD; Michiel J. Bos, MD; Albert Hofman, MD, PhD; Peter J. Koudstaal, MD, PhD; Monique M. B. Breteler, MD, PhD; Bruno H. C. Stricker, MB, PhD Background: In clinical trials, cyclooxygenase (COX)- unspecified). Current users of nonselective (HR, 1.72; 95% 2–selective nonsteroidal anti-inflammatory drugs confidence interval [CI], 1.22-2.44) and COX-2–selective (NSAIDs) were associated with an increased risk of throm- (HR, 2.75; 95% CI, 1.28-5.95) NSAIDs had a greater risk boembolic events. We studied the association between of stroke, but not users of COX-1–selective NSAIDs (HR, NSAID use and risk of stroke in the prospective, popu- 1.10; 95% CI, 0.41-2.97). Hazard ratios (95% CIs) for is- lation-based Rotterdam Study. chemic stroke were 1.68 (1.05-2.69) for nonselective and 4.54 (2.06-9.98) for COX-2–selective NSAIDs. For indi- Methods: We followed 7636 persons free of stroke at vidual NSAIDs, current use of the nonselective naproxen baseline (1991-1993) for incident stroke until Septem- (HR, 2.63; 95% CI, 1.47-4.72) and the COX-2–selective ro- ber 2004. Data on all filled prescriptions came from phar- fecoxib (HR, 3.38; 95% CI, 1.48-7.74) was associated with macy records. With Cox regression models, we calcu- a greater risk of stroke. Hazard ratios (95% CIs) for diclo- lated crude and adjusted hazard ratios (HRs) of stroke fenac (1.60 [1.00-2.57]), ibuprofen (1.47 [0.73-3.00]), and for time-dependent current use, compared with never use, celecoxib (3.79 [0.52-27.6]) were greater than 1.00 but were of NSAIDs grouped according to COX selectivity (COX-1 not statistically significant. selective, nonselective, and COX-2 selective) and indi- vidual NSAIDs. Conclusions: In the general population, we found a greater risk of stroke with current use of nonselective and Results: At baseline, the mean age of the study sample COX-2–selective NSAIDs. The risk of stroke was not lim- was 70.2 years, and 61.3% were female. During 70 063 per- ited to the use of COX-2–selective NSAIDs. son-years of follow-up (mean, 9.2 years), 807 persons de- veloped a stroke (460 ischemic, 74 hemorrhagic, and 273 Arch Intern Med. 2008;168(11):1219-1224 N CLINICAL TRIALS, THE USE OF CY- risk of ischemic stroke was found for some clooxygenase (COX)-2–selective COX-2–selective NSAIDs, but this was also nonsteroidal anti-inflammatory found for the nonselective NSAID diclo- drugs (NSAIDs) has been associ- fenac.11 Predominantly, null findings were ated with an increased risk of car- reported in 4 studies that investigated the Idiovascular events and death.1-5 Most of outcome of hemorrhagic stroke.6,8-10 the post hoc analyses of the trials focused It is supposed that selective inhibition on combined cardiovascular or cerebro- of the COX-2 enzymes by the COX-2– vascular events as the end point, without selective NSAIDs induces a prothrom- botic state, unlike NSAIDs that inhibit COX-2 to a lesser degree and additionally CME available online at inhibit COX-1.12-14 However, the cardio- Author Affiliations: www.jamaarchivescme.com vascular risk observed for the various COX- Departments of Epidemiology and questions on page 1133 2–selective NSAIDs has not been consis- and Biostatistics (Drs Haag, Bos, tent in all clinical trials.15 In a recent Hofman, Koudstaal, Breteler, further specification of the risk of stroke. placebo-controlled trial, the non–COX-2– and Stricker), Neurology Moreover, few observational studies have selective NSAID naproxen sodium was as- (Drs Bos and Koudstaal), and investigated the association between NSAID sociated with an increased cardiovascular Internal Medicine (Dr Stricker), use and the risk of stroke.6-11 Of these, one 4 Erasmus Medical Center, risk. Evidence from observational studies Rotterdam, the Netherlands; nested case-control study showed that cur- has also shown that NSAIDs differ in their 16-18 and Inspectorate for Health rent NSAID use was associated with a potential to cause cardiovascular events. Care, The Hague, the greater risk of ischemic stroke.7 In an- In all, it remains inconclusive if the greater Netherlands (Dr Stricker). other nested case-control study, a greater risk of cardiovascular events is specific for (REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 11), JUNE 9, 2008 WWW.ARCHINTERNMED.COM 1219 ©2008 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/23/2021 DRUG EXPOSURE Table 1. Classification of NSAIDs According to COX-Selective Properties and Total Person-Months of Complete information on all filled prescriptions for all per- Exposure During Follow-up sons was obtained in automated format from the pharmacies. This included the product name, international nonpropri- NSAID Class Person-Months of Exposure etary name, Anatomical Therapeutic Chemical code, total num- COX-1–selective NSAIDs ber of delivered units (eg, tablets or capsules), prescribed daily Indometacine 2181.0 number of units, date of delivery, and drug dosage. The dura- Piroxicam 3108.9 tion of a prescription was calculated as the total number of de- Ketoprofen 1306.3 livered units divided by the prescribed daily number of units. Flurbiprofen 222.4 Drug dosage was defined by the defined daily dose (DDD), the Apazone 272.5 recommended daily dosage of a drug taken by adults for the Nonselective NSAIDs main indication of the drug. Diclofenaca 15 126.7 Based on data from in vitro and clinical studies, NSAIDs were Naproxen 6776.9 classified as COX-1 selective, nonselective, and COX-2 selec- Ibuprofen 8208.6 tive according to their relative selectivity for the COX-1 and Nabumetone 1008.2 COX-2 enzymes at therapeutic dosages (Table 1).20-26 For some Sulindac 467.6 NSAIDs, COX selectivity is unknown or equivocal (ie, benzy- COX-2–selective NSAIDs damine hydrochloride, tiaprofenic acid, tolfenamic acid, phen- Rofecoxib 1882.2 ylbutazone, tenoxicam, and aceclofenac). Celecoxib 342.6 Salicylates (ie, acetylsalicylic acid and carbasalate calcium) Meloxicam 767.0 Etoricoxib 49.8 are pharmacologically related to NSAIDs and inhibit platelet Valdecoxib 12.6 aggregation via COX-1; although, contrary to NSAIDs, their ef- fects are irreversible.27,28 On these grounds, salicylates could Abbreviations: COX, cyclooxygenase; NSAIDs, nonsteroidal be regarded as COX-1–selective NSAIDs; however, they are anti-inflammatory drugs. mostly prescribed at a low dose as platelet inhibitors for the a Includes combination products of diclofenac. prevention of cardiovascular disease and stroke. We did not include salicylates in the COX-1–selective NSAID group for the following reasons: (1) they are indicated for stroke preven- COX-2–selective NSAIDs or whether other pharmacologi- tion; (2) NSAID use is cautioned in persons already using sa- licylates because of the increased risk of gastrointestinal tract cal properties of NSAIDs could cause these detrimental bleeding; and (3) some NSAIDs possibly antagonize the plate- effects. let inhibition induced by salicylates.27,28 This might obscure the We investigated the association between NSAID use protective effect of salicylates. Because of these potential sources and the risk of incident stroke in a large, prospective, of confounding by salicylates, all analyses were adjusted for the population-based cohort study and whether any ob- current use of salicylates, and the effect of salicylates on the served association was restricted to COX-2–selective association between NSAID use and stroke was studied through NSAIDs. stratification. METHODS DIAGNOSIS OF STROKE STUDY POPULATION A history of stroke at the time of enrollment into the Rotter- dam Study was assessed by asking “did you ever suffer from a The Rotterdam Study is a prospective, population-based co- stroke, diagnosed by a physician?” Positive answers to this ques- hort study of age-related disorders.19 The medical ethics com- tion were verified by reviewing the medical records. A history mittee of the Erasmus Medical Center, Rotterdam, the Neth- of transient ischemic attack was also assessed during the base- erlands, approved the study. Between 1990 and 1993, all persons line interview. After enrollment into the Rotterdam Study, par- 55 years or older living in Ommoord, a district of Rotterdam, ticipants were continuously monitored for all major events, in- were invited to participate. Of the 10 275 eligible persons, 7983 cluding cerebrovascular disease, through automated linkage of (77.7%) signed informed consent. Of these individuals, 7722 the study database with files from general practitioners. Infor- were free of stroke at baseline. Follow-up examinations were mation on vital status was obtained at regular intervals from conducted in 1993 to 1994, 1997 to 1999, and 2000 to 2004. the municipal authorities in Rotterdam. When an event or death In addition, the cohort was continuously monitored for major had been reported, additional information was obtained from disease outcomes and death through linkage with records from the general practitioner and from information in the hospital general practitioners and bimonthly updates from the munici- records (including brain imaging) and discharge letters in the pality records. This resulted in a virtually complete follow-up case of admittance or referral. In addition, nursing home phy- for stroke. Nearly all persons (99.7%) were registered at 1 or sicians’ files and files from general practitioners of partici- more of 7 automated pharmacies serving the Ommoord area. pants who moved out of the district were scrutinized. Re- Of these pharmacies, records of all filled prescriptions were avail- search physicians discussed information on all potential strokes able as of January 1, 1991. To ensure at least a 6-month his- and transient ischemic attacks with an experienced neurolo- tory of medication use, we excluded 86 persons for whom fol- gist to verify all diagnoses while blinded to drug exposure.