Long-Term Cardiovascular Risk of NSAID Use According to Time Passed After First-Time Myocardial Infarction: A Nationwide Cohort Study Anne-Marie Schjerning Olsen, Emil L. Fosbøl, Jesper Lindhardsen, Fredrik Folke, Mette Charlot, Christian Selmer, Jonas Bjerring Olesen, Morten Lamberts, Martin H. Ruwald, Lars Køber, Peter R. Hansen, Christian Torp-Pedersen and Gunnar H. Gislason Circulation. published online September 10, 2012; Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2012 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539 The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/early/2012/09/07/CIRCULATIONAHA.112.112607 Data Supplement (unedited) at: http://circ.ahajournals.org/content/suppl/2012/09/07/CIRCULATIONAHA.112.112607.DC1.html Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/ Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Long-Term Cardiovascular Risk of NSAID Use According to Time Passed After First-Time Myocardial Infarction: A Nationwide Cohort Study Running title: Olsen et al.; NSAIDs and Long-Term Risk after MI Anne-Marie Schjerning Olsen, MD1; Emil L. Fosbøl, MD, PhD1,3; Jesper Lindhardsen, MD1; Fredrik Folke, MD, PhD1; Mette Charlot, MD, PhD1; Christian Selmer, MD1; Jo Jonas ona nass Bj Bjer Bjerring e ri rinng Olesen, MD1; Morten Lamberts, MD1; Martin H. Ruwald, MD1; Lars Køber,, MD, MD, DMSc DMSc2; Peter R. Hansen, MD, PhD, DMSc1; Christian Torp-Pedersen, MD, DMSc1; Gunnar H. Gisl Gislason, las ason o , MD, PhD1 1 Dept De ept of Card Cardiology, rdio iolo io lo ogy y, Co Copenhagen ope penh nhaage nh agen en U University n versi ni sity si yH Hospital osppitaal Ge Gent Gentofte, ntof nt ofte of te, He te Hell Hellerup; ller ll e up; 2De Dept pt of of Cardiology, Card Ca rdio rd iolo io lo ogy g , the he Heart Heart Centre, Ceent n ree, Copenhagen Coppenhag agen en University Uni n ve v rssitty Hospital Hosspitaal Rigshospitalet, Rig gsh shos osspiitaleet,, Cop Copenhagen, pen nhaageen, n D Denmark; en nmarrk; 3Du Duke Clinical Research Institute, Duke University Medical Center, Durham, Clin Cl in nic ical al R esea es earc ea r h In Inst stit st itut it ute, D ukee Un uk Univ iver iv ersi er s ty si yM ediical ed ic l C en nte terr, D urha ur ham ha m, NC m, NC Address for Correspondence: Anne-Marie Schjerning Olsen MD, Research Fellow Department of Cardiology – post 635 Copenhagen University Hospital Gentofte Niels Andersens Vej 65 2900 Hellerup, Denmark Tel: (+45) 60 16 93 40 Fax: (+45) 70201283 E-mail: [email protected] Journal Subject Codes: [4] Acute myocardial infarction; [8] Epidemiology; [27] Other Treatment; [135] Risk Factors 1 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Abstract: Background - The cardiovascular risk after the first myocardial infarction (MI) declines rapidly during the first year. We analyzed whether the cardiovascular risk associated with using nonsteroidal anti-inflammatory drugs (NSAIDs) was associated with the time elapsed following first-time MI. Methods and Results - We identified patients aged 30 years or older admitted with first-time MI in 1997–2009 and subsequent NSAID use by individual-level linkage of nationwide registries of hospitalization and drug dispensing from pharmacies in Denmark. We calculated the incidence ates of death and a composite endpoint of coronary death or nonfatal f recurrent M Iss ass ssoc ss ocia oc iate ia t d te rates MIs associated with NSAID use in 1-year time intervals up to 5 years after inclusion and analyzed risk by using mult mu ltiv lt ivaaria iv aria iabl blee ad bl dju jussted st time-dependent Cox prop por o tional-hazard mo m deels ls.. Of O the 99,187 patients multivariable adjusted proportional-hazard models. nclluded, 43,608 43,6608 (44%) (44 44%) % w ere pres ere pprescribed resscr crib ibed ed NSAIDs NS SAIDss after afterr the the index inde in dexx MI. MI Th T eree we er w ere re 336,747 6 74 6, 7477 de ddeaths eaaths included, were There were an nd 28,693 28,6 28 ,693 ,6 933 coronary cor oron onnary ary deaths deaaths ath or or nonfatal nonf no nfat nf ataal al recurrent rec ecuurre reent n MIs MIs I dduring urin ur ing th in thee 5 years y ar ye ars of ffollow-up. olllo l ww-uup. up Re ela lattive tivee to and Relative trrea eatm tm men entt with w th wi h NSAIDs, NSA AID IDs, s, the he use usee of of any any NSAID NSA AID D in in the the years year ye arss following ar foll fo llow ll ow win ingg MI M was non-current treatment persistently associated with an increased risk of death (hazard ratio (HR) 1.59 (95% confidence interval (CI) 1.49–1.69) after 1 year and HR 1.63 (CI 1.52–1.74) after 5 years) and coronary death or nonfatal recurrent MI (HR 1.30 (CI 1.22–1.39) and HR 1.41 (CI 1.28–1.55). Conclusions - The use of NSAIDs is associated with persistently increased coronary risk regardless of time elapsed after first-time MI. We advise long-term caution in using NSAIDs for patients after MI. Key words: mortality; myocardial infarction; prognosis; non-steroidal anti-inflammatory drugs (NSAIDs); selective cyclooxygenase (COX)-2 inhibitors 2 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Introduction There has been much focus recently on the cardiovascular risk of commonly used nonsteroidal anti-inflammatory drugs (NSAIDs). In 2007, the American Heart Association published a focused update discouraging the use of NSAIDs for patients with established cardiovascular disease.1 Despite this, many patients with cardiovascular disease receive NSAIDs, albeit for shorter periods. We have previously demonstrated not only that NSAIDs are harmful among patients with myocardial infarction(MI) or heart failure but also that this risk is prevalent even after short treatment periods.2 Establishing the cardiovascular safety profile of NSAIDs is imperative since patients with established cardiovascular disease and the general population widely use NSAIDs. MI is associated with a high risk of death and recurrent cardiovascular carrdiov diovvasscu cula larr la events, especially immediately following MI. The riskk of death or recurrent MI after the first MI iss eelevated leva le vatted va ted in thee year yea e r after the initial event but approximates appproximates bas ap baseline sel e inee le lev levels vels after 5-10 years. Because B eccause ca using g NSAIDs NSAI AIDs D aafter fter ft er the the he first fir irst s MI st MI has haas been been n demonstrated demo mons mo nsttraated ated d too increase inccreasee the in the he risk ris iskk of death death eath h or or recurrent understanding this condition whether using ecu curr rrren e t MI, MI, un unde deers rsta taand din ing th tthee llong-term ongon g-tterm term ttrends r nd re ndss in n th his co his cond ndit nd itio it io on is iimportant: mpor mp orta tant ta nt:: wh nt whe ethe heer us usin in ng NSAIDs presents preeseent n s the the greatest grea gr eate test te st risk ris iskk inn the is he first fir irst s year st yea earr after afte af t r the te the first firs fi rsst MI aand nd ddeclines ecli ec line ness as tthe ne h years pas he pass ss or whether using NSAIDs changes the declining incidence of cardiovascular risk in the years after the first MI. If the latter is true, long-term caution with these agents after MI may need to be recommended. This uncertainty prompted us to examine the cardiovascular risk associated with episodes of NSAID use in relation to time elapsed after MI in a nationwide cohort of patients with first-time MI in Denmark. Methods Study Design and Data Sources This study was a nationwide cohort study of patients with first-time MI in Denmark. The study 3 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 period began on January 1, 1997 and ended on December 31, 2009. All residents of Denmark are provided with a permanent and unique personal identification number that enables linkage between administrative registries. Our nationwide cohort study linked national registry data relevant to hospitalization, pharmacy prescription claims and deaths. The Danish National Patient Registry has registered all hospital admissions in Denmark since 1978.3 Each admission is registered by one primary diagnosis and, if appropriate, one or more secondary diagnoses according to the International Classification of Diseases (ICD): by the 8th revision (ICD-8) until 1994 and by the 10th revision (ICD-10) after 1994. The Danish Registry of Medicinal Product Statistics has recorded all prescriptions dispensed from pharmacies in Denmark since 1995.4 The Registry classifies each eacch drug drrugg according acc ccor ordi or ding g too the international Anatomical Therapeutic Chemical (ATC) system and includes information about ab bou outt strength, stre st reeng ngth th, formulation, th fo orm rmulation, date off dispensing and and quantity dispensed. dispeens n ed d. We obtained obt btai aine ai need information info in fo orm mat atio ionn about io abou ab o t patients’ ou patieent ents’ vital vittal status status sta atuss (dead (de deaad ad orr al aliv alive) ve) ffrom ro om th thee Ci Civi Civil il Registration Registry records Regi Re gist gi stra st r ti ra tion on System Sysste tem m through th hro oug ughh Statistics Sttat atis isti is tics ti cs Denmark. Den enma mark ma rk. The rk Th he Centralized Ceent ntrrali lize li z d Civil ze Civi Ci viil Regi R egi gist sttry kkeeps eeps ee pss re ecorrdss ecor on vital status statu us and and registers regi re gist gi sterrs all st al deaths deat de a hs at h within witthi h n 14 days. days da ys. We obtained ys obt b ai aine nedd the ne the cause caus ca usee of death us dea eath th h from the Danish Registry of Causes of Death, in which immediate and underlying causes are recorded using ICD-10. The Integrated Database for Labour Market Research provided data on socioeconomic status. The Database is based on information on taxed income gathered by government tax authorities. We defined socioeconomic status according to average annual gross household income during the 5 years before the index MI excluding the year of the index MI. We divided the cohort into quintiles according to the average annual income of the patients. Participants 4 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 In the Danish National Patient Registry, we identified a cohort of all patients 30 years or older with first-time admission for MI (ICD-10 I21–I22) from 1997 to 2009. The diagnosis of MI has been validated with a specificity exceeding 90%.5 The first admission for MI implied that the Danish National Patient Registry had not registered any earlier admission for MI in the previous 19 years. This method has been used previously.2,6 To avoid selection bias in the exposure allocation caused by the high mortality associated with the MI, the cohort was restricted to individuals still alive 30 days after discharge. We followed the patients until the outcome of interest, emigration or the end of the study period (December 31, 2009), whichever came first. Drug Use pharm rm macciees are are thus thu Denmark’s health care system reimburses some medicine expenses, and all pharmacies equired to register all dispensed drug prescriptions, which ensures complete registration. During required he st stud udyy pe ud peri riodd, th ri thee only NSAID available in De Denm nm mark over the ccounter ount ntter without a prescription the study period, Denmark wass ibuprofen i uprofenn (since ib (sin (s ncee November Novem ovem embe berr 1,, 2001) be 20001) and annd onlyy iin n llow ow ((200-mg) 20020 0-mg mg)) do mg ddoses sess an se nd in llimited im mit ited e qquantity ed uaanttity was and 100 tablets) tab a le lets ts)) at each ts eacch dispensing. disppen disp ensi siing ng. Al Alll ot theer N SA AIDS AID DS analyzed annaly lyzzed ly zed we were ree aavailable vaailaable ablee bby y pr pres escrripttion es tion only. onnly nl (100 other NSAIDS prescription ed al alll fi fill lled ll ed d pprescriptions reesc s ri r pttio ions nss for or N SAID SA IDss ((ATC ID ATC AT C co code de M 0 A) iin 01 n th thee Da ani nish sh R e istry of eg We identifie identified filled NSAIDs M01A) Danish Registry Medicinal Product Statistics. We performed separate analyses for the two selective cyclooxygenase (COX)-2 inhibitors celecoxib (M01AH01) and rofecoxib (M01AH02) and the most commonly used non-selective NSAIDs: ibuprofen (M01AE01), diclofenac (M01AB05) and naproxen (M01AE02). All other NSAIDs excluding glucosamine (M01AX05) were combined in a common group called “other NSAIDs”. Concomitant treatment status was defined for the following cardiovascular drugs: beta-blockers (ATC C07), angiotensin-converting enzyme (ACE) inhibitors and angiotensin-2 receptor blockers ([ARBs] ATC C09), loop diuretics (ATC C03C) and spironolactone (ATC C03D). We defined pharmaceutically treated diabetes using 5 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 prescriptions of glucose-lowering medication (ATC A10), as done previously.7 Dose and Duration of Treatment The Danish Registry of Medicinal Product Statistics does not include information on the prescribed daily dosage of the medication. We therefore estimated the daily dosage when each new prescription was dispensed by calculating the average dosages from up to 7 consecutive prescriptions. This method allowed the dosage to change when a new prescription was dispensed. The method used to determine the dose and treatment duration has been described previously.8, 9 Comorbidity We defined comorbidity using diagnoses at discharge from index MI as specified ed d iin n th the he On Onta Ontario tari ta rio acute MI mortality prediction rules.10 To further enhance the comorbidity index, we identified di isccha harrge rge di dia agnoose sess up to 1 year before the index ex hhospitalization. ospitalization n.11 Too define de high-risk patients, discharge diagnoses wee uused w sed concom concomitant om mitaant ant us usee of lloop oopp di oo ddiuretics uretiics oorr gglucose-lowering ure luccosse-llow lower wering ng ddrugs ru ugs aass pr pro proxies oxies oxi ies fo forr he hear heart artt fail ffailure ai ure orr ddiabetes, iaabe b te tes, s, rrespectively. esppec es pectiv iveely. y.9 Study Outcom Outcome om me The study outcomes examined were: 1) all-cause death and 2) the combined endpoint of death caused by coronary artery disease (coronary death, ICD-10 codes I20–I25) or readmission for nonfatal MI (ICD-10 codes I21–I22). Statistical Analysis We calculated unadjusted rates of death and of the composite endpoint of coronary death or the incidence of nonfatal recurrent MI per 100 person-years for NSAID treatment in general. We used time-dependent Cox proportional hazard models to analyze the risk of death and the risk of the composite endpoint of coronary death or nonfatal recurrent MI associated with NSAID use. 6 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Terms for all individual NSAIDs were included, with the reference composed of those not taking any NSAIDs at that time. Exposure to NSAIDs was included as a time-dependent covariate in the models, ensuring that patients were only considered at risk when exposed to the respective drug. Each individual could have multiple independent treatment courses with the same drug but also with different drugs. To analyze time variation in risk, we defined 6 exposure periods up to 5 years after discharge from the first MI and included them as time-dependent covariates in the proportional hazard models. All models were adjusted for age, sex, year of index hospitalization, concomitant medication, comorbidity and socioeconomic status. Table 1 lists the variables. We found the assumptions on proportional hazards, the linearity of continuous variables and lack of nteraction to be valid unless otherwise indicated. interaction We performed Cox proportional hazard analysis with time-dependent variables and nciide dennce nce rates rate ra tes using te usiing us in the Stata statistical package, packag ge, e version version 11 (Stata-Corp (Staata t -C Cor orpp LP, College Station, incidence TX,, US U A). We pperformed errfo ormed aall ll oother theer st th sta atiisticaal anal lysis aand nd ddata nd ata m anagem anag emen em en nt usin uusing singg th thee SA AS TX, USA). statistical analysis management SAS tat atis isti is tica ti c l software ca soft so ftwa waaree package, pac ackkage g , version v rs ve rsio ionn 9.2 io 9.2 (SAS (SA (S AS Institute, AS Insstiitute tutee, Cary, Cary Ca ry,, NC, NC USA). U A) US A).. statistical Results Patients From 1997 to 2009, 128,418 patients were admitted with first-time MI; 99,187 (77%) were alive 30 days after discharge and were included in the study. Men comprised 64% of the study cohort, and the mean age was 69 (standard deviation (SD) 13.0) years. Table 1 and Figure 1 describe in detail the baseline characteristics of the study cohort and the distribution between NSAID groups. Of the 99,187 patients included, 43,608 (44%) filled at least one prescription of NSAIDs 7 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 during follow-up. Figure 1 in the supplemental material shows the use pattern of NSAIDs in the cohort during the study period. For each separate year, the percentages of the cohort receiving NSAID one year after the first MI increased from 1997 to 2001. From 2001 and especially after 2003, the NSAID use declined. Outcomes During follow-up, 36,747 (37%) patients died and 28,693 (29%) experienced a composite event of coronary death or nonfatal recurrent MI. The death rates per 100 person-years were calculated for NSAID treatment in general and for the individual NSAIDs compared with non-current reatment (Figure 2 and Supplemental Figure 2a–f). f Figure 2 shows a persistently persisten en ntlly increased in ncr crea ease ea sedd se treatment isk of death associated with the use of NSAIDs in general. The results from the Cox risk prrop opor orti or tion ti on nal hazard hazzar ardd model showed a consistently ly y increased inncreased risk off dea ath aamong mong patients proportional death eceeiv i ing any NSAID NSA NS AID D during du uri ring ng the the he 5 years yearss of follow-up yea follow w-up relative w-u relat elativ i e to iv o nnon-current onn-ccurrre rennt nt uuse se ooff NS NSAI AIDs AI Ds. Ds receiving NSAIDs. Use of ddiclofenac iclo ic lofe lo fena nacc was was as sso soci ciiat a ed w ith th ith he high hhighest igh hes e t rrisk iskk ((Figure Figu Fi gure gu re 33). ) ). Use associated with the cal alcu cu ulaate tedd the th he incidence inci in cideenc ci ncee rates r te ra tess of the the composite com mpo p si site te endpoint end ndpo poin po nt of of coronary coro co rona ro nary na ry deaths dea eath t s or th We calculated nonfatal recurrent MIs per 100 person-years for NSAID treatment in general and for the individual NSAIDs (Figure 4 and Supplemental Figure 3a–f). We found a persistently increased rate of the composite outcome associated with using NSAIDs relative to the noncurrent use of NSAIDs. The Cox proportional hazard model showed significantly increased risk among patients using NSAIDs in general and for each drug separately, especially for diclofenac (Figure 5). Sensitivity Analysis The effect of unmeasured confounders cannot be excluded. Our calculations showed that, if 20% 8 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 of the cohort treated with NSAIDs had an unmeasured confounder or a combination of confounders, the confounder would have to elevate the risk by a factor of 4.4–5.4 to explain the increased risk of all-cause mortality observed in our study (Supplemental Figure 4). Using the Wald test, we examined interactions between the use of NSAID and the available covariates and found no clinically important interactions. Terms for all individual NSAIDs were included, with the reference composed of those not taking any NSAIDs at that time. To determine whether the prevalence of use of the various other medicines affected the estimates, we repeated the models with the reference group only comprising the patients who did not take any NSAID at any time. The results remained the same (not shown). To analyze the impact of adjusting the Cox models, Tab ble le 1a) 1a) and and we calculated the crude hazard ratios for all-cause mortality (Supplemental Table f recurrent MI ((Supplemental Supplemental Table 2a). Further, we analyzed the coronary death or nonfatal efffeect ooff ag agee aand ndd se sexx (Supplemental Tables 1b–c 1b b–c – and and 2b–c).Addi d tion di nal ssensitivity ensitivity Cox analysis effect 2b–c).Additional demonstrated dem mo tthat hat ma ha mali l gn gnan ancy an cy, uuse see ooff lo loo loop op oorr gglucose luccose lowe llowering owerring ng ddrugs ru ugs g aatt base bbaseline ase seli line li ne w were ere mo ere more re demonstrated malignancy likely ike kely ly tto o pr pred predict edic ed ictt us usee ooff N NSAIDs SAID SA IDss. ((Supplemental ID Supp Su pp pleeme menttal a Table Tabl able le 3). 3). ). Discussion We examined the cardiovascular risk of NSAIDs in relation to the time elapsed after first-time MI in this nationwide study. We demonstrated that the proportional increase in the risk of death and of a composite endpoint of coronary death or nonfatal recurrent MI in post-MI patients receiving NSAIDs was independent of the time elapsed after their first MI. Notably, the risk associated with using NSAIDs remained virtually unchanged throughout all 5 years after discharge from hospital after the first MI. The risks of cardiovascular mortality and morbidity are well-known complications after MI. The risk is highest soon after the MI but declines as time passes and eventually corresponds 9 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 to the risk of the background population after 5–10 years.12 However, knowledge about the cardiovascular safety of NSAIDs in the years following MI is limited. Although we also previously reported an increased risk of death and reinfarction associated with using NSAID among such patients, we did not specifically analyze whether the risk associated with NSAIDs depended on the time elapsed after the primary event in that study.9 We demonstrated doserelated excess mortality associated with the use of NSAIDs, and our results indicated an acute or subacute effect of NSAIDs, since the events were closely tied to the timing of taking the drugs. To our knowledge, the present study is the first one designed to focus on the risk of death and a composite of coronary death and nonfatal MI associated with NSAIDs in relation to the time elapsed after MI. We found that the use of NSAIDs was associated with a persistently persisstent tenttly y increased inc ncre reas re a ed as cardiovascular risk in the years following MI, thus indicating thatt the cardiovascular risk of akiing N SAID SA IDs du ID duri ring the first year after MI rem mai ainns similar to tthat h t af ha fte terr 5 years. The incidence taking NSAIDs during remains after atees showed ppersistent errsiistten nt in incr crea cr eassed ea sed ab abso soolu ute rrisks iskks during duringg the the 5 years yeaarss among amo ong the the patients pattie i nts nts taking taki ta king ki ng g any any rates increased absolute NS SAI AID, D, w here he reas ass tthe h rrisk he issk am mon o g th thee pa atiien entts ts nnot o ttaking ot ak kin ng NS NSAI AIIDs ddeclines: ecli ec linnes: nes: ffor o eexample, or xaamp mplee, the the rrisk issk sk NSAID, whereas among patients NSAIDs decreases substantially suubsstaanttia iall l y after a teer the af t e first th firs fi rsst MI (Figures (Fi Figu gure gu r s 2 an re aand d 44). ) S ). ince in c tthese ce hese he se ddrugs rugs ru gs aare re w idely used id Since widely and concern about their safety is growing, further investigation is needed to clarify whether longterm caution in using these agents after MI should be recommended. It would seem prudent to limit NSAID use among patients with cardiovascular disease and to get the message out to clinicians taking care of these patients that NSAIDs are potentially harmful, even 5 years after MI. Thus, a persistent focus on the risk of NSAIDs is warranted among patients who have experienced MI. Clinical Implications Along with other recent reports of the adverse cardiovascular effects of NSAIDs, the current data 10 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 provide further evidence that using COX-2 inhibitors and non-selective NSAIDs may increase the risk of severe adverse cardiovascular events.13-23 The results from the past decades have shown that the cardiovascular safety of pharmacotherapy can have massive implications, and the risk–benefit ratio of all NSAIDs and the over-the-counter availability of nonselective NSAIDs such as diclofenac or ibuprofen in many countries should be reconsidered. Physicians should consider alternatives to NSAID therapy based on individual patient characteristics, but it is worrying that the use of NSAIDs remains high24 even though the risk of these drugs is well established in the literature, and particularly worrying that some NSAIDs remain available over the counter.9, 25 Among the conventional nonselective NSAIDs, the cardiovascular risk of naproxen has been much debated, but it is widelyy accepted that naproxen is probably proba baablly the the NSAID NSA NS AID with the safest cardiovascular risk profile, and some reports have even suggested that naproxen protects pr rottec ects ts against aga gain inst cardiovascular in cardi ard ovascular risk.22, 26, 27 In ac accordance cco c rdance with ot other the h r st stud studies, udie ud i s, we found that nnaproxen apr prox o en wass tthe h N he NSAID SAID SA ID Dw with ithh th it thee lo llowest west re wes relative elaativee ccardiovascular ard dio ova vasc sccul ulaar rrisk, isk, is k an and nd th thee re resu results sult su ltss mi lt m might ghtt gh indicate naproxen NSAID ndi dica caate t tthat haat na nap proxen pro oxen n sshould h ul ho uldd be ppreferred refe re ferr r ed iiff NS rr NSA AID D trea ttreatment reaatm tmen en nt cannot cann nnot nn ot bee avoided. av voi oide d d. de d. Nevertheless, Nev verrth theeles elesss, s naproxen was ass aassociated s oc ss ocia i te ia t d wi with th a hhigher i he ig h r ri rrisk sk ooff ggastrointestinal asstrroi oint ntes esti es tiina nal bl blee bleeding e di ding ng tthan hann ro ha rofecoxib, ofe feco coxi co xib, xi b and gastrointestinal bleeding among patients with MI is associated with poor prognosis.28 Indeed, the adverse prognostic effects of gastrointestinal bleeding further support a very conservative approach to using NSAIDs among patients with MI. The absence of large-scale comparative trials of long-term safety and efficacy hampers determining the appropriate pain relief for patients with established cardiovascular disease. Thus, large systematic reviews of the risks and benefits of a broad range of various analgesic agents are warranted.29, 30 Epidemiological studies such as ours cannot establish causality, but in a field of research with no randomized controlled trials, we find it particularly important to report associations between drugs and adverse events 11 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 that can affect patient prognosis. Strengths and Limitations The strength of this study lies in its completeness of data from a whole country. The Danish National Patient Registry and the Danish Registry of Medicinal Product Statistics are known to be accurate.3, 4 This complete registration of all residents of Denmark, including those outside the labour market, diminishes the risk of selection bias. All pharmacies in Denmark are required to register all dispensed drug prescriptions, ensuring complete registration.4,5 During our study period, the only NSAID that was available over the counter in Denmark was ibuprofen (since November 1, 2001), but only in low dosage (200 mg) and with a maximum purchase of 100 ablets. Such NSAID use having major effects in this study is therefore unlikely.. To con o fi on firm rm oour ur tablets. confirm esults, we performed sensitivity analysis with the study period ending on November 1, 2001. results, Th he re esult sult ltss re emaain ined ed unchanged (not shown). Th he m ain limitatio on is iinherent nheerent in the nh The results remained The main limitation ob bserv seer ational na natu ture re ooff the the study. sttud udy. y. Information Info nformation on about aboout important im mport porttan nt clinical clin nic ical al pparameters aram ar amet am eter erss is llacking acki king ki ng g aand nd observational nature ef ffe fect ctts of unmeasured unm nmea eaasuure redd confounders co onf nfou ou und der erss therefore theeref th eref efor oree cannot or cannnot ca not bbee eexcluded. xclu xc lu ude d d. T h eexistence he xisteence xist ence c ooff ssuch ucch ch a effects The confounder oorr co ccombination mbin mb in nattionn of cconfounders onfo on fo ounnde ders rs iiss hi hhighly ghly gh ly uunlikely nllik ikel e y bu el butt no ot im impo poss po s ib ss blee. Ou Ourr not impossible. calculations showed that, if an unmeasured confounder or a combination of confounders were present in 20% of the cohort treated with NSAIDs, the confounder would have to elevate the risk of all-cause mortality by a factor of 4.4–5.4. Confounding by indication may have contributed to the results, and we acknowledge that lack of information about the indications for NSAID treatment is a limitation of the study. However, such confounding probably cannot fully explain our results. To test for these biases, we performed sensitivity analysis that supported our results. Rheumatic diseases are common reasons for NSAID use, and previous studies have reported an increased risk of coronary artery 12 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 disease among patients with rheumatoid arthritis.31-33 We therefore performed the analysis excluding patients with rheumatoid arthritis, which did not change the results (not shown). The risk differed between the individual NSAIDs, which are used for similar indications, and the degree of COX-2 inhibition (as reported in the literature) was clearly correlated with the risk, which also indicates the predominant importance of the drugs (rather than the drug indications) for the results. Moreover, we have previously reported a clear dose-dependent increase in risk associated with NSAIDs.9 Further, given the evidence from randomized controlled trials and other observational studies of adverse cardiovascular effects of selective COX-2 inhibitors and nonselective NSAIDs,13-23 we do not believe that confounding by indication alone could have driven the results. This is further supported by the fact that an unmeasured confounder conffou undderr had had to to elevate the risk more than 4-fold to explain the risk increase by NSAIDs we demonstrated. We found fo oun undd a stronger stro st rong ro ngerr relation relat ela ion to all cause mortality ratherr rat atheerr than cause specific s eccif sp ific ic mortality, which may the cardiovascular bbee ddue u to an uunderestimation ue ndderressti tima maati tion on ooff th he ca card rdiiovaascculaar ccomponent om mpon ponennt nt iin n deat ddeath eathh rregistration egiistr strati ration on ooff th tthese e se ese patients. pati pa tien ti en nts t . We were werre studying stud st uddyiing a group gro roup up of of patients paati tien ents en tss with withh a major maajo jorr coronary coro co ro onaary event eveent and and we we therefore theereefor th eforee find fin itt likely that m most deaths this influenced coronary o t de os deat atths inn th hiss ppopulation opul op u at atio ionn ar io aree in infl flue fl uenc ue nced ed by tthe he cor oron or onar on aryy co ar ccomponent. mpon mp onen on ent. en t Thus when these patients die from pulmonary disease, infection etc –then death was influenced by their heart disease. Aspirin is available over the counter, which explains why the fraction of patients who fill prescriptions for aspirin is relatively low. Another consequence of this is that we do not have information on whether using NSAIDs may lead to prematurely discontinuing aspirin. We assume that most patients who did not fill a prescription for aspirin were treated with over-thecounter aspirin, since medication adherence has been documented to be high among patients in Denmark after MI.27,34 Our use of prescription data without knowing whether patients adhere to 13 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 treatment is another possible bias. Patients may not take their prescribed medication, and unfortunately, there is no way around this problem in observational studies. Nevertheless, nonadherence would tend to dilute the observed association between the exposure and outcome. Information bias is another limitation, as patients do not necessarily take their medicines as prescribed and in temporal sequence. Conclusion In conclusion, our study demonstrated that NSAID use among patients with first-time MI was associated with persistently increased risk of all-cause mortality and of a composite of coronary essulltss ssupport uppo up p rt po death or nonfatal recurrent MI, respectively, for att least 5 years thereafter. Thesee res results MI previous findings that NSAIDs have no apparent safe treatment window among patients with MI. Fuurt rthe herr studies he stud st udie ud ies ar ie aree warranted warranted to evaluate the cardiovascular card rdiiovascular safety rd ty y of NS NSAI A Ds, but at this point Further NSAIDs, he overall o erall evidence ov evid id dencce ce suggests suggge gessts sts advising adviisiing caution advi cautiion n in using usingg NSAIDs NSA AID I s at at all alll times tim mess after aft fter er MI. MI. the urrce c s: Anne-Marie Anne An n -M ne Mar ariee S ch hje j rn rnin in ng Ol Olse s n ha se hass re ece ceiv ived iv ed aan n un unre rest re stri st rict ri cted ct ed d ggrant rant ra n ffrom nt r m the ro Funding Sour Sources: Schjerning Olsen received unrestricted Department of Cardiology, Copenhagen University Hospital Gentofte, Denmark. The funding source had no influence on study design, interpretation of the results or the decision to submit the article. Conflict of Interest Disclosures: None. References: 1. Antman EM, Bennett JS, Daugherty A, Furberg C, Roberts H, Taubert KA. Use of nonsteroidal antiinflammatory drugs: An update for clinicians: A scientific statement from the american heart association. Circulation. 2007;115:1634-1642. 2. Schjerning Olsen AM, Fosbol EL, Lindhardsen J, Folke F, Charlot M, Selmer C, Lamberts M, 14 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Bjerring Olesen J, Kober L, Hansen PR, Torp-Pedersen C, Gislason GH. Duration of treatment with nonsteroidal anti-inflammatory drugs and impact on risk of death and recurrent myocardial infarction in patients with prior myocardial infarction: A nationwide cohort study. Circulation. 2011;123:2226-2235. 3. Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH. The danish national hospital register. A valuable source of data for modern health sciences. Dan Med Bull. 1999;46:263-268. 4. Gaist D, Sorensen HT, Hallas J. The danish prescription registries. Dan Med Bull. 1997;44:445-448. 5. Madsen M, Davidsen M, Rasmussen S, Abildstrom SZ, Osler M. The validity of the diagnosis of acute myocardial infarction in routine statistics: A comparison of mortality and hospital discharge data with the danish monica registry. J Clin Epidemiol. 2003;56:124-130. 6. Gislason GH, Abildstrom SZ, Rasmussen JN, Rasmussen S, Buch P, Gustafsson I, Friberg J, Gadsboll N, Kober L, Stender S, Madsen M, Torp-Pedersen C. Nationwide trends in the prescription of beta-blockers and angiotensin-converting enzyme inhibitors after m myocardial yoca yo card rdia iall infarction nfarction in denmark, 1995-2002. Scand Cardiovasc JJ. 2005;39:42-49. 7. Schramm TK, Gislason GH, Kober L, Rasmussen S, Rasmussen JN, Abildstrom SZ, Hansen ML, Folke F,, Buch P, Madsen M, Vaag A, Torp Torp-Pedersen p-Pedersen C. Diabetes patients pa requiring glucose-lowering gluc gl uccosse-lo e-lo lowe w ri we r ngg ttherapy herapy and nondiabetics withh a prior prior myocardial myocardiial a infarction inf nffar arcction carry the same cardiovascular study million Circulation. 2008;117:1945-1954. caard dio iovascul ullar rrisk: issk: k A ppopulation opul op u at ul atio io on st stud udyy of 33.3 .3 m illliion ppeople. eopl eo ple. pl e Ci Circ cul ulat atio ionn. 20 io 2008 08;1 08 ;117 ;1 17:1 :194 94594 5-19 1 54 19 5 . 8.. F Fosbol EL, Gislason GH, Jacobsen Abildstrom SZ, Hansen ML, Schramm TK, Folke F,, o bol EL os L, Gi isl slaasoon G H, Ja acob o seen S, Ab bild dstro om SZ Z, Ha Han nsen nM L, S chraamm chra mm T K, F olkke F ol Sorensen nonSore So rens re nsen ns en R, R, Rasmussen Rasm Rasm mus u se senn JN, JN Kober Kobe Ko berr L, be L, Madsen Mad adssen sen M, Torp-Pedersen Torrpp-Pe Pedders Pe ders rsen en n C. C. The Th he pattern patt pa t errn of uuse tt see ooff no on-steroidal anti-inflammatory drugs study million terroi oida dall an anti-i inf n laamm mmatorry dr drug ugss (n ((nsaids) said ds) fr from om 11997 9977 to 22005: 005: 00 5: A nnationwide atio at ionw nwid idee st stud udy on 4.66 m i liion il o people. Pharmacoepidemiol Pharrma maco co oep epid id dem e io ol Dr Drug ug SSaf. af 2008;17:822-833. af 2008 20 08;1 08 ;1 17: 7 82 8222 83 28333. 3. 9. Gislason GH, Jacobsen S, Rasmussen JN, Rasmussen S, Buch P, Friberg J, Schramm TK, Abildstrom SZ, Kober L, Madsen M, Torp-Pedersen C. Risk of death or reinfarction associated with the use of selective cyclooxygenase-2 inhibitors and nonselective nonsteroidal antiinflammatory drugs after acute myocardial infarction. Circulation. 2006;113:2906-2913. 10. Tu JV, Austin PC, Walld R, Roos L, Agras J, McDonald KM. Development and validation of the ontario acute myocardial infarction mortality prediction rules. J Am Coll Cardiol. 2001;37:992-997. 11. Rasmussen S, Zwisler AD, Abildstrom SZ, Madsen JK, Madsen M. Hospital variation in mortality after first acute myocardial infarction in denmark from 1995 to 2002: Lower short-term and 1-year mortality in high-volume and specialized hospitals. Med Care. 2005;43:970-978. 12. Solomon SD, Zelenkofske S, McMurray JJ, Finn PV, Velazquez E, Ertl G, Harsanyi A, Rouleau JL, Maggioni A, Kober L, White H, Van de Werf F, Pieper K, Califf RM, Pfeffer MA. Sudden death in patients with myocardial infarction and left ventricular dysfunction, heart 15 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 failure, or both. N Engl J Med. 2005;352:2581-2588. 13. Bombardier C, Laine L, Reicin A, Shapiro D, Burgos-Vargas R, Davis B, Day R, Ferraz MB, Hawkey CJ, Hochberg MC, Kvien TK, Schnitzer TJ. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. Vigor study group. N Engl J Med. 2000;343:1520-1528, 1522 p following 1528. 14. Bresalier RS, Sandler RS, Quan H, Bolognese JA, Oxenius B, Horgan K, Lines C, Riddell R, Morton D, Lanas A, Konstam MA, Baron JA. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med. 2005;352:1092-1102. 15. Solomon SD, McMurray JJ, Pfeffer MA, Wittes J, Fowler R, Finn P, Anderson WF, Zauber A, Hawk E, Bertagnolli M. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med. 2005;352:1071-1080. 16. Helin-Salmivaara A, Virtanen A, Vesalainen R, Gronroos JM, Klaukka T, Idanpaan-Heikkila JE, Huupponen R. Nsaid use and the risk of hospitalization for first myocardial infarction in the J 20 006 06;2 ;27: 7:16 1657 57general population: A nationwide case-control study from finland. Eur Heart J. 2006;27:16571663. 17. McGettigan P, Henry D. Cardiovascular risk k and inhibition of cyclooxygenase: A systematic eview of the observational studies of selective and nonselective inhibitors of cyclooxygenase 2. review JAMA. JA AMA MA.. 20 2006 2006;296:1633-1644. 06;2 06 ;296 6:1 :163 6 3-1644. 18. 18. K Kearney earney PM PM, M, Baig B Baigent aig igen nt C, C, G Godwin odwi od winn JJ,, H wi Halls allls H, E Emberson mbeers erson son JR JR, R, Patr P Patrono atron onoo C. C. D Do o se sele selective lect le ctiv ct ivee cycl ccycloycl clo-oxygenase-2 increase risk oxxyggenase-2 2 inhibitors i hibitorrs and in andd traditional traaditi t onnal non-steroidal non-ssteeroid dal anti-inflammatory an nti ti-i -inf -i n laamm nf matoryy drugs druggs in drug incre easse se tthe he ris sk of atherothrombosis? atthe hero ro oth t ro romb mbos mb osiis? Me Meta-analysis etaa-a -ana naly lysi ly siss of rrandomised si ando ando dom mise seed tria ttrials. riaalss. BM BMJ. BMJ J. 20 200 2006;332:1302-1308. 06;3 06;3 ;332 32:130 32 3022-13 2-13 1308 08 8. 19. Cannon C CP, P, C Curtis urti ur t s SP S SP,, Fi Fitz FitzGerald tzGe Gera Ge raald l G GA, A K A, Krum rum ru mH H,, Ka Kaur aur A A,, Bo Bolo Bolognese l gn gnes esee JA es JA,, Re Reic Reicin icin ic in A AS, S, Bombardier Bomb Bo mbar ardi dier er C, C Weinblatt Wein We inbl blat attt ME, ME van van der der Heijde Heij He ijde de D, D Erdmann Erdm Er dman annn E, E Laine Lai ainne L. L Cardiovascular Car ardi diov ovas ascu cula larr outcomes outc ou tcom omes es with etoricoxib and diclofenac in patients with osteoarthritis and rheumatoid arthritis in the multinational etoricoxib and diclofenac arthritis long-term (medal) programme: A randomised comparison. Lancet. 2006;368:1771-1781. 20. Juni P, Nartey L, Reichenbach S, Sterchi R, Dieppe PA, Egger M. Risk of cardiovascular events and rofecoxib: Cumulative meta-analysis. Lancet. 2004;364:2021-2029. 21. Mamdani M, Juurlink DN, Lee DS, Rochon PA, Kopp A, Naglie G, Austin PC, Laupacis A, Stukel TA. Cyclo-oxygenase-2 inhibitors versus non-selective non-steroidal anti-inflammatory drugs and congestive heart failure outcomes in elderly patients: A population-based cohort study. Lancet. 2004;363:1751-1756. 22. Solomon DH, Glynn RJ, Levin R, Avorn J. Nonsteroidal anti-inflammatory drug use and acute myocardial infarction. Arch Intern Med. 2002;162:1099-1104. 23. Strand V. Are cox-2 inhibitors preferable to non-selective non-steroidal anti-inflammatory 16 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 drugs in patients with risk of cardiovascular events taking low-dose aspirin? Lancet. 2007;370:2138-2151. 24. Http://www.Medstat.Dk/da/viewdatatables. Accessed May 29, 2012. 25. Fosbol EL, Folke F, Jacobsen S, Rasmussen JN, Sorensen R, Schramm TK, Andersen SS, Rasmussen S, Poulsen HE, Kober L, Torp-Pedersen C, Gislason GH. Cause-specific cardiovascular risk associated with nonsteroidal antiinflammatory drugs among healthy individuals. Circ Cardiovasc Qual Outcomes. 2010;3:395-405. 26. Rahme E, Pilote L, LeLorier J. Association between naproxen use and protection against acute myocardial infarction. Arch Intern Med. 2002;162:1111-1115. 27. Watson DJ, Rhodes T, Cai B, Guess HA. Lower risk of thromboembolic cardiovascular events with naproxen among patients with rheumatoid arthritis. Arch Intern Med. 2002;162:1105-1110. Jorgense senn C, Madsen Mad adse sen 28. Sorensen R, Hansen ML, Abildstrom SZ, Hvelplund A, Andersson C, Jorgensen ati tien en ntss w ithh ac it aacute ute JK, Hansen PR, Kober L, Torp-Pedersen C, Gislason GH. Risk of bleeding in pa patients with myocardial infarction treated with different combinations of aspirin, clopidogrel clopidogrel, l, an andd vi vitamin ita tami minn k mi antagonists in denmark: A retrospective analysis of nationwide registry data. Lancet. 2009;374:1967-1974. 2009;3 ;374:196 967-19 974 7 . 29. 299. Ray Ra WA WA. A. Ca Card Cardiovascular rdio rd iova vaasc s ul ular ar ssafety afet af etyy of nnsaids. et said sa ids. s. BMJ. BM MJ. 20 2011;342:c6618. 0111;3 ;342 4 :cc6618 6618. 30. Villiger PM, Egger Juni 30 0. Trelle Trelle S,, Reichenbach R ichenb Re bacch S, S, Wandel Wan ande deel S, S, Hildebrand Hilddeb debran nd P, Tschannen Tscha scha h nneen B, B, V illlig iger er P M E M, ggerr M, Ju gg uni P. C Cardiovascular ardi ar d ov di ovas ascu as cula laar sa safe safety feety y ooff no nonnon-steroidal n-st ster st eroi er oiddal dal an ant anti-inflammatory ti-i -inf -i n lamm nf mmat mm attor oryy dr drug drugs: uggs:: N Network ettwork km meta-analysis. etaet a-aanal aanallyssis is.. BM BMJ BMJ. J. 2011;342:c7086. 2011 11;3 ;342 42:c :c70 7 86 86. 31. So Solomon DH, Ka Karlson EW, Ri EB, Cannuscio CC, Mandl LA, Manson JE, St Stampfer MJ, 31 Solo lomo monn DH Karl rlso sonn EW Rimm mm E B C annu an nusc scio io C C M andl an dl L A M anso an sonn JE Stam ampf pfer er M J Curhan GC. Cardiovascular morbidity and mortality in women diagnosed with rheumatoid arthritis. Circulation. 2003;107:1303-1307. 32. Krishnan E, Lingala VB, Singh G. Declines in mortality from acute myocardial infarction in successive incidence and birth cohorts of patients with rheumatoid arthritis. Circulation. 2004;110:1774-1779. 33. Sattar N, McCarey DW, Capell H, McInnes IB. Explaining how "high-grade" systemic inflammation accelerates vascular risk in rheumatoid arthritis. Circulation. 2003;108:2957-2963. 34. Madsen JK, Grande P, Saunamaki K, Thayssen P, Kassis E, Eriksen U, Rasmussen K, Haunso S, Nielsen TT, Haghfelt T, Fritz-Hansen P, Hjelms E, Paulsen PK, Alstrup P, Arendrup H, Niebuhr-Jorgensen U, Andersen LI. Danish multicenter randomized study of invasive versus conservative treatment in patients with inducible ischemia after thrombolysis in acute myocardial infarction (danami). Danish trial in acute myocardial infarction. Circulation. 1997;96:748-755. 17 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Table 1. Baseline characteristics of the total study population and individual treatment groups. Exposure group Characteristic Total population N (%) No NSAID Overall NSAID Rofecoxib Celecoxib Ibuprofen Diclofenac Naproxen N (%) N (%) N (%) N (%) N (%) N (%) N (%) Other NSAIDs N (%) Total patients 99,187(100) 55,579(56.0) 43,608(44.0) 3,635(3.7) 3,761(3.8) 26,428(26.6) 14,543(14.7) 2,406(2.4) 13,266(13.4) Mean age (SD), y 68.9(13.2) 70.2(12.9) 66.1(12.9) 70.1(12.2) 70.2(11.8) 64.4(12.9) 64.7(12.5) 64.7( 12.4) 67.4( 12.5) Women 36,102(36.4) 20,853(37.5) 15,249(35.0) 1,760(49.0) 1,842(49.0) 8,497(32.2) 4,683(32.2) 732(30.4) 5,229(39.4) Men 63,085(63.6) 34,726(62.5) 28,359(65.0) 1,875(51.0) 1,919(51.0) 17,931(67.8) 9,860(67.8) 1,674(69.6) 8,037(60.6) Co-morbidity ty Cardiac arrhythmias 10,515(10.6) 6,977(12.6) 3,538(8.1) 352(9.7) 344(9.2) 1,873(7.1) 1,095(7.5) 156(6.5) 1,090(8.2) rrhythmias 1,095 95(7 (7.5 (7 .5)) .5 156( 15 6(6. 6( 6.5) 6. 5) Peripherall vascular disease 1,543(1.6) 1,005(1.8) 538(1.2) 55(1.5) 63(1.7) 289(1.1) 162(1.1) 174(1.3) 162( 2(1. 1 1)) 224(1.0) 4(1. 4( 1 0) 1. 0 Cerebral vascular disease 4,646(4.7) 162(4.3) 456(3.1) 456(3.4) 3,136(5.6) 1,510(3.5) 164(4.5) 799(3.0) 456( 6(33.1) 1) 665(2.7) 5(22.7) 5( 7 Diabetes with complications 4,470(4.5) 149(4.1) 472(3.6) 2,773(5.0) 1,697(3.9) 145(3.9) 1,014(3.8) 555(3.8) 91(3.8) Acute renal 854(0.9) 651(1.2) 49(0.4) nal failure 203(0.5) 23(0.7) 17(0.5) 105(0.4) 49(0.3) 9(0.4) Chronic renal 1,338(1.4) 1,044(1.9) 26(0.7) 157(0.6) 75(0.5) 71(0.6) enal failure 294(0.7) 20(0.5) 10(0.4) Malignancy 680(0.7) 481(0.9) 199(0.5) 10(0.3) 16(0.4) 109(0.4) 63(0.4) 6(0.3) 51(0.4) cy 6880( 0(0. 0.7) 10( 0((0.3) 3) 109( 9(0. 0 4) 0. Shock 1,001(1.0) 309(0.7) 32(0.9) 152(0.6) 94(0.7) 1,00 ,00 001( 1 1. 1 0) 692(1.3) 32( 2(0. 0 9)) 22(0.6) 152 2(0 (0.6 0.66) 95(0.7) 27(1.1) COPD 994(1.0) 334(0.8) 28(0.7) 167(0.6) 19(0.8) 120(0.9) 994( 99 4(1. 1.0) 0) 6660(1.2) 60(1.2 60 .2)) 334( 33 4(0.8) 8) 331(0.9) 1 0. 1( 0 9) 28(0 28 (0.7) 16 67( 7(0. 0.6) 6) 997(0.7) 7(0. 7( 0.7) 7) 19( 9(0.8) 8) Gastric ulcer 1,516(1.5) 985(1.8) 531(1.2) 82(2.3) 68(1.8) 259(1.0) 21(0.9) 179(1.4) lce c r 1,51 1, 5 6( 6 1. 1.5) 5) 985 85(1 85 (1.8 .88) 531( 53 1(1. 1.2) 82( 2 2.3) 68(1 68 ( .8 (1 .8) 8 2 9( 25 9(1. 1 0) 0) 1137(0.9) 37(0 37 (0.9 .9) 9) 21(0 21 ( .9 (0 .9)) Concomitant medical ntt medi edical treatment Beta-blockers 73,313(73.9) 40,251(72.4) 33,062(75,8) ckers ck k s 73 ,313(73. 3 9) 40,25 9) 251( 1 72 7 .4)) 33 33,062 6 (7 (75,8) 22,478(68.2) ,478(6 (68. 8 2) 22,598(69.1) ,598 ,5 98(6 98 ( 9. 9 1) 220,409(77.2) 0,409(77.2 7. ) 111,244(77.3) 1,244(77. 7 3) 3) 11,826 ,826 8 ((75.9) 75.9) 75 9) 99,926 (74.8) ACE inhibitors 44,834(45.2) 26,248(47.2) (40.7) 1,007(41.9) bito bi it rss 44,83 8 4( 4 45.2) 2) 26,2 248 48(47. 47. 7 2) 2 118,586(42.6) 8 586(42 8, (4 .6) 1, 11,434(39.5) 434(39 (39.5) 5 11,529 1, ,52 5299 (40. 0 7) 111,121(42.1) 1,121(42.1 2.1)) 55,993(41.2) 5, 993(41 99 41.2) 2) 1,00 0 7(41.9) 9) 55,546(41.8) Statins 61,269(61.8) 9,068(62.4) 1,446(60.1) 61 ,269(6 ,2 9(61. 1.8) 8 334,447(62.0) 8) 4,447( 4, 7(62 (62.0) 0) 26 226,822(61.5) ,822 ,8 2 (61. 22 61. 1 5) 11,485(40.9) ,485 ,4 85(4 ( 0. (4 0 9) 9 11,601(42.6) ,601 ,6 01(4 01 ( 2.6) 2.6) 6 117,119(64.8) 7 11 7, 1 9( 9(64.8 4.8)) 9,06 9, 0 8((62 06 6 .4 .4) 1,44 4 6(60 44 60.1) 0 ) 77,582(57.2) ASA 57,450(57.9) 1,386(38.1) 1,489(39.6) 14,578(55.2) 7,591(52.2) 577,450 450 50(5 (577.9) (5 7.9) 34,004(61.2) 34, 4 004( 4(61 1.2 .2) 2) 23,446(53.8) 2 ,446(5 23 6(5 (533.8) 3.8) 8 1,3 386 86(3 6(3 (388.1) 1) 1,4 489 8 (3 (39. 39. 9 6) 6 14, 4 57 578( 78( 8(55 55.2 2) 7,59 7,59 591( 1(52 1( 5 .2 52 2) 1,250(52.0) 1,25 250( 25 0(52 0( 52.0) 0) 6,671(50.3) 6 Clopidogrel 44,642(45.0) 27,190(48.9) 17,452(40.0) 605(16.6) 773(20.6) 11,116(42.1) 5,515(37.9) 873(36.3) rell re 44,6 44 , 42 ,6 42(4 ( 5. (4 5.0) 0)) 27 27,1 , 90 ,1 90(4 ( 8. (4 8.9) 9)) 17 17,4 , 52 ,4 52(4 ( 0. (4 0.0) 0)) 605( 60 5((16 16.6 .6)) 773( 77 3((20 20.6 .6)) 11,1 11 , 16 ,1 16(4 ( 2. (4 2.1) 1)) 5, ,51 515( 5((37 37.9 .9)) 873( 87 3((36 36.3 .3)) 44,682(35.3) Spironolactone 5,444(9.8) 2,993(6.9) 322(8.6) 1,654(6.3) 873(6.0) 889(6.7) actone 88,437(8.5) ,437 ,4 37(8 37 (8 8.5 .5)) 5, ,44 444( 4(9. 4( 9 8)) 2,9 , 933(6 ( .99) 2290(8.0) 29 0((8. 8.0) 0 0) 322 22(8 (8.6 (8 . ) .6 1,65 1, 6554( 4 6. 6 3) 873 73(6 (6.0 (6 .0 0) 165(6.9) Loop-diuretics retics 338,587(38.9) 8,5587( 587(38 38 8.9) 9) 23 223,658(42.6) ,6 658 58(42.6) 8 6) 114,929(34.2) 4 929( 4, 29(34 (34 3 .2) 2) 1, 11,647(45.3) 647( 64 7 45.3) 7( 3) 11,675(44.5) ,67 6 5( 67 5(44 4 .5) 44 5) 88,214(31.1) ,21 2 4( 21 4(31 31.1 1) 44,503(31.0) ,50 5 3( 50 3(31 3 .0 31 0) 794(33.0) 44,856(36.6) Glucose lowering drugs 12,176(12.3) 7,238(13.0) 4,938(11.3) 429(11.8) 442(11.8) 2,956(11.2) 1,653(11.4) 257(10.7) 1,510(11.4) PCI 32,094(32,4) 18,734(33.7) 13,360(30.6) 558(15.4) 607(16.1) 8,645(32.7) 4,435(30.5.) 736(30.1) 3,588(27.1) Socioeconomic factors Yearly family income in quintiles 0 19,347(19.5) 11.255(20.3) 8,092(18.6) 1,052(28.9) 1,018(27.1) 4,428(16.8) 2.478(17.0) 462(19.2) 2,895(21.8) 1 19.302(19.5) 11,362(20.4) 7,940(18.2) 883(24.3) 946(25.2) 4,297(16.3) 2,469(17.0) 401(16.7) 2,649(20.0) 2 19,793(20.0) 11,285(20.3) 8,508(19.5) 641(17.6) 655(17.4) 5,165(19.5) 2,825(19.4) 435(18.1) 2,638(19.9) 3 20,177(20.3) 10,846(19,5) 9,331(21.4) 619(17.0) 669(17.8) 6,033(22.8) 3,301(22.7) 586(24.2) 2,675(20.2) 4 (highest) 20,568(20.7) 10,831(19.5) 9,737(22.3) 440(12.1) 473(12.6) 6,505(24.6) 3,470(23.9) 522(21.7) 2,409(18.3) SD: standard deviation; MI: acute myocardial infarction; COPD: chronic obstructive pulmonary disease; ACE-inhibitors: angiotensin converting enzyme-inhibitors; ASA: acetylsalicylic acid; PCI: percutaneous coronary intervention 18 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 DOI: 10.1161/CIRCULATIONAHA.112.112607 Figure Legends: Figure 1. Study flow diagram. Figure 2. Death rates per 100 person-years during treatment with NSAIDs in general. The treatment periods are divided for each year up to 5 years after myocardial infarction (MI). The vertical bars indicate 95% confidence intervals. Figure 3. Time-dependent Cox proportional-hazard analysis of the risk of death according to the ime of NSAID treatment among patients with prior myocardial infarction (MI).. time Fiigu gure re 44.. Incidence Inci In c deenc ncee rates of coronary death (CHD HD)) or nonfatal my HD yocar arrdi dial a infarction (MI) per Figure (CHD) myocardial 100 pe pperson-years rson-yea ears rss dduring urin ur i g trea in tr rea eatm tmen tm en nt w itth N SA AIDs in gen en neral erall. Th T reaatm ment pe peri riod odds ar aaree di ivi vide d d de 100 treatment with NSAIDs general. Thee tr treatment periods divided for fo or ea each c yyear ch earr upp tto ea o 5 yyears eaars ars af after fte terr MI MI. Th The he ve ver vertical rtiica call ba bars r iindicate ndic nd icat ic atee 95 95% 5% co cconfidence nfid nf i en id ncee iintervals. nter nt ervaals er ls.. Figure 5. Time-dependent Cox proportional-hazard analysis of the risk of coronary death or nonfatal myocardial infarction (MI) according to the time of NSAID treatment among patients with prior MI. 19 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 “SUPPLEMENTAL MATERIAL” Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental figure legends Supplemental Figure 1 The utilization pattern of NSAID treatment. Percent of the cohort receiving NSAIDs, a year after discharge from myocardial infarction, in the study period 1997-2008 Supplemental Figure 2a- f. Incidence rates of death per 100 person years during treatment with the individual NSAID. Treatment periods are divided for each year up to 5 years after myocardial infarction (MI). Vertical bars indicate 95% confidence intervals. Supplemental Figure 3a-f Incidence rates of coronary death (CHD) or nonfatal myocardial infarction per 100 person years during treatment with the individual NSAID. Treatment periods are divided for each year up to 5 years after myocardial infarction (MI). Vertical bars indicate 95% confidence intervals Supplemental Figure 4 Required size for an unmeasured confounder 20%. OREC =association between drug use category and confounder; RRCD=association between confounder and all cause mortality. Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental Figure 1 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental Figure 2a-f Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental Figure 3a-f Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental Figure 4 Required size for an unmeasured confounder 20% 10 ARR=1.80 ARR=1.70 8 OREC 6 4 2 0 0 2 4 6 8 10 RRCD Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental table 1a: Crude Hazard ratios for all-cause mortality NSAID Crude Hazard p-value Confidence intervals 95 % ratio for allcause mortality Overall NSAID 1 year 1.69 <0.001 1.59 1.80 2 year 1.95 <0.001 1.81 2.11 3 year 1.88 <0.001 1.72 2.05 4 year 1.86 <0.001 1.69 2.05 5 year 1.78 <0.001 1.60 1.98 >5 years 1.60 <0.001 1.49 1.71 1 year 2.50 <0.001 2.07 3.02 2 year 2.76 <0.001 2.16 3.52 3 year 2.72 <0.001 2.04 3.61 4 year 3.07 <0.001 2.29 4.10 5 year 3.21 <0.001 2.34 4.41 >5 years 2.78 <0.001 1.97 3.94 1 year 2.68 <0.001 2.23 3.22 2 year 2.38 <0.001 1.86 3.04 3 year 2.74 <0.001 2.12 3.55 4 year 2.33 <0.001 1.71 3.18 Rofecoxib Celecoxib Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 5 year 2.33 <0.001 1.66 3.27 >5 years 2.68 <0.001 1.99 3.60 1 year 1.37 <0.001 1.23 1.50 2 year 1.71 <0.001 1.52 1.92 3 year 1.67 <0.001 1.46 1.90 4 year 1.50 <0.001 1.29 1.74 5 year 1.46 <0.001 1.24 1.72 >5 years 1.40 <0.001 1.27 1.53 1 year 2.03 <0.001 1.78 2.31 2 year 2.42 <0.001 2.07 2.83 3 year 1.91 <0.001 1.57 2.32 4 year 2.38 <0.001 1.98 2.87 5 year 2.23 <0.001 1.81 2.75 >5 years 1.74 <0.001 1.52 1.99 1 year 1.68 0.001 1.25 2.25 2 year 1.05 0.841 0.64 1.72 3 year 1.13 0.658 0.66 1.95 4 year 1.37 0.254 0.80 2.37 5 year 1.62 0.097 0.92 2.85 >5 years 1.08 0.684 0.75 1.57 Ibuprofen Diclofenac Naproxen Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Other NSAID 1 year 1.53 <0.001 1.35 1.74 2 year 1.74 <0.001 1.49 2.04 3 year 1.78 <0.001 1.49 2.12 4 year 1.83 <0.001 1.51 2.21 5 year 1.65 <0.001 1.33 2.05 >5 years 1.75 <0.001 1.55 1.98 Supplemental table 1b: Hazard ratios only adjusted of age for all cause mortality NSAID Crude Hazard p-value Confidence intervals 95 % ratio for all cause mortality Overall NSAID 1 year 1.64 <0.001 1.54 1.74 2 year 1.89 <0.001 1.75 2.05 3 year 1.87 <0.001 1.71 2.04 4 year 1.87 <0.001 1.70 2.06 5 year 1.78 <0.001 1.60 1.99 >5 years 1.67 <0.001 1.56 1.78 1 year 1.81 <0.001 1.50 2.19 2 year 2.18 <0.001 1.71 2.77 Rofecoxib Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 3 year 2.17 <0.001 1.63 2.88 4 year 2.32 <0.001 1.73 3.10 5 year 2.41 <0.001 1.76 3.31 >5 years 2.15 <0.001 1.52 3.04 1 year 2.01 <0.001 1.67 2.41 2 year 1.76 <0.001 1.37 2.25 3 year 2.12 <0.001 1.63 2.74 4 year 1.92 <0.001 1.41 2.63 5 year 1.85 <0.001 1.32 2.59 >5 years 1.91 <0.001 1.42 2.56 1 year 1.48 <0.001 1.34 1.63 2 year 1.88 <0.001 1.68 2.12 3 year 1.87 <0.001 1.64 2.13 4 year 1.67 <0.001 1.44 1.93 5 year 1.61 <0.001 1.36 1.89 >5 years 1.56 <0.001 1.42 1.72 1 year 2.32 <0.001 2.04 2.64 2 year 2.75 <0.001 2.35 3.21 3 year 2.21 <0.001 1.81 2.68 4 year 2.75 <0.001 2.28 3.31 Celecoxib Ibuprofen Diclofenac Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 5 year 2.58 <0.001 2.09 3.19 >5 years 2.08 <0.001 1.82 2.38 1 year 1.81 <0.001 1.35 2.42 2 year 1.14 <0.001 0.70 1.86 3 year 1.19 <0.001 0.69 2.05 4 year 1.41 <0.001 0.82 2.44 5 year 1.90 <0.001 1.08 3.36 >5 years 1.12 <0.001 0.77 1.63 1 year 1.34 <0.001 1.18 1.52 2 year 1.47 <0.001 1.26 1.72 3 year 1.58 <0.001 1.32 1.89 4 year 1.68 <0.001 1.39 2.03 5 year 1.53 <0.001 1.23 1.90 >5 years 1.62 <0.001 1.43 1.83 Naproxen Other NSAID Supplemental table 1c: Hazard ratios only adjusted of sex for all-cause mortality NSAID Crude Hazard p-value Confidence intervals 95 % <0.001 1.55 ratio for allcause mortality Overall NSAID 1 year 1.65 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 1.76 2 year 1.92 <0.001 1.77 2.07 3 year 1.85 <0.001 1.70 2.03 4 year 1.84 <0.001 1.67 2.02 5 year 1.76 <0.001 1.58 1.96 >5 years 1.59 <0.001 1.48 1.70 1 year 2.31 <0.001 1.91 2.80 2 year 2.59 <0.001 2.03 3.30 3 year 2.55 <0.001 1.92 3.39 4 year 2.89 <0.001 2.16 3.87 5 year 3.00 <0.001 2.19 4.12 >5 years 2.62 <0.001 1.85 3.70 1 year 2.49 <0.001 2.07 2.99 2 year 2.21 <0.001 1.73 2.83 3 year 2.56 <0.001 1.98 3.32 4 year 2.18 <0.001 1.60 2.98 5 year 2.21 <0.001 1.57 3.09 >5 years 2.51 <0.001 1.87 3.38 1 year 1.36 <0.001 1.23 1.49 2 year 1.71 <0.001 1.52 1.92 3 year 1.68 <0.001 1.47 1.92 Rofecoxib Celecoxib Ibuprofen Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 4 year 1.50 <0.001 1.29 1.74 5 year 1.47 <0.001 1.25 1.73 >5 years 1.40 <0.001 1.27 1.54 1 year 2.04 <0.001 1.79 2.32 2 year 2.45 <0.001 2.10 2.86 3 year 1.96 <0.001 1.61 2.38 4 year 2.43 <0.001 2.01 2.92 5 year 2.28 <0.001 1.85 2.81 >5 years 1.77 <0.001 1.55 2.03 1 year 1.68 0.001 1.25 2.25 2 year 1.06 0.805 0.65 1.74 3 year 1.14 0.635 0.66 1.97 4 year 1.38 0.245 0.80 2.38 5 year 1.63 0.090 0.93 2.88 >5 years 1.08 0.679 0.75 1.57 1 year 1.48 <0.001 1.30 1.68 2 year 1.68 <0.001 1.43 1.97 3 year 1.71 <0.001 1.43 2.05 4 year 1.76 <0.001 1.46 2.13 5 year 1.59 <0.001 1.28 1.97 Diclofenac Naproxen Other NSAID Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 >5 years 1.70 <0.001 1.50 1.92 Supplemental table 2a: Crude Hazard ratios for coronary death or Re-MI NSAID Crude Hazard p-value Confidence intervals 95 % ratio for coronary death or recurrent myocardial infarction Overall NSAID 1 year 1.38 <0.001 1.30 1.48 2 year 1.76 <0.001 1.60 1.93 3 year 1.66 <0.001 1.48 1.86 4 year 1.45 <0.001 1.27 1.65 5 year 1.42 <0.001 1.22 1.65 >5 years 1.41 <0.001 1.29 1.56 1 year 1.96 <0.001 1.59 2.41 2 year 3.11 <0.001 2.39 4.05 3 year 2.21 <0.001 1.50 3.25 4 year 2.40 <0.001 1.58 3.65 5 year 2.61 <0.001 1.64 4.16 >5 years 2.19 0.005 1.27 3.77 Rofecoxib Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Celecoxib 1 year 2.05 <0.001 1.67 2.51 2 year 2.11 <0.001 1.57 2.85 3 year 2.23 <0.001 1.57 3.18 4 year 1.78 0.012 1.13 2.79 5 year 1.77 0.027 1.07 2.95 >5 years 1.17 0.629 0.63 2.17 1 year 1.21 <0.001 1.10 1.34 2 year 1.57 <0.001 1.37 1.82 3 year 1.37 <0.001 1.15 1.63 4 year 1.24 0.035 1.02 1.52 5 year 1.19 0.129 0.95 1.50 >5 years 1.33 <0.001 1.16 1.52 1 year 1.46 <0.001 1.26 1.70 2 year 1.83 <0.001 1.49 2.24 3 year 1.44 <0.001 1.09 1.88 4 year 1.94 <0.001 1.50 2.50 5 year 1.51 0.012 1.09 2.08 >5 years 1.58 <0.001 1.31 1.91 1.55 0.004 1.16 2.09 Ibuprofen Diclofenac Naproxen 1 year Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 2 year 1.54 0.066 0.97 2.45 3 year 1.00 0.998 0.50 2.00 4 year 1.12 0.766 0.53 2.35 5 year 1.49 0.295 0.71 3.12 >5 years 1.15 0.575 0.71 1.85 1 year 1.27 <0.001 1.11 1.45 2 year 1.50 <0.001 1.24 1.83 3 year 1.91 <0.001 1.56 2.36 4 year 1.29 0.071 0.98 1.71 5 year 1.54 0.003 1.16 2.05 >5 years 1.25 0.028 1.02 1.52 Other NSAID Supplemental table 2b: Hazard ratios only adjusted of age for coronary death or Re-MI Overall NSAID Hazard ratio for p-value Confidence intervals 95 % coronary death or recurrent myocardial infarction 1 year 1.36 <0.001 1.27 1.45 2 year 1.72 <0.001 1.56 1.88 3 year 1.65 <0.001 1.47 1.84 4 year 1.44 <0.001 1.26 1.65 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 5 year 1.42 <0.001 1.22 1.65 >5 years 1.44 <0.001 1.31 1.58 1 year 1.57 <0.001 1.27 1.93 2 year 2.61 <0.001 2.00 3.40 3 year 1.85 0.002 1.25 2.72 4 year 1.95 0.002 1.28 2.97 5 year 2.09 0.002 1.31 3.33 >5 years 1.79 0.036 1.04 3.09 1 year 1.67 <0.001 1.36 2.05 2 year 1.71 <0.001 1.27 2.31 3 year 1.85 0.001 1.30 2.64 4 year 1.52 0.069 0.96 2.39 5 year 1.54 0.095 0.93 2.56 >5 years 0.93 0.821 0.50 1.73 1 year 1.28 <0.001 1.16 1.41 2 year 1.66 <0.001 1.45 1.91 3 year 1.46 <0.001 1.23 1.74 4 year 1.32 0.006 1.08 1.62 5 year 1.26 0.047 1.00 1.58 >5 years 1.41 <0.001 1.24 1.62 Rofecoxib Celecoxib Ibuprofen Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Diclofenac 1 year 1.58 <0.001 1.36 1.83 2 year 1.97 0.001 1.61 2.42 3 year 1.57 <0.001 1.20 2.07 4 year 2.10 <0.001 1.63 2.70 5 year 1.64 0.002 1.19 2.26 >5 years 1.76 <0.001 1.46 2.13 1 year 1.60 0.002 1.19 2.15 2 year 1.61 0.043 1.01 2.56 3 year 1.02 0.956 0.51 2.05 4 year 1.13 0.738 0.54 2.38 5 year 1.63 0.198 0.78 3.42 >5 years 1.17 0.515 0.73 1.89 1 year 1.15 0.041 1.01 1.32 2 year 1.34 0.003 1.10 1.63 3 year 1.76 <0.001 1.43 2.17 4 year 1.21 0.179 0.92 1.60 5 year 1.44 0.012 1.08 1.92 >5 years 1.16 0.137 0.95 1.41 Naproxen Other NSAID Supplemental table 2c: Hazard ratios only adjusted of sex for coronary death or Re-MI Overall NSAID Hazard ratio for p-value Confidence intervals 95 % Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 coronary death or recurrent myocardial infarction 1 year 1.37 <0.001 1.28 1.47 2 year 1.74 <0.001 1.58 1.91 3 year 1.65 <0.001 1.47 1.85 4 year 1.44 <0.001 1.26 1.64 5 year 1.41 <0.001 1.21 1.64 >5 years 1.41 <0.001 1.28 1.55 1 year 1.89 <0.001 1.53 2.33 2 year 3.02 <0.001 2.32 3.92 3 year 2.14 <0.001 1.46 3.15 4 year 2.33 <0.001 1.53 3.54 5 year 2.52 <0.001 1.59 4.01 >5 years 2.12 0.007 1.23 3.66 1 year 1.98 <0.001 1.61 2.42 2 year 2.04 <0.001 1.51 2.75 3 year 2.16 <0.001 1.51 3.07 4 year 1.72 0.019 1.09 2.70 5 year 1.72 0.036 1.04 2.87 Rofecoxib Celecoxib Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 >5 years 1.13 0.694 0.61 2.11 1 year 1.21 <0.001 1.09 1.34 2 year 1.57 <0.001 1.37 1.80 3 year 1.37 <0.001 1.15 1.63 4 year 1.24 0.034 1.02 1.52 5 year 1.19 0.126 0.95 1.50 >5 years 1.33 <0.001 1.16 1.52 1 year 1.46 <0.001 1.26 1.70 2 year 1.84 <0.001 1.50 2.26 3 year 1.45 0.007 1.11 1.90 4 year 1.95 <0.001 1.51 2.51 5 year 1.52 0.011 1.10 2.10 >5 years 1.59 <0.001 1.32 1.92 1 year 1.56 0.003 1.16 2.09 2 year 1.55 0.063 0.98 2.47 3 year 1.00 1.000 0.50 2.00 4 year 1.12 0.767 0.53 2.35 5 year 1.49 0.292 0.71 3.13 >5 years 1.15 0.578 0.71 1.84 Ibuprofen Diclofenac Naproxen Other NSAID Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 1 year 1.25 0.001 1.09 1.43 2 year 1.48 <0.001 1.22 1.79 3 year 1.88 <0.001 1.53 2.31 4 year 1.27 0.094 0.96 1.68 5 year 1.51 0.005 1.13 2.01 >5 years 1.23 0.043 1.00 1.49 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Supplemental Table 3 Cox analysis of factors predicting NSAID use. Supplemental Table 3 Characteristic Hazard Ratio 95 % confidence Limits 1998 0.91 0.75 1.11 1999 0.94 0.77 1.14 2000 0.97 0.80 1.18 2001 0.99 0.82 1.21 2002 0.94 0.77 1.15 2003 0.93 0.76 1.13 2004 0.87 0.71 1.06 2005 0.85 0.70 1.04 2006 0.80 0.65 0.98 2007 0.77 0.63 0.94 2008 0.70 0.57 0.85 2009 0.69 0.56 0.86 Age 0.99 0.99 0.99 Sex 0.98 0.96 1.00 0.87 0.84 0.91 disease 1.02 0.94 1.11 Cerebral vascular disease 0.90 0.85 0.95 Year Co-morbidity Cardiac arrhythmias Peripheral vascular Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012 Diabetes with complications 0.97 0.91 1.04 Acute renal failure 0.83 0.72 0.95 Chronic renal failure 0.75 0.67 0.84 Malignancy 1.33 1.16 1.54 Shock 1.02 0.91 1.15 COPD 0.98 0.88 1.10 Gastric ulcer 0.91 0.83 0.99 Concomitant medical treatment Beta-blockers 0.95 0.93 0.97 ACE inhibitors 0.94 0.92 0.96 Statins 0.99 0.97 1.01 ASA 1.02 1.00 1.04 Clopidogrel 1.02 0.99 1.05 Spironolactone 0.92 0.88 0.96 Loop-diuretics 1.07 1.04 1.09 Glucose lowering drugs 1.05 1.01 1.10 PCI 0.97 0.94 1.00 Socioeconomic factors 0.97 0.96 0.97 Downloaded from http://circ.ahajournals.org/ by guest on September 11, 2012