Nonsteroidal anti-inflammatory drugs (NSAIDs) are

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2013;11: Failure to Renew Prescriptions for Gastroprotective Agents to Patients on Continuous Nonsteroidal Anti-inflammatory Drugs Increases Rate of Upper Gastrointestinal Injury ISABELLE LE RAY,*, ALAN N. BARKUN, FRANÇOISE VAUZELLE KERVROËDAN, and MARC BARDOU*,,,# *INSERM Centre d Investigations Cliniques Plurithématique 803, Centre Hospitalier Universitaire du Bocage, Dijon, France; Faculté de Médecine de Dijon, Université de Bourgogne, Dijon, France; Division of Gastroenterology, McGill University Health Center, McGill University, Montréal, Québec, Canada; Laboratoire Ethypharm, Saint-Cloud, France; Service d Hépato-Gastroentérologie, CHU du Bocage, Dijon, France; # Centre de Recherche INSERM, U866, Dijon, France BACKGROUND & AIMS: METHODS: RESULTS: CONCLUSIONS: Patients with risk factors for gastrointestinal (GI) disorders who continuously use nonsteroidal anti-inflammatory drugs (NSAIDs) also should take gastroprotective agents (GPAs), such as proton pump inhibitors (PPIs). However, it is not clear how many physicians continue to prescribe GPAs to these patients, and whether stopping the GPA prescription increases GI complications. We performed a retrospective, observational, longitudinal study using a validated electronic database of representative general practitioners in France. We analyzed data for 1856 patients at risk for GI events (>65 y, past history of GI ulcer, or receiving antiplatelet agents) who received prescriptions for an NSAID and PPI from 2007 to Kaplan Meier curves were used to determine the probability of still being prescribed a GPA at 12 and 24 months after the first prescription. Multivariate logistic regression analysis was used to identify factors associated with nonpersistence. GI complication rates were compared using the Student t test. The probability of still being prescribed a PPI along with an NSAID 1 year after the study began was 77.5% (95% confidence interval [CI], 75.6% 79.4%) and 68.3% after 2 years (95% CI, 66.1% 70.4%). Risk factors for no longer receiving a prescription for a PPI included switching to a cyclooxygenase-2 selective inhibitor (hazard ratio [HR], 2.50; 95% CI, ; P <.001) or to a nonselective NSAID (HR, 1.63; 95% CI, ; P <.001), and female sex (HR, 1.25; 95% CI, ; P <.05). In 50% of these cases, the PPI was reintroduced within 6 months, without a specific reason in 70% of the cases. The risk for upper GI injury was higher among patients with discontinued prescriptions for PPIs (odds ratio, 1.45; 95% CI, ; P.02). Within 2 years after prescribing a PPI, physicians do not renew this prescription for approximately 33% of patients receiving continuous NSAIDs. This increases the risk for GI adverse events among these patients. Keywords: Stomach; Protection; Bleeding. See editorial on page 505. Nonsteroidal anti-inflammatory drugs (NSAIDs) are one of the most widely prescribed class of medications in the world. 1 NSAID use has been associated with increased upper gastrointestinal (GI) complications, especially upper-gastrointestinal (UGI) bleeding, 2 4 as well as with lower GI events, 5,6 with a relative risk greater than 2 for both oral and parenteral formulations (odds ratio [OR], 2.26; 95% confidence interval [CI], ; and OR, 5.64; 95% CI, , respectively), 7 generating clinical and economic consequences. 8 The association between nonselective NSAID use, including low-dose acetylsalicylic acid, and upper-gi adverse events is well documented When examined endoscopically, 26% of patients treated for 12 weeks with naproxen for rheumatoid arthritis present with ulcers. 13 However, the incidence of clinically significant GI events (bleeding, obstruction, and perforation) caused by NSAIDs is much lower, occurring in only 3% 4.5% of patients taking NSAIDs. 14 Overall, serious gastric and duodenal ulcer related adverse events, including perforation and bleeding, affect approximately 1% 2% of patients receiving these medications. 15 Two studies conducted in Spain found that the Abbreviations used in this paper: CI, confidence interval; COX2, cyclooxygenase-2; CSD, Cegedim strategic data; GI, gastrointestinal; GP, general practitioners; GPA, gastroprotective agent; HR, hazard ratio; LPD, Longitudinal Patient Database; NSAID, nonsteroidal antiinflammatory drug; OR, odds ratio; PPIs, proton pump inhibitors; UGI, upper-gastrointestinal by the AGA Institute /$

2 500 LE RAY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 5 incidence of hospital admission owing to major GI events of the gastrointestinal tract (upper and lower) was events/ 100,000 persons/year, mostly affecting the upper GI tract. 7,14 Several risk factors for GI complications have been identified, including patient-related characteristics such as age; a history of prior GI events; medication-related determinants including the type of NSAID; and co-prescription of low-dose aspirin, steroids, or even acetaminophen. 3,6,16 19 Current guidelines recommend co-prescription with a gastroprotective agent (GPA), mostly a proton pump inhibitor (PPI) in case of chronic NSAID treatment for patients at high GI risk because this practice has been shown to reduce (UGI) adverse events. 10,18,20,21 Unfortunately, 2 major limiting factors have been reported pertaining to the effectiveness of gastroprotection among patients receiving NSAIDs: the first relates to adherence to guidelines by the physicians. Indeed, a study conducted among 82,766 US veterans found that only 27.2% were prescribed an appropriate strategy, and that this percentage increased only to a modest 41.8% among patients with 3 risk factors for NSAID-related GI adverse events. 22 The second issue is patient adherence to the GPA co-prescription; a recent cohort study conducted in Spain showed that, among patients receiving NSAIDs plus GPA (for most 30 d), optimal adherence was reported in only 79.7% of patients. Furthermore, more adverse events occurred among patients who reported nonoptimal adherence to GPA (22.1% vs 1.9%; P.0001). 18 This result is congruent with the earlier findings that only 68% of patients receiving concomitant NSAID and PPI therapy had adherence rates of 80% or more, and that suboptimal adherence to a PPI was associated with a significantly higher risk of upper-gi ulcers/complications in NSAID-treated patients (OR, 2.4; 95% CI, ). 23 More specifically, it has been described that the risk of an NSAID-related UGI complication among NSAID users increases 16% for every 10% decrease in adherence. 24 Also, the risk of UGI events decreased from a hazard ratio (HR) of 3.0 (95% CI, ) when a PPI was prescribed 0% 20% of the time to 1.1 (95% CI, ) when a PPI was prescribed 80% 100% of the time. 25 Interestingly, compliance rates decrease with the number of prescription renewals, suggesting poor adherence for patients treated chronically with NSAIDs. 26 In addition to these major issues, another possible cause factor influencing GPA co-prescription is the loss of PPI prescription renewals over time among chronic NSAID users (ie, physicians renewing an NSAID but not a GPA prescription), a phenomenon that has not yet been assessed appropriately. The present study aimed to describe the persistence of PPI co-prescription among chronic NSAID users at high risk for GI complications, and to assess factors associated with PPI interruption, as well as the impact on GI adverse events. Methods Data Selection We used the Longitudinal Patient Database (LPD) from Cegedim Strategic Data (CSD; Boulogne Billancourt, France). 27,28 This database collects systematic data from a representative group of 1200 general practitioners (GPs) in France. Participating physicians are provided with medical software and, on a voluntary basis, transmit anonymous data about their patients daily, extracted directly from the medical record and the actual electronic prescriptions. Analyses performed using this database have been approved by the Commission Nationale de l Informatique et des Libertés. GPs participating in the LPD CSD network are selected to be representative of the French GP population according to 3 main criteria: age, sex, and geographic area. Activity and prescription habits of the GP cohort also have been shown to be representative of national data when compared a posteriori. 27 The CSD LPD has been used for several publications, for example, in combination with the General Practice Research Database and the Integrated Primary Care Information database (The Netherlands) to assess risks of UGI events in NSAID users according to GPA nonadherence. 28 The database includes routinely collected records for more than 2.6 million patients. Since 2005, every subject covered by one of the French social insurance systems is required by law to have a single referent GP to limit medical roaming and multiple consultations for the same reason. Patient data collected by GPs since 2005 thus can be considered exhaustive and nonredundant. For each patient, information on disease status and medication prescriptions is entered directly into the database by the physician at the time of the consultation. However, no information as to the reasons for making individual diagnostic or prescription choices is provided. The disease status is coded using terms from a specific thesaurus of symptoms and disease entities adapted from the International Classification of Diseases 10th revision system. Prescription data contain the dispensed drug names (brand name or international common denomination), Anatomical Therapeutic Chemical classification categories, dose regimens, prescription durations, and number of renewals. Patients and Events Our observational cohort included all patients, as follows: (1) who received an NSAID prescription for more than 7 days in 2007, (2) had an NSAID prescription associated with a PPI coprescription at inclusion, and (3) a renewal of their NSAID within the 6 months after the last prescription over a 2-year period spanning 2007 to Included patients also were to have been followed up regularly by their GP (ie, to have had at least one outpatient visit during the year preceding the index NSAID prescription). An initial prescription for at least 7 days and a renewal within at least 6 months defined our patient population with chronic NSAID treatment. The LPD CSD database does not provide compliance data as is the case for most, if not all, electronic databases. Patients were considered as having risk factors for NSAIDinduced GI complications if they were aged 65 years or older, and/or had a history of upper-gi disease (gastralgia, gastritis, ulcer and ulcer complications), and/or associated conditions likely to increase ulcer risk such as rheumatoid arthritis, coprescription with an anticoagulant or antiplatelet agent, or if they were considered to be at higher risk for an adverse outcome after an episode of GI bleeding caused by a comorbid condition (eg, heart failure or severe coronary artery disease). Our study was restricted only to patients with risk factors (ie, patients for whom the need for a GPA associated with NSAID use was accepted according to published guidelines) The inclusion date was defined as the first NSAID prescription for more than 7 days issued in Persistence, usually used to assess length of therapy, fill refill ratio, and discontinuation rate, 32 was defined as a systematic PPI prescription concomitant with that for an NSAID prescription (ie, by an overlap in the periods covered by the prescriptions).

3 May 2013 GPA LOSS IN CHRONIC NSAID USERS 501 their GP before the index prescription. Of these, 35,458 had an initial PPI prescription. Among the 3036 patients receiving a chronic NSAID treatment with a follow-up evaluation between 2007 and 2009, 1856 had risk factors for GI complications and fulfilled the criteria of our target population (Figure 1). Most of the patients (1195 of 1856; 74.4%) were older than age 65 years, and 63.8% were female. Supplementary Table 1 describes the criteria for which the patients were considered at high risk for a GI event; 44.6% had a past history of GI events. Table 1 presents associated comorbidities and shows that only 27.7% of the patients had no comorbidities, whereas about one third of patients (32.7%) presented with 2 or more comorbidities. Figure 1. Patient flow chart. In the LPD CSD database, the distinction between a new event and the recall of a past event is not indicated clearly. To avoid the potential bias of past GI event recall after the index prescription and to assess the risk for GI event associated with nonpersistent PPI prescription we only considered new types of GI events entered by the GP after the index date. Because this approach reduces the crude rate of GI events by omitting recurrent events, we performed a sensitivity analysis that accounted for all events. Statistics Descriptive statistics included proportions with 95% CIs for categoric variables, and mean and standard deviations for continuous variables. Survival curves for persistence of PPI prescription were performed using the Kaplan Meier method and persistence percentages were calculated along with their 95% CIs. Univariable Cox models were used to select the covariates associated with persistence among the following: NSAID molecule, sex, age class, and number of co-prescribed treatments. All covariates with a P value of.20 or less were included in a corresponding multivariate model, and the most parsimonious was selected using a descendant stepwise selection technique based on likelihood ratios. The rate of reintroduction of a PPI within 6 months after its disruption, expressed as a percentage, was assessed together with the associated diagnosis at the time of reintroduction. Complication rates then were compared between persistent and nonpersistent patients. Univariate logistic regression was performed. Data were considered significantly different at a P value of less than.05; results are expressed as ORs and 95% CIs. All analyses were performed using SAS software 9.1 (SAS Institute, Cary, NC). Results Population Among patients treated by NSAIDs during 2007, there were 91,619 patients who had regular follow-up evaluation with Nonsteroidal Anti-Inflammatory Drugs More than 14 different NSAIDs were prescribed at inclusion: 4 of them, diclofenac, ketoprofen, piroxicam, and celecoxib, accounted for 64% of all NSAID prescriptions (Supplementary Table 2). The median number of NSAID prescription renewals during the 2-year study period was 8 (range, 2 24), with a median duration of 30 days (range, d) for each prescription. The median NSAID treatment time was days per patient. Most patients (77.6%) did not experience any change in daily dose or type of NSAID molecule during the study period. Persistence of Proton Pump Inhibitor Prescription PPI persistence probability (ie, the likelihood of still having a combined prescription of both an NSAID and a PPI), was 0.77 (95% CI, ) after 12 months and 0.68 (95% CI, ) for patients treated for 24 months (Figure 2). Univariate and Multivariate Cox Analyses of Proton Pump Inhibitor Co-prescription Interruption Among the variables considered for univariate analysis, 3 factors were found to be associated with PPI co-prescription interruption (or nonpersistence) with a P value of.20 or less: sex; number of co-prescribed treatments; and change in NSAID molecule, for either a cyclooxygenase-2 (COX2) inhibitor or another NSAID. The multivariate analysis identified these same 3 factors as independent predictors of PPI co-prescription interruption (Table 2). The highest HR was noted for a change of a nonselective NSAID molecule for a COX2 inhibitor, which more than doubled Table 1. Baseline Characteristics Variable Patients, n (%) (N 1856) Female 1184 (63.8) Male 672 (36.2) Age, 65 y 661 (35.6) Age, 65 y 1195 (74.4) Hypertension 1020 (55.1) Hypercholesterolemia 756 (40.8) History of stroke 170 (9.2) Cardiovascular disease 31 (1.7) Cardiac failure 59 (3.2) Alcohol abuse 7 (0.4)

4 502 LE RAY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 5 Table 3. PPI Reintroduction After Disruption Figure 2. Persistence of PPI co-prescription. Variables New PPI prescription within 6 months after disruption, n(%) Change in the NSAID, n (%) Time to reintroduction of PPI, d SD, standard deviation. Patients with PPI disruption (N 588) Yes 294 (50.0) No 294 (50.0) Yes, to a COX2 61 (10.4) inhibitor Yes, to another NSAID 127 (21.6) No 400 (68) Mean (SD) 84.5 (43.73) Median (range, minimum maximum) 77.5 (11 179) the risk of PPI co-prescription interruption (multivariate HR, 2.50; 95% CI, ), but also was increased significantly when the switch was for another nonselective NSAID (HR, 1.63; 95% CI, ). The risk was decreased according to the number of co-prescribed treatments (HR, 0.94 for each additional treatment; 95% CI, ), and was higher for women (HR, 1.25; 95% CI, ). Proton Pump Inhibitor Reintroduction Table 3 shows that among the 588 patients whose PPI treatment was stopped prematurely, 294 (50%) had PPI treatment resumed within 6 months after interruption. The median interruption duration was 77.5 days (range, d). Of the 588 nonpersistent patients, 61 (10.4%) had their treatment switched to a COX2 inhibitor. The most frequently reported diagnoses that could explain PPI reintroduction were addition of a new antiplatelet or anticoagulant treatment, and GI adverse events (18.0% and 15.3%, respectively; Supplementary Table 3), whereas in two thirds of patients, no reason to resume PPI therapy was specified. Gastrointestinal Adverse Events GI adverse events are detailed in Table 4. During follow-up evaluation, a GI adverse event was reported in 379 patients, among which 155 were classified as new according to the Methods section, including 18 ulcers with or without complications. Patients who were not persistent for a PPI prescription were significantly more likely to experience a GI adverse event than those who had an optimal NSAID and PPI coprescription (OR, 1.45; 95% CI, ; P.02). The absolute risk reduction that was associated with a continuous prescription of a PPI with NSAIDs, in at-risk patients, was 3.2% (Supplementary Table 4). The analysis accounting for all GI events instead of only new GI events provided similar results, even if the magnitude of the risk was reduced slightly (OR, 1.29; 95% CI, ; P.036). Discussion This observational study showed that only two thirds of a cohort of high-risk patients on chronic NSAID therapy who initially received a PPI are still prescribed a PPI after 2 years of follow-up evaluation. We characterized the persistence of a prescription (ie, systematic renewal of the GPA when the NSAID was being renewed) over time. Several studies have provided information about physician compliance to guidelines, 17,18,22,33 but modification of the prescription over time remained to be explored. It is widely recognized that the use of a PPI, 34,35 or H2-receptor antagonist, 36 reduces GI adverse events in high-risk patients receiving long-term NSAID therapy, and guidelines clearly state that high-risk patients should receive gastroprotection as long as they remain on NSAID treatment. 29 Although underprescription of GPAs, sometimes associated with a bad knowledge of risk factors, 24,26 or poor compliance, 24,26 already has been assessed, little is known about the persistence of GPA co-prescription in high-risk patients receiving long-term NSAIDs. The main risk factor for PPI co-prescription discontinuation appears to be a change in the NSAID molecule. As expected, the Table 2. Univariate and Multivariate Analysis of Factors Associated With PPI Co-prescription Interruption Variables Crude HR (95% CI) Adjusted HR (95% CI) P value Female sex 1.20 ( ) 1.25 ( ).009 Number of co-prescribed treatments (per treatment) 0.94 ( ) 0.93 ( ).001 Age, 65 y 1.10 ( ) a NS Change in NSAID molecule.001 For a COX inhibitor 2.56 ( ) 2.50 ( ) For other than a COX inhibitor 1.67 ( ) 1.63 ( ) a Age category was not retained for multivariate analysis because of a P.20.

5 May 2013 GPA LOSS IN CHRONIC NSAID USERS 503 Table 4. Outcomes Outcome Cumulative NSAID prescription (196.09) duration, mean, d ( SD) GI complications, total 155 Gastralgia, n (%) 81 (38.71) Gastritis, n (%) 44 (28.39) Ulcer, n (%) 18 (11.6) Others, n (%) 12 (7.74) SD, standard deviation. highest risk was observed when the treatment was switched to a COX-2 selective NSAID, but this is still of concern because international guidelines recommend co-prescription of a PPI for at-risk patients, even when using a COX-2 selective agent In addition, it is unlikely that the reason for PPI interruption could have been owing to PPI-related side effects, but it was impossible to address this issue directly. Interestingly, PPI coprescription was resumed within 6 months in half of these patients, suggesting that it was unintended. In support of that assumption, for two thirds of the patients who resumed taking a PPI, no specific associated diagnosis that could justify a PPI prescription was noted at the time of its reintroduction. Another key finding of our study was that the absence of systematic simultaneous prescription of both a PPI and NSAID significantly increased the risk of a GI complication by 45%. This is consistent with a recent study that showed higher rates of GI adverse events and shorter NSAID treatment among patients who were noncompliant with their PPI. 18 Another investigation, based on a randomized trial cohort, showed that among patients receiving an NSAID and suffering from gastroesophageal reflux disease, compliance with PPI therapy was associated with longer NSAID treatment. 30 This finding suggests that NSAID treatment may be optimized by the effective alleviation of GI side effects. This is consistent with the findings of the study by Lanas et al, 37 showing that the mortality rate from NSAIDrelated lower or upper GI adverse events are similar (around 5.6%), but that events related to the upper-gi tract are 6 times more frequent. Our data link the GI complications to the prescription behavior. The fact that GI adverse events are more frequent among patients without appropriate PPI coverage highlights the necessity of optimizing the PPI prescription. Coté et al 38 showed that physician education can improve appropriate PPI discharge prescription practice, especially if it is combined with a computer alert. In this study, the overall protection of high-risk patients reaches 67% in the education and computer alert group compared with 45% in the control group. 38 Although effective, this strategy is not completely satisfactory because a third of patients remain unprotected. Underprescription could be avoided with a combination preparation. It has been suggested that a combination pill containing an NSAID and PPI may be more cost effective than an NSAID alone or prescribing an NSAID and PPI separately, given the lower patient adherence rate. 39 One of the strengths of our study was the use of a large, prospective, and well-validated database. We also were able to retrieve information about comorbidities, which usually are absent from prescription or delivery databases. The validity of these data are enhanced further by the structure of the French health care system because all French citizens have equal access to health care services and are cared for by a GP as part of the health care system. The referent GP possesses exhaustive data about his/her patients. N This study had some intrinsic limitations. First, even if data were collected prospectively, this was not a randomized trial. Consequently, it cannot infer causation, but rather can identify associations. Multivariate analyses are used to avoid this bias; however, as in every observational study, we were not able to guarantee there were no unmeasured confounders. Although we had access to prescription data, we did not have access to dispensation and patient adherence to the prescribed medications. However, because there is no reason to expect that patients who lost their GPA coverage were less likely to have either purchased or taken their PPI, we believe that we have truly identified an additional, and amendable, cause of underprotection in chronic NSAID users at GI risk. Another limitation of our study was that we did not capture over-the-counter use of either NSAIDs or PPIs. However, in France, there is almost universal coverage of prescribed drugs by the National Health Insurance system. It is therefore not likely that chronically treated patients might have bought their medications over the counter, and even in that case it would have driven the results for GI events toward the null hypothesis. In conclusion, our study showed that about one fifth and one third of at-risk patients who start a co-prescription of an NSAID and a PPI lose their PPI after 1 and 2 years of follow-up evaluation, respectively. The patients who do not receive the indicated PPI co-prescription experience greater GI complication rates. These data suggest that optimizing GPA coverage, mostly with a PPI, remains a major health issue among chronic NSAID users. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi.org/ /j.cgh References 1. Paulose-Ram R, Hirsch R, Dillon C, et al. Frequent monthly use of selected non-prescription and prescription non-narcotic analgesics among U.S. adults. Pharmacoepidemiol Drug Saf 2005;14: García Rodríguez LA, Jick H. Risk of upper gastrointestinal bleeding and perforation associated with individual non-steroidal antiinflammatory drugs. Lancet 1994;343: Rahme E, Bardou M, Dasgupta K, et al. Hospitalization for gastrointestinal bleeding associated with non-steroidal anti-inflammatory drugs among elderly patients using low-dose aspirin: a retrospective cohort study. Rheumatology (Oxford) 2007;46: García Rodríguez LA, Barreales Tolosa L. Risk of upper gastrointestinal complications among users of traditional NSAIDs and COXIBs in the general population. Gastroenterology 2007;132: Laine L, Smith R, Min K, et al. Systematic review: the lower gastrointestinal adverse effects of non-steroidal anti-inflammatory drugs. Aliment Pharmacol Ther 2006;24: Rahme E, Barkun A, Nedjar H, et al. Hospitalizations for upper and lower GI events associated with traditional NSAIDs and acetaminophen among the elderly in Quebec, Canada. Am J Gastroenterol 2008;103: Chang CH, Lin JW, Chen HC, et al. Non-steroidal anti-inflammatory drugs and risk of lower gastrointestinal adverse events: a nationwide study in Taiwan. Gut 2011;60: Rahme E, Joseph L, Kong SX, et al. Gastrointestinal health care resource use and costs associated with nonsteroidal antiinflammatory drugs versus acetaminophen: retrospective cohort study of an elderly population. Arthritis Rheum 2000;43: Lanas A, García-Rodríguez LA, Arroyo MT, et al. Risk of upper gas-

6 504 LE RAY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 5 trointestinal ulcer bleeding associated with selective cyclo-oxygenase-2 inhibitors, traditional non-aspirin non-steroidal anti-inflammatory drugs, aspirin and combinations. Gut 2006;55: Chan FK, Abraham NS, Scheiman JM, et al. Management of patients on nonsteroidal anti-inflammatory drugs: a clinical practice recommendation from the First International Working Party on Gastrointestinal and Cardiovascular Effects of Nonsteroidal Anti-inflammatory Drugs and Anti-platelet Agents. Am J Gastroenterol 2008;103: Jones R, Rubin G, Berenbaum F, et al. Gastrointestinal and cardiovascular risks of nonsteroidal anti-inflammatory drugs. Am J Med 2008;121: De Berardis G, Lucisano G, D Ettorre A, et al. Association of aspirin use with major bleeding in patients with and without diabetes. JAMA 2012;307: Simon LS, Weaver AL, Graham DY, et al. Anti-inflammatory and upper gastrointestinal effects of celecoxib in rheumatoid arthritis: a randomized controlled trial. JAMA 1999;282: Laine L. Approaches to nonsteroidal anti-inflammatory drug use in the high-risk patient. Gastroenterology 2001;120: Langman MJ, Jensen DM, Watson DJ, et al. Adverse upper gastrointestinal effects of rofecoxib compared with NSAIDs. JAMA 1999;282: Laine L, Curtis SP, Cryer B, et al. Risk factors for NSAID-associated upper GI clinical events in a long-term prospective study of arthritis patients. Aliment Pharmacol Ther 2010;32: Lanas A, Garcia-Tell G, Armada B, et al. Prescription patterns and appropriateness of NSAID therapy according to gastrointestinal risk and cardiovascular history in patients with diagnoses of osteoarthritis. BMC Med 2011;9: lanas A, Polo-Tomás M, Roncales P, et al. Prescription of and adherence to non-steroidal anti-inflammatory drugs and gastroprotective agents in at-risk gastrointestinal patients. Am J Gastroenterol 2012;107: Tielemans MM, Eikendal T, Jansen JB, et al. Identification of NSAID users at risk for gastrointestinal complications: a systematic review of current guidelines and consensus agreements. Drug Saf 2010;33: Rahme E, Roussy JP, Lafrance JP, et al. Concordance with guideline recommendations: previous and more recent nonsteroidal anti-inflammatory drug prescriptions in Quebec, Canada. Pharmacoepidemiol Drug Saf 2012;21: Miyamoto M, Haruma K, Okamoto T, et al. Continuous proton pump inhibitor treatment decreases upper gastrointestinal bleeding and related death in rural area in Japan. J Gastroenterol Hepatol 2012;27: Abraham NS, El-Serag HB, Johnson ML, et al. National adherence to evidence-based guidelines for the prescription of nonsteroidal antiinflammatory drugs. Gastroenterology 2005;129: Goldstein JL, Howard KB, Walton SM, et al. Impact of adherence to concomitant gastroprotective therapy on nonsteroidal-related gastroduodenal ulcer complications. Clin Gastroenterol Hepatol 2006;4: van Soest EM, Sturkenboom MC, Dieleman JP, et al. Adherence to gastroprotection and the risk of NSAID-related upper gastrointestinal ulcers and haemorrhage. Aliment Pharmacol Ther 2007; 26: Abraham NS, Hartman C, Castillo D, et al. Effectiveness of national provider prescription of PPI gastroprotection among elderly NSAID users. Am J Gastroenterol 2008;103: Sturkenboom MC, Burke TA, Tangelder MJ, et al. Adherence to proton pump inhibitors or H2-receptor antagonists during the use of non-steroidal anti-inflammatory drugs. Aliment Pharmacol Ther 2003;18: Bouée S, Charlemagne A, Fagnani F, et al. Changes in osteoarthritis management by general practitioners in the COX2-inhibitor era-concomitant gastroprotective therapy. Joint Bone Spine 2004;71: van Soest EM, Valkhoff VE, Mazzaglia G, et al. Suboptimal gastroprotective coverage of NSAID use and the risk of upper gastrointestinal bleeding and ulcers: an observational study using three European databases. Gut 2011;60: Rostom A, Moayyedi P, Hunt R, et al. Canadian consensus guidelines on long-term nonsteroidal anti-inflammatory drug therapy and the need for gastroprotection: benefits versus risks. Aliment Pharmacol Ther 2009;29: Scheiman JM, Hindley CE. Strategies to optimize treatment with NSAIDs in patients at risk for gastrointestinal and cardiovascular adverse events. Clin Ther 2010;32: Burmester G, Lanas A, Biasucci L, et al. The appropriate use of non-steroidal anti-inflammatory drugs in rheumatic disease: opinions of a multidisciplinary European expert panel. Ann Rheum Dis 2011;70: El-Serag HB, Fitzgerald S, Richardson P. The extent and determinants of prescribing and adherence with acid-reducing medications: a national claims database study. Am J Gastroenterol 2009;104: Doherty GA, Cannon MD, Lynch KM, et al. Co-prescription of gastro-protectants in hospitalized patients: an analysis of what we do and what we think we do. J Clin Gastroenterol 2010;44: e51 e Dubois RW, Melmed GY, Henning JM, et al. Guidelines for the appropriate use of non-steroidal anti-inflammatory drugs, cyclooxygenase-2-specific inhibitors and proton pump inhibitors in patients requiring chronic anti-inflammatory therapy. Aliment Pharmacol Ther 2004;19: Lanza FL, Chan FK, Quigley EM, et al. Guidelines for prevention of NSAID-related ulcer complications. Am J Gastroenterol 2009; 104: Laine L, Kivitz AJ, Bello AE, et al. Double-blind randomized trials of single-tablet ibuprofen/high-dose famotidine vs. ibuprofen alone for reduction of gastric and duodenal ulcers. Am J Gastroenterol 2012;107: Lanas A, Perez-Aisa MA, Feu F, et al. A nationwide study of mortality associated with hospital admission due to severe gastrointestinal events and those associated with nonsteroidal antiinflammatory drug use. Am J Gastroenterol 2005;100: Coté GA, Rice JP, Bulsiewicz W, et al. Use of physician education and computer alert to improve targeted use of gastroprotection among NSAID users. Am J Gastroenterol 2008;103: De Groot NL, Siersemal PD, de Wit NJ, et al. Gastroprotective strategies in chronic NSAID users: a cost-effectiveness analysis comparing a single tablet formulation with individual component strategies. Gut 2011;60:A159. Reprint requests Address requests for reprints to: Marc Bardou, MD, INSERM CIC-P 803, CHU de Dijon, 14, bd Gaffarel, BP 77908, Dijon, France. Marc.bardou@u-bourgogne.fr; fax: (33) Conflicts of interest These authors disclose the following: Françoise Vauzelle works for Ethypharm, and Marc Bardou has served as a consultant for Ethypharm. The remaining authors disclose no conflicts. Funding The study was funded by Ethypharm, Saint Cloud, France.

7 504.e1 LE RAY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 5 Supplementary Table 1. Risk Factors at Study Entry Patients, n(%) (N 1856) History of GI pathology 828 (44.6) Risk factors of complications in case of hemorrhage 76 (4.1) Anticoagulant and/or anti-aggregant treatment 420 (22.6) Age, 65 y 1195 (74.4) Supplementary Table 2. NSAID Molecule at Inclusion NSAID molecule Patients, n (%) (N 1856) Diclofenac 455 (24.5) Ketoprofen 282 (15.2) Piroxicam 264 (14.2) Celecoxib 187 (10.1) Naproxen 134 (7.2) Nimesulide 135 (7.3) Aceclofenac 115 (6.1) Meloxicam 56 (3.0) Nabumetone 52 (2.8) Flurbiprofen 46 (2.5) Tenoxicam 39 (2.1) Indomethacin 30 (1.6) Etodolac 17 (0.9) Other treatments 44 (2.4) Supplementary Table 3. Factors Associated With PPI Reintroduction Factors mentioned at PPI reintroduction Patients with PPI disruption, n(%) (N 588) GI side effects 45 (15.3) Anticoagulant/antiplatelet 53 (18.0) prescription Change in age class 7 (2.4) No factor mentioned 197 (67.0) Supplementary Table 4. GI Adverse Events Patient category Persistence of PPI prescription GI adverse event, n(%) (n 155) GI adverse event, OR (95% CI) Yes 62 (10.5) 1 (reference).02 No 93 (7.3) 1.45 ( ) P