SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Aliment Pharmacol Ther 2010; 32: 1240–1248

Summary

Background  Nonsteroidal anti-inflammatory drugs (NSAID)-related GI effects vary based on patient characteristics.

Aims  To assess risk factors for NSAID-associated upper GI clinical events and dyspepsia.

Methods  Patients ≥50 years with osteoarthritis or rheumatoid arthritis were randomized to etoricoxib or diclofenac in a prespecified intent-to-treat analysis of three double-blind randomized trials. Potential risk factors for upper GI clinical events (bleeding, perforation, obstruction, or ulcer), complicated events (perforation, obstruction, bleeding) and discontinuations due to dyspepsia were assessed with Cox proportional hazard models.

Results  Significant predictors of clinical events and complicated events included age ≥65 years [hazards ratios (HRs) = 2.25 (1.84–2.76), 4.09 (2.82–5.92)], prior event [HRs = 2.57 (1.94–3.39), 3.23 (2.09–5.00)], low-dose aspirin [HRs = 2.34 (1.87–2.92), 3.41 (2.33–5.00)] and corticosteroid [HRs = 1.85 (1.41–2.43), 2.09 (1.29–3.38)]. Predictors of discontinuation due to dyspepsia included prior dyspepsia [HR = 1.78 (1.44–2.00)], prior event [HR = 1.78 (1.40–2.27)] and age ≥65 years [HR = 1.35 (1.16–1.57)].

Conclusions  Assessment for age ≥65 years, prior upper GI clinical events and low-dose aspirin use are key in identifying patients who should either avoid NSAIDs or employ management strategies to reduce NSAID-associated upper GI events. Prior dyspepsia or upper GI clinical events and age ≥65 years also predict an increased risk of developing dyspepsia severe enough to necessitate discontinuation of NSAIDs.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Upper gastrointestinal (GI) clinical events such as bleeding and ulcers, as well as upper GI symptoms such as dyspepsia, are significantly increased with the use of traditional nonsteroidal anti-inflammatory drugs (NSAID).1 These side effects have important clinical implications in terms of quality of life, morbidity and mortality, and they increase healthcare resource utilization such as diagnostic testing, doctor visits, hospitalizations and medications. The cost of NSAID-associated upper GI side effects is substantial. North American studies have indicated that for every dollar spent on NSAIDs, $0.66 to $1.25 may be spent on GI side effects,2, 3 and that nearly one-third of total medical costs in arthritis patients may be related to GI adverse events.4

Strategies recommended to decrease the risk of NSAID-associated upper GI clinical events include medical co-therapy with misoprostol or proton pump inhibitors (PPIs), and/or the use of cyclooxygenase (COX)-2 selective inhibitors.5–7 Determining the baseline risk of patients taking NSAIDs is crucial in deciding which patients are candidates for one of these protective strategies. Although these strategies may decrease the number of clinical events regardless of baseline patient risk, they generally are not employed in low-risk patients due to the small absolute benefit. Unfortunately, a large proportion of higher risk patients who might benefit from protective therapy do not receive it.8, 9 Identification of specific risk factors and precise determination of their risk increase are therefore key in developing appropriate management strategies for patients taking NSAIDs.

A number of observational studies have assessed risk for development of upper GI clinical events with NSAID use. Although these studies have the benefit of assessing real-world data in large populations, they suffer from shortcomings including a lack of predefined GI endpoints, uncertainty regarding diagnoses, inconsistent recording of data, and unrecorded or unmeasured characteristics of interest. Prospective outcome studies which have predefined GI endpoints, full recording of patient characteristics and study results, and blinded independent adjudication of endpoints by a central independent panel should help better define risk factors for GI events. However, relatively few such data have been published.10–12 In addition, no prior prospective study has simulated ‘real-world’ practice by allowing use of both low-dose aspirin and PPIs, and no outcome study has assessed risk factors for the development of dyspepsia with NSAID use. We report the risk factor analysis for upper GI clinical events and discontinuations due to dyspepsia from the MEDAL (Multinational Etoricoxib and Diclofenac Arthritis Long-Term) Program, a double-blind randomized comparison of the COX-2 selective inhibitor etoricoxib and the traditional NSAID diclofenac in 34 701 arthritis patients treated for a mean of 18 months.13–15

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study design

The design of the MEDAL Program and the primary results for cardiovascular and upper GI outcomes have been presented in detail elsewhere.13–15 The programme was conducted from June 2002 to May 2006 at 1380 sites in 46 countries. The MEDAL Program was prospectively designed to pool data from three randomized, double-blind clinical trials: the MEDAL (Multinational Etoricoxib and Diclofenac Arthritis Long-Term) study, the EDGE (Etoricoxib vs. Diclofenac Sodium Gastrointestinal Tolerability and Effectiveness) study and the EDGE II study (ClinicalTrials.gov identifiers: NCT00092703; NCT00092742; NCT00250445). The ethics committee for each study site approved the trial at that site and all patients provided written informed consent prior to their participation.

Patients ≥50 years of age with osteoarthritis of the knee, hip, hand, or spine or with rheumatoid arthritis were eligible for enrolment if they required chronic therapy with an NSAID. Patients meeting entry criteria were randomly assigned using concealed allocation to treatment in equal proportions within each study site using a different computer-generated randomization schedule for each of the three component trials. The first 4333 osteoarthritis patients and all rheumatoid arthritis patients in the MEDAL study received etoricoxib 90 mg daily or diclofenac 75 mg twice daily. The subsequent osteoarthritis patients enrolled in the MEDAL study received etoricoxib 60 mg daily (the current recommended dose for osteoarthritis) or diclofenac 75 mg twice daily. In EDGE and EDGE II, patients received etoricoxib 90 mg daily, diclofenac 50 mg thrice daily (EDGE), or diclofenac 75 mg twice daily (EDGE II). A double-dummy design provided blinding to treatment assignment.

Low-dose aspirin (≤100 mg/day) was recommended in patients with established cardiovascular, peripheral arterial, or cerebrovascular disease as well as diabetes. Use of medical co-therapy (proton pump inhibitors or misoprostol) was also recommended for patients considered at high risk of upper GI clinical events (age >65; prior history of GI ulcer or haemorrhage; concurrent use of corticosteroid, anticoagulant, or antiplatelet therapy). If low-dose aspirin or a PPI or misoprostol was not given to a patient meeting these criteria, investigators were contacted and required to state their reasons for not providing the medication. For the MEDAL study, omeprazole and low-dose aspirin were provided free of charge; low-dose aspirin was provided free of charge in the EDGE and EDGE II trials. Patients returned for visits every 4 months and a scheduled telephone contact was made between visits. Patient compliance with study medication was assessed by pill count. An independent data and safety monitoring board monitored emerging safety data from all three trials at regular intervals.

Endpoints

Potential upper GI clinical events (bleeding, perforation, obstruction, ulcer diagnosed on clinical work-up) were identified through active surveillance of reported adverse events, and were adjudicated by an independent blinded committee using predefined criteria.15 The subset of complicated events included those with perforation, obstruction and complicated bleeding. Uncomplicated events included uncomplicated bleeding and uncomplicated ulcer. Uncomplicated bleeding was defined as occult blood-positive stool associated with a documented upper GI lesion judged to be the source of the bleeding and stigmata of recent bleeding (visible vessel, pigmented spot, or clot) at endoscopy, but no clinically significant bleeding (hypotension, orthostatic changes in heart rate or blood pressure, haemoglobin drop ≥2 g/dL, or transfusion), or patient-reported haematemesis, melaena, or haematochezia associated with a documented upper GI lesion judged to be the source of the bleeding and stigmata of recent bleeding at endoscopy, but no clinically significant bleeding. Uncomplicated ulcers were defined as gastric or duodenal ulcers documented on clinical evaluation by endoscopy, surgery, upper GI contrast radiography, or autopsy.

Discontinuations due to upper GI symptoms consistent with dyspepsia were also assessed. The definition of dyspepsia required pain or discomfort in the upper abdomen (including ‘epigastric’, or ‘stomach’) or nausea. Patients were not queried regarding specific symptoms of dyspepsia; these symptoms represent spontaneous reports by patients. As reports of mild GI adverse events are extremely common and may be variably reported by different sites and investigators, we assessed discontinuations as a marker of clinically meaningful upper GI symptoms to study patients whose symptoms more clearly had an adverse effect on their quality of life.

Statistical analysis

A list of 13 baseline clinical variables, which were considered potential risk factors for NSAID-associated clinical events, were specified for assessment along with treatment group in a multivariable analysis. Rates of upper GI clinical events and complications per 100-patient-years as well as rates of patient discontinuations for dyspepsia were calculated and a Cox proportional hazard model was applied with inclusion of the specified covariates in the model for all upper GI clinical events, the subset of complicated upper GI events, and discontinuations for dyspepsia. All analyses were performed in the intention-to-treat population of all patients randomized followed up until 14 days after discontinuation of study medication. The treatment effects within subgroups based on the individual risk factors were also assessed. The subgroup analyses were performed by adding terms for the subgroup factor and interaction with treatment to the Cox model.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

A total of 34 701 patients were enrolled in the MEDAL Program (MEDAL study, 23 504 patients; EDGE, 7111 patients; EDGE II, 4086 patients), including 24 913 (71.8%) with osteoarthritis and 9787 (28.2%) with rheumatoid arthritis. As previously reported,15 the baseline characteristics were comparable in the two treatment groups: 41% were ≥65 years, 7% had a prior upper GI clinical event, 35% used low-dose aspirin, 17% took systemic corticosteroids, and 39% used PPIs. Mean treatment duration was 18 months with a maximum of 40 months.

Upper GI clinical events occurred in 422 patients. As previously reported,15 the rates for etoricoxib and diclofenac were 0.67 and 0.97 per 100 patient-years [hazards ratio (HR) = 0.69 (95% CI, 0.57–0.83)]. Complicated upper GI events occurred in 160 patients and there was no significant difference for etoricoxib vs. diclofenac [0.30 vs. 0.32 events per 100 patient-years; HR = 0.91 (0.67–1.24)].

The results of the multivariable analysis assessing the association of baseline characteristics with upper GI clinical events are shown in Table 1. Age ≥65 years, prior upper GI event and baseline low-dose aspirin use increased the relative risk of overall upper GI events by ∼2½ fold, and increased the relative risk of complicated events by ∼3 to 4-fold. Use of systemic corticosteroids at baseline also increased the relative risk of overall and complicated events – by approximately 2-fold. Dyspepsia and atherosclerotic cardiovascular disease showed a borderline significant increase in relative risk of less than 50% in overall clinical events, but not in complicated events. The use of PPIs at baseline was associated with just over a one-third relative risk decrease in overall upper GI clinical events and just over a 50% decrease in complicated events.

Table 1.   The association of individual baseline characteristics with all upper GI clinical events and with complicated upper GI clinical events: hazards ratios (95% CIs) from multivariable analysis
 All upper GI clinical eventsComplicated upper GI events
Age ≥652.25 (1.84–2.76)4.09 (2.82–5.92)
Prior UGI clinical event2.57 (1.94–3.39)3.23 (2.09–5.00)
Low-dose aspirin use2.34 (1.87–2.92)3.41 (2.33–5.00)
Steroid use1.85 (1.41–2.43)2.09 (1.29–3.38)
History of dyspepsia1.45 (1.07–1.98)1.17 (0.66–2.07)
Atherosclerotic cardiovascular disease1.30 (1.01–1.68)1.35 (0.92–1.98)
Methotrexate use1.36 (0.98–1.87)1.67 (0.88–3.17)
Smoker1.18 (0.84–1.64)1.14 (0.64–2.00)
Diabetes1.13 (0.85–1.50)1.16 (0.75–1.81)
Traditional NSAID use1.06 (0.78–1.45)1.28 (0.76–2.13)
PPI use0.63 (0.51–0.77)0.44 (0.31–0.62)
Female gender0.75 (0.59–0.94)0.77 (0.53–1.10)
Osteoarthritis (vs. Rheumatoid arthritis)0.91 (0.66–1.24)1.70 (0.93–3.12)

We also examined the treatment effect for etoricoxib vs. diclofenac related to the presence or absence of the significant baseline predictors of overall upper GI clinical events (Table 2) and complicated upper GI events (Table 3); numbers-needed-to-treat are provided for overall events because the treatment effect was significant, but not for complicated events for which the treatment effect was not significant. No treatment-by-subgroup interaction was significant for any individual risk factor related to either overall events or complicated events, indicating that the treatment effects for etoricoxib vs. diclofenac were consistent in patients with and without the individual risk factors. We then determined the risk of events with differing numbers of the four major risk factors: age ≥65 years, prior upper GI clinical event, low-dose aspirin use, and systemic corticosteroid use (Table 4). As the number of risk factors increased from 0 to 4, the rate of events also increased for both etoricoxib and diclofenac.

Table 2.   Rates of upper GI clinical events and treatment effects related to significant baseline predictors of upper GI events
 Rates per 100 patient-years (n/N; %)RR (95% CI)NNT (1-year)
EtoricoxibDiclofenac
  1. PPI, proton pump inhibitor; ASCVD, atherosclerotic cardiovascular disease; NNT, number-needed-to-treat.

Age ≥651.03 (106/7234; 1.47%)1.49 (146/7162; 2.04%)0.69 (0.54–0.89)217
Age <650.44 (70/10 178; 0.69%)0.64 (100/10 127; 0.99%)0.68 (0.50–0.92)500
Prior event1.69 (29/1127; 2.57%)2.00 (32/1133; 2.82%)0.85 (0.52–1.41)323
No prior event0.60 (147/16 285; 0.90%)0.90 (214/16 156; 1.32%)0.66 (0.53–0.81)333
Aspirin1.11 (100/6030; 1.66%)1.48 (127/5976; 2.13%)0.75 (0.58–0.98)270
No aspirin0.44 (76/11 382; 0.67%)0.71 (119/11 313; 1.05%)0.62 (0.46–0.82)370
PPIs0.58 (65/6742; 0.96%)0.98 (104/6664; 1.56%)0.59 (0.43–0.81)250
No PPIs0.73 (111/10 670; 1.04%)0.96 (142/10 625; 1.34%)0.76 (0.59–0.97)435
Steroid1.09 (49/2685; 1.82%)1.60 (71/2705; 2.62%)0.68 (0.48–0.98)196
No steroid0.58 (127/1472; 0.86%)0.84 (175/14 584; 1.20%)0.69 (0.55–0.87)385
Dyspepsia0.86 (18/1372; 1.31%)1.34 (28/1399; 2.00%)0.67 (0.37–1.20)208
No dyspepsia0.65 (158/16 040; 0.99%)0.94 (218/15 890; 1.37%)0.69 (0.56–0.85)345
ASCVD1.34 (38/2014; 1.89%)1.71 (47/2010; 2.34%)0.79 (0.51–1.21)270
No ASCVD0.59 (138/15 398; 0.90%)0.88 (199/15 279; 1.30%)0.66 (0.53–0.83)345
Female0.72 (141/12 925; 1.09%)0.99 (187/12 823; 1.46%)0.73 (0.58–0.90)370
Male0.51 (35/4487; 0.78%)0.91 (59/4466; 1.32%)0.56 (0.37–0.85)250
Table 3.   Rates of upper GI complicated events and treatment effects related to significant baseline predictors of upper GI clinical events
 Rates per 100 patient-years (n/N; %)RR (95% CI)
EtoricoxibDiclofenac
  1. PPI, proton pump inhibitor; ASCVD, atherosclerotic cardiovascular disease.

Age ≥650.59 (61/7234; 0.84%)0.60 (59/7162; 0.82%)0.99 (0.69–1.41)
Age <650.11 (17/10 178; 0.17%)0.15 (23/10 127; 0.23%)0.71 (0.38–1.33)
Prior event0.76 (13/1127; 1.15%)0.81 (13/1133; 1.15%)0.95 (0.44–2.05)
No prior event0.26 (65/16 285; 0.40%)0.29 (69/16 156; 0.43%)0.90 (0.64–1.27)
Aspirin0.55 (50/6030; 0.83%)0.62 (53/5976; 0.89%)0.90 (0.61–1.32)
No aspirin0.16 (28/11 382; 0.25%)0.17 (29/11 313; 0.26%)0.94 (0.56–1.57)
PPIs0.22 (25/6742; 0.37%)0.29 (31/6664; 0.47%)0.76 (0.45–1.29)
No PPIs0.35 (53/10 670; 0.50%)0.34 (51/10 625; 0.48%)1.01 (0.69–1.48)
Steroid0.40 (18/2685; 0.67%)0.38 (17/2705; 0.63%)1.05 (0.54–2.04)
No steroid0.27 (60/14 727; 0.41%)0.31 (65/14 584; 0.45%)0.88 (0.62–1.25)
Dyspepsia0.33 (7/1372; 0.51%)0.29 (6/1399; 0.43%)1.23 (0.41–3.66)
No dyspepsia0.29 (71/16 040; 0.44%)0.33 (76/15 890; 0.48%)0.89 (0.64–1.23)
ASCVD0.78 (22/2014; 1.09%)0.65 (18/2010; 0.90%)1.19 (0.64–2.22)
No ASCVD0.24 (56/15 398; 0.36%)0.28 (64/15 279; 0.42%)0.84 (0.58–1.20)
Female0.31 (60/12 925; 0.46%)0.32 (60/12 823; 0.47%)0.96 (0.67–1.38)
Male0.26 (18/4487; 0.40%)0.34 (22/4466; 0.49%)0.77 (0.41–1.44)
Table 4.   Rates of upper GI clinical events and upper GI complicated events with 0–4 significant predictors* of upper GI events
# Risk factorsUpper GI clinical events rates per 100 patient-years, 95% CI (n/N; %)Complicated upper GI events rates per 100 patient-years, 95% CI (n/N; %)
EtoricoxibDiclofenacRR (95% CI)EtoricoxibDiclofenacRR (95% CI)
  1. * Prior upper GI event, age ≥65 years, aspirin use, steroid use.

00.24, 0.15–0.37 [20/5528 (0.36%)]0.35, 0.24–0.51 [29/5467 (0.53%)]0.67 (0.38–1.19)0.04, 0.01–0.10 [3/5528 (0.05%)]0.05, 0.01–0.13 [4/5467 (0.07%)]0.73 (0.16–3.25)
10.50, 0.37–0.64 [56/7295 (0.77%)]0.79, 0.63–0.98 [86/7291 (1.18%)]0.63 (0.45–0.88)0.19, 0.11–0.28 [21/7295 (0.29%)]0.23, 0.15–0.34 [25/7291 (0.34%)]0.81 (0.45–1.45)
21.26, 0.99–1.58 [74/4011 (1.84%)]1.90, 1.56–2.30 [106/3943 (2.69%)]0.66 (0.49–0.89)0.73, 0.53–0.98 [43/4011 (1.07%)]0.77, 0.56–1.04 [43/3943 (1.09%)]0.96 (0.63–1.46)
33.04, 1.95–4.53 [24/553 (4.34%)]2.94, 1.84–4.45 [22/553 (3.98%)]1.03 (0.58–1.84)1.13, 0.52–2.15 [9/553 (1.63%)]1.20, 0.55–2.28 [9/553 (1.63%)]0.95 (0.38–2.39)
45.56, 0.67–20.10 [2/25 (8.00%)]6.42, 1.32–18.77 [3/35 (8.57%)]1.07 (0.18–6.38)5.56, 0.67–20.10 [2/25 (8.00%)]2.09, 0.05–11.66 [1/35 (2.86%)]3.18 (0.29–35.15)

Patient discontinuations due to dyspepsia occurred in 751 patients. As previously reported,15 the rate was significantly lower with etoricoxib than with diclofenac [1.25 per 100 patient-years vs. 1.69 per 100 patient-years; HR = 0.75 (0.65–0.87)]. The results of the multivariable analysis assessing the association of baseline characteristics with patient discontinuations due to dyspepsia are shown in Table 5. Prior dyspepsia and prior upper GI clinical event were the two strongest predictors, associated with approximately an 80% relative risk increase in development of dyspepsia leading to discontinuation. Older age (≥65 years) also was associated with a modest 35% increase in relative risk. The risk was decreased in men (28% relative risk reduction) and in patients with traditional NSAID use or systemic steroid use at baseline (approximately 20% relative risk reduction). The discontinuations for dyspepsia for etoricoxib vs. diclofenac related to the presence or absence of the significant baseline predictors of discontinuation are shown in Table 6.

Table 5.   The association of individual baseline characteristics with patient discontinuations due to dyspepsia: hazards ratios (95% CIs) from multivariable analysis
History of dyspepsia1.78 (1.44–2.20)
Prior UGI clinical event1.78 (1.40–2.27)
Age ≥651.35 (1.16–1.57)
Low-dose aspirin use1.18 (0.99–1.40)
Smoker1.04 (0.82–1.30)
Diabetes1.04 (0.83–1.31)
Male (vs. Female)0.72 (0.60–0.86)
Traditional NSAID use0.81 (0.67–0.98)
Systemic steroid use0.77 (0.60–0.99)
Methotrexate use0.78 (0.59–1.03)
Osteoarthritis (vs. Rheumatoid arthritis)0.85 (0.68–1.07)
PPI use0.89 (0.77–1.04)
Atherosclerotic cardiovascular disease0.96 (0.76–1.21)
Table 6.   Rates of discontinuation due to dyspepsia and treatment effects related to significant baseline predictors of discontinuation due to dyspepsia
 Rates per 100 patient-years (n/N; %)RR (95% CI)NNT (1-year)
EtoricoxibDiclofenac
  1. NNT, number-needed-to-treat.

Prior UGI event2.00 (34/1127; 3.02)2.91 (46/1133; 4.06)0.72 (0.46, 1.12)110
No prior UGI event1.20 (293/16285; 1.80)1.61 (378/16156; 2.34)0.76 (0.65, 0.88)244
Age ≥651.52 (155/7234; 2.14)2.15 (207/7162; 2.89)0.72 (0.59, 0.89)159
Age <651.08 (172/10178; 1.69)1.41 (217/10127; 2.14)0.78 (0.64, 0.95)303
Female1.34 (259/12925; 2.00)1.80 (336/12823; 2.62)0.75 (0.64, 0.88)217
Male1.01 (68/4487; 1.52)1.37 (88/4466; 1.97)0.75 (0.55, 1.03)278
Steroid1.14 (51/2685; 1.90)1.00 (44/2705; 1.63)1.16 (0.77, 1.73)714
No steroid1.28 (276/14727; 1.87)1.84 (380/14584; 2.61)0.70 (0.60, 0.82)179
Traditional NSAID1.26 (269/14209; 1.89)1.59 (328/14174; 2.31)0.80 (0.68, 0.94)303
No traditional NSAID1.22 (58/3203; 1.81)2.18 (96/3115; 3.08)0.57 (0.41, 0.79)104

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The results of our analysis in a large and long prospective randomized comparison indicate that prior upper GI events, increasing age and use of low-dose aspirin are the most important baseline risk factors, with ∼3 to 4 fold increase in upper GI complications (bleeding, perforation, obstruction) and ∼2–2.5 fold increase in overall upper GI clinical events (complicated events plus uncomplicated symptomatic ulcers or mild bleeding). Use of systemic corticosteroids also increased relative risk by about 2-fold. Dyspepsia and established atherosclerotic cardiovascular disease had a modest association with upper GI clinical events (approximately 50% relative risk increase), but not with the more serious upper GI complications. In addition, simply counting the number of major risk factors (prior upper GI event, older, age, aspirin use, and steroid use) allows further stratification, with increasing numbers of factors above one markedly increasing the risk of an upper GI event.

The importance of a prior history of upper GI events and increasing age has been well described in previous observational studies and prospective randomized trials, and corticosteroid therapy has also been reported as a risk factor in prior reports.1, 10–12 Although observational studies indicate that the addition of an NSAID to low-dose aspirin increases the risk of upper GI bleeding by approximately 2–4 fold as compared with low-dose aspirin alone,1 the risk of adding low-dose aspirin to an NSAID (as compared with the NSAID alone) has not been clearly delineated. A recent Cochrane review of randomized controlled trials found that upper GI complications among nonselective NSAID arms were not significantly higher among those taking aspirin vs. those not taking aspirin, although the risk in the coxib arms was increased with aspirin use (RR = 4.1, 2.4–7.1).16 Importantly, these analyses represent non-randomized comparisons within a traditional NSAID or coxib treatment arm and confounding factors other than aspirin use may influence the results; no endoscopic or outcome trial (including ours) directly compares patients randomized to an NSAID alone vs. an NSAID plus low-dose aspirin. In the MEDAL Program, we found that low-dose aspirin was an important independent predictor of upper GI complications with a relative risk of 3.4 (2.3–5.0), and the relative increase among aspirin users was similar in both the traditional NSAID and coxib arms.

Dyspepsia with NSAID use is not generally considered a particularly useful predictor for GI events because most complications occur without antecedent symptoms,17, 18 although prior observational studies have reported mixed results regarding whether dyspepsia is a significant risk factor.19, 20 We found that a history of dyspepsia had a modest association with development of upper GI clinical events (approximately 50% relative risk increase), but no significant association with complicated events. This association was independent of a history of prior ulcer disease. As discussed below, a prior history of dyspepsia is predictive of recurrent dyspepsia on NSAIDs. When severe enough, dyspepsia may trigger endoscopic evaluation. Some may argue that dyspepsia triggers more endoscopies, and that the increase in clinical events with dyspepsia merely represents the known increase in endoscopic ulcers with a traditional NSAID. If this were true, we would expect a lower relative risk for clinical events with etoricoxib vs. diclofenac in the subgroup with dyspepsia (more endoscopic procedures performed and relatively more ulcers discovered in the diclofenac group) than in the subgroup without dyspepsia. However, the relative risks were nearly identical in those with and without baseline dyspepsia. Thus, prior dyspepsia does appear to be independently associated with development of upper GI clinical ulcer disease.

Patients with atherosclerotic cardiovascular disease had higher rates of clinical but not complicated events than those without cardiovascular disease. Some, but not most, investigators have reported an association of cardiovascular disease and NSAID-related upper GI complications.10 However, our multivariable analysis failed to show a significant association with upper GI complications despite the much higher rate of events in the population with cardiovascular disease. This is likely because patients with cardiovascular disease commonly take low-dose aspirin, and it is the aspirin rather than the presence of the cardiovascular disease that is primarily responsible for the development of upper GI bleeding. Certainly, patients with cardiovascular disease who develop upper GI complications such as bleeding have an increased mortality as compared with those without cardiovascular disease.21

The rates of upper GI clinical events were much lower in the patients enrolled in the MEDAL Program than in previous GI outcome studies,10, 22–24 leading to smaller absolute risk differences between the treatment groups and larger numbers-needed-to-treat. This is most notably seen in the assessment of multiple risk factors. Analysis of a prior GI outcomes study revealed annualized rates of upper GI clinical events with naproxen of 8.6% with age ≥65 years, 5.7% with corticosteroid use, and ∼15% with a prior clinical event.11 The rate among patients with multiple risk factors was at least additive: e.g. 28.8% with older age and prior event, 40.6% with all three factors. In the MEDAL Program, more modest increases were seen. This may be related at least in part to the greater use of co-therapy with PPIs in patients at increasing risk: PPIs were used more commonly in patients with GI risk factors15 and PPI use increased with more than one risk factor.9 Greater adherence to protective co-therapy in patients with GI risk factors is associated with a decreased risk of upper GI clinical events.25

The most common side effect with NSAID therapy is dyspepsia (upper abdominal pain or discomfort). Weekly dyspepsia may occur in up to approximately 30% of NSAID users, and daily symptoms may occur in around 15%.26 Although not life-threatening, development of dyspepsia with NSAID use leads to decreased quality of life and increased healthcare resource utilization. We found that close to 2% of patients taking the traditional NSAID diclofenac in a trial setting discontinued their medication due to dyspepsia annually. Predictors of the development of NSAID-associated dyspepsia have not been formally assessed in previous prospective trials to our knowledge. Our analysis revealed that prior dyspepsia or prior upper GI event had the strongest associations, with 78% relative risk increases in dyspepsia severe enough to necessitate discontinuation of therapy. Age ≥65 years was associated with a modest, but significant relative risk increase of 35%. Two of these three predictors (prior event and older age) are also the key predictors of upper GI clinical events. Furthermore, men and patients not taking traditional NSAIDs at baseline had decreased risk of developing dyspepsia requiring discontinuation of therapy. It is not surprising that patients who previously have shown that they can tolerate NSAIDs are less likely to develop clinically significant dyspepsia with future NSAID therapy.

In conclusion, precise assessment of the risk of GI complications is key in the management of patients requiring NSAID therapy.1, 5–7 Medical co-therapy with proton pump inhibitors or misoprostol and/or the use of COX-2 selective inhibitors are(is) recommended to decrease GI complications5–7 in patients at increased GI risk, as the absolute benefit of these strategies is greatest in higher risk patients. Older age, prior GI clinical events and low-dose aspirin increase the risk of upper GI complications approximately 3 to 4 fold, and corticosteroid use increases risk approximately 2-fold. The risk also increases markedly as the number of risk factors increases. When one or more of these risk factors is present, physicians and patients should first reaffirm the need for NSAID therapy. If NSAIDs are felt to be necessary, then strategies to decrease upper GI complications should be instituted.5–7

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Declaration of personal interests: L. Laine has served as a consultant for AstraZeneca, Horizon, Santarus, Eisai, Novartis, Pozen, Nicox, Logical Therapeutics; has received research funding from Takeda Pharmaceuticals; has served on a data safety and monitoring board for Merck, Pfizer, and Bayer. S. Curtis and A. Kaur are employees of Merck and own stock and/or stock options. B. Cryer has served as a consultant for Merck, Pfizer, AstraZeneca, and Takeda Pharmaceuticals. C. Cannon has served as a consultant for Audiomedics; has received research funding from Accumetrics, AstraZeneca, Bristol-Myers Squibb/Sanofi partnership, GlaxoSmithKline, Merck/Schering Plough partnership; and has equity in Audiomedics Medical Systems. The authors thank Dr Yijie Zhou of Merck Research Laboratories for assistance with statistical analysis of the manuscript. Declaration of funding interests: This study was funded by Merck & Co., Inc., Whitehouse Station, NJ. The authors performed the analyses and wrote the manuscript without additional outside assistance or support.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Laine L. GI risk and risk factors of NSAIDs. J Cardiovasc Pharmacol 2006; 47(Suppl 1): S606.
  • 2
    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: 91724.
  • 3
    Smalley WE, Griffin MR, Fought RL, et al. Excess costs from gastrointestinal disease associated with nonsteroidal anti-inflammatory drugs. J Gen Intern Med 1996; 11: 4619.
  • 4
    Bloom BS. Direct medical costs of disease and gastrointestinal side effects during treatment for arthritis. Am J Med 1986; 84(Suppl 2A): 204.
  • 5
    Wilcox CM, Allison J, Benzuly K, et al. Consensus development conference on the use of nonsteroidal anti-inflammatory agents, including cyclooxygenase-2 enzyme inhibitors and aspirin. Clin Gastroenterol Hepatol 2006; 4: 10829.
  • 6
    Rostom A, Moayyedi P, Hunt R. Canadian Association of Gastroenterology Consensus Group Canadian consensus guidelines on long-term nonsteroidal anti-inflammatory drug therapy and the need for gastroprotection: benefits versus risks. Aliment Pharmacol Ther 2009; 29: 48196.
  • 7
    Lanza FL, Chan FKL, Quigley EMM. Guidelines for prevention of NSAID-related ulcer complications. Am J Gastroenterol 2009; 104: 72838.
  • 8
    Abraham NS, El-Serag HB, Johnson ML, et al. National adherence to evidence-based guidelines for the prescription of nonsteroidal anti-inflammatory drugs. Gastroenterology 2005; 129: 11718.
  • 9
    Laine L, Connors L, Griffin MR, et al. Prescription rates of protective co-therapy for NSAID users at high GI risk and results of attempts to improve adherence to guidelines. Aliment Pharmacol Ther 2009; 30: 76774.
  • 10
    Silverstein FE, Graham DY, Senior JR, et al. Misoprostol reduces serious gastrointestinal complications in patients with rheumatoid arthritis receiving nonsteroidal anti-inflammatory drugs. A randomized, double-blind, placebo-controlled trial. Ann Int Med 1995; 123: 2419.
  • 11
    Laine L, Bombardier C, Hawkey CJ, et al. Stratifying the risk of NSAID-related upper gastrointestinal clinical events: results of a double-blind outcomes study in patients with rheumatoid arthritis. Gastroenterology 2002; 123: 100612.
  • 12
    Hawkey CJ, Weinstein WM, Smalley W, et al. Effect of risk factors on complicated and uncomplicated ulcers in the TARGET lumiracoxib outcomes study. Gastroenterology 2007; 133: 5764.
  • 13
    Cannon CP, Curtis SP, Bolognese JA, et al. Clinical trial design and patient demographics of the Multinational Etoricoxib and Diclofenac Arthritis Long-term (MEDAL) Study Program: cardiovascular outcomes with etoricoxib vs diclofenac in patients with osteoarthritis and rheumatoid arthritis. Am Heart J 2006; 152: 23745.
  • 14
    Cannon CP, Curtis SP, FitzGerald GA, et al. Cardiovascular outcomes 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: 177181.
  • 15
    Laine L, Curtis SP, Cryer B, et al. Assessment of upper gastrointestinal safety of 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 2007; 369: 46573.
  • 16
    Rostom A, Muir K, Dube C, et al. The gastrointestinal toxicity of COX-2 inhibitors: a Cochrane Collaboration Systematic Review. Clin Gastroenterol Hepatol 2007; 5: 81828.
  • 17
    Armstrong CP, Blower AL. Non-steroidal anti-inflammatory drugs and life threatening complications of peptic ulceration. Gut 1987; 28: 52732.
  • 18
    Holvoet J, Terriere L, Van Hee W, et al. Relation of upper gastrointestinal bleeding to non-steroidal anti-inflammatory drugs and aspirin: a case control study. Gut 1991; 32: 7304.
  • 19
    Hansen JM, Hallas J, Lauritsen JM, et al. Non-steroidal anti-inflammatory drugs and complications: a risk factor analysis for clinical decision-making. Scand J Gastroenterol 1996; 31: 12630.
  • 20
    Singh G, Ramey DR, Morfeld D, et al. Gastrointestinal tract complications of nonsteroidal anti-inflammatory drug treatment in rheumatoid arthritis. A prospective observational cohort study. Arch Intern Med 1996; 156: 15306.
  • 21
    Rockall TA, Logan RF, Devlin HB, et al. Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996; 38: 31621.
  • 22
    FDA Arthritis Advisory Committee Meeting. February 7, 2001. Available at: http://www.fda.gov/ohrms/dockets/ac/01/briefing/3677b1.htm.
  • 23
    Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000; 343: 15208.
  • 24
    Schnitzer TJ, Burmester GR, Mysler E, et al. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), reduction in ulcer complications: randomised controlled trial. Lancet 2004; 364: 66574.
  • 25
    Van Soest EM, Sturkenboom MCJM, Dieleman JP, Verhamme KMC, Siersema PD, Kuipers EJ. Adherence to gastroprotection and the risk of NSAID-related upper gastrointestinal ulcers and haemorrhage. Aliment Pharmacol Ther 2007; 26: 26575.
  • 26
    Larkai EN, Smith JL, Lidsky MD, et al. Dyspepsia in NSAID users: the size of the problem. J Clin Gastroenterol 1989; 11: 15862.