PPI risc.pdf

Published on October 2019 | Categories: Documents | Downloads: 15 | Comments: 0 | Views: 329
of 9
Download PDF   Embed   Report

Comments

Content

GASTROENTEROLOGY 2007;133:790–798

Risk of Peptic Ulcer Hospitalizations in Users of NSAIDs With Gastroprotective Cotherapy Versus Coxibs A L   I    M E    C  N L   T   I    N A I    R   C  Y  A T   L   R  – A  C  T  

WAYNE A. RAY,* ,‡ CECILIA P. CHUNG,§ C. MICHAEL STEIN, §, WALTER E. SMALLEY, ‡,¶ KATHI HALL,* PATRICK G. ARBOGAST, # and MARIE R. GRIFFIN* ,‡,** *Division of Pharmacoepidemiology, Department of Preventive Medicine, Divisions of   § Rheumatology,   Clinical Pharmacology,  ¶ Gastroenterolo Gastroenterology, gy, and **Internal  Medicine, Department of Medicine, and   # Department of Biostatistics, Vanderbilt University School of Medicine, Nashville; and   ‡ Veterans’ Administration Tennessee Valley Health Care System, Geriatric Research, Education and Clinical Center, Nashville, Tennessee

See Maiden L et al on See Maiden al  on page 1040 in the September 2007 issue of CGH  of  CGH .

Background Background & Aims:   The primar primaryy strateg strategies ies to reduce the risk of serious gastropathy caused by traditional nonsteroidal nonsteroidal anti-inflammatory anti-inflammatory drugs (NSAIDs) (NSAIDs) are use of a coxib or concurrent use of a proton pump inhibitor or double-dose histamine-2 receptor antagonist. However, the relative clinical effectiveness of these therapeutic alternatives is understudied.  Methods:  We  We studied peptic ulcer hospitalizations in a cohort of Tennessee Medicaid enrollees between 1996 and 2004. To decrea decrease se potent potential ial “channe “channeling ling” ” bias, bias, the study study included only new episodes of prescribed NSAID or coxib use and controlled for multiple baseline risk factors for upper gastrointestinal disease. There were 234,010 and  48,710 new episodes of NSAID and coxib use, respectively, with 363,037 person-years of follow-up and 1223 peptic ulcer hospitalizations.  Results:  Current   Current users of  NSAIDs with no gastroprotective cotherapy had an ad justed  justed inciden incidence ce of peptic peptic ulcer ulcer hospitaliz hospitalizatio ations ns of 5.65 5.65 per 1000 person-years person-years,, 2.76 (95% confidence confidence interval, 2.35–3.23) times greater than that for persons not currently rently using using either either NSAIDs NSAIDs or coxibs. coxibs. This risk risk was reduced by 39% (16%–56%, 95% CI) for current users of  NSAIDs NSAIDs with gastro gastropro protec tective tive cother cotherapy apy and 40% (23%– (23%– 54%) 54%) for curr curren entt user userss of coxib coxibss witho without ut such such cother therap apy. y. Conc Concur urre rent nt user userss of NS NSAI AIDs Ds and and prot proton on pump pump inhibitors had a 54% (27%–72%) risk reduction, very  similar to the 50% (27%–66%) reduction for concurrent  user userss of prot proton on pump ump inhi inhibi bito tors rs and and coxi coxibs bs.. Conclusions:   These findings suggest that coprescribing a proton pump inhibitor with an NSAID is as effective as use of a coxib for reducing the risk of NSAIDinduced gastropathy.

N

onsteroidal anti-inflammatory anti-inflammatory drugs are among the most most common commonly ly prescr prescribe ibed d medica medicatio tions ns in the 1 world. Howeve However, r, the tradit tradition ional al drugs drugs in this this class class (NSAIDs) substantially increase the risk of serious upper

gastrointe gastrointestina stinall disease disease (NSAID-in (NSAID-induce duced d gastropat gastropathy), hy), including peptic ulcers, perforations, and upper gastrointestinal hemorrhages.1,2  At present, there are 2 primary  strategies to reduce this risk: use of a coxib or concurrent use of medications that protect the gut from the adverse effects of NSAIDs (gastroprotective cotherapy). 3–5 Coxibs confer a 40%–60% lower risk of ulcer complications than the NSAIDs6 – 8 but also can cause serious cardiovascular disease. 7,9–11 The synthetic prostaglandin misoprostol reduces ulcer complications in NSAID users by 40%12 but is poorly tolerated and now infrequently used. The histamine-2 receptor antagonists (in doses twice those recommended for ulcer healing) and the proton pump inhibitors tors are well-t well-tole olerat rated ed medica medicatio tions ns that that reduce reduce the occurrence of NSAID-associated peptic ulcers identified by endoscopic examination by 60%–80%. 13 There are critical unanswered questions regarding the effective effectiveness ness of these these strategie strategiess to prevent prevent NSAID-in NSAID-in-duced gastropathy. The clinical trials of proton pump inhibitors and double-dose histamine-2 receptor antagonists had endoscopic lesions as an end point, 14–16 so it remains uncertain whether or not these agents prevent the clinically relevant ulcer complications. Despite this lack lack of data, data, expert expert bodies bodies have have recomm recommen ended ded use of  gastroprotective cotherapy for high-risk NSAID users.3–5 However, these guidelines are frequently not followed in practice,17,18 perhaps because of the limited information on which they are based.  At present, no adequately powered gastrointestinal outcomes trials have directly compared a coxib with an NSAID NSAID with gastropro gastroprotect tective ive cotherap cotherapy. y. Although Although the 6–8 pivotal coxib trials were powered to study ulcer complications, they compared these agents with NSAID use without gastroprotective cotherapy. Unlike earlier studies, the recently completed trial comparing etoricoxib and diclofenac permitted proton pump inhibitor use; however, there was no reduction in ulcer complications for  Abbreviations used in this paper: CI, confidence interval; IRR, inci-

dence rate ratio. 2007 by the AGA Institute 0016-5085/07/$32.00 doi:10.1053/j.gastro.2007.06.058 ©

September 2007

the etoricoxib arm and no randomized comparison of  this agent with diclofenac plus a proton pump inhibitor.19 The trials that have directly compared these 2 therapeutic modalities20,21 were not powered to study  ulcer complications. However, it is increasingly unlikely that large clinical trials of gastroprotective cotherapy in NSAID users will be conducted. Proton pump inhibitors are available generically, which limits the incentives for pharmaceutical manufacturers to fund such studies. Furthermore, because NSAID-induced gastropathy is frequent and potentially life-threatening,1 the ethics of randomizing patients to NSAID use without gastroprotective cotherapy are questionable. For example, all patients in the in-progress PRECISION trial comparing celecoxib, naproxen, and ibuprofen in patients at high risk of cardiovascular disease will receive a proton pump inhibitor (www.clinicaltrials.gov ,   NCT00346216), and thus this trial will not provide information on the relative efficacy of a coxib alone versus an NSAID with gastroprotective cotherapy. Thus, observational studies may provide the best current alternative to obtaining data that are essential to guide clinical practice. We conducted a large cohort study to directly compare the effect of an NSAID with gastroprotective cotherapy with that of a coxib on the incidence of  peptic ulcer hospitalizations.

Materials and Methods Cohort and Follow-up Study data were obtained from computerized files of TennCare, Tennessee’s expanded Medicaid program.22,23 Study files (enrollment, hospital, outpatient, nursing home, linked death certificate) allowed identification of the study cohort, classification of baseline upper gastrointestinal disease risk and other medical comorbidity, tracking of current use of NSAIDs/coxibs and gastroprotective cotherapy, and ascertainment of end points.22,24 The cohort consisted of persons with a new episode of  prescribed NSAID or coxib use, which should reduce several types of potential bias.25 It included those who filled a prescription for an NSAID (excluding low-dose aspirin) or coxib between January 1, 1996, and December 31, 2004, and who had no prior use of these drugs in the 365 days preceding the date the prescription was filled (t0). As of t0, cohort members were 40 years of age or older (younger persons were not included because of lower incidence of peptic ulcer disease) and had enrollment for at least 730 days preceding t 0  (lapses 7 days allowed) in a category with full drug benefits, providing a 2-year period to characterize patients’ baseline risk of upper gastrointestinal disease. To assure that cohort members had active medical surveillance, we also required filling of  at least 1 prescription in each of the first and second years preceding t0.

GASTROPROTECTIVE COTHERAPY IN NSAID USERS

791

The cohort excluded persons with serious illnesses in the past 730 days (cancer other than nonmelanoma skin cancer, human immunodeficiency virus infection, renal failure, liver disease, respiratory failure, or transplantation), because these were considered likely to have shortened follow-up or to have disease-related upper gastrointestinal complications. Patients also were excluded if  they had serious upper gastrointestinal diseases that cause bleeding unrelated to NSAID use (liver cirrhosis, esophageal varices, chronic alcoholism, bariatric/other surgery resulting in gastrojejunal anastomosis) or who were in the hospital in the preceding 30 days. Follow-up began on t0  and extended through the first of the end of the study, death, occurrence of a study end point, loss of TennCare enrollment, or diagnosis of an exclusion illness. Follow-up also ceased 365 days following last NSAID/coxib use or when the patient switched to a drug in the other category. Patients who left the cohort could subsequently reenter if they met the study criteria. Person-time in the hospital and within 30 days of discharge was excluded because in-hospital events were not considered study end points and because we did not have information on in-hospital medications (eg, heparin) and up to 30 days could be required to detect medication changes made in the hospital.

 Medication Exposure TennCare pharmacy records were used to classify  each person-day of follow-up according to probable use of NSAIDs, coxibs, gastroprotective cotherapy, and other medications. These included the date prescription was dispensed, drug, quantity, dose, and days of supply (usually limited to 30 days), edited to resolve infrequent discrepancies with drug quantity. Computerized pharmacy records are an excellent source of medication data  because they are not subject to information bias22 and have high concordance with patient self-report of medication use.26–28  Although many study medications are available over the counter, the low-income Medicaid patients had an economic incentive to obtain these via  prescription, and, during the study period, the TennCare program had a relatively generous drug benefit with no deductible or co-payment. Although there will be some misclassification due to nonprescription drugs, our past experience suggests that this is limited 29–31 and is most likely to bias toward the null. “Current use” was defined as the period between the filling of the prescription and the end of the days of  supply (for aspirin, this period extended an additional 7 days, the typical platelet half-life), “indeterminate use” as that from the end of the days of supply through 90 days following the filling of the prescription, and “former use” as the remainder of the 365 days following the fill date. The most frequently prescribed study drugs were naproxen (38% of current NSAID use), ibuprofen (28%), celecoxib (47% of current coxib use), and rofecoxib (40%),

    T     C     A    –    R     T     L     A    Y     C    R     I     A     N     T     I     L    N     C    E     M     I     L     A

792

A L   I    M E    C  N L   T   I    N A I    R   C  Y  A T   L   R  – A  C  T  

RAY ET AL

for which individual analyses were performed. Valdecoxib accounted for 12% of coxib current use, and other NSAIDs/coxibs each accounted for 5% or less of current use. NSAID and coxib current use was classified as low or usual dose, with the latter defined as doses at or above those studied in key clinical trials for osteoarthritis or rheumatoid arthritis (eg, 1000 mg for naproxen, 2400 mg for ibuprofen, 200 mg for celecoxib, 25 mg for rofecoxib). Usual doses accounted for 66% of NSAID and 87% of  coxib current use. Gastroprotective cotherapy 13 included misoprostol, the proton pump inhibitors (omeprazole, lansoprazole, rabeprazole, pantoprazole, esomeprazole), in any dose, and the histamine-2 receptor antagonists in twice the doses recommended for ulcer healing (cimetidine 1600 mg, ranitidine 600 mg, famotidine 80 mg, nizatidine 600 mg). The study cohort included 234,010 new episodes of  NSAID use and 48,710 new episodes of coxib use, with 363,037 person-years of follow-up. There were 67,657 (22% of total follow-up) and 20,987 (41%) person-years for current use of NSAIDs or coxibs, respectively. NSAID and coxib users had gastroprotective cotherapy for 16% and 34% of current use, respectively. Proton pump inhibitors accounted for 71% of gastroprotective cotherapy, double-dose histamine-2 receptor antagonists for 22%, misoprostol for 1%, and use of multiple agents for 6%. Gastroprotective cotherapy use was present before t 0  for 68% of proton pump inhibitor use during follow-up and 86% of double-dose histamine-2 receptor antagonist use.

 End Points  The study end point was hospitalization with peptic ulcer disease, identified from a primary discharge diagnosis code for peptic ulcer (International Classification of Diseases, Ninth Revision, Clinical Modification codes of 531, 532, 533), gastritis with hemorrhage (535.01, 535.41, 535.51), or gastrointestinal hemorrhage (578). The latter code, which could indicate bleeding from either the upper or lower gastrointestinal tract, was accepted only if a secondary diagnosis indicated a peptic ulcer or there were no diagnoses for lower gastrointestinal disease. The date of end point occurrence was considered the day of the hospital admission or the prior day  if there was an emergency department visit on that day. We assessed the accuracy of the end point definition in a random sample of 135 cases. Blinded adjudication using the case definition from the TARGET study 8,32 classified 90% as either definite/probable (lesion identified by diagnostic procedure) or possible (signs and symptoms consistent with upper gastrointestinal disease and no evidence from diagnostic procedures of lower gastrointestinal hemorrhage) cases. We thus identified 1223 peptic ulcer hospitalizations during cohort follow-up, or 3.4 per 1000 person-years. Incidence increased with age, from 1.6 per 1000 per year

GASTROENTEROLOGY Vol. 133, No. 3

for persons 40–44 years of age to 18.2 per 1000 for those 90 years of age or older. Persons with a prior peptic ulcer hospital admission subsequently had a 5.6-fold increase in the age-adjusted rate of ulcer hospitalizations ( P   .001).

 Potential Confounders  To control for channeling of high-risk patients to either coxibs or gastroprotective cotherapy, we identified an extensive set of baseline characteristics of the cohort members as well as changes in some of these during follow-up. Baseline characteristics included demographic factors (age, gender, TennCare uninsured enrollment, race, urban residence, calendar year, living in nursing home) as well as factors indicating increased risk for upper gastrointestinal disease. The latter included history  of peptic ulcer, gastritis or duodenitis, other upper gastrointestinal symptoms, use of drugs to treat peptic ulcer disease (proton pump inhibitor, histamine-2 receptor antagonist, other antiulcer drug, prior   Helicobacter pylori eradication therapy 33), and other medications that can cause gastrointestinal bleeding (prescribed low-dose aspirin, other antiplatelet drugs [dipyridamole, ticlopidine, clopidogrel], anticoagulants, or systemic corticosteroids). To control for the effect of general medical frailty 1 on the risk of peptic ulcer hospitalizations, we constructed a  baseline summary medical comorbidity score from prior prescriptions and medical care encounters for lower gastrointestinal, cardiovascular, metabolic, musculoskeletal, neurologic, mental, respiratory, and infectious diseases as well as indicators of general frailty (any other nongastrointestinal hospital admission or emergency department  visit, fall-related injuries, prescription of wheelchair or other mobility aid, prescription for home oxygen). To calculate the score, we performed a Poisson regression analysis among former users of NSAIDs/coxibs with no gastroprotective cotherapy using a model that included demographic factors and the medical comorbidity indicators listed previously. The score was then defined for the entire cohort as the deciles of the sum 1*M1  2*M2  . . .  k*Mk, where the  M  values are the indi vidual variables indicating medical comorbidity and the   values are the regression coefficients estimated from the Poisson model. After controlling for demographic factors, there was a greater than 5-fold difference in end point incidence between the lowest (0) and highest (9)  value of the comorbidity score. This technique to summarize the effect of multiple risk factors 34 yields estimates similar to those obtained by including the individual score components in regression models.35 Potential time-dependent confounders were use of  low-dose aspirin, other antiplatelet drugs, anticoagulants or systemic corticosteroids, new residence in a nursing home, and new nongastrointestinal hospital admission. The analysis did not control for gastrointestinal disease diagnosed after cohort entry because this was potentially 

September 2007

Table 1.   Baseline

GASTROPROTECTIVE COTHERAPY IN NSAID USERS

793

Cohort Characteristics, According to NSAID/Coxib Status NSAID

Characteristics at baseline New episodes of use, N Age, mean  SD ( y ) 65 years of age or older (% ) Female (% ) TennCare enrollment uninsured ( % ) White race (% ) Residence in Standard Metropolitan Statistical Area ( % ) Year cohort entry 2002 or later ( % ) In nursing home ( % ) Medical care in the 2 years before baseline Diagnosed upper gastrointestinal disease ( % ) Peptic ulcer hospitalization ( % ) Peptic ulcer, outpatient (% ) Gastritis (% ) Other upper gastrointestinal symptoms ( % ) Helicobacter pylori  eradication treatment ( % ) Drugs for upper gastrointestinal disease ( % ) Proton pump inhibitor ( % ) Histamine-2 receptor antagonist ( % ) Other drug for ulcer treatment ( % ) Antacid (% ) Any medication that can cause gastrointestinal bleeding ( % ) Low-dose aspirin (% ) Other antiplatelet drug ( % )b  Anticoagulant (% ) Systemic corticosteroid ( % ) Medical comorbidity score, meanc 

All

Gastroprotective cotherapy a 

Coxib

234,010 55.8  12.0 21.9 71.1 39.5 69.7 52.7 30.9 5.3

27,927 55.9  11.0 20.5 69.9 39.0 77.6 47.8 58.6 5.1

48,710 61.2  12.8 38.6 70.7 43.0 75.2 50.7 61.0 5.5

27.8 1.8 7.3 9.2 19.3 1.6 46.0 15.8 37.9 2.0 1.7 42.5 14.5 5.4 3.9 27.7 4.6

48.6 4.1 14.8 21.4 32.5 4.5 100.0 70.4 67.1 4.3 3.5 57.1 21.6 10.0 5.3 38.2 5.8

30.7 2.2 8.5 11.0 20.7 2.6 58.0 34.1 40.2 2.2 2.2 51.3 18.3 8.2 6.5 32.5 5.2

NOTE. Data (except for number of episodes and age) adjusted for age by the method of Brenner et al. 43 a Defined as use of a proton pump inhibitor, histamine-2 receptor antagonist (double dose), or misoprostol within 30 days of the beginning of  follow-up. b Dipyridamole, ticlopidine, or clopidogrel. c Summary score calculated from medical care for nongastrointestinal illness that ranges from 0 (least comorbidity) to 9 (most comorbidity).

on the causal pathway for NSAID-induced gastropathy. Because low-dose aspirin may be obtained over the counter and thus not be recorded in the TennCare data, some analyses controlled for eligibility for low-dose aspirin use (history of cardiovascular disease for which the Food and Drug Administration recommends long-term aspirin prophylaxis7); these had identical findings as the primary analysis.

 Analysis  Estimates of incidence rate ratios (IRRs) adjusted for potential differences between study groups were calculated from Poisson regression models that included NSAID or coxib exposure, gastroprotective cotherapy, and baseline as well as time-dependent potential confounders. In some analyses, adjusted incidence was calculated by multiplying the unadjusted rate for the reference category by the adjusted IRR. Planned subgroup analyses were performed for patients who met current guidelines for use of gastroprotective cotherapy,3–5 lowdose aspirin users, and the most frequently prescribed

individual NSAIDs/coxibs. Because the study began before the availability of the coxibs, we conducted an analysis restricted to 1999 through 2004; findings did not differ materially from those of the primary analysis (data  not shown). All analyses were performed with SAS version 9.0 (SAS Institute Inc, Cary, NC). All  P  values are 2 sided. The study was approved by the Vanderbilt Committee for the Protection of Human Subjects as well as the State of Tennessee Bureau of TennCare and Department of  Health and Environment. The study was funded by a  grant from the Agency for Healthcare Quality and Research. The sponsor of the study had no role in study  design, data collection, data analysis, data interpretation, or writing of the report.

Results Coxib users had a mean age 5 years greater than that of NSAID users (61 vs 56 years; Table 1) but otherwise had comparable demographic characteristics. After

    T     C     A    –    R     T     L     A    Y     C    R     I     A     N     T     I     L    N     C    E     M     I     L     A

794

RAY ET AL

GASTROENTEROLOGY Vol. 133, No. 3

A L   I    M E    C  N L   T   I    N A I    R   C  Y  A T   L   R  – A  C  T  

Figure 1.   Adjusted incidence of peptic ulcer–related hospitalizations for NSAID and coxib users, according to gastroprotective cotherapy.  The left y-axis shows the adjusted rate of hospitalizations per 1000 person-years (PY) of follow-up; the right y-axis shows the IRR. Vertical bars are 95% CIs. The reference group is former users of either NSAIDs or coxibs with no gastroprotective cotherapy. All rates and IRRs are adjusted for baseline demographic factors, baseline history of upper gastrointestinal disease, baseline medical comorbidity, use of medications that cause upper gastrointestinal bleeding (baseline and during follow-up), and hospital or nursing home admission during follow-up.  The person-years of follow-up were as follows: 57,032 (NSAID, no gastroprotective cotherapy), 10,625 (NSAID, gastroprotective cotherapy), 13,962 (coxib, no gastroprotective cotherapy), 7,025 (coxib, gastroprotective cotherapy), and 135,758 (former). Data shown in the figure exclude 120,640 person-years for indeterminate users and 17,995 for former users with concurrent gastroprotective cotherapy.

adjusting for age, coxib users had more frequent use of  medications for ulcer treatment or upper gastrointestinal symptoms (58% vs 46%). Coxib users also had more frequently received other drugs that cause gastrointestinal bleeding (51% vs 42%) and had a slightly higher medical comorbidity score (5.2 vs 4.6). However, NSAID users with baseline gastroprotective cotherapy had an

even greater baseline prevalence of upper gastrointestinal disease, use of medications that cause gastrointestinal bleeding, and other medical comorbidity than the coxib users (Table 1). Current users of NSAIDs with no gastroprotective cotherapy had an adjusted incidence of peptic ulcer hospitalizations of 5.65 per 1000 person-years, nearly 3 times greater than that of 2.04 per 1000 for comparable persons who had stopped using either NSAIDs or coxibs (Figure 1; IRR, 2.76; 95% confidence interval [CI], 2.35– 3.23;  P   .0001). In this group, peptic ulcer hospitalization incidence increased with NSAID dose (IRRs for usual and low doses: 3.27 [2.74–3.90] and 2.10 [1.69– 2.62], respectively). Gastroprotective cotherapy or use of a  coxib reduced, but did not eliminate, this increased risk (Figure 1). Among current users of NSAIDs with gastroprotective cotherapy and current users of coxibs without or with such cotherapy, the respective IRRs were 1.68 (1.20–2.34), 1.65 (1.28–2.14), and 1.37 (0.94–1.98). When compared with NSAID use alone, use of either an NSAID with gastroprotective cotherapy or a coxib was associated with a similar reduction in the incidence of  peptic ulcer hospitalizations (Table 2). For current users of either NSAIDs with gastroprotective cotherapy or coxibs without or with such cotherapy, the respective risk reductions were 39% (16%–56%), 40% (23%–54%), and 50% (28%–66%). Among current users of NSAIDs, concurrent users of proton pump inhibitors had the greatest risk reduction (54% [27%–72%]), which was very similar to that for concurrent users of proton pump inhibitors and coxibs (50% [27%–66%]). We performed a dose-adjusted analysis for naproxen, ibuprofen, rofecoxib, and celecoxib, the most frequently  used individual drugs in the study cohort (Table 3). There were 2 primary comparisons. In the first, current use of naproxen (the most frequently used NSAID) with

Table 2.  Incidence

of Peptic Ulcer Hospitalizations for Current Users of NSAIDs or Coxibs, According to Use of  Gastroprotective Cotherapy 

NSAID No gastroprotective cotherapy Any gastroprotective cotherapy a  Proton pump inhibitor Histamine-2 receptor antagonist, double dose Misoprostol Coxib No gastroprotective cotherapy Any gastroprotective cotherapy a  Proton pump inhibitor

Person-years

Rate/1000, Adjusted

IRR

57,032 10,625 6227 3077 405

5.65 3.44 2.57 4.72 3.94

1 0.608 0.456 0.835 0.697

Reference 0.44–0.84 0.28–0.73 0.50–1.40 0.17–2.80

13,962 7025 6027

3.38 2.80 2.80

0.598 0.495 0.496

0.46–0.77 0.34–0.72 0.34–0.73

95% CI

P   value

.003 .0011 .4959 .6106 .0001

.0002 .0004

NOTE. “Double dose” represents twice the dose recommended for ulcer healing. All rates and IRRs are adjusted for baseline demographic factors, baseline history of upper gastrointestinal disease, baseline medical comorbidity, use of medications that cause upper gastrointestinal bleeding (baseline and during follow-up), and hospital or nursing home admission during follow-up. a The total for any gastroprotective cotherapy includes person-time with concurrent use of more than a single type of therapy, which is excluded from the specific gastroprotective cotherapy therapy analyses. For the coxibs, sample size was insufficient to assess concurrent use of  double-dose histamine-2 receptor antagonists or misoprostol.

September 2007

GASTROPROTECTIVE COTHERAPY IN NSAID USERS

795

Table 3.   Incidence

of Peptic Ulcer Hospitalizations for Current Users of Frequently Prescribed Individual NSAIDs or Coxibs, Adjusted for Dose and According to Concurrent Use of Proton Pump Inhibitors Comparison with naproxen, no gastroprotective cotherapy 

Naproxen No gastroprotective cotherapy Concurrent proton pump inhibitor Ibuprofen No gastroprotective cotherapy Concurrent proton pump inhibitor Rofecoxib No gastroprotective cotherapy Concurrent proton pump inhibitor Celecoxib No gastroprotective cotherapy Concurrent proton pump inhibitor

Comparison with naproxen, concurrent proton pump inhibitor

Person-years

Rate/1000, adjusted

IRR

95% CI

P   value

P   value

22,559 1925

7.80 4.09

1 0.524

Reference 0.27–1.03

.0616

Reference

16,900 1387

4.57 2.03

0.585 0.261

0.44–0.77 0.08–0.82

.0002 .0216

.7575 .2958

5700 2321

4.13 3.77

0.530 0.483

0.37–0.76 0.28–0.83

.0006 .0086

.9759 .8481

6690 2760

2.66 1.40

0.341 0.179

0.24–0.49 0.09–0.37

.0001

.2565 .0275

.0001

NOTE. All rates and IRRs are adjusted for baseline demographic factors, baseline history of upper gastrointestinal disease, baseline medical comorbidity, use of medications that cause upper gastrointestinal bleeding (baseline and during follow-up), and hospital or nursing home admission during follow-up. Adjustment for dose performed by adding a term for dose of current use (low versus usual) to the Poisson regression model

no gastroprotective cotherapy was the reference category. Relative to this group, concurrent users of naproxen and a proton pump inhibitor had a 48% ( 3% to 73%; P   .0616) reduction in the incidence of peptic ulcer hospitalizations and there were reductions of 47% (24%–63%;  P   .0006) for rofecoxib and 66% (51%–76%) for celecoxib, each with no gastroprotective cotherapy. In the second comparison, concurrent use of naproxen with a proton pump inhibitor was the reference category, which tested whether or not the other drugs, with or without a proton pump inhibitor, conferred an additional protective effect. None of these had a statistically significant further reduction in risk, with the exception of concurrent users of  celecoxib and a proton pump inhibitor ( P   .0275). The gastroprotective effect of concurrent proton pump inhibitor or coxib use was present in subgroups of patients with factors that increase risk of NSAID-induced gastropathy (Figure 2). These included increased NSAID dose, age 65 years or older, past history of peptic ulcer disease, medical frailty, use of low-dose aspirin, and use of other antiplatelet drugs or oral anticoagulants. Within each subgroup, the gastroprotective strategies were associated with reduced incidence of peptic ulcer hospitalizations, relative to that for current NSAID use with no gastroprotective cotherapy, although in some of the smaller subgroups the differences were not statistically  significant. For patients with prescribed aspirin, the only  gastroprotective strategy associated with a statistically  significant risk reduction was concurrent coxib and proton pump inhibitor use.

Discussion The findings from this large cohort study show  that use of an NSAID in conjunction with a proton

pump inhibitor has a gastrointestinal safety advantage over NSAID use alone comparable to that of a coxib, with respective reductions in the risk of peptic ulcer hospitalizations of 54% and 40%. This is similar to the 40%–60% reduction in ulcer complications reported from the pivotal coxib trials.6 – 8 Because the present study included substantial numbers of high-risk patients generally excluded from the coxib trials, such as those with recent peptic ulcers or other upper gastrointestinal disease or use of medications that cause gastrointestinal bleeding, our findings further suggest that the benefits of these strategies for preventing NSAID-induced gastropathy  will be present in the patients for whom they would be most important but who have not been well studied. Because the currently available guidelines for pre vention of NSAID-induced gastropathy did not distinguish between the individual agents recommended for cotherapy, 3–5,13 our initial analysis grouped proton pump inhibitors and double-dose histamine-2 receptor antagonists. However, proton pump inhibitors provide more complete suppression of gastric acid and greater efficacy for healing of NSAID-related peptic ulcers 36 and presently are the most commonly used gastroprotective cotherapy. Thus, we conducted a planned analysis of the different types of gastroprotective cotherapy, which showed greater efficacy for the proton pump inhibitors. Our findings are consistent with those from other recently published epidemiologic studies. Höer et al found that concomitant prescribing of a proton pump inhibitor with diclofenac reduced the odds ratio of an ulcer hospitalization from 2.4 to 1.3.37 García-Rodríguez et al reported that coxibs conferred a lower risk of serious upper gastrointestinal complications than did NSAIDs

    T     C     A    –    R     T     L     A    Y     C    R     I     A     N     T     I     L    N     C    E     M     I     L     A

796

RAY ET AL

GASTROENTEROLOGY Vol. 133, No. 3

A L   I    M E    C  N L   T   I    N A I    R   C  Y  A T   L   R  – A  C  T  

Figure 2. Subgroup analyses. Adjusted incidence of peptic ulcer–related hospitalizations in users of NSAIDs with concurrent proton pump inhibitor (PPI) use or users of coxibs without or with concurrent PPI use. The chart depicts the IRR and 95% CIs, with the reference category that of NSAID users with no gastroprotective cotherapy. All rates andIRRs are adjusted for baseline demographic factors, baseline history of upper gastrointestinal disease, baseline medical comorbidity, use of medications that cause upper gastrointestinal bleeding (baseline and during follow-up), and hospital or nursing home admission during follow-up. The rate in parentheses after the subgroup description is the adjusted rate in that subgroup of peptic ulcer–related hospitalizations per 1000 person-years (PY) among NSAID users with no gastroprotective cotherapy. “Past ulcer” is defined as diagnosis of peptic ulcer disease or gastritis in the 2 years preceding start of follow-up, “medical comorbidity” is defined as the upper quintile of  summary medical comorbidity score, and “low dose aspirin” is defined as current use of low-dose aspirin; antiplatelet drugs are dipyridamole, ticlopidine, or clopidogrel.

but that addition of gastroprotective cotherapy reduced the risk associated with NSAIDs by nearly 40%. 38 Lanas et al noted that among current users of NSAIDs, concurrent use of histamine-2 receptor antagonists or proton pump inhibitors was associated with a 35% and 67% reduction in the risk of hospitalization for upper gastrointestinal bleeding.39 Our analysis of individual drugs and high-risk patient subgroups included several interesting findings. Naproxen without gastroprotective cotherapy was associated with the greatest risk of peptic ulcer hospitalizations. This was reduced by approximately half with concurrent use of a proton pump inhibitor or with use of  rofecoxib alone, although the former finding did not reach conventional levels of statistical significance.  Among patients with no gastroprotective cotherapy, the incidence of peptic ulcer hospitalizations was lower for use of ibuprofen than for naproxen and was comparable to that for rofecoxib. This is consistent with the findings of numerous epidemiologic studies that ibuprofen is less likely than other NSAIDs to cause peptic ulcers. 40 The

best gastroprotection was associated with concurrent use of celecoxib and a proton pump inhibitor, which was significantly better than either naproxen alone or naproxen with a proton pump inhibitor. This finding is consistent with the recently reported clinical trial of  Chan et al,41 in which patients not taking low-dose aspirin randomized to celecoxib with esomeprazole had lower risk of recurrent peptic ulcer bleeding than did patients treated with celecoxib alone. We conducted an analysis according to current use of  prescribed low-dose aspirin. Although the numbers of  such patients in the cohort were small, neither a concurrent proton pump inhibitor nor a coxib alone provided statistically significant gastroprotection. The latter finding is consistent with findings from the randomized controlled trials.6,8 However, we did find a marked and statistically significant protective effect in patients treated with aspirin who received a coxib with a proton pump inhibitor. Further study of how to best provide gastroprotection to patients who take both an NSAID and low-dose aspirin is needed.

September 2007

The primary limitation of this nonrandomized study is possible confounding due to “channeling” of patients at high risk of peptic ulcer disease to use of either a coxib or gastroprotective cotherapy. The study design thus included several measures to reduce this potential bias. We required cohort members to have a new episode of prescribed NSAID or coxib use, preceded by a drug-free period, and stopped follow-up if a patient switched to a  drug in the other class. This effectively excluded patients who had switched from an NSAID to a coxib because of  gastrointestinal symptoms, a type of channeling that is particularly difficult to address with traditional designs. We also controlled for an extensive set of baseline risk factors for upper gastrointestinal disease in the statistical analysis, which should further reduce channeling bias. Because we measured these factors before NSAID therapy  began, they would not be markers for NSAID-induced gastropathy and controlling for them would not introduce the bias ordinarily caused by controlling for variables on the causal pathway between exposure and end point.42  At baseline, coxib users were 5 years older than NSAID users, and, after adjusting for age, were slightly more likely to have a history of upper gastrointestinal or other medical comorbidity. This raises the possibility that residual confounding could cause the gastrointestinal benefit of coxibs to be underestimated. However, baseline NSAID users with gastroprotective cotherapy had an even greater prevalence of upper gastrointestinal and other medical comorbidity. Thus, any residual confounding should understate the gastrointestinal benefits of  both coxibs and proton pump inhibitors, and, in the comparison of these 2 strategies for preventing NSAIDinduced gastropathy, favor the coxibs.  Another study limitation was the use of a computerized database of medical care encounters to define exposure to study medications. Although automated pharmacy dispensing records are an excellent unbiased source of information on drug use,22,26–28 some NSAIDs, omeprazole, and several histamine-2 receptor antagonists were available over the counter, use that would not have been recorded in the prescription database. However, Medicaid paid for over-the-counter drugs when prescribed, and thus patients had strong economic incentive to obtain these drugs by prescription. Our previous studies suggest that in this circumstance, over-the-counter use is minimal.29–31 However, some exposure misclassification is inevitable and should bias toward the null. Similarly, aspirin could be obtained both via prescription and over the counter. However, the results of an analysis that excluded patients for whom low-dose aspirin prophylaxis is recommended, the group in which misclassification would be of greatest concern, did not differ from those of the primary analysis. In conclusion, our findings strongly support concurrent use of a proton pump inhibitor for NSAID-treated

GASTROPROTECTIVE COTHERAPY IN NSAID USERS

797

patients with high risk of upper gastrointestinal disease, because this strategy conferred gastroprotection equivalent to that of a coxib. They thus underscore the need for future clinical trials of coxibs to directly compare these agents to an NSAID with a proton pump inhibitor. Finally, further investigation is urgently needed of the benefits of adding a proton pump inhibitor to celecoxib and other coxibs, because our findings suggest this strategy provided the best gastroprotection and was effective for patients taking low-dose aspirin. References 1. Wolfe MM, Lichtenstein DR, Singh G. Gastrointestinal toxicity of  nonsteroidal antiinflammatory drugs. N Engl J Med 1999;340: 1888–1899. 2. Hernandez-Diaz S, Garcia-Rodriguez LA. Epidemiologic assessment of the safety of conventional nonsteroidal anti-inflammatory  drugs. Am J Med 2001;110:20S–27S. 3. Recommendations for the medical management of osteoarthritis of the hip and knee: 2000 update. Arthritis Rheum 2000;43: 1905–1915. 4. Lanza FL. A guideline for the treatment and prevention of NSAIDinduced ulcers. Am J Gastroenterol 1998;93:2037–2046. 5. MacLean CH. Quality indicators for the management of osteoarthritis in vulnerable elders. Ann Intern Med 2001;135:711–721. 6. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celocoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS Study: a randomized controlled trial. JAMA 2000;284:1247–1255. 7. 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;343:1520–1528. 8. 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; 364:665–674. 9. Bresalier RS, Sandler RS, Quan H, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med 2005;352:1–11. 10. Solomon SD, McMurray JJV, Pfeffer MA, et al. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med 2005;352:1–10. 11. Nussmeier NA, Whelton AA, Brown MT, et al. Complications of  the COX-2 inhibitors parecoxib and valdecoxib after cardiac surgery. N Engl J Med 2005;352:11. 12. 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 Intern Med 1995;123:241–249. 13. Rostom A, Wells G, Tugwell P, et al. Prevention of NSAID-induced gastroduodenal ulcers. Cochrane Database Syst Rev 2000;(4): CD002296. Review. Update in: Cochrane Database Syst Rev. 2002;(4):CD002296. 14. Yeomans ND, Tulassay Z, Juhasz L, et al. A comparison of  omeprazole with ranitidine for ulcers associated with nonsteroidal antiinflammatory drugs. Acid Suppression Trial: Ranitidine versus Omeprazole for NSAID-associated Ulcer Treatment (ASTRONAUT) Study Group. N Engl J Med 1998;338:719–726. 15. Hawkey CJ, Karrasch JA, Szczepanski L, et al. Omeprazole compared with misoprostol for ulcers associated with nonsteroidal antiinflammatory drugs. Omeprazole versus misoprostol for NSAID-induced ulcer management. N Engl J Med 1998;338:727– 734.

    T     C     A    –    R     T     L     A    Y     C    R     I     A     N     T     I     L    N     C    E     M     I     L     A

798

A L   I    M E    C  N L   T   I    N A I    R   C  Y  A T   L   R  – A  C  T  

RAY ET AL

16. Taha AS, Hudson N, Hawkey CJ, et al. Famotidine for the prevention of gastric and duodenal ulcers caused by nonsteroidal antiinflammatory drugs. N Engl J Med 1996;334:1435–1439. 17. Smalley W, Stein CM, Arbogasts PG, et al. Underutilization of  gastroprotective measures in patients receiving nonsteroidal anti-inflammatory drugs. Arthritis Rheum 2002;46:2195–2199. 18. 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;129:1171– 1178. 19. 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;369:465–473. 20. Chan FKL, Hung LCT, Suen BY, et al. Celecoxib versus diclofenac and omeprazole in reducing the risk of recurrent ulcer bleeding in patients with arthritis. N Engl J Med 2002;347:2104–2110. 21. Lai KC, Chu KM, Hui WM, et al. Celecoxib compared with lansoprazole and naproxen to prevent gastrointestinal ulcer complications. Am J Med 2005;118:1271–1278. 22. Ray WA, Griffin MR. Use of Medicaid data for pharmacoepidemiology. Am J Epidemiol 1989;129:837–849. 23. Ruther M, Pagan BA, Rinkle V, et al. Program statistics: Medicare and Medicaid data book, 1984. Baltimore, MD: Health Care Financing Administration, 1986. 24. Ray WA. Population-based studies of adverse drug effects. N Engl J Med 2003;349:1592–1594. 25. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol 2003;158:915–920. 26. Landry JA, Smyer MA, Tubman JG, et al. Validation of two methods of data collection of self-reported medicine use among the elderly. Gerontologist 1988;28:672–676. 27. West SL, Savitz DA, Koch G, et al. Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol 1995;142:1103–1110. 28. Johnson RE, Vollmer WM. Comparing sources of drug data about the elderly. J Am Geriatr Soc 1991;39:1079–1084. 29. Griffin MR, Piper JM, Daugherty JR, et al. Nonsteroidal antiinflammatory drug use and increased risk for peptic ulcer disease in elderly persons. Ann Intern Med 1991;114:257–263. 30. Smalley WE, Ray WA, Daugherty J, et al. Use of nonsteroidal anti-inflammatory drugs and incidence of colorectal cancer. Arch Intern Med 1999;159:161–166. 31. Griffin MR, Yared A, Ray WA. Nonsteroidal anti-inflammatory  drugs and acute renal failure in elderly persons. Am J Epidemiol 2000;151:488–496. 32. Hawkey CJ, Farkouh M, Gitton X, et al. Therapeutic arthritis research and gastrointestinal event trial of lumiracoxib—study  design and patient demographics. Aliment Pharmacol Ther 2004; 20:51–63. 33. de Boer WA, Tytgat GNJ. Regular review: treatment of Helicobacter pylori infection. BMJ 2000;320:31–34.

GASTROENTEROLOGY Vol. 133, No. 3

34. Wang TJ, Gona P, Larson MG, et al. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med 2006;355:2631–2639. 35. Sturmer T, Schneeweiss S, Brookkhart MA, et al. Analytic strategies to adjust confounding using exposure propensity scores and disease risk scores: nonsteroidal antiinflammatory drugs and short-term mortality in the elderly. Am J Epidemiol 2005;161: 891–898. 36. Yeomans ND, Tulassay Z, Juhasz L, et al. A comparison of  omeprazole with ranitidine for ulcers associated with nonsteroidal antiinflammatory drugs. N Engl J Med 1998;338:719–726. 37. Höer A, Gothe H, Schiffhorst G, et al. Comparison of the effects of diclofenac or other non-steroidal anti-inflammatory drugs (NSAIDs) and diclofenac or other NSAIDs in combination with proton pump inhibitors (PPI) on hospitalisation due to peptic ulcer disease. Pharmacoepidemiol Drug Saf 2007;16:854–858. 38. García-Rodríguez LA, Tolosa LB. Risk of upper gastrointestinal complications among users of traditional NSAIDs and COXIBs in the general population. Gastroenterology 2007;132:498–506. 39. Lanus A, Garcia-Rodriguez LA, Arroyo MT, et al. Effect of antisecretory drugs and nitrates on the risk of ulcer bleeding associated with nonsteroidal anti-inflammatory drugs, antiplatelet agents, and anticoagulants. Am J Gastroenterol 2007;102:507–515. 40. Henry D, L-Y Lim L, Garcia Rodriguez LA, et al. Variability in risk of  gastrointestinal complications with individual non-steroidal antiinflammatory drugs: results of a collaborative meta-analysis. Br Med J 1996;312:1563–1566. 41. Chan FKL, Wong VWS, Suen BY, et al. Combination of a cyclooxygenase-2 inhibitor and a proton-pump inhibitor for prevention of recurrent ulcer bleeding in patients at very high risk: a doubleblind, randomised trial. Lancet 2007;369:1621–1626. 42. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven, 1998. 43. Brenner H, Arndt V, Gefeller O, et al. An alternative approach to age adjustment of cancer survival rates. Eur J Cancer 2004;40: 2317–2322.

Received April 16, 2007. Accepted June 11, 2007. Address requests for reprints to: Wayne A. Ray, PhD, Department of Preventive Medicine, 1500 21st Avenue South, Suite 2600, Nashville, Tennessee 37212. e-mail: [email protected]; fax: (615) 3430962. Supported by the Agency for Healthcare Research and Quality, Centers for Education and Research on Therapeutics cooperative agreement (grant HS1-0384). The sponsor had no role in the design or conduct of the study. No support from any pharmaceutical company was provided for this study. W.A.R. has received research support from Pfizer and is a consultant for plaintiffs’ attorneys in rofecoxib litigation. W.E.S. is a consultant for Novartis. M.R.G. is a consultant for Merck and has received research support from Pfizer. All other authors report no conflict of interest. The authors thank the Tennessee Bureau of TennCare and Department of Health, which provided study data.

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close