Correction to: Impact of risk minimisation measures on the use of strontium ranelate in Europe: a multi-national cohort study in 5 EU countries by the EU-ADR Alliance.

The original version of this article, published on 26 November 2019 contained a mistake.

Impact of risk minimisation measures on the use of strontium ranelate in Europe: a multi-national cohort study in 5 EU countries by the EU-ADR Alliance.

INTRODUCTION: In May 2013 and March 2014, the European Medicines Agency (EMA) issued two decisions restricting the use of strontium ranelate (SR). These risk minimisation measures (RMM) introduced new contraindications and limited the indications of SR therapy. The EMA required an assessment of the impact of RMMs on the use of SR in Europe. Methods design: multi-national, multi-database cohort Setting: electronic medical record databases based on hospital (Denmark) and primary care provenance (Italy, Spain, the Netherlands, UK). PARTICIPANTS: the database source populations were included for population-based analyses, and SR users for patient-level analyses. INTERVENTION: New RMMs included contraindications (ischaemic heart disease, peripheral arterial disease, cerebrovascular disease, uncontrolled hypertension) and restricted SR indication to severe osteoporosis with initiation by experienced physician and not as first line anti-osteoporosis therapy. METHODS: Prevalence and incidence rates of SR use in the population; prevalence of contraindications and restricted indications in SR users, plus 1-year therapy persistence. Drug use measures were calculated in three periods for comparison: reference (2004 to May 2013), transition (June 2013 to March 2014) and assessment (from April 2014 to end 2016). RESULTS: The study population included 143 million person-years(PY) of follow-up and 76,141 incident episodes of SR treatment. Average monthly prevalence rates of SR use dropped by 86.4% from 62.6/10,000 PY (95 CI 62.4-62.9) in the reference to 8.5 (8.5-8.6) in the assessment period. Similarly, the incidence rate of SR use fell by 97.3% from 7.4/10,000 PY (7.4-7.4) to 0.2 (0.2-0.2) between the reference and assessment period. The prevalence of any contraindication decreased, whilst the prevalence of restricted indications increased in these periods. One-year persistence decreased in the assessment compared with reference period. CONCLUSIONS: Our study demonstrates a substantial impact of the regulatory action to restrict use of SR in Europe: SR utilisation overall decreased strongly. The proportion of patients fulfilling the restricted indications, without contraindications, increased after the proposed RMMs.

Hormone replacement therapy and incidence of acute myocardial infarction. A population-based nested case-control study.

BACKGROUND: Epidemiological studies suggest a decreased risk of coronary heart disease (CHD) in healthy women taking hormonal replacement therapy (HRT). Whether this effect is shared by oral and transdermal preparations is unknown. METHODS AND RESULTS: We conducted a population-based case-control study nested in a cohort of women 50 to 74 years of age without cardiovascular disease history in the United Kingdom. Among 164 769 women from the General Practice Research Database (January 1, 1991, to December 31, 1995), we identified 1242 first acute myocardial infarctions (AMI) and confirmed 1013 after medical record review. We randomly selected 5000 age-frequency-matched control subjects. AMI incidence was 1.6 per 1000 person-years; 13% and 17% of cases and control subjects used HRT within 6 months before the index date. Risk factor and comorbidity-adjusted OR of AMI for current-recent HRT users compared with nonusers was 0.72 (95% CI 0.59 to 0.89). The OR was similar within 30 days before the index date. The beneficial effect was present after 1 year of use (OR 0. 68; 95% CI 0.53 to 0.86), with no increase in risk within the first year. ORs for unopposed and opposed therapy were 0.52 (95% CI 0.35 to 0.78) and 0.79 (95% CI 0. 59 to 1.08); 79% and 21% used oral and transdermal therapy. The protective effect was present at medium-high doses of estrogens with ORs for oral and transdermal therapy of 0.63 (95% CI 0.46 to 0.86) and 0.62 (95% CI 0.37 to 1.06) and ceased after 2 to 3 years since stopping HRT. CONCLUSIONS: Results are consistent with those previously reported in women without CHD who were taking oral HRT and, although based on few users, suggest that transdermal therapy might have similar cardioprotective effects.

A cohort study on the risk of acute liver injury among users of ketoconazole and other antifungal drugs.

AIMS: The aim of this cohort study was to estimate the risk of clinical acute liver injury among users of oral antifungals identified in the general population of the General Practice Research Database in UK. METHODS: The cohort included 69 830 patients, 20-79 years old, free of liver and systemic disease, who had received at least one prescription for either oral fluconazole, griseofulvin, itraconazole, ketoconazole, or terbinafine between 1991 and 1996. RESULTS: Sixteen cases of acute liver injury were identified and validated. Ten cases occurred during nonuse of oral antifungals with a background rate of 0.6 per 100,000 person-months (95% confidence interval 0.3,1.1). Five cases occurred during current use of oral antifungals. Two were using ketoconazole, another two itraconazole, and one terbinafine. Incidence rates of acute liver injury were 134.1 per 100 000 person-months (36.8,488.0) for ketoconazole, 10.4 (2.9-38.1) for itraconazole, and 2.5 (0.4,13. 9) for terbinafine. The remaining case was associated with past use of fluconazole. Ketoconazole was the antifungal associated with the highest relative risk, 228.0 (95% confidence interval 33.9,933.0), when compared with the risk among nonusers, followed by itraconazole and terbinafine with relative risks of 17.7 (2.6,72.6) and 4.2 (0.2, 24.9), respectively. CONCLUSIONS: Ketoconazole and itraconazole were the two oral antifungal associated with a marked increase of clinical acute liver injury. The risk associated with ketoconazole should be taken into account when prescribing it as initial treatment for uncomplicated fungal infections.

Characteristics of users of inhaled long-acting beta 2-agonists in a southern European population.

We characterized the population of users of inhaled long-acting beta 2-agonists in the region of Friuli-Venezia Giulia, in Italy, and assessed changes in asthma treatment and control after initiating long-acting beta 2-agonists. All residents using formoterol or salmeterol between 1992 and 1996 were identified in the regional Health Databases. Utilization rates of asthma medications and hospitalization rates for asthma were computed for the year before and after the date of the first long-acting beta 2-agonist prescription. There were 3803 users of formoterol and 20,054 users of salmeterol. Overall, 65% of users were older than 54 years of age. All formoterol users and 86% of salmeterol users received their first prescription for the respective drug during the study period (new users). Among these new users, 50% had not received any asthma drug during the 4 months preceding the start of long-acting beta 2-agonist administration. Prior 1 yr utilization rates of asthma medications and hospitalization rates for asthma were greater among new users of long-acting beta 2-agonists than among new users of salbutamol and xanthines. In addition, formoterol new users had higher prior use of asthma drugs than new users of salmeterol. One year prior hospitalization rates for asthma were also higher among formoterol than salmeterol new users with rate ratios of 1.7 (95% CI 1.3-2.2) for patients younger than 45 and 1.5 (1.2-1.9) for older patients. Use of short-acting beta 2-agonists, oral steroids and xanthines significantly declined after starting formoterol, whereas the use of inhaled steroids increased after the start of either formoterol or salmeterol. Asthma hospitalizations decreased by 32% in patients under age 45, by 43% in older patients, during the year following the start of formoterol, and by 15% and 24%, respectively, after the start of salmeterol. We conclude that long-acting beta 2-agonists were mainly prescribed to middle-aged and elderly patients and that formoterol appeared to be preferentially prescribed to patients with more severe asthma than salmeterol. Changes in asthma treatment and reduction in hospitalization rates for asthma after starting formoterol and salmeterol are compatible with an improvement in the control of asthma.

Low-dose diclofenac, naproxen, and ibuprofen cohort study.

The risk of a newly diagnosed episode of upper gastrointestinal bleeding, acute liver and renal failure, agranulocytosis, aplastic anemia, severe skin disorders, and anaphylaxis was examined within 30 days after the first prescription for a low dose of diclofenac, naproxen, or ibuprofen in a cohort in the United Kingdom. We identified 22,146 persons using diclofenac (< or = 75 mg), 46,919 using naproxen (< or = 750 mg), and 54,830 using ibuprofen (< or = 1200 mg). Age, gender, and comorbidity were similar in the three cohorts. Overall 64 potential cases were identified, and 20 were confirmed by medical record review. Incidence rates (95% CI) of upper gastrointestinal bleeding/10,000 people using diclofenac, naproxen, and ibuprofen were 1.8 (0.5-4.6), 2.3 (1.2-4.2), and 0.4 (0.04-1.3), respectively. There were three cases of hepatic injury, one with naproxen and two with ibuprofen. Although low, the incidence of gastrointestinal toxicity remains the main serious adverse event for all study drugs.

Antiepileptics and blood dyscrasias: a cohort study.

We conducted a cohort study to investigate the frequency of serious blood dyscrasias in patients age 10-74 years, taking antiepileptic drugs between January 1, 1990, and October 31, 1994. Main outcome measures were validated diagnoses of neutropenia, agranulocytosis, hemolytic anemia, thrombocytopenia, bicytopenia, pancytopenia, or aplastic anemia. A total of 29,357 recipients of antiepileptic therapy received 684,706 prescriptions. Among them there were 21 cases of serious blood dyscrasia of which only 18 could be considered to have a temporal relationship to drug use. Seventeen cases occurred in current users of carbamazepine, phenobarbital, phenytoin or valproate, and 7 in patients taking two or more drugs. Twenty of the 21 patients recovered. The overall rate of blood dyscrasias was 3-4/100,000 prescriptions. The rate in those age less than 60 years was 2.0 (range 0.9-3.6)/100,000 prescriptions compared with 4.0 (range 1.6-8.2) for those age 60 or older. The overall rate of neutropenia was 1.2 (0.5-2.3)/100,000 prescriptions, compared with 0.9 (0.3-1.9) for thrombocytopenia and 0.4 (0.1-1.3) for hemolytic anemia. Rates did not differ among the four drugs. Serious blood dyscrasias are rare in patients taking antiepileptic agents.

Use of the UK General Practice Research Database for pharmacoepidemiology.

The last decade has seen a surge in the use of computerized health care data for pharmacoepidemiology. Of all European databases, the General Practice Research Database (GPRD) in the UK, has been the most widely used for pharmacoepidemiological research. Since 1994, this database has belonged to the UK Department of Health, and is maintained by the Office of National Statistics (ONS). Currently, around 1500 general practitioners with a population coverage in excess of 3 million, systematically provide their computerized medical data anonymously to ONS. Validation studies of the GPRD have documented the recording of medical data into general practitioners' computers to be near to complete. The GPRD collects truly population-based data, has a size that makes it possible to follow-up large cohorts of users of specific drugs, and includes both outpatient and inpatient clinical information. The access to original medical records is excellent. Desirable improvements to the GPRD would be additional computerized information on certain variables and linkage to other health care databases. Most published studies to date have been in the area of drug safety. The General Practice Research Database has proved that valuable data can be collected in a general practice setting. The full potential of this rich computerized database has yet to come. This experience should serve to encourage others to develop similar population-based data in other countries.

Hormone replacement therapy and the risk of hospitalization for venous thromboembolism: a population-based study in southern Europe.

The authors evaluated the risk of venous thromboembolism associated with hormone replacement therapy in a cohort of 265,431 women aged 45-79 years who did not have major risk factors for venous thromboembolism. Through review of hospital charts, 171 cases were confirmed (pulmonary embolism = 77; deep venous thrombosis = 94). Ten thousand controls were randomly sampled. The risk of venous thromboembolism among nonusers of hormone replacement therapy was 1.3 per 10,000 women per year. Current users of hormone replacement therapy had 2.3 times higher risk of venous thromboembolism (95 percent confidence interval 1.0-5.3) compared with nonusers. The increased risk was restricted to the first year of treatment.

Recent epidemiological studies of the association between hormone replacement therapy and venous thromboembolism. A review.

The association between use of hormone replacement therapy (HRT) and the risk of venous thromboembolism (VTE) has been assessed in relatively few epidemiological studies. Evidence from the earliest studies did not support an increased risk of VTE among HRT users. However, methodological limitations in most studies, including small sample size and inadequate control of confounding, did not allow firm conclusions to be made. Most of these limitations have been overcome in 5 recent studies which consistently show that the risk of VTE among women currently using HRT is 2 to 3 times higher than among women not using HRT. The overall relative risk of VTE for women currently using HRT obtained from these studies was 2.6 (95% confidence interval 1.6 to 4.2). This association is unlikely to be explained by confounding or other potential biases affecting observational studies. The risk appears to be more prominent during the first year of HRT use, and in 2 studies the risk disappeared after the first year of therapy. A dose-response relationship, with a doubling of risk among users of high doses of estrogens, was shown in 2 of these studies. No major differences were observed with the different types of therapy, but users of unopposed estrogen therapy and transdermal therapy might be at lower risk than users of opposed regimens and oral preparations. Evidence from these new studies indicates that, among healthy post-menopausal women, between 1 and 2 additional cases of VTE per 10,000 women can be annually attributed to current use of HRT. The Committee on Safety of Medicines in the UK evaluated this risk as small and considered that it does not change the overall benefit-risk profile of HRT for most women.

Hormone replacement therapy and risk of venous thromboembolism: population based case-control study.

OBJECTIVE: To evaluate the association between use of hormone replacement therapy and the risk of idiopathic venous thromboembolism. DESIGN: Population based case-control study. SETTING: Population enrolled in the General Practice Research Database, United Kingdom. SUBJECTS: A cohort of 347,253 women aged 50 to 79 without major risk factors for venous thromboembolism was identified. Cases were 292 women admitted to hospital for a first episode of pulmonary embolism or deep venous thrombosis; 10,000 controls were randomly selected from the source cohort. MAIN OUTCOME MEASURES: Adjusted relative risks estimated from unconditional logistic regression. RESULTS: The adjusted odds ratio of venous thromboembolism for current use of hormone replacement therapy compared with non-users was 2.1 (95% confidence interval 1.4 to 3.2). This increased risk was restricted to first year users, with odds ratios of 4.6 (2.5 to 8.4) during the first six months and 3.0 (1.4 to 6.5) 6-12 months after starting treatment. No major risk differences were observed between users of low and high doses of oestrogens, unopposed and opposed treatment, and oral and transdermal preparations. The risk of idiopathic venous thromboembolism among non-users of replacement therapy was estimated to be 1.3 per 10,000 women per year. Among current users, idiopathic venous thromboembolism occurs at two to three times the rate in non-users, resulting in one to two additional cases per 10,000 women per year. CONCLUSIONS: Current use of hormone replacement therapy was associated with a higher risk of venous thromboembolism, although the risk seemed to be restricted to the first year of use.

Nonsteroidal anti-inflammatory drugs and the risk of hospitalization for acute renal failure.

BACKGROUND: Few studies have evaluated the association between use of nonsteroidal anti-inflammatory drugs (NSAIDs) and the risk of idiopathic acute renal failure (ARF) in the general population. METHODS: Population-based case-control study among persons in the Canadian province of Saskatchewan who received at least 1 NSAID prescription between January 1, 1982, and December 31, 1986. Health department databases were used for case detection, as the sampling frame for selecting controls (n = 1997), and as the primary source of information on drug use and comorbidity. A total of 306 hospital records were reviewed. Twenty-eight patients who were hospitalized fulfilled the diagnostic criteria for ARF. RESULTS: The incidence rate of hospitalization for ARF among the general population not exposed to NSAIDs was 2 per 100000 person-years. Current exposure to NSAIDs, acetylsalicylic acid and other nephrotoxic drugs, male gender, increasing age, cardiovascular comorbidity, and recent hospitalization for disorders other than renal were found to be independent risk factors for ARF. Current NSAID users had an adjusted odds ratio for ARF of 4.1 (95% confidence interval, 1.5-10.8). The risk of ARF was especially high during the first month of use (odds ratio, 8.5). For prescribed dose, we found that users of high daily doses of NSAIDs experienced an odds ratio of 9.8 for ARF. CONCLUSIONS: In the general population, hospitalizations for ARF were found to be a rare condition. The 4-fold increase in risk associated with NSAID use was dose-dependent and occurred especially during the first month of therapy. Concurrent comedication with other potentially nephrotoxic agents should be prescribed with care, especially in the elderly.

Variability in risk of gastrointestinal complications with individual non-steroidal anti-inflammatory drugs: results of a collaborative meta-analysis.

OBJECTIVE: To compare the relative risks of serious gastrointestinal complications reported with individual non-steroidal anti-inflammatory drugs. DESIGN: Systematic review of controlled epidemiological studies that found a relation between use of the drugs and admission to hospital for haemorrhage or perforation. SETTING: Hospital and community based case-control and cohort studies. MAIN OUTCOME MEASURES: (a) Estimated relative risks of gastrointestinal complications with use of individual drugs, exposure to ibuprofen being used as reference; (b) a ranking that best summarised the sequence of relative risks observed in the studies. RESULTS: 12 studies met the inclusion criteria. 11 provided comparative data on ibuprofen and other drugs. Ibuprofen ranked lowest or equal lowest for risk in 10 of the 11 studies. Pooled relative risks calculated with exposure to ibuprofen used as reference were all significantly greater than 1.0 (interval of point estimates 1.6 to 9.2). Overall, ibuprofen was associated with the lowest relative risk, followed by diclofenac. Azapropazone, tolmetin, ketoprofen, and piroxicam ranked highest for risk and indomethacin, naproxen, sulindac, and aspirin occupied intermediate positions. Higher doses of ibuprofen were associated with relative risks similar to those with naproxen and indomethacin. CONCLUSIONS: The low risk of serious gastrointestinal complications with ibuprofen seems to be attributable mainly to the low doses of the drug used in clinical practice. In higher doses ibuprofen is associated with a similar risk to other non-steroidal anti-inflammatory drugs. Use of low risk drugs in low dosage as first line treatment would substantially reduce the morbidity and mortality due to serious gastrointestinal toxicity from these drugs.

Positive predictive value of ICD-9 codes in the identification of cases of complicated peptic ulcer disease in the Saskatchewan hospital automated database.

We examined the positive predictive value of the International Classification of Diseases, 9th revision (ICD-9), codes used to identify cases of complicated peptic ulcer disease from the Saskatchewan Hospital automated database. Nonspecific and site- and lesion-specific codes were used in case detection to increase the sensitivity of the initial computer search. Review of the hospital records of 1,762 potential cases resulted in 1,375 confirmed cases. The positive predictive value of site and lesion-specific codes was about 90% and was about 70% for the nonspecific codes. Almost 50% of cases, however, would have been missed if the nonspecific codes had not been used.

The increased risk of hospitalizations for acute liver injury in a population with exposure to multiple drugs.

We conducted a nested case-control study to estimate and compare the relative risks for hospitalizations for newly diagnosed acute liver injury associated with the use of non-steroidal antiinflammatory drugs (NSAIDs) and other hepatotoxic drugs and their interaction. The source population comprised 228,392 members of the Saskatchewan Health Plan from 1982 to 1986. We used hospital records and the databases of the Department of Health. Thirty-four cases with confirmed liver injury were hospitalized. We randomly selected 500 controls from the source population. Crude risks ranged from 1 case per 100,000 prescriptions in current users of methyldopa, ampicillin, or NSAIDs to 14 cases per 100,000 prescriptions in current users of erythromycin estolate. The age-adjusted odds ratios for current users of NSAIDs was 1.8 [95% confidence interval (CI) = 0.8-3.7] and for other hepatotoxic drugs 5.9 (95% CI = 2.8-12.4). The adjusted relative excess risk due to the interaction between current exposure to both categories of drugs was 3.6, accounting for 31% of the cases of acute liver injury among those with exposure to both types of drugs. We conclude that the risk of hospitalization for acute noninfectious liver injury is different among users of various individual potentially hepatotoxic drugs. Concomitant current exposure to two or more drugs increases this risk above what would merely be expected from the sum of the individual risks.

Risk factors for coronary artery disease in patients with systemic lupus erythematosus.

PURPOSE: To estimate the frequency of and examine risk factors for coronary artery disease (CAD) in patients with systemic lupus erythematosus (SLE) in a prospective longitudinal study. PATIENTS AND METHODS: Patients were SLE are enrolled in The Johns Hopkins Lupus Cohort, a prospective study of outcomes in 229 subjects with SLE. CAD was defined as angina, myocardial infarction, or sudden death. Data on CAD risk factors were obtained prospectively every 3 months and were analyzed using univariate and multiple logistic regression. RESULTS: CAD occurred in 19 (8.3%) of 229 patients with SLE and accounted for 3 (30%) of 10 deaths as of December 31, 1990. Compared to subjects without CAD, those with CAD were more likely to have been older at both diagnosis of SLE (37.1 years versus 28.9 years, p = 0.004) and at entry into the cohort (47.1 years versus 34.7 years, p < 0.0001), to have a longer mean duration of SLE (12.3 years versus 8.1 years, p = 0.013) and a longer mean duration of prednisone use (14.3 years versus 7.2 years, p < 0.0001), to have a higher mean serum cholesterol (271.2 mg/dL versus 214.9 mg/dL, p < 0.0001) or a cholesterol level greater than 200 mg/dL (odds ratio [OR] 14.5, 95% confidence intervals [CI] 1.9, 112.1), and to have both a history of hypertension (OR 3.5, 95% CI 1.3, 9.6) and a history of use of antihypertensive medications (OR 5.5, 95% CI 1.8, 17.2). There were no significant associations with other known CAD risk factors such as smoking, diabetes, family history of CAD, race, or sex, or variables related to steroid therapy including the presence of cushingoid features or ever use of corticosteroids. The best multiple logistic regression model for CAD included age at diagnosis, duration of prednisone use, requirement for antihypertensive treatment, maximum cholesterol level, and obesity (using NHANES-II [National Health and Nutrition Examination Survey] definitions). CONCLUSION: Primary and secondary prevention strategies directed at hypertension, hypercholesterolemia, and obesity, as well as other known CAD risk factors, should be routinely employed in the management of patients with SLE.

The role of non-steroidal anti-inflammatory drugs in acute liver injury.

OBJECTIVE: To investigate the association between use of non-steroidal anti-inflammatory drugs and serious, acute non-infectious liver injury. DESIGN: Retrospective cohort study, cross over design. SETTING: Health records from provincial database in Saskatchewan, Canada, 1982-6. SUBJECTS: 228,392 adults who contributed 645,456 person years. All were either using or had used non-steroidal anti-inflammatory drugs. MAIN OUTCOME MEASURES: Number and type of prescriptions for non-steroidal anti-inflammatory drugs. Admission to hospital for newly diagnosed acute liver injury. RESULTS: There were 34 admissions to hospital; 16 among subjects currently using non-steroidal anti-inflammatory drugs and 18 among subjects who were not. The incidence rate among current users was 9 per 100,000 person years (95% confidence interval 6 to 15 per 100,000 person years). Subjects currently using non-steroidal anti-inflammatory drugs had twice the risk of newly diagnosed liver injury as those not currently taking these drugs (rate ratio 2.3; 95% confidence interval 1.1 to 4.9) and an excess risk of 5 per 100,000 person years. The age and sex adjusted risk ratio was 1.7 (0.8 to 3.7). The strength of the association increased when only cases with no concomitant use of other hepatotoxic drugs were considered (4.0; 0.9 to 19.0). The rate ratio for people having received one to nine prescriptions was constant. There was no increased risk with long duration of treatment (1.0; 0.3 to 3.5). CONCLUSIONS: There is a small excess risk of serious, acute non-infectious liver injury associated with the use of non-steroidal anti-inflammatory drugs.

Nonsteroidal antiinflammatory drugs and gastrointestinal hospitalizations in Saskatchewan: a cohort study.

We evaluated the association between individual nonsteroidal antiinflammatory drugs (NSAIDs) and gastrointestinal (GI) toxicity in a retrospective cohort study aimed at examining and comparing the incidence of serious gastrointestinal disorders among NSAIDs users. We observed 2,302 GI hospitalizations among diclofenac, indomethacin, naproxen, piroxicam, sulindac, and other NSAIDs users in the province of Saskatchewan, Canada, from 1982 to 1986 for 228,392 persons who contributed 679,075 person-years of follow-up and filled close to 1.5 million NSAID prescriptions. Current NSAID users presented an increased risk of GI hospitalization [rate ratio (RR) = 3.9, 95% confidence interval = 3.5-4.4]. RRs decreased as time since the last prescription increased: 2.2 (1.9-2.6) for recent past users and 1.3 (1.1-1.5) for less recent past users. Among current users, RRs were the highest in indomethacin users (5.1, 4.3-6.0), and the lowest in sulindac users (3.1, 2.3-4.2). All of these results are adjusted for calendar time, sex, and age. Age showed a particularly strong association with the risk of GI hospitalization.

The impact of research quality and study design on epidemiologic estimates of the effect of nonsteroidal anti-inflammatory drugs on upper gastrointestinal tract disease.

BACKGROUND: Considerable differences in the estimates of the risk of upper gastrointestinal tract disease associated with treatment with nonsteroidal anti-inflammatory drugs (NSAIDs) have been observed. We conducted a meta-analysis of epidemiologic studies of upper gastrointestinal tract disease related to NSAIDs to answer the following research questions: Are study characteristics (study design and quality) associated with different estimates of risk, for both aspirin and nonaspirin NSAIDs? Does the risk increase for particular groups of patients, such as women and the elderly? METHODS: Thirty-four studies addressing severe upper gastrointestinal tract disease associated with NSAIDs (including aspirin and nonaspirin NSAIDs) were examined and scored according to a quality checklist that we designed for this review. RESULTS: Only 44% of the studies controlled for major confounding variables (age and sex). While exposure was defined as current in 40% and 78% of the studies for nonaspirin NSAIDs and aspirin, respectively, few investigations checked for history of ulcer disease and concurrent diseases or other comedications known to be risk factors for upper gastrointestinal tract bleeding. The overall risk ratio, by means of a random-effects regression model, was 3.0 (95% confidence interval, 1.9 to 4.7). The individual estimates for aspirin and nonaspirin NSAIDs were similar. Both with and without control for quality, cohort studies provided a lower risk ratio estimate than did case-control studies. CONCLUSIONS: The design and quality of the studies appear to be strong independent predictors of the risk estimate; cohort studies were associated with lower risk estimates than case-control studies, and satisfactory studies were associated with lower risk estimates than unsatisfactory studies.