Statistics from Altmetric.com
Epidemiological studies of patients taking prescribed non-steroidal anti-inflammatory drugs (NSAIDs) show that these drugs are associated with a three to 10-fold increased risk of bleeding, perforation, hospitalisation and death.1 Primary studies and meta-analyses also show that NSAIDs differ in their risk. Ibuprofen is consistently associated with lower risks than the group as a whole whereas other NSAIDs such as piroxicam and azapropazone seem to be associated with relatively high risks.
In recent years a number of NSAIDs have become available over the counter. This may partly reflect patent expiry, anticipation of a new generation of NSAIDs (COX-2 selective or NO-NSAIDs) and recognition that the absolute risks of short term, low dose treatment are relatively low.
The problem has been to assess the safety of drugs in this setting. In their study Strom et al have attempted to quantify the risks of over the counter naproxen sodium by reference to a prescribing situation that tried to replicate intended patterns of over the counter use. Information was gathered from the Computerized On-Line Medical Pharmaceutical Analysis and Surveillance System (COMPASS) database for Michigan and Ohio, USA. Patients were eligible for study if they had received benefits from Ohio or Michigan Medicaid and had received naproxen, sodium or ibuprofen but not both drugs. Age, sex, race, US state, in patient and outpatient diagnoses, drugs dispensed from outpatients, and any over the counter drugs that were paid for by Medicaid were extracted from COMPASS. In order to mimic potential over the counter use only adverse events occurring within 14 days after the first prescription were counted.
Of a cohort of 378 919 patients, 59 were admitted to hospital within 14 days of the first prescription of naproxen sodium (n= 26) or ibuprofen (n=33). This was more likely to occur after naproxen sodium, with an adjusted relative risk of 2.0 (95% confidence interval 1.1 to 3.8) compared with ibuprofen. The authors also report that the increase in the adjusted absolute risk for naproxen sodium compared with ibuprofen was 0.0011% (−0.0001% to 0.0023%). Because this aspect of the results fell just short of statistical significance, they drew the curious conclusion that there was an increase in relative but not absolute risk. It seems more likely that there was a true increase in risk with naproxen sodium at borderline levels of statistical significance which achieve conventional levels when analysed as relative risk and fell just short of this when analysed as absolute risk.
This relation showing a higher level of risk with naproxen than ibuprofen is similar to that seen studies of all comers and it reinforces findings that NSAIDs differ intheir gastrointestinal toxicity. It may well be that this is largely attributable to differences in potency.
Naproxen sodium did not lead to substantially more upper gastrointestinal tract bleeding than ibuprofen during short term use as an analgesic
Do patients who use naproxen sodium as an over the counter analgesic have an increased risk of upper gastrointestinal tract bleeding compared with those who use ibuprofen?
A case cohort study in which exposure to naproxen sodium and ibuprofen was compared in patients from a claims database who had upper gastrointestinal tract bleeding and in a subcohort of patients from the database.
Michigan and Ohio, USA.
Patients were eligible if they had received benefits from Ohio Medicaid or Michigan Medicaid and had received naproxen sodium or ibuprofen, but not both drugs. From a cohort of 378 919 patients in the Computerized On-Line Medical Pharmaceutical Analysis and Surveillance System (COMPASS), 59 patients (54% men) who were admitted to hospital for upper gastrointestinal tract bleeding within 14 days of their first prescription for naproxen sodium (n=26) or ibuprofen (n=33) were selected as case patients; a random sample of 37 891 patients (71% women) was selected as a subcohort to determine relative prescription rates (n=10 024 for naproxen sodium; n=27 867 for ibuprofen).
Assessment of risk factors
Age, sex, race, US state, inpatient and outpatient diagnoses, drugs dispensed for outpatients, and any over the counter drugs that were paid for by Medicaid were extracted from COMPASS.
Main outcome measures
Inpatient diagnoses recorded in COMPASS were used to identify patients who had upper gastrointestinal tract bleeding within 14 days of their first prescription of ibuprofen or naproxen sodium. The relative risk of upper gastrointestinal tract bleeding that required hospital admission was calculated. A modified version of the Cox proportional hazards was used to adjust for the fact that the case group and the subcohort were not independent.
The incidence of inpatient upper gastrointestinal tract bleeding was 0.026% (95% CI 0.017% to 0.038%) for patients who were prescribed naproxen sodium and 0.012% (CI 0.008% to 0.017%) for patients who were prescribed ibuprofen. Patients with uncertain Medicaid eligibility, chronic conditions, long term use of non-steroidal anti-inflammatory agents (NSAIDs), or concomitant use of other NSAIDs were excluded from the analysis (n=10 for the case patients; n=7989 for the subcohort) and the results were adjusted for the state of residence, age group, sex, and presence of pre-existing gastrointestinal tract bleeding. Compared with patients who received ibuprofen, those who received naproxen sodium had an increase in the adjusted relative risk of upper gastrointestinal tract bleeding that required hospital admission (adjusted relative risk 2.0, 95% CI 1.1 to 3.8), but no increase in the adjusted absolute risk occurred (adjusted absolute risk difference 0.0011%, CI −0.0001% to 0.0023%).
The incidence of upper gastrointestinal tract bleeding was low for naproxen sodium and ibuprofen. Although patients who used naproxen sodium as an analgesic had an increase in the relative risk of upper gastrointestinal tract bleeding that required hospital admission, no increase occurred in the absolute risk.
Source of funding: Syntex Corporation.
For article reprint: Dr B L Strom, Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, 824 Blockley Hall, Philadelphia, PA 19104–6021, USA.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.