Compared with patients who were not prescribed an NSAID, the risk of mortality was twice as high among those prescribed diclofenac, 75% higher among those prescribed celecoxib, 70% higher among those prescribed rofecoxib, 31% higher among those prescribed ibuprofen, 22% higher among those prescribed naproxen, and 28% higher among those prescribed other NSAIDs. The number needed to treat to cause one additional death was 9 for rofecoxib, 11 for diclofenac, 14 for celecoxib, 51 for naproxen, and 53 for ibuprofen, indicating that naproxen and ibuprofen were safer than rofecoxib, celecoxib, and diclofenac. Risks were dose dependent, but even low doses of diclofenac, rofecoxib, and celecoxib were associated with an increased risk of mortality. Ibuprofen at a dose of 1200 mg/d or less and naproxen at a dose of 500 mg/d or less were not associated with an increased risk of death. The risk of death associated with NSAID use was comparable in subgroups that at baseline were predicted to have a high or low risk of death. A similar pattern of risk was found for hospitalizations for CHF and myocardial infarction.
This study is important because it extends the range of complications of NSAID use among patients with CHF to include not only more frequent exacerbations of HF, but also an increased risk of death. The large population-based sample, validated coding of discharge diagnoses, and near-complete data on medication use enhance the validity of the findings. The investigators did not have information on the clinical indications for which NSAIDs were prescribed, raising the question of whether the underlying condition was the true risk factor, rather than the medication itself, or whether the indications for treatment differed among users of different NSAIDs. Only 1.7% of patients had connective tissue disease included among their hospital discharge diagnoses, suggesting that the most common indication for the use of an NSAID was for mechanical musculoskeletal pain. The study did not include a comparison group that received prescriptions for analgesics, which would have helped isolate the associations with mortality and morbidity better to the medications, rather than the condition being treated. The source group of patients had all been hospitalized and had a discharge diagnosis of CHF. We do not know how severe their HF was, how heterogeneous these patients were with respect to their HF, or whether the NSAID-associated risks extend to those with milder HF.
Despite these questions, this study highlights the need to review medication choices for patients with CHF. Although low-dose naproxen or ibuprofen may be safer than other NSAIDs, this study cautions us to consider alternatives to NSAIDs for this group of patients.