Current issues of ACP Journal Club are published in Annals of Internal Medicine


NSAIDs but not aspirin or acetaminophen reduced the risk for Alzheimer disease

ACP J Club. 1997 Sep-Oct;127:46. doi:10.7326/ACPJC-1997-127-2-046

Related Content in this Issue
• Companion Abstract and Commentary: NSAIDs, smoking, and alcohol were risk factors for gastrointestinal perforation

Source Citation

Stewart WF, Kawas C, Corrada M, Metter EJ. Risk of Alzheimer's disease and duration of NSAID use. Neurology. 1997 Mar;48:626-32.



To determine whether nonsteroidal anti-inflammatory drug (NSAID) use was associated with the development of Alzheimer disease.


Cohort study with up to 16 years of follow-up (Baltimore Longitudinal Study on Aging [BLSA]).


Baltimore-Washington area of the United States.


The BLSA, which started in 1958, is a cohort study of normal aging including 1686 participants (58% men, 75% college educated, 45% < 55 y at enrollment, 93% white) who had ≥ 1 year of follow-up between 1980 and 1995. Participants were evaluated for 2.5 days every second year.

Assessment of risk factors

All medications were assessed at each visit (type, schedule, frequency, duration, brand name). Aspirin, other NSAIDs, and acetaminophen were reported separately and analyzed in categories of users and nonusers and use of 0 years, < 2 years, and ≥ 2 years. Aspirin use was relatively stable at 39% to 44%, whereas use of other NSAIDs tripled during the study. Nonaspirin NSAID categories were ibuprofen (50%), naproxen (15%), indomethacin (8%), fenoprofen (2%), flurbiprofen (1%), and others (23%).

Main outcome measures

Dementia and possible and probable Alzheimer disease based on the National Institute for Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association criteria. Diagnosis was assessed using data from neurologic examination, neurophysiologic testing, the Dementia Questionnaire, and personal medical records without knowledge of medication history.

Main results

81 persons developed Alzheimer disease during follow-up. A reduced risk for Alzheimer disease was associated with use of nonaspirin NSAIDs (relative risk [RR] 0.50, 95% CI 0.30 to 0.85) but not aspirin (RR 0.81, CI 0.52 to 1.28) or acetaminophen (RR 1.23, CI 0.73 to 2.07). Longer use of nonaspirin NSAIDs was associated with a reduced risk for Alzheimer disease (RR for ≥ 2 years use 0.40, CI 0.19 to 0.84). Longer use of aspirin and acetaminophen did not change the risk for Alzheimer disease. Analysis of latency (time from initiation of drug until its effects were able to be evaluated) did not substantially change any findings.


Use of nonaspirin nonsteroidal anti-inflammatory drugs, but not aspirin or acetaminophen, was associated with a reduced risk for the development of Alzheimer disease.

Source of funding: In part, National Institute on Aging.

For article reprint: Dr. W.F. Stewart, Department of Epidemiology, School of Hygiene and Public Health, 615 North Wolfe Street, Baltimore, MD 21205, USA. FAX 410-955-3906.


Aspirin has been used since 1982 to prevent heart disease and even before that to treat rheumatic disorders and other aches and pains. With new evidence that NSAIDs might also protect against the development of Alzheimer disease, the use of anti-inflammatory drugs could increase. In addition to benefits, aspirin and other NSAIDs have associated risks. These 2 studies from very different journals and areas of medicine once again open the debate on the need for balance between the risks and benefits of NSAID use.

The hypothesis that NSAIDs might protect against Alzheimer disease arose from observations that the neuritic plaques, a cardinal feature of the disease, contain proteins that are characteristic of an inflammatory process. Thus, anti-inflammatory drugs might influence the disease process. In support of this theory, observational studies have found reduced NSAID use in patients with Alzheimer disease.

By design, these observational studies were not able to exclude the obvious explanation for apparently reduced NSAID use (underreporting of drug use by patients with Alzheimer disease) or the less likely possibility of selection bias. Patients with Alzheimer disease who are still alive may have been healthier at disease onset (i.e., not NSAID users) than those patients who have already died of the disease.

Both these biases were avoided in the cohort study by Stewart and colleagues because data on drug use were collected prospectively over a 16-year period. The diagnosis of Alzheimer disease was based on a standardized neuropsychological assessment applied to all participants in the cohort. However, the authors did not collect details of the dose of aspirin and precise duration of use of NSAIDs. Nevertheless, they did categorize use as < 2 years and ≥ 2 years and showed a lower risk for Alzheimer disease with longer use of nonaspirin NSAIDs. In contrast, aspirin use showed only a weak inverse association with the development of Alzheimer disease. This association may be because of the lower dose used for heart disease prophylaxis in the 1980s. The anti-inflammatory dose of aspirin tends to be higher than that used for heart disease or stroke prophylaxis. Another possibility is that aspirin was taken on an over-the-counter basis and therefore used only occasionally, whereas other NSAIDs tend to be prescribed more often and are therefore used more regularly. The association with acetaminophen is consistent with its lack of anti-inflammatory effects.

Most of what is known about GI risks of NSAID use comes from studies of patients with bleeding peptic ulcers. The study by Lanas and colleagues confirms that use of aspirin and other NSAIDs increases the risk for both upper and lower GI perforation, with the increase in risk being similar to that for upper GI bleeding.

An unusual feature of the study by Lanas and colleagues was testing for aspirin use with thromboxane B2. They detected unreported aspirin use in 21% of patients compared with 12% of community controls and 5% of hospital controls. We do not know the amount taken and whether some of the excess use might have been to relieve pain from the GI perforation. Nevertheless, it is reassuring that the use of aspirin < 500 mg/d was not associated with perforation.

GI perforation is much less frequent than GI bleeding, and this study is not large enough to detect whether risks differ with particular nonaspirin NSAIDs. Overall, the level of aspirin use by patients (47%) was high compared with the 22% use by participants who were community controls. 60% of this use was nonprescribed. Thus, attempts to decrease ulcer complications must address both prescription and nonprescription use.

Lower GI perforation is even less frequent than upper perforation (only 16 patients in this study) and was predominantly perforated colonic diverticula. Nevertheless, an association with NSAID use was as strong for lower perforation as for upper perforation, which suggests systemic rather than local effects of NSAIDs. The findings add to previous evidence that shows associations with ileal and colonic ulceration, bleeding and perforated diverticular disease, and exacerbations of inflammatory bowel disease. In all these situations, physicians need to be aware of the possible contribution of aspirin and other NSAIDs.

Richard F. Logan, MD
University of Nottingham Medical SchoolNottingham, England, UK

Richard F. Logan, MD
University of Nottingham Medical School
Nottingham, England, UK