Introduction

With the recent publication of the Rotterdam Study findings showing an association between nonsteroidal anti-inflammatory drugs (NSAIDs) medication and a lowered risk of Alzheimer disease, health professionals everywhere are likely to be besieged by their patients asking if they should start taking NSAIDs as a prophylactic measure. Here's a summary of the Netherland study results and the accompanying editorial¹ in the New England Journal of Medicine.

Inflammation occupies a central position in the pathological cascade of events in developing Alzheimer disease². Several case-control studies conducted before 1997 indicated that anti-inflammatory medication, in particular NSAIDs, are associated with delayed or reduced incidence of Alzheimer disease. The Baltimore Longitudinal Study of Aging found a relative risk of 0.5 in people taking NSAIDs regularly.

Since then, however, Mayo Clinic investigators reported they were unable to find such an association, and neither were the Rotterdam scientists in their first study. In trying to reconcile these disparate findings, the timing and duration of NSAID use was identified as being of likely importance. The present report from Rotterdam goes some way towards resolving this aspect.

In the Netherlands, virtually complete medication records are obtainable from computerized pharmacy records. This enabled enrollees in the Rotterdam Study to be analyzed for associations between dementia and the use of NSAIDs. All persons over 55 living in a Rotterdam suburb were invited to enroll. Of the 10,000-plus eligible subjects, almost 8,000 agreed to participate. Pharmacy records were available for 99.7% of these.

After exclusions for pre-existing dementia and inadequate follow up, there were 6,989 subjects available for analysis who had screenings for dementia at three intervals during the eight years January 1991 to December 1998.

Dementia was diagnosed after screening with the Mini-Mental State Examination and the Geriatric Mental State Schedule. The Cambridge Mental Disorders of the Elderly Examination was used, together with neurologist and neuro-psychologist consults and a brain MRI. A panel used the DSMMD (3rd edition revised) criteria for a final diagnosis of dementia, with subdiagnoses of Alzheimer disease or vascular dementia, based on appropriate NIH guidelines.

Use of NSAIDs and oral salicylates was categorized into: nonuse, short-term use (1 month or less of cumulative use), intermediate-term use (1-24 months of cumulative use), and long-term use (24 months or more of cumulative use). A Cox proportional-hazards model was used to calculate the age-specific incidence of Alzheimer disease and vascular dementia in relation to the use of these drugs.

Results

The 4 commonest NSAIDs prescribed were diclofenac (43%), ibuprofen (22%), naproxen (18%), and piroxicam (7%). NSAID use for any time-period, compared with nonuse, was associated with a relative risk (RR) of Alzheimer disease of 0.86 (95% confidence interval, 0.66-1.09). The categorized use of NSAIDs gave the following relative risks for Alzheimer disease and vascular dementia:

NSAID use	Entire cohort	No. of dementia	RR Alzheimer's	RR vasc. dementia
No exposure	2,553	210	1.00	1.00
1 month or less	2,001	88	0.95 (0.70-1.29)	1.25 (0.63-2.53)
1 month -- 23 months	2,202	93	0.83 (0.62-1.11)	1.36 (0.70-2.64)
24 months or more	233	3	0.20 (0.05-0.83)	0.99 (0.13-7.58)

There was a clear-cut, statistically significant, reduced risk of Alzheimer disease with NSAID use for 24 months or longer. This risk reduction could not be attributed to one particular NSAID.

No NSAID association was found for the risk of vascular dementia. On the other hand, while salicylate use had no association with Alzheimer disease risk, it was associated with a 3- to 5-fold increased risk of vascular dementia when taken for more than 1 month.

Further analyses showed that age over or less than 80, dosage of the NSAID, non-narcotic analgesic use, steroid use, pre-existing rheumatoid arthritis, apolipoprotein E genotype, or estrogen use by women, had no influence on the associations described above.

Implications

Why should we pay more attention to the results of this study than those from earlier, less conclusive studies? First, the cohort patients enrolled were found to be free of dementia at the study baseline. Second, medication use was clearly defined in this population. Third, follow-up information was almost complete. Finally, the duration-response effect shown goes some way towards explaining discrepancies between the results of earlier studies.

The findings suggest there may be a critical period during which taking NSAIDs may protect against Alzheimer disease. This period ends about two years before the point where Alzheimer's can be diagnosed with some certainty. The lack of dose-dependency is encouraging if there is consideration of long-term prophylactic use of an NSAID, in view of the known gastrointestinal side effects.

Most physicians are wary about advising the use of a potentially toxic medication over a long period in the absence of a clear-cut medical indication. Refreshingly, the Johns Hopkins editorial writers do not make any specific cautionary remarks; they merely point out that prospective randomized controlled trials to prove the value of NSAIDs in this indication will take several years. Until such results are available, some physicians may feel justified in recommending otherwise-healthy elderly patients to take an NSAID in low doses, prophylactically, while watching carefully for potential adverse effects.

Note: A large well-controlled study published in JAMA 2003;289:2819-26 has shown that treating mild-to-moderate Alzheimer's patients with naproxen or rofecoxib (a COX-2 inhibitor) was not effective in slowing progression of the disease, once it's established. Side effects were more common in the people taking the anti-inflammatory drugs than in the placebo controls. Robert Griffith.