Get Newsletter
AlzRisk Paper Detail

Reference: Stewart, 1997
Cohort: Baltimore Longitudinal Study of Aging
Risk Factor: Non-Steroidal Anti-Inflammatory Drugs

Average Follow-up Time Detail
"A total of 1,686 BLSA participants were followed for 1 or more years between 1980 and 1995. Eighty-one individuals were diagnosed with AD during the 16-year follow-up period. Of the 1,686 participants, 45% enrolled before 1980. About 80% of the subjects were followed for 5 or more years, and 90% completed two or more visits between 1980 and 1995."

Exposure Detail
Interviewers ascertained exposure information at each biennial examination, where participants were asked to list all medication used since their last visit. Information on medicine type, schedule of use, frequency of use, and duration of time use were reported. The investigators reported results separately for ever vs. never any NSAID use, ever vs. never ASA NSAID use, duration of any NSAID use, and duration of ASA use, all with 0 or 2 years of lagging relative to the dementia assessments. This entry pertains to results on duration of ASA NSAID use at two years prior to dementia assessment (i.e., an analysis of time-updated NSAID use with 2 years’ lag between the medication report and dementia assessment; see Covariates & Analysis Detail).

The investigators compared incident AD risk in three groups: the group of participants who used ASA NSAIDs for less than 2 years ("<2 years"), the group of participants who used ASA NSAIDs for at least 2 years ("2+ years"), and the reference group of participants who did not use ASA NSAIDs at any time ("Never used"). The distribution of ASA NSAID use by duration was not reported.

Age Detail
At the start of enrollment 60% of participants were younger than 55 years old, 13% of participants were ages 55-64, 7% were ages 65-69, 8% were ages 70-74, 6% were ages 75-79, and 6% were at least 80 years old. A total of 52% of the person-time between 1980 and 1995 occurred when participants were younger than age 65.

Screening and Diagnosis Detail
Screening Method:
BlessedBlessed Information-Memory-Concentration Test
MMSEMini-Mental State Examination (Folstein 1975)
Pfeffer FAQPfeffer Functional Activities Questionnaire
TMTTrail Making Test (Reitan 1958)

AD Diagnosis:
DSM IIIR Diagnostic and Statistical Manual III-Revised
NINCDS ADRDA National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer's Disease and Related Disorders Association Criteria (McKhann 1984)

Covariates & Analysis Detail
Analysis Type:
Cox proportional hazards regression

"Use of these specific pain medications was defined as a time-dependent binary variable (non-user versus ever user during follow-up) and a time-dependent categorical duration-of-use variable (0 years, <2 years, 2-t years). The cut-points for duration of medication use were defined to ensure an adequate number of individuals in each group for each type of drug. Relative risk estimates for aspirin, NSAIDs, and acetaminophen were derived with and without adjustment for each other.
In the analysis, lagging,(31,32) method for assessing latency
in person-time analysis, was used to determine whether there was a minimum latency between the time of reported drug use and protection against AD. By using this method, follow-up time is lagged by some assumed latency period of “y” years (e.g., 2 years) relative to exposure time. Effectively, the first y-years of follow-up are ignored. The exposure level assigned to the last year of follow-up is that which accumulated y-years earlier. We examined the effect of 0 to 5 years of lagging. Since the results did not differ substantially when exposure was lagged from 2 to 5 years, we display findings for no lagging and 2 years of lagging."

AD Covariates:
FUTfollow up time

"Education was not retained in the regression model because it was neither a significant predictor of AD risk (the BLSA cohort has a high level of education compared with other cohorts) and has no effect on the coefficients for reported use of NSAIDS, aspirin, or acetaminophen."

Follow-up time was represented by the calendar year at follow-up (a time dependent variable in the cox model).