The study identified 2,185 study participants who were cognitively intact and had baseline medication information at the 2001 evaluation. The describes the flow of participants through each wave of the study. At the first follow-up evaluation (2004), 544 participants were lost to follow-up for various reasons including death and participant refusal. However, 11 participants from this group later rejoined the study and were included in the second follow-up evaluation in 2007. Therefore, this study includes 1,652 participants who had at least 1 follow-up evaluation, with 339 developing incident cognitive impairment during the 6-year study period (194 identified in the first follow-up evaluation and 145 in the second follow-up). The remaining 1,313 participants had preserved cognition.
Figure Description of study sample
The population demographics, education, and medication use are described in . This community-dwelling, African American population in Indianapolis had a mean age of 81.8 ± 5.3 years and included over two-thirds women (69.1%). At baseline, this population used a mean of 4.9 ± 3.3 medications daily, with 53.3% reporting the use of at least 1 medication with possible anticholinergic properties, and 10.8% using at least 1 medication with definite anticholinergic properties (). describes baseline demographic and comorbidity information of the study participants stratified by the exposure of definite anticholinergic medications.
Table 1 Baseline characteristics of the study participants (n = 1,652)
Table 2 Population characteristics by baseline exposure to definite anticholinergic medications
The majority of anticholinergic medications used by this population were classified as possible anticholinergics. The most commonly used possible anticholinergics in this population are frequently prescribed for cardiovascular disease. The most frequently used definite anticholinergic medications were oxybutynin and meclizine, although the general frequency was low.
describes the association between the use of anticholinergic medications and incident cognitive impairment. Model 1 examined the use of possible anticholinergic medications with the likelihood of incident cognitive impairment. After adjusting for age, gender, education, and baseline CSI-D score, the OR of developing incident cognitive impairment in those using medications with possible anticholinergic properties was 0.87 (95% CI 0.67–1.13) when compared with those who did not use any type of anticholinergic medications. There was no significant association between the number of possible anticholinergic medications used per participant and incident cognitive impairment (model 2).
Table 3 Use of anticholinergic medications at baseline and risk of incident cognitive impairment
Models 3 and 4 examine the association between the use of definite anticholinergic medications and incidence of cognitive impairment. When compared with those not using definite anticholinergics, the use of definite anticholinergics was associated with the development of cognitive impairment, resulting in an OR of 1.43 (95% CI 0.98–2.07). The number of definite anticholinergics used per study participant was evaluated as the exposure parameter in model 4, and resulted in an OR of 1.46 (95% CI 1.07–1.99). We evaluated a list of comorbidities as potential confounders and found that only a history of stroke had a significant influence on both the outcome and exposure. Adjusting for stroke, in addition to the baseline characteristics of age, sex, education, and baseline cognitive function score, revealed an OR for model 4 for the number of definite anticholinergic medications of 1.40 (95% CI 1.02–1.92; p = 0.0397). We also evaluated the cumulative use of anticholinergic medications, measured by adding the sum of ACB scores for any anticholinergic medication. The comparison of the total ACB score with the incidence of cognitive impairment revealed an OR of 1.04 (95% CI 0.96–1.12; p = 0.3188).
APOE ε4carrier status was available from 1,206 participants included in the study. Stratification by APOE ε4 carrier status () indicates a trend of definite anticholinergic exposure in increasing the risk of cognitive impairment in the group identified as carriers of the APOE ε4 allele, though statistical significance is not retained. Interestingly, for those participants without the APOE ε4 allele, the risk of cognitive impairment with the use of definite anticholinergic medications becomes stronger (OR 1.77, 95% CI 1.03–3.05; p = 0.04).
Logistic regression models were used for further subgroup analysis of the cognitive effect of anticholinergic medications on the separate outcomes of dementia. We compared the use of anticholinergics between those who met criteria for dementia (n = 57) with those who were cognitively normal (n = 1,313). The OR for developing incident dementia was 1.08 (95% CI 0.47–2.49; p = 0.8517) adjusting for age at baseline, years of education, gender, and baseline cognitive score.
We next evaluated the use of definite anticholinergic medications over time (baseline and 3- and 6-year evaluations). Due to missing medication data during at least 1 follow-up visit, 1,578 participants were available for this analysis. Of these, 72 participants were continuous users and an additional 251 participants were intermittent users during the 6-year study period. The OR for developing incident cognitive impairment in those exposed to continuous use of definite anticholinergics was 1.40 (95% CI 0.77–2.54; p = 0.2651) compared to nonusers after adjusting for age at baseline, gender, education, and baseline CSI-D score. The OR for developing incident cognitive impairment in participants who were intermittent users of definite anticholinergics was 1.63 (95% CI 1.17–2.28; p = 0.0042) when compared to nonusers.
To ensure our results were not biased by those lost to follow-up, we compared demographic and medical conditions between the 1,652 study participants included in the analysis and the 533 participants who were lost to follow-up after the baseline assessment. Those lost to follow-up were older, had less education, had worse cognitive scores, were more likely male, and more often had a history of depression, Parkinson disease, alcohol use, or smoking (p < 0.05). However, the lost to follow-up group did not differ significantly from those with follow-up evaluations in the proportions of participants using possible anticholinergics (56.8% vs 53.3%, p = 0.1563) or in the proportions of participants using definite anticholinergics (11.8% vs 10.8%, p = 0.5288). Thus it is possible that the poorer cognitive function observed at baseline in the lost to follow-up group was the result of comorbid conditions and not due to the greater burden of anticholinergic use.