Modest Evidence for an Association but Not for Causality
September 2, 2019
Journal of Clinical Psychiatry 2019 August 6, 80 (4)
Many observational studies published during the past 15 years have found an association between anticholinergic drug exposure and the risk of incident dementia. Animal data suggest plausible causal mechanisms for this finding. The results of a recent, large, and well-conducted study on the subject were widely disseminated by the lay and scientific media. This study addressed protopathic bias by examining anticholinergic drug exposure in time windows 1-11, 3-13, and 5-20 years before the identification of dementia. In brief, the study found that, pooling anticholinergic drug exposure across 11 drug categories, exposure in smallest to largest cumulative dosing groups was significantly associated with incident dementia risk, with an apparently dose-dependent relationship in unadjusted as well as adjusted analyses in all time windows prior to dementia identification. Whereas these findings appear compelling, there are at least 4 elephants in the room. First, only 3 of 11 anticholinergic drug categories were consistently associated with an increased risk of dementia; this suggests that anticholinergic activity may be an irrelevant common denominator. Second, for 2 of these 3 categories (antidepressant and antipsychotic drugs), confounding by indication seemed a distinct possibility. Third, in many analyses it seemed that exposure for as little as the equivalent of 1-90 days sufficed to increase the risk of dementia at a time interval of up to 20 years later; a causal mechanism here would need to have strong neurotoxic effects to result in the widespread brain changes that characterize dementia. Finally, the associations were almost uniformly stronger for vascular dementia than for Alzheimer's disease, making the identification of a causal mechanism even more challenging. Deprescribing anticholinergics to reduce state-dependent cognitive impairment, or to reduce the risk of delirium in vulnerable demographic and medical populations, is reasonable. Deprescribing anticholinergics to reduce the risk of future dementia is presently unwarranted.