Sedation-Associated Medications at Dementia Diagnosis, Their Receptor Activity, and Associations With Adverse...

We aimed to investigate whether sedative medications are associated with adverse outcomes in people with dementia, and whether specific characteristics of these medications predict a higher risk of harm.

Retrospective cohort study.

15,210 patients diagnosed with dementia between 2008 and 2017 in South London.

From recorded medications at dementia diagnosis, we ascertained those with drowsiness listed as a side effect (termed "sedative" hereafter) and subdivided them by frequency and strength of sedation, receptor profile, half-life, and whether they were psychotropics. Multivariable Cox regression models were applied to determine risk of mortality and emergency hospitalization, and generalized estimating equations to investigate cognitive decline. Final models were adjusted for 19 potential confounders, including measures of physical and mental health, functioning, and central anticholinergic burden.

At diagnosis, 70.4% of patients with dementia were receiving at least 1 sedative medication. Median survival time was 4.0 years and median time to first hospitalization 1.4 years. After controlling for potential confounders, receipt of any sedative medication at dementia diagnosis was associated with accelerated cognitive decline and a higher hospitalization risk, but only medications with a cautionary warning yielded an increased mortality hazard. Medications acting through γ-aminobutyric acid agonism, psychotropic sedatives, and those with a short half-life were associated with a higher risk of mortality. γ-aminobutyric acid agonists, N-methyl-d-aspartate receptor antagonists, and nonpsychotropic sedatives were associated with an increased hospitalization risk. α₁ antagonist, antihistamines, N-methyl-d-aspartate receptor antagonists, psychotropic sedatives, and those with the shortest or longest half-life were associated with accelerated cognitive decline.

Receipt of any sedative agent was associated with hospitalization and accelerated cognitive decline. Differences in hazard appear to exist between frequency and strength of sedation, receptor profiles, half-life, and prescribing indication. These differences should be taken into consideration in medication reviews at the time of dementia diagnosis.