Association between sedating medications and delirium in older inpatients

Objectives To examine the association between Beers criteria sedative medications and delirium in a large cohort of hospitalized elderly adults with common medical conditions. Design Retrospective cohort and nested case-control studies. Setting 374 U.S. hospitals. Participants All individuals aged 65 and older admitted to the hospital between September 2003 and June 2005 with one of six principal diagnoses (acute myocardial infarction, chronic obstructive pulmonary disease, community-acquired pneumonia, congestive heart failure, ischemic stroke, urinary tract infection). Measurements Primary outcome was presumed hospital-acquired delirium, defined as initiation of an antipsychotic medication or restraints on hospital Day 3 or later. Logistic and proportional hazards regression were used to model the associations between sedative exposure and delirium. Results The dataset contained 225,028 participants (median age 82; 58% female). Four percent fit the definition of hospital-acquired delirium (median onset Day 5). In all, 38,883 (17%) participants received one or more sedative medications. In the cohort study, diphenhydramine (adjusted odds ratio (AOR) = 1.22, 95% confidence interval (CI) = 1.09-1.36) and short-acting benzodiazepines (AOR = 1.18, 95% CI = 1.03-1.34) were associated with greater risk of subsequent delirium. In the nested case-control study, diphenhydramine, short- and long-acting benzodiazepines and promethazine were associated with delirium. Amitriptyline and muscle relaxants were not associated with delirium in either study. Confounding by indication could not be excluded for drugs that are sometimes used improperly to treat delirium. Conclusion An association was found between several Beers criteria sedative medications and delirium in hospitalized medical patients. Given the prevalence of these medications and the morbidity associated with delirium, further investigation into the appropriateness of such prescribing is warranted.