Association of Sedating vs. Non‐Sedating Antihistamines With Delirium Among Hospitalized Older Adults

Sedating antihistamines may precipitate delirium, but a recent systematic review found that evidence of this association is inconclusive [1]. Despite the potential risks, sedating antihistamines are commonly prescribed in the inpatient setting [2]. We sought to clarify the association between sedating antihistamines and delirium among hospitalized older adults using non-sedating antihistamines as an active comparator. We included older adults (≥ 65 years) admitted to general medical wards at 17 hospitals in Ontario contributing to the GEMINI hospital data research network between April 1, 2015 and March 31, 2022 (the years for which reliable data was available). The 17 hospitals in this analysis had validated medication data for the study period [3]. We identified hospitalized older adults with an inpatient order for a sedating or non-sedating antihistamine (Table S1). In the primary analysis, we classified people prescribed both a sedating and non-sedating antihistamine as sedating antihistamine users; in a sensitivity analysis, we excluded those patients. We tabulated patient characteristics, including age, sex, history of dementia (identified using ICD-10 codes on admission, Table S2), whether they were admitted from the intensive care unit or a nursing home, Charlson comorbidity index, and modified laboratory-based acute physiology score (mLAPS) on admission [4]. We used those variables, hospitalization year, and hospital site to calculate each older adult's propensity of being prescribed sedating or non-sedating antihistamines, which we then used to calculate overlap weights [5]. Using those weights, we calculated the risk difference (RD) and risk ratio (RR) for delirium between older adults prescribed sedating versus non-sedating antihistamines, with empirical 95% confidence intervals (CI) derived using 2000 bootstrap iterations. Our primary definition of delirium was GEMINI's validated machine learning delirium identification tool that has 68% sensitivity and 78% positive predictive value [6]. In a secondary analysis, we defined delirium with ICD-10 codes. We performed subgroup analyses by sex and age (≤ 75 vs. > 75 years old). We identified 12,938 older adults prescribed sedating antihistamines and 3106 prescribed non-sedating antihistamines (Table 1, Figure S1). The groups were similar in age, and both groups included more women than men. Those prescribed sedating antihistamines were less likely to have dementia or be admitted from a nursing home but were more likely to be admitted from ICU and have a Charlson score ≥ 2. After weighting, the groups were similar across characteristics (Figures S2 and S3). Delirium occurred in 39.6% of hospitalizations associated with sedating antihistamines and 35.9% of admissions associated with non-sedating antihistamines (Table 2). After weighting, delirium occurred in 4.0% more admissions with sedating antihistamines than non-sedating antihistamines (RD 4.0%, 95% CI 2.4%–5.7%). The corresponding relative risk was 1.11 (95% CI 1.07–1.17). The RD was attenuated, but still in the same direction and statistically significant when defining delirium using ICD codes (RD 1.1%, 95% CI 0.5%–1.6%), and the relative risk was higher (1.50, 95% CI 1.19–1.99). Findings were similar in analyses stratified by sex and age and excluding hospitalizations (n = 1256) with orders for both sedating and non-sedating antihistamines. Our study adds to previous literature by providing an estimate of the risk of delirium associated with sedating antihistamines versus an active non-sedating antihistamine control. A systematic review identified one small case–control study (n = 245) and one small cohort study (n = 426), each of which found non-significantly increased risk of delirium associated with diphenhydramine [1, 7, 8]. The case–control study among postoperative patients found an odds ratio (95% CI) of 1.8 (0.7–4.5) [7] and the cohort study among older medical patients found a relative risk of 2.1 (95% CI 0.9–4.7) [8]. In a cross-sectional study of hospitalized older adults (n = 328,140), every 1% increase in hospitalist's sedating antihistamine prescribing was associated with an odds ratio (95% CI) of 1.08 (1.05–1.10) for delirium [2]. Absolute rates of delirium in the present study (39.6% and 35.9%) are somewhat higher than in GEMINI patients not selected for antihistamine use (34.8%) [2]. Future studies on whether non-sedating antihistamines are also associated with delirium are warranted. This study is limited by a lack of data on medication administration (only prescribing was collected) and inability to timestamp delirium diagnoses. Reverse causation is unlikely because physicians do not prescribe sedating antihistamines to treat delirium. We do not know the indication for antihistamine use for each patient, so confounding by indication is possible; for example, sedating antihistamines may be more likely used for drug reactions, whereas non-sedating antihistamines may be more likely used for allergic rhinitis. Our findings demonstrate that hospitalized older adults prescribed sedating antihistamines were more likely than those prescribed non-sedating antihistamines to experience delirium. Clinicians should avoid sedating antihistamines whenever possible for hospitalized older adults. Funding: A.M.D., A.C.B., A.A.V., and F.R. Study conception: A.C.B., A.M.D., A.A.V., and F.R. Study methodology: A.M.D., A.A.V., S.B.R., and F.R. Data analysis: S.B.R. and M.A.I. Writing of manuscript: A.M.D. Review and editing of manuscript: All authors. We would like to acknowledge Anne Loffler (GEMINI Senior Data Scientist) for analytical support and sharing code to implement overlap weighting. This study was supported by GEMINI, a research program based out of Unity Health Toronto. The development of GEMINI's data platform has been supported by several funding partners, which can be found listed at https://geminimedicine.ca/partners/. All analyses, results, conclusions, and opinions presented in this paper are solely those of the listed authors and are independent from GEMINI's funding sources. No endorsement by GEMINI's funding sources is intended, nor should it be inferred. This project was supported by the WCHAMSG Innovation Fund of the Alternative Funding Plan for the Academic Health Sciences Centres of Ontario (WCH-23-02). The funder had no role in the design, conduct, or reporting of this study. Dr. Drucker has received compensation from the British Journal of Dermatology (reviewer and Section Editor), American Academy of Dermatology (guidelines writer), Canadian Dermatology Today (manuscript writer), and Canadian Agency for Drugs and Technologies in Health (consultant). Dr. Drucker has received research grants to his institution from the National Eczema Association, Eczema Society of Canada, Canadian Dermatology Foundation, Canadian Institutes for Health Research, US National Institutes of Health, and Physicians Services Incorporated Foundation. Dr. Tadrous received consulting fees from Health Canada and Canada's Drug Agency unrelated to this work. Dr. Fralick is an advisor for SIGNAL1, a start-up company deploying machine learning models to improve inpatient care; has been an expert witness on content unrelated to this work; and holds a provisional patent for an AI model unrelated to the current work. Dr. Verma holds minority interests in Signal1 as a co-inventor of an AI early warning system, outside of the submitted work. Dr. Verma receives salary support from the Temerty Professorship of AI Research and Education in Medicine, University of Toronto. Dr. Verma is a part-time employee of Ontario Health, outside of the submitted work. Fahad Razak receives salary support from the Canada Research Chairs program. Dr. Razak is a part-time employee of Ontario Health, outside of the submitted work. Data S1: jgs70323-sup-0001-supinfo.pdf. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.