Abstract
Antipsychotic use carries many risks, including increased mortality. Although interventions exist to reduce inappropriate prescribing, strategies tailored to the underlying drivers of use remain unexplored. This study investigates how resident and facility factors influence antipsychotic use in Canadian Long-Term Care (LTC) homes, focusing on age and gender disparities. We conducted a retrospective longitudinal analysis of residents aged 65 and older without psychosis in LTC homes across six provinces and territories, using data from the first quarters of 2010-2022. Antipsychotic initiation was associated with male gender, cognitive and behavioural impairment, specific psychiatric diagnoses, non-English/French language, and residence in larger LTC facilities. The highest odds were observed among residents aged 65-74 and male residents with aggressive behaviour. These findings highlight the need for gender-responsive clinical strategies and facility-level interventions to promote equitable prescribing. Reduced use among oldest residents is encouraging, but elevated rates in younger seniors indicate a subgroup requiring targeted approaches.
Introduction
The inappropriate use of antipsychotics in Long-Term Care (LTC) has been a long-standing concern and continues to persist, as evidenced by ongoing national surveillance in Canada and the establishment of explicit reduction targets to curb use in residents without psychosis.1-4 Antipsychotics are associated with a range of adverse side effects, including sedation, delirium, parkinsonism, extrapyramidal symptoms oedema, chest infections, and stroke.5-7 Further, older persons prescribed antipsychotic medications have a substantially increased risk of mortality, with approximately 14% dying within 6 months of treatment initiation. 8 This elevated risk is most pronounced during the early phase of therapy, irrespective of the clinical care setting.8,9 Over the years, numerous interventions have been undertaken to reduce the inappropriate use of antipsychotics, including educational programs and training,10,11 psychiatric liaison and multidisciplinary interventions,12,13 medication reviews,14,15 and quality improvement initiatives with mentoring programmes. 16
Antipsychotic medications are often used in LTC homes to control the behavioural symptoms and agitation of residents with dementia. 17 Aggressive behaviour is evident in about 50% of residents with dementia and about 21% have instances of wandering at least once in the last 7 days. 17 It has been reported that men are more likely than women to be using antipsychotics at the time of nursing home admission and are more likely to have antipsychotic treatment initiated during the follow-up period. 18 Historically, over 30% of older care home residents with documented dementia but no severe mental illness or qualifying diagnosis received antipsychotics.19-21 Prescription patterns were shown to be lower among older age groups in LTC homes such that residents older than 95 years of age are least likely to get antipsychotics prescribed for them.22-24 Quality improvement initiatives to reduce potentially inappropriate use of antipsychotics were effective prior to the COVID-19 pandemic 25 ; however, in the aftermath of the pandemic these rates have rebounded to higher levels. 26
The Canadian Institute for Health Information (CIHI) provides public reporting on nine risk-adjusted Quality Indicators (QIs) that use the assessment data and were developed by interRAI, including a QI on the potential inappropriate use of antipsychotics.3,27 The CIHI indicator excludes individuals receiving end-of-life care and cases where there may be an appropriate clinical justification for antipsychotic use (e.g., schizophrenia, Huntington’s disease, active delusions or hallucinations). 3
The aim of this study is to assess equity in prescribing practices by considering the personal and facility-level factors affecting potentially inappropriate antipsychotic prescribing patterns with a particular focus on age and gender-based analysis.
Methods
Design, Population, Participant Recruitment, and Variables
We conducted longitudinal analyses of residents living in LTC homes across Alberta, British Columbia, Manitoba, Newfoundland and Labrador, Ontario, and the Yukon Territory using deidentified interRAI Minimum Data Set (MDS 2.0) assessments from the first quarters of 2010-2022, to ensure comparability and consistency across years. These standardized assessments are completed quarterly as part of routine care and submitted to CIHI; the interRAI MDS 2.0 has well-documented reliability and validity.28-32 They capture comprehensive information on health status, functional abilities, and care needs and are used to plan care, evaluate quality, and allocate funding based on individual resident needs.33-35 The MDS dataset contains no missing values because built-in field validations and quality assurance processes ensure complete data entry. Moreover, assessments are submitted to CIHI, which mandates the full completion of all required data elements; this requirement effectively eliminates missingness at the time of submission.
The study cohort comprised 413,753 LTC residents assessed between January 1, 2010, and March 31, 2022. We included residents aged 65 years or older and excluded those with a diagnosis of psychosis. There was one observation per person used in our logistic regression model (first and second assessment paired); therefore, repeated measures within residents were not present in the final analytic sample. We incorporated established risk adjusters, including impaired or moderately impaired decision-making ability, long-term memory problems, the Cognitive Performance Scale, a diagnosis of Alzheimer’s disease or other dementias, and age under 65. Leveraging these routinely collected data allowed us to track changes in resident health over time and examine system-level patterns in care delivery and outcomes across jurisdictions.
Our primary outcomes were antipsychotic initiation (new use among residents untreated at baseline) and antipsychotic cessation (discontinuation among baseline users), assessed at routine three-month follow-ups. Independent variables included demographic and clinical factors, with age and gender as key predictors; gender was categorized as male vs non-male, using the non-male group as the reference due to preliminary evidence of higher potentially inappropriate antipsychotic use in males. To evaluate the robustness of our gender classification, we conducted a sensitivity analysis in which gender was re-specified as female vs. non-female, while keeping all covariates and the model structure unchanged.
Analysis
We used logistic regression models to identify predictors of antipsychotic initiation and cessation. We then assessed interactions between aggressive behaviour, measured using the Aggressive Behaviour Scale (ABS), 36 and demographic variables, including age and gender. Risk adjusters included in the interaction model were impaired or moderately impaired decision-making ability, long-term memory problems, motor agitation, the Cognitive Performance Scale (CPS), 37 a diagnosis of Alzheimer’s disease or other dementias (ADRD), and age under 65.
All analyses were conducted using SAS version 9.4.
Ethics
Use of these data and the processes in place to protect patient privacy and confidentiality were approved by the relevant institutional ethics committee. Obtaining informed consent from participants was not necessary for this study, as it exclusively utilized secondary analysis of deidentified clinical assessment records. 38
Results
Odds Ratio for Initiation and Cessation of Antipsychotics among Residents of Canadian Long-Term Care Homes, 2010–2022
ac-stat 0.70.
bc-stat 0.60.
Bold text indicates a statistically significant odds ratio
Similarly, in the final model examining cessation of antipsychotic therapy, males exhibited higher odds of stopping treatment; however, this association was not statistically significant (OR 0.95, 95% CI 0.86-1.05) (Table 1). The sensitivity analysis, using female vs non-female as the comparison, likewise demonstrated a non-significant higher odds of cessation (OR 1.06, 95% CI 0.95-1.17). Overall, the consistency in the magnitude and direction of these associations across alternative gender classifications supports the robustness of the findings.
The statistically significant predictors of an antipsychotic being initiated (Table 1) after admission to a LTC facility are male residents (OR 1.11, 95% CI 1.02-1.21), and residents speaking other languages than English or French (OR 1.20, 95% CI 1.07-1.35).
Among the variables that showed higher odds of initiating an antipsychotic were ADRD (OR 1.83, 95% CI 1.65-2.03), Bipolar Disorder (BPD) (OR 3.73, 95% CI 2.72-5.13), ABS (OR 2.09, 95% CI 1.92-2.28), medium facilities (OR 1.42, 95% CI 1.22-1.66), and large facilities (OR 1.30, 95% CI 1.11-1.52).
We also found strong predictors for discontinuation of an antipsychotic (cessation) (Table 1), including age with the effect greater for age 95+ (OR 2.54, 95% CI 1.86-3.48), and large facilities (OR 1.52, 95% CI 1.26-1.82).
Finally, we used logistic regression to explore the interaction between age and any aggressive behaviour (Figure 1) as well as gender and any aggressive behaviour (Figure 2) in the initiation and cessation of antipsychotics. Our results show a curvilinear interaction such that residents in the 65-75 age category have the highest odds of an antipsychotic being initiated (Figure 1) if they have an aggressive behaviour (1.93) after which the odds decrease with advancing age. On the other hand, the odds ratio for stopping an antipsychotic (Figure 1) increase with age with the highest odds found in the 95+ age category (1.46). We also found a significant gender and aggressive behaviour interaction. For residents without aggressive behaviour present, the gender differences were minor; however, with aggressive behaviour present the odds of antipsychotic initiation rose to 1.08 for non-male residents and were magnified to 2.14 among male residents (Figure 2). Further, for antipsychotic cessation (Figure 2) we found that males had slightly lower odds to have an antipsychotic stopped among those with and without aggressive behaviour present when compared to their counterparts. Odds Ratio for Antipsychotic Initiation and Cessation by Age and Any Aggressive Behaviour in Canadian Long-Term Care Homes, 2010-2022 Odds Ratio for Antipsychotic Initiation and Cessation by Gender and Any Aggressive Behaviour in Canadian Long-Term Care Homes, 2010-2022
Discussion
By conducting this analysis, we aimed to understand whether age or gender of LTC home residents in Canada predicted antipsychotics initiation or cessation with or without aggressive behaviours.
Our findings indicate that certain resident characteristics are associated with higher odds of initiating antipsychotic therapy in LTC. These include being a male, speaking a language other than English or French, having Alzheimer’s disease or related dementias, bipolar disorder, long-term memory impairment, impaired decision-making, motor agitation, or any form of aggressive behaviour. Additionally, residents living in medium or large LTC homes were more likely to have antipsychotics started. Previous studies have shown results similar to ours regarding gender differences, impaired cognition, and dementias.18-21 Canada’s bilingual framework, recognizing English and French as its official languages, may inadvertently contribute to disparities in mental healthcare. Individuals who primarily speak other languages could face significant communication barriers when interacting with healthcare providers. These challenges may hinder their ability to express emotions or concerns effectively, potentially leading to misinterpretations of distress or confusion as aggression, thereby increasing the risk of inappropriate antipsychotic prescriptions. Performance on risk-adjusted quality measures varies by language group rather than favouring any single group overall. 39 This pattern supports a context-dependent communication-barrier mechanism, further reflected in worse outcomes under language discordance, such as greater end-of-life hospitalizations and higher pain burden among Chinese-speaking residents. 40 Similar disparities are observed among francophone residents in non-designated homes, who experience higher depressive symptoms, antipsychotic use, and frequent falls.40-42 These findings highlight the need for linguistically responsive, guideline-concordant dementia care that prioritizes non-pharmacological approaches. 43 It is plausible that language barriers impede the effective implementation of non-pharmacological interventions, thereby prompting the earlier-than-necessary initiation of antipsychotic treatment. 42 Behavioural symptoms often lead to antipsychotic use in dementia; however, the Canadian Coalition for Seniors’ Mental Health (CCSMH) recommends a structured, interdisciplinary, and personalized approach that includes provider education, individualized care plans, meaningful activities, and preferred music, with conditional support for several non-pharmacological therapies. 43 Given the high need for psychiatric services in Ontario LTC homes and the limited, inequitable access to such care, 44 focusing on behavioural antecedents and consequences rather than routine pharmacological responses is essential for equitable, guideline-concordant dementia care. Additional effective non-pharmacological strategies include psychosocial interventions such as group cognitive stimulation therapy, which improves quality of life, 45 and broader cognitive stimulation approaches that help maintain cognitive and functional abilities.46-49 In addition, optimizing environmental settings and care processes is recommended to further enhance overall clinical outcomes.46,50
We also examined age and gender-based differences on the prescribing and deprescribing of antipsychotics among those residents with any aggressive behaviours and those with none. Regarding age, and in agreement with the literature,22,23 one of the most interesting patterns we observed was how age interacts with aggressive behaviour in antipsychotic prescribing. Residents aged 65-74 who displayed aggressive behaviour had higher odds of having an antipsychotic initiated, while those aged 95 or older with similar behaviours had higher odds of having the medication stopped compared to other age groups, especially those younger than 85. This contrast may reflect how staff perceive risk: younger residents are often more mobile and physically stronger, so their aggression may feel more threatening, prompting initiation as a safety measure. In contrast, for the oldest residents, frailty and medical complexity likely make clinicians more cautious about adverse effects, leading to deprescribing. These findings highlight how perceptions of strength and vulnerability can shape treatment decisions in LTC. Gender adds another layer to the story. We found that males with any aggressive behaviour had higher odds of having an antipsychotic initiated compared to non-male residents. Interestingly, this difference did not extend to cessation where both male and non-male residents had similarly low odds of stopping an antipsychotic once prescribed. This raises critical questions: Are antipsychotics more frequently used in males because their aggressive behaviour is perceived as more dangerous, potentially due to physical strength? Or do staff believe non-pharmacological strategies are less effective for male residents? These findings suggest that gendered perceptions may influence prescribing decisions, warranting closer examination of how safety concerns and treatment biases shape care in LTC settings.
Taken together, these patterns point to important opportunities for improving care. The higher odds of antipsychotic initiation among younger, more physically capable residents with aggression, and among male residents, suggest that perceptions of risk and safety may be driving prescribing decisions as much as clinical need. Conversely, the tendency to deprescribe in the oldest residents reflects growing awareness of frailty and medication harms. These results point to a clear need for gender-sensitive clinical strategies and enhanced staff training to manage aggression without defaulting to antipsychotics. The decline in use among the oldest residents is promising, but elevated initiation rates in younger seniors, especially those with aggressive behaviour, signal an urgent need for tailored, non-pharmacological approaches in everyday practice. Multi-component, person-centred non-pharmacological interventions such as the Well-being and Health for People with Dementia (WHELD) and Dementia Care Mapping (DCM) offer modest but consistent reductions in agitation and improvements in quality of life when supported by trained staff and adequate organizational resources.51,52 Additional evidence indicates that approaches like aromatherapy with massage, validation therapy, psycho-educational support, music therapy, and structured exercise can reduce neuropsychiatric symptoms and agitation, although effects on quality of life remain mixed.43,51,53,54
Strengths and Limitations
Our analysis adopted a pan-Canadian lens, incorporating data from LTC homes across seven provinces, with a focus on age and gender-based patterns as well as facility-level characteristics. We employed the CIHI-defined QI for potentially inappropriate antipsychotic use, deliberately excluding residents with schizophrenia, Huntington’s disease, or those receiving end-of-life care, populations for whom antipsychotic treatment may be clinically justified due to symptoms such as hallucinations or delusions. This approach ensured that our analysis concentrated solely on residents for whom antipsychotic use is generally considered inappropriate.
However, several limitations must be acknowledged. First, our dataset captured the primary language of residents but lacked information on the language environment of the LTC facilities themselves, an omission that may have influenced our findings, particularly in the context of language-concordant care. Second, the observational design precludes causal inference; residual confounding from unmeasured clinical or organizational factors may remain despite adjustment. Third, the model predicting antipsychotic cessation showed limited discriminative ability (c-statistic = 0.60), suggesting that the variables available in administrative and assessment data capture only part of the factors influencing deprescribing. Key elements that shape discontinuation such as facility prescribing culture, staffing capacity to deliver non-pharmacological care, clinician risk tolerance, and family involvement were not measured and may have contributed to the model’s reduced performance. In addition, some misclassification of cessation (e.g., temporary holds, illness-related interruptions, or documentation variability) is possible. Consequently, the model should be interpreted as identifying population-level patterns rather than providing reliable individual-level prediction. Fourth, the logistic regression model did not account for clustering of residents within facilities. Future analyses using mixed-effects or GEE models could more appropriately address facility-level and within-resident dependency.
Conclusion
The findings emphasize the need for targeted training and adaptation of clinical interventions that account for gender differences, in order to promote more equitable prescribing practices. Significant facility-level effects were observed, suggesting that additional support should be directed towards facilities lacking adequate resources to effectively manage QI efforts. The lower use of antipsychotics in the oldest age groups is encouraging; however, further efforts are needed to reduce prescribing rates among the younger-old population, who may represent a clinically distinct cohort requiring a tailored approach to care. Although several non-pharmacological approaches show potential as alternatives for managing behavioural symptoms, the evidence base remains mixed and their uptake across care settings is inconsistent. Consequently, further research is needed to refine and standardize these interventions and to evaluate their long-term clinical value as part of broader initiatives to reduce inappropriate antipsychotic use and improve dementia care.
Footnotes
Ethical Approval
Use of these data and the processes in place to protect patient privacy and confidentiality were approved by the University of Waterloo’s Office of Research Ethics (ORE# 30372).
Consent to Participate
Obtaining informed consent from participants was not necessary for this study, as it exclusively utilized secondary analysis of deidentified clinical assessment records.
Authors’ Contributions
RTM performed the literature search and drafted the initial manuscript. JPH verified the conceptual as well as analytical methods. RTM performed the analysis. All authors reviewed and revised the work. All authors approved of the final version.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded by Unity Health Toronto. Portions of this material are based on work developed with financial support from CDA-AMC (C-242502000). The analyses, conclusions, opinions, and statements expressed in this material are those of the author and do not necessarily reflect the view of CDA-AMC.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The data supporting the findings of this study are available from the Canadian Institute for Health Information (CIHI). Access to this data is restricted and was granted under license for this study.