Development and usability testing of an online support tool to identify models and frameworks to inform implementation.

BACKGROUND: Theories, models and frameworks (TMFs) are useful when implementing, evaluating and sustaining healthcare evidence-based interventions. Yet it can be challenging to identify an appropriate TMF for an implementation project. We developed and tested the usability of an online tool to help individuals who are doing or supporting implementation practice activities to identify appropriate models and/or frameworks to inform their work. METHODS: We used methods guided by models and evidence on implementation science and user-centered design. Phases of tool development included applying findings from a scoping review of TMFs and interviews with 24 researchers/implementers on barriers and facilitators to identifying and selecting TMFs. Based on interview findings, we categorized the TMFs by aim, stage of implementation, and target level of change to inform the tool's algorithm. We then conducted interviews with 10 end-users to test the usability of the prototype tool and administered the System Usability Scale (SUS). Usability issues were addressed and incorporated into the tool. RESULTS: We developed Find TMF, an online tool consisting of 3-4 questions about the user's implementation project. The tool's algorithm matches key characteristics of the user's project (aim, stage, target change level) with characteristics of different TMFs and presents a list of candidate models/frameworks. Ten individuals from Canada or Australia participated in usability testing (mean SUS score 84.5, standard deviation 11.4). Overall, participants found the tool to be simple, easy to use and visually appealing with a useful output of candidate models/frameworks to consider for an implementation project. Users wanted additional instruction and guidance on what to expect from the tool and how to use the information in the output table. Tool improvements included incorporating an overview figure outlining the tool steps and output, displaying the tool questions on a single page, and clarifying the available functions of the results page, including adding direct links to the glossary and to complementary tools. CONCLUSIONS: Find TMF is an easy-to-use online tool that may benefit individuals who support implementation practice activities by making the vast number of models and frameworks more accessible, while also supporting a consistent approach to identifying and selecting relevant TMFs.

Edible Cannabis Legalization and Cannabis Poisonings in Older Adults.

This cross-sectional study examines the association between edible cannabis legalization and emergency department visits for cannabis poisonings in older adults.

Evaluating methods for risk prediction of Covid-19 mortality in nursing home residents before and after vaccine availability: a retrospective cohort study.

BACKGROUND: SARS-CoV-2 vaccines are effective in reducing hospitalization, COVID-19 symptoms, and COVID-19 mortality for nursing home (NH) residents. We sought to compare the accuracy of various machine learning models, examine changes to model performance, and identify resident characteristics that have the strongest associations with 30-day COVID-19 mortality, before and after vaccine availability. METHODS: We conducted a population-based retrospective cohort study analyzing data from all NH facilities across Ontario, Canada. We included all residents diagnosed with SARS-CoV-2 and living in NHs between March 2020 and July 2021. We employed five machine learning algorithms to predict COVID-19 mortality, including logistic regression, LASSO regression, classification and regression trees (CART), random forests, and gradient boosted trees. The discriminative performance of the models was evaluated using the area under the receiver operating characteristic curve (AUC) for each model using 10-fold cross-validation. Model calibration was determined through evaluation of calibration slopes. Variable importance was calculated by repeatedly and randomly permutating the values of each predictor in the dataset and re-evaluating the model's performance. RESULTS: A total of 14,977 NH residents and 20 resident characteristics were included in the model. The cross-validated AUCs were similar across algorithms and ranged from 0.64 to 0.67. Gradient boosted trees and logistic regression had an AUC of 0.67 pre- and post-vaccine availability. CART had the lowest discrimination ability with an AUC of 0.64 pre-vaccine availability, and 0.65 post-vaccine availability. The most influential resident characteristics, irrespective of vaccine availability, included advanced age (≥ 75 years), health instability, functional and cognitive status, sex (male), and polypharmacy. CONCLUSIONS: The predictive accuracy and discrimination exhibited by all five examined machine learning algorithms were similar. Both logistic regression and gradient boosted trees exhibit comparable performance and display slight superiority over other machine learning algorithms. We observed consistent model performance both before and after vaccine availability. The influence of resident characteristics on COVID-19 mortality remained consistent across time periods, suggesting that changes to pre-vaccination screening practices for high-risk individuals are effective in the post-vaccination era.

Impact of a vaccine passport on first-dose SARS-CoV-2 vaccine coverage by age and area-level social determinants of health in the Canadian provinces of Quebec and Ontario: an interrupted time series analysis.

BACKGROUND: In Canada, all provinces implemented vaccine passports in 2021 to reduce SARS-CoV-2 transmission in non-essential indoor spaces and increase vaccine uptake (policies active September 2021-March 2022 in Quebec and Ontario). We sought to evaluate the impact of vaccine passport policies on first-dose SARS-CoV-2 vaccination coverage by age, and area-level income and proportion of racialized residents. METHODS: We performed interrupted time series analyses using data from Quebec's and Ontario's vaccine registries linked to census information (population of 20.5 million people aged ≥ 12 yr; unit of analysis: dissemination area). We fit negative binomial regressions to first-dose vaccinations, using natural splines adjusting for baseline vaccination coverage (start: July 2021; end: October 2021 for Quebec, November 2021 for Ontario). We obtained counterfactual vaccination rates and coverage, and estimated the absolute and relative impacts of vaccine passports. RESULTS: In both provinces, first-dose vaccination coverage before the announcement of vaccine passports was 82% (age ≥ 12 yr). The announcement resulted in estimated increases in coverage of 0.9 percentage points (95% confidence interval [CI] 0.4-1.2) in Quebec and 0.7 percentage points (95% CI 0.5-0.8) in Ontario. This corresponds to 23% (95% CI 10%-36%) and 19% (95% CI 15%-22%) more vaccinations over 11 weeks. The impact was larger among people aged 12-39 years. Despite lower coverage in lower-income and more-racialized areas, there was little variability in the absolute impact by area-level income or proportion racialized in either province. INTERPRETATION: In the context of high vaccine coverage across 2 provinces, the announcement of vaccine passports had a small impact on first-dose coverage, with little impact on reducing economic and racial inequities in vaccine coverage. Findings suggest that other policies are needed to improve vaccination coverage among lower-income and racialized neighbourhoods and communities.

The association between frailty, long-term care home characteristics and COVID-19 mortality before and after SARS-CoV-2 vaccination: a retrospective cohort study.

BACKGROUND: The relative contributions of long-term care (LTC) resident frailty and home-level characteristics on COVID-19 mortality has not been well studied. We examined the association between resident frailty and home-level characteristics with 30-day COVID-19 mortality before and after the availability of SARS-CoV-2 vaccination in LTC. METHODS: We conducted a population-based retrospective cohort study of LTC residents with confirmed SARS-CoV-2 infection in Ontario, Canada. We used multi-level multivariable logistic regression to examine associations between 30-day COVID-19 mortality, the Hubbard Frailty Index (FI), and resident and home-level characteristics. We compared explanatory models before and after vaccine availability. RESULTS: There were 11,179 and 3,655 COVID-19 cases in the pre- and post-vaccine period, respectively. The 30-day COVID-19 mortality was 25.9 and 20.0% during the same periods. The median odds ratios for 30-day COVID-19 mortality between LTC homes were 1.50 (95% credible interval [CrI]: 1.41-1.65) and 1.62 (95% CrI: 1.46-1.96), respectively. In the pre-vaccine period, 30-day COVID-19 mortality was higher for males and those of greater age. For every 0.1 increase in the Hubbard FI, the odds of death were 1.49 (95% CI: 1.42-1.56) times higher. The association between frailty and mortality remained consistent in the post-vaccine period, but sex and age were partly attenuated. Despite the substantial home-level variation, no home-level characteristic examined was significantly associated with 30-day COVID-19 mortality during either period. INTERPRETATION: Frailty is consistently associated with COVID-19 mortality before and after the availability of SARS-CoV-2 vaccination. Home-level characteristics previously attributed to COVID-19 outcomes do not explain significant home-to-home variation in COVID-19 mortality.

Identifying Unexpected Deaths in Long-Term Care Homes.

OBJECTIVES: Predicting unexpected deaths among long-term care (LTC) residents can provide valuable information to clinicians and policy makers. We study multiple methods to predict unexpected death, adjusting for individual and home-level factors, and to use as a step to compare mortality differences at the facility level in the future work. DESIGN: We conducted a retrospective cohort study using Resident Assessment Instrument Minimum Data Set assessment data for all LTC residents in Ontario, Canada, from April 2017 to March 2018. SETTING AND PARTICIPANTS: All residents in Ontario long-term homes. We used data routinely collected as part of administrative reporting by health care providers to the funder: Ontario Ministry of Health and Long-Term Care. This project is a component of routine policy development to ensure safety of the LTC system residents. METHODS: Logistic regression (LR), mixed-effect LR (mixLR), and a machine learning algorithm (XGBoost) were used to predict individual mortality over 5 to 95 days after the last available RAI assessment. RESULTS: We identified 22,419 deaths in the cohort of 106,366 cases (mean age: 83.1 years; female: 67.7%; dementia: 68.8%; functional decline: 16.6%). XGBoost had superior calibration and discrimination (C-statistic 0.837) over both mixLR (0.819) and LR (0.813). The models had high correlation in predicting death (LR-mixLR: 0.979, LR-XGBoost: 0.885, mixLR-XGBoost: 0.882). The inter-rater reliability between the models LR-mixLR and LR-XGBoost was 0.56 and 0.84, respectively. Using results in which all 3 models predicted probability of actual death of a resident at <5% yielded 210 unexpected deaths or 0.9% of the observed deaths. CONCLUSIONS AND IMPLICATIONS: XGBoost outperformed other models, but the combination of 3 models provides a method to detect facilities with potentially higher rates of unexpected deaths while minimizing the possibility of false positives and could be useful for ongoing surveillance and quality assurance at the facility, regional, and national levels.

Ontario's COVID-19 Modelling Consensus Table: mobilizing scientific expertise to support pandemic response.

SETTING: COVID-19 has highlighted the need for credible epidemiological models to inform pandemic policy. Traditional mechanisms of commissioning research are ill-suited to guide policy during a rapidly evolving pandemic. At the same time, contracting with a single centre of expertise has been criticized for failing to reflect challenges inherent in specific modelling approaches. INTERVENTION: This report describes an alternative approach to mobilizing scientific expertise. Ontario's COVID-19 Modelling Consensus Table (MCT) was created in March 2020 to enable rapid communication of credible estimates of the impact of COVID-19 and to accelerate learning on how the disease is spreading and what could slow its transmission. The MCT is a partnership between the province and academic modellers and consists of multiple groups of experts, health system leaders, and senior decision-makers. Armed with Ministry of Health data, the MCT meets once per week to share results from modelling exercises, generate consensus judgements of the likely future impact of COVID-19, and discuss decision-makers' priorities. OUTCOMES: The MCT has enabled swift access to data for participants, a structure for developing consensus estimates and communicating these to decision-makers, credible models to inform health system planning, and increased transparency in public reporting of COVID-19 data. It has also facilitated the rapid publication of research findings and its incorporation into government policy. IMPLICATIONS: The MCT approach is one way to quickly draw on scientific advice outside of government and public health agencies. Beyond speed, this approach allows for nimbleness as experts from different organizations can be added as needed. It also shows how universities and research institutes have a role to play in crisis situations, and how this expertise can be marshalled to inform policy while respecting academic freedom and confidentiality.

RéSUMé: LIEU: La COVID-19 a mis en évidence le besoin de modèles épidémiologiques crédibles pour éclairer la politique pandémique. Les mécanismes habituels pour commander des travaux de recherche sont peu propices à orienter les politiques lors d'une pandémie qui évolue rapidement. En même temps, la passation de contrats avec un seul centre d'expertise est critiquée, car elle ne tient pas compte des difficultés inhérentes de certaines approches de modélisation. INTERVENTION: Le présent rapport décrit une approche de rechange pour mobiliser le savoir scientifique. L'Ontario a créé en mars 2020 une Table de concertation sur la modélisation (TCM) qui permet de communiquer de façon rapide et fiable les estimations des effets de la COVID-19 et d'apprendre plus vite comment la maladie se propage et ce qui pourrait en ralentir la transmission. La TCM, un partenariat entre les modélisateurs de la province et des milieux universitaires, est composée de plusieurs groupes d'experts, de dirigeants du système de santé et de décideurs de haut niveau. Armée des données du ministère de la Santé, la TCM se réunit une fois par semaine pour partager les résultats d'exercices de modélisation, générer des jugements consensuels sur les futurs effets probables de la COVID-19 et discuter des priorités des décideurs. RéSULTATS: La TCM rend possible un accès rapide aux données pour les participants, une structure pour élaborer des estimations consensuelles et les communiquer aux décideurs, des modèles fiables pour éclairer la planification du système de santé, ainsi qu'une transparence accrue dans la communication des données sur la COVID-19 au public. Elle facilite aussi la publication rapide des résultats de recherche et leur intégration dans la politique gouvernementale. CONSéQUENCES: L'approche de la TCM est un moyen d'obtenir rapidement des conseils scientifiques à l'extérieur du gouvernement et des organismes de santé publique. Au-delà de sa rapidité, cette approche offre une grande souplesse, car des experts de différents organismes peuvent être ajoutés au besoin. Elle montre aussi que les universités et les établissements de recherche ont un rôle à jouer dans les situations de crise, et qu'il est possible de mobiliser leurs compétences pour éclairer les politiques tout en respectant la liberté et la confidentialité des milieux de la recherche et de l'enseignement.

Facteurs de risque d'éclosion de SRAS-CoV-2 dans les résidences pour aînés en Ontario, au Canada: étude de cohorte à l'échelle de la population.

CONTEXTE: L'épidémiologie de l'infection au SRAS-CoV-2 dans les résidences pour aînés (offrant une aide à la vie autonome), est pour une bonne part inconnue. Nous avons étudié le lien entre les caractéristiques des résidences et des communautés avoisinantes et le risque d'éclosion de SRAS-CoV-2 dans les résidences pour aînés depuis le début de la première vague de la pandémie de COVID-19. MÉTHODES: Nous avons procédé à une étude de cohorte rétrospective dans la population des résidences pour aînés certifiées en Ontario, au Canada, entre le 1er mars et le 18 décembre 2020. Notre paramètre principal était toute éclosion de SRAS-CoV-2 (≥ 1 cas confirmé parmi les résidents ou le personnel au moyen d'un test d'amplification des acides nucléiques). Nous avons utilisé la méthode des risques proportionnels avec prédicteurs chronologiques pour modéliser les liens entre les caractéristiques des résidences et des communautés avoisinantes et les éclosions de SRAS-CoV-2. RÉSULTATS: Notre cohorte a inclus l'ensemble des 770 résidences privées pour aînés (RPA) certifiées en Ontario qui hébergeaient 56 491 résidents. On a dénombré 273 (35,5 %) résidences pour aînés qui ont connu 1 éclosion de SRAS-CoV-2 ou plus; 1944 résidents (3,5 %) et 1101 employés (3,0 %) ont contracté l'infection. Ces cas étaient inégalement distribués entre les résidences. En effet, 2487 cas parmi les résidents et le personnel (81,7 %) sont survenus dans 77 résidences (10 %). Le rapport de risque ajusté d'une éclosion de SRAS-CoV-2 dans une résidence a été clairement associé aux établissements qui avaient une grande capacité d'accueil, qui comportaient des unités de soins de longue durée, qui appartenaient à de plus grandes bannières et offraient plusieurs services sur place, qui se trouvaient dans des régions marquées par une hausse de l'incidence régionale de SRAS-CoV-2 et où la concentration ethnique à l'échelle de la communauté était supérieure. INTERPRÉTATION: Certaines caractéristiques facilement identifiables des résidences pour aînés sont associées de manière indépendante aux éclosions de SRAS-CoV-2 et peuvent faciliter l'évaluation des risques et orienter la priorisation de la vaccination.

Risk factors for outbreaks of SARS-CoV-2 infection at retirement homes in Ontario, Canada: a population-level cohort study.

BACKGROUND: The epidemiology of SARS-CoV-2 infection in retirement homes (also known as assisted living facilities) is largely unknown. We examined the association between home-and community-level characteristics and the risk of outbreaks of SARS-CoV-2 infection in retirement homes since the beginning of the first wave of the COVID-19 pandemic. METHODS: We conducted a population-based, retrospective cohort study of licensed retirement homes in Ontario, Canada, from Mar. 1 to Dec. 18, 2020. Our primary outcome was an outbreak of SARS-CoV-2 infection (≥ 1 resident or staff case confirmed by validated nucleic acid amplification assay). We used time-dependent proportional hazards methods to model the associations between retirement home- and community-level characteristics and outbreaks of SARS-CoV-2 infection. RESULTS: Our cohort included all 770 licensed retirement homes in Ontario, which housed 56 491 residents. There were 273 (35.5%) retirement homes with 1 or more outbreaks of SARS-CoV-2 infection, involving 1944 (3.5%) residents and 1101 staff (3.0%). Cases of SARS-CoV-2 infection were distributed unevenly across retirement homes, with 2487 (81.7%) resident and staff cases occurring in 77 (10%) homes. The adjusted hazard of an outbreak of SARS-CoV-2 infection in a retirement home was positively associated with homes that had a large resident capacity, were co-located with a long-term care facility, were part of larger chains, offered many services onsite, saw increases in regional incidence of SARS-CoV-2 infection, and were located in a region with a higher community-level ethnic concentration. INTERPRETATION: Readily identifiable characteristics of retirement homes are independently associated with outbreaks of SARS-CoV-2 infection and can support risk identification and priority for vaccination.

Barriers and facilitators to uptake of systematic reviews by policy makers and health care managers: a scoping review.

BACKGROUND: We completed a scoping review on the barriers and facilitators to use of systematic reviews by health care managers and policy makers, including consideration of format and content, to develop recommendations for systematic review authors and to inform research efforts to develop and test formats for systematic reviews that may optimise their uptake. METHODS: We used the Arksey and O'Malley approach for our scoping review. Electronic databases (e.g., MEDLINE, EMBASE, PsycInfo) were searched from inception until September 2014. Any study that identified barriers or facilitators (including format and content features) to uptake of systematic reviews by health care managers and policy makers/analysts was eligible for inclusion. Two reviewers independently screened the literature results and abstracted data from the relevant studies. The identified barriers and facilitators were charted using a barriers and facilitators taxonomy for implementing clinical practice guidelines by clinicians. RESULTS: We identified useful information for authors of systematic reviews to inform their preparation of reviews including providing one-page summaries with key messages, tailored to the relevant audience. Moreover, partnerships between researchers and policy makers/managers to facilitate the conduct and use of systematic reviews should be considered to enhance relevance of reviews and thereby influence uptake. CONCLUSIONS: Systematic review authors can consider our results when publishing their systematic reviews. These strategies should be rigorously evaluated to determine impact on use of reviews in decision-making.

The relationship between variations in antipsychotic prescribing across nursing homes and short-term mortality: quality of care implications.

BACKGROUND: High rates of antipsychotic drug prescribing in nursing homes can signal poor quality processes, but also raise concerns about drug safety due to the mortality risk of this therapy. Determining the extent to which variations in antipsychotic use are a symptom of facility-level quality problems as compared with a drug safety issue is important for selecting the correct interventions to effect change. OBJECTIVE: To determine whether nursing homes with higher rates of antipsychotic dispensing had higher rates of short-term mortality among their residents. METHODS: This population-based study examined 60,105 older adults newly admitted to nursing homes in Ontario between April 1, 2000 and March 31, 2004. Using adjusted Cox proportional hazard models, we explored the relationship between facility-level dispensing rates and mortality, controlling for resident characteristics. Facilities were grouped into quintiles according to mean antipsychotic rate. All-cause mortality at 30 and 120 days after admission was stratified by recent hospital discharge and analyzed by quintile. RESULTS: Average antipsychotic dispensing ranged from 11.6% in the lowest quintile to 30.0% in the highest. Among residents with no recent hospitalization, all-cause mortality at 30 days was 2.5% in the lowest compared with 3.3% in the highest quintile (adjusted hazard ratio: 1.28, confidence interval: 1.06-1.56) and at 120 days was 9.3% compared with 11.7% (adjusted hazard ratio: 1.25, confidence interval: 1.13-1.39). CONCLUSION: Residents were at increased risk of death simply by being admitted to a facility with a higher intensity of antipsychotic drug use, despite similar clinical characteristics at admission.

A change from subcutaneous to intravenous erythropoietin increases the cost of anemia therapy.

BACKGROUND: It seems that more erythropoietin (EPO) is required when given intravenously (IV) than when given subcutaneously (SC). Estimates of the magnitude of this difference vary widely, impeding development of economic models in this area. Concerns about pure red cell aplasia led our program to switch from SC to IV EPO, so we studied the impact of this change on the cost of anemia therapy. METHODS: All in-center hemodialysis patients who had received EPO for at least 3 months prior to and following conversion to IV EPO were studied. Data was obtained retrospectively for 1 year prior to and prospectively for 1 year following conversion. The costs of anemia therapy (EPO, transfusions and iron) were calculated from the hospital's perspective. RESULTS: 158 patients were studied. One month after switching, the hemoglobin fell significantly, reaching a nadir at 3 months. This triggered more use of EPO, iron and transfusions. By month 7 hemoglobin levels had returned to initial levels, with a median rise in EPO dose of 1,250 units/week (p < 0.001). After the switch, the median rise in total anemia therapy costs was 13.1% (CAD 665/patient-year, p < 0.01). CONCLUSIONS: Conversion of EPO from SC to IV dosing increased the costs of anemia therapy at our center.

Skilled nursing facility rehabilitation and discharge to home after stroke.

OBJECTIVE: To determine the relation between rehabilitation therapy (RT) intensity and time to discharge home for stroke patients in skilled nursing facilities (SNFs). DESIGN: Retrospective cohort study. We used regression analyses, stratified by expected outcome, and propensity score adjustment. Setting All SNFs in Ohio, Michigan, and Ontario, Canada. PARTICIPANTS: A cohort of residents, aged 65 and over, admitted from hospitals to SNFs with a diagnosis of stroke (N=23,824). INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURE: Time to discharge home from an SNF. RESULTS: RT was given to more than 95% of residents for whom discharge was expected within 90 days and to more than 60% of residents for whom discharge was uncertain or not expected. RT increased the likelihood of discharge to the community for all groups except those expected to be discharged within 30 days. The dose-response relation was strongest for residents with either an uncertain discharge prognosis or no discharge expected. CONCLUSIONS: Postacute residents with an uncertain prognosis are an important target population for intensive RT.

Nursing home profit status and quality of care: is there any evidence of an association?

This article critically reviews the association between the profit status of North American nursing homes and the quality of care. Studies were identified by searching MEDLINE (January 1990-October 2002), reference lists, letters, commentaries, and editorials. The quality indicator(s) used to measure quality of care, and its relationship to profit status, was extracted from each publication. The study design and risk-adjustment methodologies used were also extracted. The interrater reliability for the extraction of these three items was determined to be 1.0, 0.6, and 0.8, respectively. A qualitative systematic review was performed using Donabedian's framework of structure, process, and outcome for analyzing medical quality of care. Empirical research in the past 12 years has found that systematic differences exist between for-profit and not-for-profit nursing homes. For-profit nursing homes appear to provide lower quality of care in many important areas of process and outcome.

Representation of patients with dementia in clinical trials of donepezil.

OBJECTIVES: To evaluate the representation of frail older adults in randomized controlled trials (RCTs), and to assess consequences of under representation by analyzing drug discontinuation rates. METHODS: A cohort of older adults newly dispensed donepezil in Ontario between September 2001 and March 2002 was constructed using administrative data. A systematic review of the literature identified RCTs of donepezil. Patients dispensed donepezil were then compared to clinical trial subjects. Discontinuation rates were examined for patients with and without potential contraindications to this drug. RESULTS: There were 6,424 older adults in the Ontario cohort with new claims for donepezil. Ten RCTs evaluating the use of donepezil were identified (n = 3,423). Between 51% and 78% of the Ontario cohort would have been ineligible for RCT enrollment. Patients dispensed donepezil were older (80.3 vs. 73.7 years, p < 0.001) and more likely to be in long-term care (14.1 vs. 7.1%, p < 0.001) than RCT subjects. Overall, 27.8% of the Ontario cohort discontinued donepezil within seven months of initial prescription. Discontinuation rates were significantly higher for patients with a history of obstructive lung disease, active cardiovascular disease, or Parkinsonism. CONCLUSIONS: Fewer than half of the older adults dispensed donepezil in Ontario would have been eligible to participate in the RCTs that established the efficacy of this drug. Discontinuation rates were higher among patient groups not represented in the trials. Clinicians should carefully assess the potential risks and benefits of such drug therapies for older patients with dementia.

Atypical antipsychotic drugs in the treatment of behavioural and psychological symptoms of dementia: systematic review.

OBJECTIVE: To review the role of oral atypical antipsychotic drugs in the management of the behavioural and psychological symptoms of dementia (BPSD). DATA SOURCES: Medline, Embase, and the Cochrane Library. Reference lists were reviewed and experts were contacted to identify additional trials. STUDY SELECTION: Double blind randomised controlled trials that evaluated the four oral atypical antipsychotic therapies for BPSD. REVIEW METHODS: Two reviewers assessed trial validity independently. DATA EXTRACTION: Demographics of patients, study duration, dose of antipsychotic, primary end points, adverse events. RESULTS: 77 abstracts were reviewed. Five randomised trials (1570 patients) evaluating risperidone and olanzapine were identified. The quality of trials was generally good. Most participants were in an institution (> 96%), elderly (weighted mean 82.3 years), and had Alzheimer's disease (76.3%). Trials lasted 6-12 weeks. Treatment with atypical antipsychotic drugs was superior to placebo for the primary end point in three of the five trials. Two trials comparing risperidone with haloperidol did not find any differences in the primary measures of efficacy. Adverse events were common and included extrapyramidal symptoms, somnolence, and abnormal gait. CONCLUSIONS: Although atypical antipsychotic drugs are being used with increasing frequency, few randomised trials have evaluated their use for BPSD. Limited evidence supports the perception of improved efficacy and adverse event profiles compared with typical antipsychotic drugs.