Factors associated with plans for early retirement among Ontario family physicians during the COVID-19 pandemic: a cross-sectional study.

BACKGROUND: Higher numbers of family physicians (FPs) stopped practicing or retired during the COVID-19 pandemic, worsening the family doctor shortage in Canada. Our study objective was to determine which factors were associated with FPs' plans to retire earlier during the COVID-19 pandemic. METHODS: We administered two cross-sectional online surveys to Ontario FPs asking whether they were "planning to retire earlier" as a result of the pandemic during the first and third COVID-19 pandemic waves (Apr-Jun 2020 and Mar-Jul 2021). We used logistic regression to determine which factors were associated with early retirement planning, adjusting for age. RESULTS: The age-adjusted proportion of FP respondents planning to retire earlier was 8.2% (of 393) in the first-wave and 20.5% (of 454) in the third-wave. Planning for earlier retirement during the third-wave was associated with age over 50 years (50-59 years odds ratio (OR) 5.37 (95% confidence interval (CI):2.33-12.31), 60 years and above OR 4.18 (95% CI: 1.90-10.23)), having difficulty handling increased non-clinical responsibilities (OR 2.95 (95% CI: 1.79-4.94)), feeling unsupported to work virtually (OR 1.96 (95% CI: 1.19-3.23)) or in-person (OR 2.70 (95% CI: 1.67-4.55)), feeling unable to provide good care (OR 1.82 (95% CI: 1.10-3.03)), feeling work was not valued (OR 1.92 (95% CI: 1.15-3.23)), feeling frightened of dealing with COVID-19 (OR 2.01 (95% CI: 1.19-3.38)), caring for an elderly relative (OR 2.36 (95% CI: 1.69-3.97)), having difficulty obtaining personal protective equipment (OR 2.00 (95% CI: 1.16-3.43)) or difficulty implementing infection control practices in clinic (OR 2.10 (95% CI: 1.12-3.89)). CONCLUSIONS: Over 20% of Ontario FP respondents were considering retiring earlier by the third-wave of the COVID-19 pandemic. Supporting FPs in their clinical and non-clinical roles, such that they feel able to provide good care and that their work is valued, reducing non-clinical (e.g., administrative) responsibilities, dealing with pandemic-related fears, and supporting infection control practices and personal protective equipment acquisition in clinic, particularly in those aged 50 years or older may help increase family physician retention during future pandemics.

Multimorbidity and Blood Pressure Control in Patients Attending Primary Care in Canada.

BACKGROUND: There has been conflicting evidence on the association between multimorbidity and blood pressure (BP) control. This study aimed to investigate this associations in people with hypertension attending primary care in Canada, and to assess whether individual long-term conditions are associated with BP control. METHODS: This was a cross-sectional study in people with hypertension attending primary care in Toronto between January 1, 2017 and December 31, 2019. Uncontrolled BP was defined as systolic BP ≥ 140 mmHg or diastolic BP ≥ 90 mmHg. A list of 11 a priori selected chronic conditions was used to define multimorbidity. Multimorbidity was defined as having ≥1 long-term condition in addition to hypertension. Logistic regression models were used to estimate the association between multimorbidity (or individual long-term conditions) with uncontrolled BP. RESULTS: A total of 67 385 patients with hypertension were included. They had a mean age of 70, 53.1% were female, 80.6% had multimorbidity, and 35.7% had uncontrolled BP. Patients with multimorbidity had lower odds of uncontrolled BP than those without multimorbidity (adjusted OR = 0.72, 95% CI 0.68-0.76). Among the long-term conditions, diabetes (aOR = 0.73, 95%CI 0.70-0.77), heart failure (aOR = 0.81, 95%CI 0.73-0.91), ischemic heart disease (aOR = 0.74, 95%CI 0.69-0.79), schizophrenia (aOR = 0.79, 95%CI 0.65-0.97), depression/anxiety (aOR = 0.91, 95%CI 0.86-0.95), dementia (aOR = 0.87, 95%CI 0.80-0.95), and osteoarthritis (aOR = 0.89, 95%CI 0.85-0.93) were associated with a lower likelihood of uncontrolled BP. CONCLUSION: We found that multimorbidity was associated with better BP control. Several conditions were associated with better control, including diabetes, heart failure, ischemic heart disease, schizophrenia, depression/anxiety, dementia, and osteoarthritis.

Developing an Audit and Feedback Dashboard for Family Physicians: User-Centered Design Process.

BACKGROUND: Audit and feedback (A&F), the summary and provision of clinical performance data, is a common quality improvement strategy. Successful design and implementation of A&F-or any quality improvement strategy-should incorporate evidence-informed best practices as well as context-specific end user input. OBJECTIVE: We used A&F theory and user-centered design to inform the development of a web-based primary care A&F dashboard. We describe the design process and how it influenced the design of the dashboard. METHODS: Our design process included 3 phases: prototype development based on A&F theory and input from clinical improvement leaders; workshop with family physician quality improvement leaders to develop personas (ie, fictional users that represent an archetype character representative of our key users) and application of those personas to design decisions; and user-centered interviews with family physicians to learn about the physician's reactions to the revised dashboard. RESULTS: The team applied A&F best practices to the dashboard prototype. Personas were used to identify target groups with challenges and behaviors as a tool for informed design decision-making. Our workshop produced 3 user personas, Dr Skeptic, Frazzled Physician, and Eager Implementer, representing common users based on the team's experience of A&F. Interviews were conducted to further validate findings from the persona workshop and found that (1) physicians were interested in how they compare with peers; however, if performance was above average, they were not motivated to improve even if gaps compared to other standards in their care remained; (2) burnout levels were high as physicians are trying to catch up on missed care during the pandemic and are therefore less motivated to act on the data; and (3) additional desired features included integration within the electronic medical record, and more up-to-date and accurate data. CONCLUSIONS: We found that carefully incorporating data from user interviews helped operationalize generic best practices for A&F to achieve an acceptable dashboard that could meet the needs and goals of physicians. We demonstrate such a design process in this paper. A&F dashboards should address physicians' data skepticism, present data in a way that spurs action, and support physicians to have the time and capacity to engage in quality improvement work; the steps we followed may help those responsible for quality improvement strategy implementation achieve these aims.

Breast Cancer Screening Among Females With and Without Schizophrenia.

Importance: Breast cancer screening with mammography is recommended in Ontario, Canada, for females 50 years or older. Females with schizophrenia are at higher risk of breast cancer, but in Ontario it is currently unknown whether breast cancer screening completion differs between those with vs without schizophrenia and whether primary care payment models are a factor. Objective: To compare breast cancer screening completion within 2 years after the 50th birthday among females with and without schizophrenia, and to identify the association between breast cancer screening completion and different primary care payment models. Design, Setting, and Participants: This case-control study analyzed Ontario-wide administrative data on females with and without schizophrenia who turned 50 years of age between January 1, 2010, and December 31, 2019. Those with schizophrenia (cases) were matched 1:10 to those without schizophrenia (controls) on local health integration network, income quintile, rural residence, birth dates, and weighted Aggregated Diagnosis Group score. Data analysis was performed from November 2021 to February 2023. Exposures: Exposures were schizophrenia and primary care payment models. Main Outcomes and Measures: Outcomes included breast cancer screening completion among cases and controls within 2 years after their 50th birthday and the association with receipt of care from primary care physicians enrolled in different primary care payment models, which were analyzed using logistic regression and reported as odds ratios (ORs) and 95% CIs. Results: The study included 11 631 females with schizophrenia who turned 50 years of age during the study period and a matched cohort of 115 959 females without schizophrenia, for a total of 127 590 patients. Overall, 69.3% of cases and 77.1% of controls had a mammogram within 2 years after their 50th birthday. Cases had lower odds of breast cancer screening completion within 2 years after their 50th birthday (OR, 0.67; 95% CI, 0.64-0.70). Cases who received care from a primary care physician in a fee-for-service (OR, 0.57; 95% CI, 0.53-0.60) or enhanced fee-for-service (OR, 0.79; 95% CI, 0.75-0.82) payment model had lower odds of having a mammogram than cases whose physicians were paid under a Family Health Team model. Conclusions and Relevance: This case-control study found that, in Ontario, Canada, breast cancer screening completion was lower among females with schizophrenia, and differences from those without schizophrenia may partially be explained by differences in primary care payment models. Widening the availability of team-based primary care for females with schizophrenia may play a role in increased breast cancer screening rates.

Quality indices for topic model selection and evaluation: a literature review and case study.

BACKGROUND: Topic models are a class of unsupervised machine learning models, which facilitate summarization, browsing and retrieval from large unstructured document collections. This study reviews several methods for assessing the quality of unsupervised topic models estimated using non-negative matrix factorization. Techniques for topic model validation have been developed across disparate fields. We synthesize this literature, discuss the advantages and disadvantages of different techniques for topic model validation, and illustrate their usefulness for guiding model selection on a large clinical text corpus. DESIGN, SETTING AND DATA: Using a retrospective cohort design, we curated a text corpus containing 382,666 clinical notes collected between 01/01/2017 through 12/31/2020 from primary care electronic medical records in Toronto Canada. METHODS: Several topic model quality metrics have been proposed to assess different aspects of model fit. We explored the following metrics: reconstruction error, topic coherence, rank biased overlap, Kendall's weighted tau, partition coefficient, partition entropy and the Xie-Beni statistic. Depending on context, cross-validation and/or bootstrap stability analysis were used to estimate these metrics on our corpus. RESULTS: Cross-validated reconstruction error favored large topic models (K ≥ 100 topics) on our corpus. Stability analysis using topic coherence and the Xie-Beni statistic also favored large models (K = 100 topics). Rank biased overlap and Kendall's weighted tau favored small models (K = 5 topics). Few model evaluation metrics suggested mid-sized topic models (25 ≤ K ≤ 75) as being optimal. However, human judgement suggested that mid-sized topic models produced expressive low-dimensional summarizations of the corpus. CONCLUSIONS: Topic model quality indices are transparent quantitative tools for guiding model selection and evaluation. Our empirical illustration demonstrated that different topic model quality indices favor models of different complexity; and may not select models aligning with human judgment. This suggests that different metrics capture different aspects of model goodness of fit. A combination of topic model quality indices, coupled with human validation, may be useful in appraising unsupervised topic models.

Tolerability of bedtime diuretics: a prospective cohort analysis.

OBJECTIVES: We sought to validate, or refute, the common belief that bedtime diuretics are poorly tolerated due to nocturia. DESIGN: Prespecified prospective cohort analysis embedded within the randomised BedMed trial, in which hypertensive participants are randomised to morning versus bedtime antihypertensive administration. SETTING: 352 community family practices across 4 Canadian provinces between March 2017 and September 2020. PARTICIPANTS: 552 hypertensive patients (65.6 years old, 57.4% female) already established on a single once-daily morning antihypertensive and randomised to switch that antihypertensive to bedtime. Of these, 203 used diuretics (27.1% thiazide alone, 70.0% thiazide/non-diuretic combinations) and 349 used non-diuretics. INTERVENTION: Switching the established antihypertensive from morning to bedtime, and comparing the experience of diuretic and non-diuretic users. PRIMARY AND SECONDARY OUTCOME MEASURES: Primary outcome: Adherence to bedtime allocation time at 6 months (defined as the willingness to continue with bedtime use, not an assessment of missed doses). Secondary 6-month outcomes: (1) nocturia considered to be a major burden and (2) increase in overnight urinations/week. All outcomes were self-reported and additionally collected at 6 weeks. RESULTS: At 6 months: Adherence to bedtime allocation time was lower in diuretic users than non-diuretic users (77.3% vs 89.8%; difference 12.6%; 95% CI 5.8% to 19.8%; p<0.0001; NNH 8.0), and more diuretic users considered nocturia a major burden (15.6% vs 1.3%; difference 14.2%; 95% CI 8.9% to 20.6%; p<0.0001; NNH 7.0). Compared with baseline, diuretic users experienced 1.0 more overnight urinations/week (95% CI 0.0 to 1.75; p=0.01). Results did not differ between sexes. CONCLUSIONS: Switching diuretics to bedtime did promote nocturia, but only 15.6% found nocturia a major burden. At 6 months, 77.3% of diuretic users were adherent to bedtime dosing. Bedtime diuretic use is viable for many hypertensive patients, should it ever become clinically indicated. TRIAL REGISTRATION NUMBER: NCT02990663.

Identify and classify interprofessional primary care performance indicators: a scoping review protocol.

INTRODUCTION: Measuring the performance of interprofessional primary care is needed to examine whether this model of care is achieving its desired outcomes on patient care and health system effectiveness as well as to guide quality improvement initiatives. The aim of this scoping review is to map the literature on primary care performance measurement indicators to determine the extent to which current indicators capture or could be adapted to capture processes, outputs and outcomes that reflect interprofessional primary care. METHODS AND ANALYSIS: The review will be guided by the six-stage framework by Arksey and O'Malley (2005). MEDLINE, Embase, CINAHL, grey literature and the reference list of key studies will be searched to identify any study, published in English or French between 2000 and 2022, related to the concepts of performance indicators, frameworks, interprofessional teams and primary care. Two reviewers will independently screen all abstracts and full-text studies for inclusion. Eligible indicators will be classified according to process, output and outcome domains proposed by two validated frameworks. This study started in November 2022 and is expected to be completed by July 2023. ETHICS AND DISSEMINATION: This review does not require ethical approval. The results will be disseminated through a peer-reviewed publication, conference presentations and presentations to stakeholders.

The Influence of Age, Sex, and Socioeconomic Status on Glycemic Control Among People With Type 1 and Type 2 Diabetes in Canada: Patient-Led Longitudinal Retrospective Cross-sectional Study With Multiple Time Points of Measurement.

BACKGROUND: Clinical guidelines for most adults with diabetes recommend maintaining hemoglobin A1c (HbA1c) levels ≤7% (≤53 mmol/mol) to avoid microvascular and macrovascular complications. People with diabetes of different ages, sexes, and socioeconomic statuses may differ in their ease of attaining this goal. OBJECTIVE: As a team of people with diabetes, researchers, and health professionals, we aimed to explore patterns in HbA1c results among people with type 1 or type 2 diabetes in Canada. Our research question was identified by people living with diabetes. METHODS: In this patient-led retrospective cross-sectional study with multiple time points of measurement, we used generalized estimating equations to analyze the associations of age, sex, and socioeconomic status with 947,543 HbA1c results collected from 2010 to 2019 among 90,770 people living with type 1 or type 2 diabetes in Canada and housed in the Canadian National Diabetes Repository. People living with diabetes reviewed and interpreted the results. RESULTS: HbA1c results ≤7.0% represented 30.5% (male people living with type 1 diabetes), 21% (female people living with type 1 diabetes), 55% (male people living with type 2 diabetes) and 59% (female people living with type 2 diabetes) of results in each subcategory. We observed higher HbA1c values during adolescence, and for people living with type 2 diabetes, among people living in lower income areas. Among those with type 1 diabetes, female people tended to have lower HbA1c levels than male people during childbearing years but higher HbA1c levels than male people during menopausal years. Team members living with diabetes confirmed that the patterns we observed reflected their own life courses and suggested that these results be communicated to health professionals and other stakeholders to improve the treatment for people living with diabetes. CONCLUSIONS: A substantial proportion of people with diabetes in Canada may need additional support to reach or maintain the guideline-recommended glycemic control goals. Blood sugar management goals may be particularly challenging for people going through adolescence or menopause or those living with fewer financial resources. Health professionals should be aware of the challenging nature of glycemic management, and policy makers in Canada should provide more support for people with diabetes to live healthy lives.

Emulating a Target Trial Using Primary-Care Electronic Health Records: Sodium-Glucose Cotransporter 2 Inhibitor Medications and Hemoglobin A1c.

Substantial effort has been dedicated to conducting randomized controlled experiments to generate clinical evidence for diabetes treatment. Randomized controlled experiments are the gold standard for establishing cause and effect. However, due to their high cost and time commitment, large observational databases such as those comprised of electronic health record (EHR) data collected in routine primary care may provide an alternative source with which to address such causal objectives. We used a Canadian primary-care data repository housed at the University of Toronto (Toronto, Ontario, Canada) to emulate a randomized experiment. We estimated the effectiveness of sodium-glucose cotransporter 2 inhibitor (SGLT-2i) medications for patients with diabetes using hemoglobin A1c (HbA1c) as a primary outcome and marker for glycemic control from 2018 to 2021. We assumed an intention-to-treat analysis for prescribed treatment, with analyses based on the treatment assigned rather than the treatment eventually received. We defined the causal contrast of interest as the net change in HbA1c (percent) between the group receiving the standard of care versus the group receiving SGLT-2i medication. Using a counterfactual framework, marginal structural models demonstrated a reduction in mean HbA1c level with the initiation of SGLT-2i medications. These findings provided effect sizes similar to those from earlier clinical trials on assessing the effectiveness of SGLT-2i medications.

Using ICD-9 diagnostic codes for external validation of topic models derived from primary care electronic medical record clinical text data.

Background/Objectives: Unsupervised topic models are often used to facilitate improved understanding of large unstructured clinical text datasets. In this study we investigated how ICD-9 diagnostic codes, collected alongside clinical text data, could be used to establish concurrent-, convergent- and discriminant-validity of learned topic models. Design/Setting: Retrospective open cohort design. Data were collected from primary care clinics located in Toronto, Canada between 01/01/2017 through 12/31/2020. Methods: We fit a non-negative matrix factorization topic model, with K = 50 latent topics/themes, to our input document term matrix (DTM). We estimated the magnitude of association between each Boolean-valued ICD-9 diagnostic code and each continuous latent topical vector. We identified ICD-9 diagnostic codes most strongly associated with each latent topical vector; and qualitatively interpreted how these codes could be used for external validation of the learned topic model. Results: The DTM consisted of 382,666 documents and 2210 words/tokens. We correlated concurrently assigned ICD-9 diagnostic codes with learned topical vectors, and observed semantic agreement for a subset of latent constructs (e.g. conditions of the breast, disorders of the female genital tract, respiratory disease, viral infection, eye/ear/nose/throat conditions, conditions of the urinary system, and dermatological conditions, etc.). Conclusions: When fitting topic models to clinical text corpora, researchers can leverage contemporaneously collected electronic medical record data to investigate the external validity of fitted latent variable models.

Estimation of marginal structural models under irregular visits and unmeasured confounder: calibrated inverse probability weights.

Clinical information collected in electronic health records (EHRs) is becoming an essential source to emulate randomized experiments. Since patients do not interact with the healthcare system at random, the longitudinal information in large observational databases must account for irregular visits. Moreover, we need to also account for subject-specific unmeasured confounders which may act as a common cause for treatment assignment mechanism (e.g. glucose-lowering medications) while also influencing the outcome (e.g. Hemoglobin A1c). We used the calibration of longitudinal weights to improve the finite sample properties and to account for subject-specific unmeasured confounders. A Monte Carlo simulation study is conducted to evaluate the performance of calibrated inverse probability estimators using time-dependent treatment assignment and irregular visits with subject-specific unmeasured confounders. The simulation study showed that the longitudinal weights with calibrated restrictions improved the finite sample bias when compared to the stabilized weights. The application of the calibrated weights is demonstrated using the exposure of glucose lowering medications and the longitudinal outcome of Hemoglobin A1c. Our results support the effectiveness of glucose lowering medications in reducing Hemoglobin A1c among type II diabetes patients with elevated glycemic index ([Formula: see text]) using stabilized and calibrated weights.

Factors Associated With Initiation of Antidepressant Medication in Adults With Type 1 and Type 2 Diabetes: A Primary Care Retrospective Cohort Study in Ontario, Canada.

OBJECTIVES: Depression in patients with diabetes mellitus is common and associated with poorer outcomes. This study aims to identify demographic, socioeconomic and medical factors associated with the initiation of antidepressant medication after a diagnosis of diabetes in adult patients without a previous prescription for antidepressants. We also examined frequency of primary care visits in the year after antidepressant initiation compared with the year before treatment began. METHODS: This was a retrospective cohort study using routinely collected electronic medical record data spanning January 2011 to December 2019 from the University of Toronto Practice-based Research Network (UTOPIAN) Data Safe Haven. Our primary outcome was a first prescription for an antidepressant in patients with diabetes. We used a mixed-effects logistic regression model to identify sociodemographic and medical factors associated with this event. RESULTS: Among 22,750 patients with diabetes mellitus, 3,055 patients (13.4%) began taking an antidepressant medication. Increased odds of antidepressant initiation were observed in younger patients (odds ratio [OR], 1.77; 95% confidence interval [CI], 1.39 to 2.26), females (OR, 1.60; 95% CI, 1.46 to 1.7), those receiving insulin treatment (OR, 1.59; 95% CI, 1.43 to 1.78) and cases of polypharmacy (OR, 3.67; 95% CI, 3.29 to 4.11). There was an increase in the mean number of primary care visits from 4.6 to 5.9 per year after antidepressant initiation. CONCLUSIONS: In patients with diabetes, age, sex and medical characteristics were associated with the initiation of antidepressants. These patients accessed primary care more frequently. Screening and prevention of depression, particularly in these subgroups, could reduce its personal and systemic burdens.

Essential requirements for the governance and management of data trusts, data repositories, and other data collaborations.

Introduction: Around the world, many organisations are working on ways to increase the use, sharing, and reuse of person-level data for research, evaluation, planning, and innovation while ensuring that data are secure and privacy is protected. As a contribution to broader efforts to improve data governance and management, in 2020 members of our team published 12 minimum specification essential requirements (min specs) to provide practical guidance for organisations establishing or operating data trusts and other forms of data infrastructure. Approach and Aims: We convened an international team, consisting mostly of participants from Canada and the United States of America, to test and refine the original 12 min specs. Twenty-three (23) data-focused organisations and initiatives recorded the various ways they address the min specs. Sub-teams analysed the results, used the findings to make improvements to the min specs, and identified materials to support organisations/initiatives in addressing the min specs. Results: Analyses and discussion led to an updated set of 15 min specs covering five categories: one min spec for Legal, five for Governance, four for Management, two for Data Users, and three for Stakeholder & Public Engagement. Multiple changes were made to make the min specs language more technically complete and precise. The updated set of 15 min specs has been integrated into a Canadian national standard that, to our knowledge, is the first to include requirements for public engagement and Indigenous Data Sovereignty. Conclusions: The testing and refinement of the min specs led to significant additions and improvements. The min specs helped the 23 organisations/initiatives involved in this project communicate and compare how they achieve responsible and trustworthy data governance and management. By extension, the min specs, and the Canadian national standard based on them, are likely to be useful for other data-focused organisations and initiatives.

Comprehensiveness of care for women with depression: Association between primary care use and cervical cancer screening.

OBJECTIVE: To explore comprehensiveness of care in patients with depression by examining associations between a diagnosis of depression, frequency of primary care visits, and Papanicolaou test completion. DESIGN: Cross-sectional retrospective survey using electronic medical record data from the Canadian Primary Care Sentinel Surveillance Network. SETTING: Primary care practices in Ontario. PARTICIPANTS: Women aged 21 to 69 eligible to receive Pap tests in 2015. MAIN OUTCOME MEASURES: Associations between 2 predictors (depression and number of primary care visits in 2015) and Pap test completion were measured. RESULTS: Overall, 125,258 women were included: 20.5% completed a Pap test and 16.4% had a diagnosis of depression. Having a diagnosis of depression was associated with lower likelihood of Pap test completion (adjusted odds ratio [AOR]=0.92, 95% CI 0.88 to 0.95). A greater number of primary care visits was associated with a higher likelihood of Pap test completion; this association was stronger in women with a diagnosis of depression (AOR=4.9, 95% CI 4.16 to 5.69) than in those without (AOR=3.4, 95% CI 3.25 to 3.60). CONCLUSION: While depression was associated with fewer completed Pap tests, women with depression who saw their family doctors more often were more likely to be screened for cervical cancer. More primary care visits for depression treatment may be associated with an improved likelihood of screening for cervical cancer.

Socioeconomic and immigration status and COVID-19 testing in Toronto, Ontario: retrospective cross-sectional study.

BACKGROUND: Preliminary evidence suggests that individuals living in lower income neighbourhoods are at higher risk of COVID-19 infection. The relationship between sociodemographic characteristics and COVID-19 risk warrants further study. METHODS: We explored the association between COVID-19 test positivity and patients' socio-demographic variables, using neighborhood sociodemographic data collected retrospectively from two COVID-19 Assessment Centres in Toronto, ON. RESULTS: Eighty-three thousand four hundred forty three COVID-19 tests completed between April 5-September 30, 2020, were analyzed. Individuals living in neighbourhoods with the lowest income or highest concentration of immigrants were 3.4 (95% CI: 2.7 to 4.9) and 2.5 (95% CI: 1.8 to 3.7) times more likely to test positive for COVID-19 than those in highest income or lowest immigrant neighbourhoods, respectively. Testing was higher among individuals from higher income neighbourhoods, at lowest COVID-19 risk, compared with those from low-income neighbourhoods. CONCLUSIONS: Targeted efforts are needed to improve testing availability in high-risk regions. These same strategies may also ensure equitable COVID-19 vaccine delivery.

Health system use among patients with mental health conditions in a community based sample in Toronto, Canada: A retrospective cohort study.

OBJECTIVE: To identify hospital and primary care health service use among people with mental health conditions or addictions in an integrated primary-secondary care database in Toronto, Ontario. METHOD: This was a retrospective cohort study of adults with mental health diagnoses using data from the Health Databank Collaborative (HDC), a primary care-hospital linked database in Toronto. Data were included up to March 31st 2019. Negative binomial and logistic regression were used to evaluate associations between health care utilization and various patient characteristics and mental health diagnoses. RESULTS: 28,482 patients age 18 or older were included. The adjusted odds of at least one mental health diagnosis were higher among younger patients (18-30 years vs. 81+years aOR = 1.87; 95% CI:1.68-2.08) and among female patients (aOR = 1.35; 95% CI: 1.27-1.42). Patients with one or more mental health diagnoses had higher adjusted rates of hospital visits compared to those without any mental health diagnosis including addiction (aRR = 1.74, 95% CI: 1.58-1.91) and anxiety (aRR = 1.28, 95% CI: 1.23-1.32). 14.5% of patients with a psychiatric diagnosis were referred to the hospital for specialized psychiatric services, and 38% of patients referred were eventually seen in consultation. The median wait time from the date of referral to the date of consultation was 133 days. CONCLUSIONS: In this community, individuals with mental health diagnoses accessed primary and hospital-based health care at greater rates than those without mental health diagnoses. Wait times for specialized psychiatric care were long and most patients who were referred did not have a consultation. Information about services for patients with mental health conditions can be used to plan and monitor more integrated care across sectors, and ultimately improve outcomes.

Marginal Structural Models Using Calibrated Weights With SuperLearner: Application to Type II Diabetes Cohort.

As different scientific disciplines begin to converge on machine learning for causal inference, we demonstrate the application of machine learning algorithms in the context of longitudinal causal estimation using electronic health records. Our aim is to formulate a marginal structural model for estimating diabetes care provisions in which we envisioned hypothetical (i.e. counterfactual) dynamic treatment regimes using a combination of drug therapies to manage diabetes: metformin, sulfonylurea and SGLT-2i. The binary outcome of diabetes care provisions was defined using a composite measure of chronic disease prevention and screening elements [27] including (i) primary care visit, (ii) blood pressure, (iii) weight, (iv) hemoglobin A1c, (v) lipid, (vi) ACR, (vii) eGFR and (viii) statin medication. We used several statistical learning algorithms to describe causal relationships between the prescription of three common classes of diabetes medications and quality of diabetes care using the electronic health records contained in National Diabetes Repository. In particular, we generated an ensemble of statistical learning algorithms using the SuperLearner framework based on the following base learners: (i) least absolute shrinkage and selection operator, (ii) ridge regression, (iii) elastic net, (iv) random forest, (v) gradient boosting machines, and (vi) neural network. Each statistical learning algorithm was fitted using the pseudo-population generated from the marginalization of the time-dependent confounding process. Covariate balance was assessed using the longitudinal (i.e. cumulative-time product) stabilized weights with calibrated restrictions. Our results indicated that the treatment drop-in cohorts (with respect to metformin, sulfonylurea and SGLT-2i) may have improved diabetes care provisions in relation to treatment naïve (i.e. no treatment) cohort. As a clinical utility, we hope that this article will facilitate discussions around the prevention of adverse chronic outcomes associated with type II diabetes through the improvement of diabetes care provisions in primary care.

Primary Care Severe Asthma Registry and Education Project (PCSAR-EDU): Phase 1 - an e-Delphi for registry definitions and indices of clinician behaviour.

INTRODUCTION: Although most asthma is mild to moderate, severe asthma accounts for disproportionate personal and societal costs. Poor co-ordination of care between primary care and specialist settings is recognised as a barrier to achieving optimal outcomes. The Primary Care Severe Asthma Registry and Education (PCSAR-EDU) project aims to address these gaps through the interdisciplinary development and evaluation of both a 'real-world' severe asthma registry and an educational programme for primary care providers. This manuscript describes phase 1 of PCSAR-EDU which involves establishing interdisciplinary consensus on criteria for the: (1) definition of severe asthma; (2) generation of a severe asthma registry and (3) definition of an electronic-medical record data-based Clinician Behaviour Index (CBI). METHODS AND ANALYSIS: In phase 1, a modified e-Delphi activity will be conducted. Delphi panellists (n≥13) will be invited to complete a 30 min online survey on three separate occasions (i.e., three separate e-Delphi 'rounds') over a 3-month period. Expert opinion will be collected via an open-ended survey ('Open' round 1) and 5-point Likert scale and ranking surveys ('Closed' round 2 and 3). A fourth and final Delphi round will occur via synchronous meeting, whereby panellists approve a finalised ideal 'core criteria list', CBI and corresponding item weighting. ETHICS AND DISSEMINATION: Ethical approval has been obtained for the activities involved in phase 1 from the University of Toronto's Human Research Ethics Programme (approval number 39695). Future ethics approvals will depend on information gathered in the proceeding phase; thus, ethical approval for phase 2 and 3 of this study will be sought sequentially. Findings will be disseminated through conference presentations, peer-reviewed publications and knowledge translation tools.

Improving Health Equity by Screening for Poverty: A Survey of Family Physician Screening Behaviors and Perceptions in Toronto, Canada.

Purpose: Given the importance of socioeconomic status in both directly and indirectly influencing one's health, "poverty screening" by family physicians (FPs) may be one viable option to improve patient health. However, rates of screening for poverty are low, and reported barriers to screening are numerous. This study sought to collate and investigate reasons for refraining from screening among FPs, many of whom had opted into a Targeted Poverty Screening (TPS) Program, to be able to enhance uptake of the intervention. The TPS Program is a "targeted screening and referral process," whereby medical charts of adult patients residing in "deprived neighborhoods," as determined by postal code, were flagged for screening for FPs who elected to partake in the program. Methods: A survey containing 15 questions was developed through an iterative process with pilot-testing by faculty physicians. The survey was administered to FPs registered in the North York Family Health Team (NYFHT) using Qualtrics© research software. Results: Half of the respondents (n=19/38; 50%) indicated that they enrolled in the TPS program. Irrespective of enrollment in the TPS Program, the majority of respondents (n=31/38; 81.6%) stated that they elect to screen their patients for poverty using the evidence-based question of "do you have difficulty making ends meet at the end of the month?." Among those not enrolled in the program, 84.2% (n=16/19) of respondents indicated that they screened their patients for poverty and 15.8% (n=3/19) indicated they did not. Among respondents who said they did not screen (n=7/38; 18.4%), the reasons for not screening patients were as follows: forgot (n=2; 28.6%); time constraints/feel uncomfortable asking (n=1; 14.3%); and "feel I know patients well" (n=1; 14.3%). For the remaining respondents, a nurse or locum did the screening as part of a periodic health review (i.e., patient was screened, but not by the FP completing the survey (n=3). Conclusion: This study yielded numerous insights, such as barriers faced by FPs in undertaking poverty screening that differs from the literature. The findings suggest that (1) barriers faced by FPs in poverty screening can be mitigated, (2) there is a need to integrate screening into routines and normalize the activity, and (3) there is a need for enhanced training to support patients of lower socioeconomic status.