Physician Health Care Visits for Mental Health and Substance Use During the COVID-19 Pandemic in Ontario, Canada

Key Points Question Has the incidence of physicians seeking outpatient care for mental health and substance use changed during the COVID-19 pandemic? Findings In a cohort study of 34 055 physicians, the rate of outpatient visits for mental health and substance use increased on average by 13% per physician during the first 12 months of the pandemic compared with the prior 12 months. Meaning These findings suggest that the COVID-19 pandemic is associated with greater mental health services use among physicians.

This supplementary material has been provided by the authors to give readers additional information about their work.

eMethods 1. Cohort and Data Sources Information Physician Linkage between CPSO and ICES
Physicians were linked to health care visits using unique, encoded identifiers from the CPSO. Deterministic followed by probabilistic linkage (based on name, date of birth, and sex) was performed by a small, specialized group at ICES (formerly known as the Institute for Clinical and Evaluative Sciences). All identifying information was removed before data were sent to the study team. This is done to mitigate any privacy breaches. ICES is an independent, non-profit research institute that houses routinely collected health data from Ontario's publicly funded health care system. ICES is a prescribed entity under section 45 of Ontario's Personal Health Information Protection Act. Section 45 authorizes ICES to collect personal health information, without consent, for the purpose of health system evaluation and improvement. Projects conducted under section 45, by definition, do not require review by a specific institutional research ethics board. This project was conducted under section 45, and approved by ICES's Privacy and Legal Office.

Data Sources
Physician demographic and speciality information was collected through the CPSO Registration Database and from the ICES Physicians Database. Data on physician characteristics and health care use were obtained through the following linked databases at ICES: 1) the Ontario Registered Persons Database, which captured demographic information including age and sex; 2) the postal code conversion file which contains information on the rurality of an individual's home address; 3) the OHIP Claims Database, which captured all outpatient claims for mental health visits, including virtual visits, in Ontario; and 4) the National Ambulatory Care Reporting System (NACRS), Ontario Mental Health Reporting System (OMHRS), and Discharge Abstract Database (DAD), which were used to capture acute care use as part of the definition for past history of mental health or substance use.

eTable 1. Definition of MHSU Visits and List of OHIP Codes
List of all mental health and substance use (MHSU) related OHIP billing and fee codes used to determine outpatient visits. We identified MHSU visits using the methods from the Mental Health and Addictions Scorecard and Evaluation Framework indicator (MHASEF). We telemedicine and virtual care visits when a physicians' OHIP claim included either 1) a mental health-related diagnostic or fee code in combination with a telemedicine flag or 2) a mental health-related diagnostic and a corresponding virtual fee code. Substance abuseextended assessment

Mental Health and Substance Use Codes Description of Code
To avoid double-counting, patients with multiple mental health-related claims from one provider on the same day only contributed to one visit. However, we allowed patients who saw multiple providers in one day to contribute one visit per provider (i.e., a patient who saw a primary care physician for a mental health-related reason and a psychiatrist on the same day would contribute two visits). We excluded tobacco use (OHIP diagnostic code 305) and mental health codes related to pediatric presentations (OHIP Diagnostic Codes 313, 314, 315, and 319) from the MHASEF indicator.

eMethods 2. ARIMA Model Building Summary General explanation:
We used the ARIMA procedure in SAS to produce descriptive output to identify an ARIMA model to forecast expected rates. We only used data up to the point of forecasting to build the model. We used time series output to determine the need for an ARIMA model and the differencing required to stabilize the series. We used autocorrelation function (ACF) and partial autocorrelation function (PACF) plots, and white noise autocorrelation tests to guide the selection of model terms. Terms were retained if they were significant. We identified better performing models using AIC. Model building stopped when there was no indication that further terms needed to be added (e.g. no patterns in the ACF/PACF and no significant autocorrelations).

Model 1:
We fit an ARIMA model to the biweekly rates of physician mental health visits (per 1,000 physicians) and compared the predicted rates post COVID to observed values. Model building steps: • Step 1: identify if the series is stationary. o Time series plot show a seasonal trendsuggests series should be seasonally differenced. o Non-seasonal autoregressive term (p=1) and non-seasonal moving average term (q=1) are non-significant when added to the model and did not improve model AIC. o White noise probabilities at higher order lags were approaching significancesuggests a seasonal AR or MA term could improve the model.