Article Text
PDF
Abstract
Objectives The aim of this study was to investigate physician benzodiazepine (BZD) self-use pre-COVID-19 pandemic and to examine changes in BZD self-use during the first year of the pandemic.
Design Population-based retrospective cohort study using linked routinely collected administrative health data comparing the first year of the pandemic to the period before the pandemic.
Setting Province of Ontario, Canada between March 2016 and March 2021.
Intervention Onset of the COVID-19 pandemic in March 2020.
Outcomes measures The primary outcome measure was the receipt of one or more prescriptions for BZD, which was captured via the Narcotics Monitoring System.
Results In a cohort of 30 798 physicians (mean age 42, 47.8% women), we found that during the year before the pandemic, 4.4% of physicians had 1 or more BZD prescriptions. Older physicians (6.8% aged 50+ years), female physicians (5.1%) and physicians with a prior mental health (MH) diagnosis (12.4%) were more likely than younger (3.7% aged <50 years), male physicians (3.8%) and physicians without a prior MH diagnosis (2.9%) to have received 1 or more BZD prescriptions. The first year of the COVID-19 pandemic was associated with a 10.5% decrease (adjusted OR (aOR) 0.85, 95% CI: 0.80 to 0.91) in the number of physicians with 1 or more BZD prescriptions compared with the year before the pandemic. Female physicians were less likely to reduce BZD self-use (aORfemale=0.90, 95% CI: 0.83 to 0.98) compared with male physicians (aORmale=0.79, 95% CI: 0.72 to 0.87, pinteraction=0.046 during the pandemic. Physicians presenting with an incident MH visit had higher odds of filling a BZD prescription during COVID-19 compared with the prior year.
Conclusions Physicians’ BZD prescriptions decreased during the first year of the COVID-19 pandemic in Ontario, Canada. These findings suggest that previously reported increases in mental distress and MH visits among physicians during the pandemic did not lead to greater self-use of BZDs.
- EPIDEMIOLOGY
- Health policy
- MEDICAL EDUCATION & TRAINING
- PUBLIC HEALTH
Data availability statement
Data-sharing agreements prohibit ICES from making the data set publicly available. However, access may be granted at the following website https://www-ices-on-ca.proxy.bib.uottawa.ca/DAS to those who meet pre-specified criteria for confidential access.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
We used a population-level cohort of all practising physicians in Ontario who registered between 1990 and 2018 (N=30 798).
We captured all prescriptions for BZDs filled in Ontario through a validated monitoring system.
Our study included outcomes before and during the COVID-19 pandemic, allowing us to quantify pandemic-associated changes.
The monitoring system does not include data on other non-BZD medications used for mental health concerns such as antidepressants.
Our data does not indicate whether a medication was actually taken by the recipient (ie, medication prescription vs use).
Introduction
Even before the start of the COVID-19 pandemic, multiple studies and surveys have documented high levels of self-reported stress, burnout and mental health (MH) and substance use disorders among physicians in Canada and internationally.1–5 As the pandemic emerged, there were immediate concerns about the potential negative impact on the MH of physicians due to added occupational stressors such as fear of infection, possibility of moral injury and increased workloads.6 7
A growing body of international evidence has demonstrated the validity of these concerns, with surveys of physicians finding increases in self-reported symptoms of burnout, mental distress and insomnia.8–11 A national survey of Canadian physicians (n=1300) in late 2020 found that the majority of respondents reported being extremely stressed most days during COVID-19.12 Two surveys of physicians in Ontario, Canada, conducted in March 2020 (n=1407) and a year later in March 2021 (n=2649) found that the proportion of respondents reporting feeling ‘completely burned out’ increased from 10.6% to 14.0%.13 Moreover, several studies have found that physicians directly caring for patients infected with COVID-19 have reported higher increases in anxiety and depressive symptoms compared with those not directly involved.8 10
Corresponding with these reports of MH symptoms among physicians, emerging data suggests that there has been an increase in the self-use of MH services by physicians. We recently used provincial health administrative data to detect that rates of outpatient mental healthcare visits increased by 27% during the first year of the COVID-19 pandemic compared with the previous year among 34 000 physicians in Ontario, Canada.14 Previous research has also reported short-term increases in benzodiazepine (BZD) use in the general population following mass trauma events.15–18 To date, there has been conflicting evidence on changes in BZD use in the general population since the start of the COVID-19 pandemic with studies from the USA suggesting increases in use, while data from Canada suggesting no increase in BZD prescriptions during the first 10 months of the pandemic.19–21 Consequently, it is unclear whether pandemic-related MH symptoms have contributed to an increase in BZD self-use in the physician population. In addition, it is unclear how common BZD self-use among physicians was pre pandemic.
To address these gaps, we used health administrative data to examine population-level changes in self-use of BZD among Canadian physicians during the first year of the COVID-19 pandemic compared with the prior year. In addition, we examined prepandemic predictors of self-use of BZD in physicians. We hypothesised that there would be an increase in the number of physicians who filled a BZD prescription during the pandemic compared with pre pandemic in response to pandemic-related stress impacting anxiety and insomnia.
Methods
Study design and setting
We conducted a cohort study of practising physicians in Ontario, Canada, using linked health administrative data from the province’s universal healthcare system. We used data from the College of Physicians and Surgeons of Ontario (CPSO) to identify 45 835 physicians who registered to practise medicine in Ontario between 1990 and 2018. Physicians were assessed for cohort eligibility between 1 March 2016 and 31 March 2021 and were excluded from follow-up during time periods in which they were not living in Ontario, eligible for the provinces universal healthcare system—the Ontario Health Insurance Plan (OHIP), or deceased (n=11 043 excluded from entire study). We used unique coded deidentified physician data obtained from the CPSO to link physicians to their healthcare records including filling a prescription for a controlled medication, held at ICES (formerly known as the Institute for Clinical and Evaluative Sciences). The publically available website https://datadictionary.ices.on.ca/Applications/DataDictionary/ contains a description of databes housed at ICES. See online supplemental table 1 for additional methodological details on data linkage, cohort creation and privacy protection.
Supplemental material
Our study time frame captured the first wave of COVID-19 and state of emergency in Ontario (March 2020 to May 2020), along with a phased reopening during a period of declining and then relatively low case and COVID-19 hospitalisation incidence (June 2020 to September 2020). This period was followed by a second wave of increasing cases and hospitalisations starting in mid-September 2020, which continued until March 2021.22 During our study time frame, 0.9% of physicians in Ontario tested positive for COVID-19.23 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting observational studies.
Patient and public involvement
No patient involved.
Outcome
Our primary outcome was the proportion of physicians who filled one or more prescriptions for a short-acting or long-acting BZD captured by the Narcotics Monitoring System (NMS). We did not include non-BZD hypnotic medications such as zopiclone, as the NMS does not fully capture these medications (eg, see online supplemental table 2 for a full list of BZDs captured in this study).
The NMS captures all prescriptions for all narcotics and other controlled drugs dispensed in Ontario including stimulants, opioids, BZDs and barbiturates, regardless of payment type (eg, private insurance vs public drug programme or out of pocket). The NMS is routinely used to monitor prescription drug dispensing at the population level in Ontario and has been validated has having an error rate less than 1%.19 Our data sources did not have access to population-level prescriptions of other psychotropic medications of interest (eg, antidepressants, non-BZD sleep medications) and we did not include these as outcomes in our study.
Physician subgroups and covariates
We decided a priori to examine changes in BZD prescriptions among physicians by key characteristics of interest. We examined changes between men and women, age (<50 compared with 50 years or older) and physicians living in an urban (>100 000 residents) or rural (<100 000 residents) setting.24 We compared changes among physicians with and without a prior MH history (defined as one or more MH and substance use (MHSU)-related outpatient, emergency department (ED) or hospitalisation encounters in the prior 2 years prior to index data).14 We then compared BZD prescriptions among physicians who did or did not have an incident MHSU visit (defined as having a new MHSU visit with no prior MHSU visits in prior 2 years). Finally, we compared changes among physicians who did or did not provide any direct acute care (in the ED or inpatient setting; ‘front line’) for patients with suspected or confirmed COVID-19 using physician billing location and COVID diagnostic codes during the first year of COVID-19. In terms of physician COVID-19 infections, we did not include this as a covariate as our team has previously explored physician COVID-19 infections during the first year of the pandemic and found that less than 1% of Ontario physicians tested positive for COVID-19.23 For this analysis, physician residents and fellows were excluded from the denominator because they do not bill OHIP and consequently we could not determine who they provided care to. We also obtained physician specialty from the CPSO registration and the ICES Physician Database. We used specialty to characterise our cohort and for adjustment in analysis but decided a priori to not examine differences in BZD self-use by specialty.
Statistical analysis
We plotted the observed data to visualise trends over time before and during the first year of the pandemic. We used an autoregressive integrated moving average model applied to the monthly proportion of physicians with one or more BZD prescriptions to forecast the expected proportion of physicians with a BZD prescription during COVID-19 in the absence of the pandemic (13 monthly intervals) along with 95% CIs as based on 47 monthly intervals pre COVID-19. The denominator was the number of physicians alive and eligible for OHIP coverage during each month. We selected the best-fitting model, specified as (0,0,0) x (0,1,1)26(1,0,0), based on the lowest values for the Akaike information criterion and presented the results graphically. No seasonal trend was identified, and thus seasonality was not accounted for.
We conducted logistic regression models with generalised estimating equations using the individual physician as the unit of analysis to examine changes in BZD prescriptions during the pandemic. Physicians were compared with themselves if they were eligible in both time periods and to the physician average in the other time period if only eligible during COVID-19 or pre COVID-19. The dependent variable was a binary indicator for BZD prescription (yes/no) in each time period of interest (pre vs post pandemic). We defined the time periods using a binary indicator to represent the pre-COVID (11 March 2019 to 10 March 2020) and post-COVID periods (11 March 2020 to 10 March 2019). To understand if physician’s characteristics modify the relationship between changes in BZD self-use during the COVID-10 pandemic, we included an interaction term between the COVID-19 variable and each physician characteristic to explore differences across the time periods. We specified separate models for each interaction.
Models were adjusted for age (continuous), sex, rurality, medical specialty, prior MH history and exposure to COVID-19 patients (binary indicator).
Results
Table 1 summarises the characteristics of 30 798 physicians who met the study eligibility criteria during the year before the pandemic (11 March 2019 to 10 March 2020) or the first year of the pandemic 11 March 2020 to 10 March 2021). In both years, 52% of the cohort were men, the mean age was 42.8 years, 95% of physicians lived in an urban setting and 16.9% of physicians had 1 or more MH visits in the 2 prior years. The proportion of physicians who treated 1 or more patients with COVID-19 in an ED or inpatient setting during the first wave of the pandemic was 11.2% (2898/25 891).
Characteristics of the physician population in first before COVID-19 and during the first year of COVID-19*
Figure 1 shows the observed and forecasted trends in the proportions of physicians who filled at least one BZD prescription month by month from March 2016 to March 2021. During the first month of the pandemic, there was a 31% increase in the proportion of physicians who filled a BZD prescription compared with the prior year average. During the next 11 months, the proportions of physicians with a BZD prescription were consistent with forecasted proportions based on historic data.
Autoregressive integrated moving average (ARIMA) model showing predicted versus observed monthly proportion of physicians who received one or more benzodiazepine prescriptions between 1 March 2016 and 10 March 2021. The vertical grey line represents the declaration of the COVID-19 pandemic on 11 March 2020. The forecasted numbers and 95% CIs were generated from an ARIMA model, which was specified as (0,0,0) x (0,1,1)26(1,0,0).
Tables 2 and 3 compare the number and crude proportion of physicians who filled at least one BZD prescription and the odds of filling a prescription, respectively, during the year before and the first year of the pandemic. Pre pandemic, there were little differences in the proportion of physicians with 1 or more BZD prescriptions by sex (4.5% female, 5.1% male), or rurality (4.5% urban, 4.1% rural). However, a greater proportion of older physicians filled BZD prescriptions compared with younger physicians (3.9% aged <50 vs 6.8% aged >50). Physicians with a prior MH visit in the past 2 years were much more likely to fill 1 or more BZD prescriptions compared with those without (12.4% prior MH, 2.9% no prior MH). Overall, 52% of physicians who filled a BZD prescription did not have a recorded outpatient or acute care visit with an MH diagnosis in the prior 2 years.
Number of physicians with one or more benzodiazepine prescriptions during the year before and during COVID-19
Adjusted ORs for physician benzodiazepine prescriptions by time period
During the first year of the pandemic, 4.0% of physicians filled at least one BZD prescription, a 10.5% decrease compared with 4.4% in the year before the pandemic (adjusted OR 0.85, 95% CI: 0.80 to 0.91, p<0.0001). There was no significant difference in the decrease in filling a prescription between older and younger physicians (pinteraction=0.193), rural and urban physicians (pinteraction=0.077) and physicians who cared for one or more patients with COVID-19 and those who did not (pinteraction=0.764). During COVID-19, men were significantly less likely to fill a BZD prescription compared with women (aORfemale=0.90, 95% CI: 0.83 to 0.98 vs aORmale=0.79, 95% CI: 0.72 to 0.87, pinteraction=0.046). In addition, physicians presenting with an incident MH visit during COVID-19 had higher odds of filling a BZD prescription compared with physicians with an incident MH visit in the year prior to COVID-19 (aOR=1.20, 95% CI: 0.92 to 1.56). This increase was significant when compared with decreases observed in physicians who did not have an incident MH visit (aOR=0.84, 95% CI: 0.79 to 0.90, pinteraction=0.004).
Table 4 displays the type of BZD prescriptions self-used by physicians during the study. In the year before COVID-19, the majority (69.8%) had a prescription for lorazepam, followed by 21.8% of physicians who had a prescription for clonazepam. A lower proportion of physicians had a prescription for diazepam (4.2%), alprazolam (4.0%) or another BZD (6.5%). During the first year of COVID-19, the proportion of physicians with a lorazepam prescription decreased to 66.6%, while the proportion of physicians with a clonazepam prescription increased to 27.3%. The total number of pills dispensed for the two most commonly prescribed BZDs increased by 5.4% during the first year of the pandemic.
Number of physicians with one or more BZD prescriptions by type and quantity during the year before and during COVID-19
Discussion
In this population-level study of over 30 000 practising physicians in Ontario, we found that almost 1 in 20 physicians (4.4%) in the year prior to the pandemic filled a BZD prescription. Physicians over 50 years old were more likely to fill a prescription (6.8%) than physicians younger than 50 (3.9%), as were physicians with an MH visit in the previous 2 years (12.4%) compared with physicians without an MH visit (2.8%). We found that the COVID-19 pandemic was associated with a 10.5% decrease in the proportion of physicians filling a BZD prescription compared with the year prior. After adjusting for demographic and physician characteristics, we found that physicians on average had a 15% (95% CI: 9% to 20%) lower odds of filling a BZD prescription during the pandemic. We did not find signficant differences in the changes of self-use of BZDs during COVID-19 among physicians by age, rurality, or by whether physicians were directly caring for patient with COVID-19 in acute care settings. We observed that women were more likely to continue BZD self-use during the pandemic compared with men. In addition, physicians with an incident outpatient visit for a mental concern issue during the pandemic were more likely to receive a BZD prescription than physicians with an incident outpatient visit for a mental concern prior to the pandemic.
Despite large increases in self-reported anxiety, depression and stress among physicians13 and documented increases in outpatient MH care,14 physicians in Ontario filled fewer prescriptions for BZD during the pandemic than during the previous year. There are a number of possible explanations for these findings. First, these findings may speak to the resiliency of physicians who during a period of intense challenge did not require an increase in the self-use of sedatives or antianxiety medications. Second, BZDs are generally not considered first-line medications for the treatment of stress-related symptoms and physicians as patients may not have wanted BZDs and as providers may have not prescribed them for COVID-19-related stressors.
These findings are consistent with prescriptions for the general population in Ontario who similarly did not fill more BZD prescriptions than expected during the first 10 months of the pandemic.25 The one increase in BZD prescriptions in our study during March 2020 likely reflects the impact of policies enacted in Ontario at the start of the pandemic to limit medication dispensing to 30 days or less due to concerns of drug shortages rather than an increase in new prescriptions.26 Collectively, our results suggest Canadian physicians were not widely prescribing BZDs due to presumed pandemic distress or MH concerns. Further work identifying the type of care that physicians did receive when seeking health services during COVID-19 (eg, antidepressant medications or counselling services) is required.
We noted important trends related to overall BZD self-use in physician’s pre pandemic. First, physician BZD self-use pre pandemic (4.4%) was lower compared with the general population of Ontario in 2019 (5.5%). Second, male and female physicians had relatively similar levels of BZD self-use, despite a twofold higher rate of MH visits in female physicians.14 27 There are several possible explanations for this finding including prescribing patterns by physicians based on the sex of the patient (eg, gender based biases in terms of symptom evaluation and prescribing) and the acceptance of BZD prescriptions by male versus female patients (eg, greater acceptance of BZD self-use by male physicians). Third, older physicians were two times as likely to self-use a BZD than younger physicians. This practice likely reflects a trend of fewer of individuals taking BZDs in Ontario over time.27 Fourth, over half of physicians who filled a BZD prescription in the year before COVID-19 lacked an outpatient or acute care visit related to MH or substance use (including codes for sleep disturbances) in the past 2 years. Some of these prescriptions are likely for primary insomnia, other non-MH-related sleep disturbances or neurological conditions (eg, seizures). However, these finding may reflect a reluctance of physicians to diagnose other physicians with MH concerns due to stigma or potential licensing implications, a lack of appropriate self-use of MH services by physicians or both. Our findings raise concern regarding BZD self-use in the physician population as they are not optimal for long-term use and may lead to dependency. Evidence-based alternatives, targeting the potential root cause for their use, may be more suitable and be prescribed with appropriate MH visits and follow-up.
Strengths and limitations
Our study has a number of strengths including following over 30 000 physicians over time and capturing all BZD prescriptions in this population. However, our study has several limitations. First, our data does not capture changes in other MH-related prescription medications (eg, Selective serotonin reuptake inhibitors) and medications for chronic conditions not-related to MH among physicians. Consequently, while we are confident that the pandemic was not associated with an increase in BZD self-use, we cannot contextualise this change against other drugs or exclude increases in other MH-related medications such as antidepressant, anxiolytic medications or non-BZD sleep medications. Second, our study only examined the first 12 months of the pandemic and it is possible that physician MH and consequent BZD self-use were impacted in later stages of the pandemic, particularly with the recent Omicron variant wave, which has placed large strain on the health system. Finally, indications for BZD prescription were unknown (eg, sleep aid, antianxiety) and further research could investigate in greater detail physician-level predictors of BZD prescriptions.
Conclusions
Overall physician BZD self-use decreased in Ontario during the first 12 months of the pandemic. These findings suggest that previously documented increased mental distress and consequent MH visits among physicians did not lead to greater self-use of BZD. Importantly, physicians with an incident MH concern during the pandemic were more likely to receive a BZD prescription, which may suggest that some physicians had high mental distress during the pandemic. Future research should focus on longer-term changes in BZD use during the pandemic and explore in greater details risk factors for prepandemic patterns of BZD self-use among physicians.
Data availability statement
Data-sharing agreements prohibit ICES from making the data set publicly available. However, access may be granted at the following website https://www-ices-on-ca.proxy.bib.uottawa.ca/DAS to those who meet pre-specified criteria for confidential access.
Ethics statements
Patient consent for publication
Ethics approval
This project was conducted under section 45 of Ontario’s Personal Health Information Protection Act and approved by ICES’s Privacy and Legal Office (TRIM 2021 0901 284 000). ICES is an independent, non-profit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyse healthcare and demographic data for health system evaluation and improvement.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
TK and PT are joint senior authors.
Twitter @cmilani_
Contributors DM and TK conceived the study idea. All authors contributed to the study design. MP gathered the data and performed data analyses. All authors interpreted the data analyses. DM and CM cowrote and revised the manuscript for intellectual content based on feedback from all authors. JMH, MS and CEK provided clinical interpretations of the study data. All authors provided their final approval for manuscript submission. All authors agree to be accountable for all aspects of the work. TK and PT contributed equally as cosenior authors. DM accepts full responsibility for the work and/or the conduct of the study, had access to the data, and controlled the decision to publish.
Funding This work was supported by the University of Ottawa site of ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and Ministry of Long-Term Care (MLTC). This study also received funding from: The Canadian Institutes for Health Research (CIHR) Operating Grant ‘Protecting and improving the mental health of physicians during and after the COVID-19 pandemic’ (Grant number MS3-173107); and grant support from the Academic Medical Organization of Southwestern Ontario.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.