Association between substance use and professional outcomes among medical students: Findings from a Canadian cross-sectional survey

Abstract

Background:

While substance use is common among medical students, there is limited research on this topic involving Canadian medical students or exploring its associations with professional outcomes. The present study examines the association between Canadian medical students’ substance use and related counselling attitudes and practices, career satisfaction, academic/clinical workload, and the medical school environment.

Methods:

We sent an electronic cross-sectional survey to students attending all 17 Canadian medical schools between November 2015 and March 2016. A total of 4,438 participants completed the survey across four years of study, with a participation rate of 40.2%. We considered four categories of substance use: cannabis, alcohol, non-medical use of prescription stimulants (NPS), and cigarettes. Covariates included professional attitudes (e.g., career satisfaction, distress, patient counselling on alcohol or smoking cessation), specialty of interest, learner mistreatment, and perceived medical school support. We used multivariate logistic regression models, generating adjusted odds ratios (AORs), to examine covariates associated with substance use and how substance use (as a covariate) was associated with different professional outcomes.

Results:

Individuals more interested in “lifestyle” specialties (AOR, 1.81; 95% CI, 1.08-3.05) and surgical specialties (AOR, 1.69; 95% CI, 1.16-2.47) were more likely to report excessive alcohol use. Those interested in primary care were more likely to report cannabis use in the past 12 months (AOR, 1.85; 95% CI, 1.14-3.00). We did not identify significant associations between specialty of interest and current cigarette or NPS use in the past 12 months. However, excessive alcohol use was associated with greater career satisfaction (AOR, 1.24; 95% CI: 1.04-1.49), whereas NPS in the past 12 months was associated with poorer career satisfaction (AOR, 0.63; 95% CI: 0.42-0.93). In addition, there was a negative association between NPS use and the ability to handle workloads due to physical (AOR, 0.31; 95% CI, 0.18-0.54) or mental health issues (AOR, 0.46; 95% CI, 0.30-0.71), but not for other substances. We also found significant negative associations between current cigarette use and the perceived relevance of smoking cessation counselling (AOR, 0.48; 95% CI: 0.29-0.80) and alcohol cessation counselling (AOR, 0.42; 95% CI: 0.25-0.70).

Conclusions:

These findings suggest that specific patterns of substance use in medical students appeared to be significantly associated with some professional outcomes, specialty of interest, and attitudes towards addiction-related clinical practice. Encouraging medical students to practise healthy habits, including minimizing harmful substance use behaviours, could be an important target for improving medical students’ health and their patient care.

INTRODUCTION

Substance use is highly prevalent among post-secondary students. In particular, the prevalence of the non-medical use of prescription stimulants (NPS), such as Ritalin, Adderall, and Vyvanse, has risen among post-secondary students over the past two decades¹, with an overall use rate of approximately 17%.² Cannabis use has also increased notably in American post-secondary students since the early 2000s, with nearly 30% of college students reporting that they have used cannabis at school.^1,3,4 According to U.S. national surveys, 53% of full-time college students drank alcohol, with 33% binging in the past month.⁵ As well, 26% reported past-year tobacco use, with 5% reporting daily cigarette use.⁶

Among post-secondary student populations, medical student substance use has been a particularly important area of research.^7–10 International reviews have indicated that medical students’ binge drinking and cannabis use rates range from 7% to 56%¹¹, and 17% to 47%⁸, respectively. Similarly, tobacco and NPS consumption rates range from 9% to 27%^12–14, and 5% to 29%^2,15–19, respectively. However, a 2015 study of French medical students found that tobacco use was not lower than the general population of the same age, with more than a third of students endorsing current smoking (35%), while 21% reported daily and 14% reported occasional smoking.²⁰ Previous reviews have demonstrated that high-frequency substance use appears to be associated with several harmful outcomes compared to students who do not report substance use. For example, a 2016 study of US medical students found that nearly one-fifth reported excess drinking, which was associated with depression, suicidality, and other substance use.²¹ In an online survey of US medical students, Ayala et al. found that medical trainees who endorsed alcohol or drug use reported several harmful consequences, including interpersonal altercations, suicidal ideation, cognitive deficits, and driving under the influence of substances.⁷ Furthermore, Ayala et al. found that 40% of trainees were unaware of their medical institution’s substance-use policies.⁷

Substance use by medical trainees also has important implications for professional behaviours and outcomes.^12,21–23 For example, previous surveys have found that medical students who self-report substance use also report lenient attitudes towards self-prescribing^24–26 and towards the duty to report impaired colleagues^25,27; impacts on their academic performance⁷; and polysubstance use.^28,29 Substance use in medical students has also been related to aspects of their professional practice. For example, in a large sample of U.S. medical students, those who reported using tobacco products were 23% less likely to offer smoking counselling to their patients than trainees who did not report tobacco use.¹² A 2008 study of US medical students found that those endorsing excessive alcohol use were less likely to counsel patients or think it relevant.¹¹ A 2013 study of US medical students found that over one-third reported excessive past-month drinking, and 5% reported NPS use during the past year; those who endorsed such behaviour were significantly less likely to view it as unprofessional and warranting intervention.³⁰ Worryingly, many students seemed unfamiliar with how to help a classmate with an NPS use problem.³⁰ Ultimately, this is not a surprising finding as research has shown that the personal health behaviours of physicians in practice also predict their prevention counselling practices.^31,32

Research findings in general post-secondary Canadian settings indicate high rates of substance use.^33–35 However, there are many gaps in our understanding of factors associated with Canadian medical student substance use, underscoring the importance of further research on this population. For example, while research has indicated an association between substance use and burnout^36–44, depression^34,42–52, suicidal ideation^{38,51,53–55}, and distress^{14,34,48,56–62} in medical professionals, there is limited Canadian literature to support policy and curriculum changes that would enhance wellness and resilience in medical students and future physicians.^{10,34,61,63–65}

The present study aims to address some specific gaps in the literature. First, there is very limited Canadian literature on medical student substance use in general, relative to the United States.^{10,34,61,63–65} Second, few studies have examined professional outcomes associated with medical student substance use, such as specialty of interest or substance use counselling attitudes and practices. Third, there is limited research on factors potentially predisposing medical students to substance use, such as conditions in the medical school environment and their perceptions of medical school demands.

To address these gaps, we conducted a nationally representative survey of Canadian medical students investigating students’ mental and physical health, nutrition and physical activity behaviours, substance use, perceived stigma, health resource use, learning environment, and learner mistreatment. While some results of the survey have been published^10,60, the current study expands on prior substance use statistics by exploring whether substance use in Canadian medical students is associated with factors within the medical school environment, such as learner mistreatment, perceived medical school support, and learners’ current specialties of greatest interest. Additionally, we sought to identify whether substance use behaviours were associated with academic and professional outcomes, such as career satisfaction, the impact of physical and mental health on their workload, and the relevance and frequency of patient counselling attitudes and practices.

METHODS

Overview

We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement guidelines in the design and reporting of the present study.⁶⁶

Sample and survey instrument

Study details on the CFMS annual survey, a large, nationally representative survey of Canadian medical students, have been previously published.^10,60 In brief, we administered an anonymous, self-reported electronic survey (in English and French) to all 17 Canadian medical schools during separate two-week recruitment waves in 2015 and 2016. We used purposive sampling to achieve a total population sample of Canadian medical students, with a target population of 11,469. The full survey is supplied in Appendix 1.

Demographics.

Demographic measures included age (in years), sex (males vs. females), year of study (1, 2, 3, or 4), relationship status, and campus site (distributed vs. main site).

Substance use.

Measures for the prevalences of alcohol, tobacco, cannabis, and NPS use have been described in the primary publication of this survey.¹⁰ Briefly, we classified past-month alcohol consumption (excessive, non-excessive, or none) using the U.S. National Institute on Alcohol Abuse and Alcoholism criteria.^1,2 We categorized cannabis and NPS consumption use into four groups: “never,” “not in the past 12 months,” “not in the past 30 days,” or “used in the past 30 days.” For smoking, we used three categories (never, past, and current use) per the Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance Survey.⁶⁷ For this survey, NPS was defined as using any pharmaceutical-grade psychostimulant (e.g., Ritalin, Dexedrine, or Adderall) outside of a doctor’s orders.^68,69

Measures

We adapted questions verbatim from previous paper-based, self-administered survey instruments conducted with U.S. medical students^70,71 and Canadian physicians⁷² to assess and compare learner mistreatment, perceived medical school support, career satisfaction, the impact of mental and physical health and work and academic duties, and patient counselling attitudes and practices (Appendix 2).

Learner mistreatment.

Respondents were asked to report the frequency of belittlement/harassment (none/some/severely) from five sources: other students, residents/fellows, preclinical professors, attending physicians, and patients. We dichotomized responses for each source (e.g., 1=some/severely; 0=none) and scored using unweighted summation across all five sources, creating two continuous variables ranging from 0 to 5 for harassment and belittlement, respectively.^10,60

Perceived medical school support.

Respondents were asked about their school’s health promotion environment, including questions on stress management, healthy eating, exercise habits, and perceived support.⁷¹ Responses were scored on a five-point Likert scale, from ‘strongly disagree’ (0) to ‘strongly agree’ (4), and included in regression analyses as a continuous variable.⁷¹

Career satisfaction.

Respondents were asked, “If you re-lived your life, would you still want to become a physician?”. Responses were scored on a five-point Likert scale, from ‘definitely not’ to ‘definitely’ (4), and subsequently dichotomized at the Likert score of 3 or higher.

Impact of mental and physical health on work and academic duties.

Respondents were asked, “In the past four weeks, how often did your physical [or mental] health make it difficult for you to handle your workload as a medical student?” Responses were scored on a five-point Likert scale, from ‘all of the time’ (0) to ‘none of the time’ (4), and the score was included in regression analyses as a continuous variable.

Patient counselling attitudes and practices towards alcohol and smoking cessation.

Respondents were asked, “How relevant do you think this will be in your intended practice: 1) Smoking cessation; and 2) Alcohol?”; and “With a typical general medicine patient, how often do you talk about: 1) Smoking cessation (among smokers), and 2) Alcohol?” Responses were scored on a three-point Likert scale, from ‘never/rarely’ (0) to ‘usually/always’ (2), and this was further dichotomized (1=usually/always/sometimes; 0=never/rarely). Respondents who had not yet interacted with patients (i.e., preclinical/pre-clerkship students) were asked to respond ‘N/A’ (not applicable) (n = 2779 [unweighted]; n=2375 [weighted]).

Specialty of Interest.

Respondents were asked to select their current specialty of interest, and responses were collapsed into eight categories: “lifestyle,” primary care, surgical, internal medicine, pediatrics, psychiatry, other, and undecided (Appendix 2). We applied this categorization per previous studies.^73,74

Data analysis

We conducted our analyses in RStudio using R’s survey package (version 3.5.3).^71,72 We adjusted for early-versus-late respondents based on the timing of survey completion and compared these two waves for significant differences.^73,74 We also weighted all data via raking ratios for non-response⁷⁵ and matched respondents’ gender, year of study, medical school, and study site with national enrollment data.⁷⁶ We used multivariate logistic regression models, generating adjusted odds ratios (AOR), to first examine covariates associated with dichotomized substance use and then examined how substance use (as a covariate) was associated with different professional outcomes (e.g., career satisfaction, distress, patient counselling on alcohol or smoking cessation), current specialty of interest, learner mistreatment, and perceived medical school support, using a forced-entry approach after controlling for gender, age, and year of study (Tables 1 & 2). In all models for each outcome, statistical significance was set at an overall alpha level of 0.05.

Table 1.

Results of Multivariable Logistic Regression of Factors Associated with Alcohol Use, Tobacco Use, Cannabis Use, and NPS

Independent Variable	Dependent Variable
	Excessive alcohol		Current cigarette use		Cannabis in past 12 months		NPS in past 12 months
	OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value
Specialty of interest
“Lifestyle”	1.81 (1.08, 3.05)	0.02	1.05 (0.36, 3.04)	0.08	1.53 (0.78,3.01)	0.22	0.99 (0.30, 3.23)	0.98
Internal Medicine	1.37 (0.93, 2.01)	0.11	1.06 (0.44, 2.57)	0.13	1.62 (0.98,2.68)	0.06	0.76 (0.33,1.79)	0.53
Pediatrics	1.41 (0.92, 2.16)	0.11	0.89 (0.30, 2.58)	0.82	1.32 (0.74,2.36)	0.35	0.40 (0.12, 1.30)	0.13
Primary care	1.44 (1.00, 2.07)	0.05	0.88 (0.38, 2.06)	0.77	1.85 (1.14,3.00)	0.01	0.55 (0.25, 1.22)	0.14
Psychiatry	0.96 (0.57, 1.61)	0.87	1.52 (0.56, 4.13)	0.41	1.30 (0.69,2.46)	0.41	0.83 (0.28,2.41)	0.73
Surgical	1.69 (1.16, 2.47)	0.01	1.59 (0.67, 3.77)	0.30	1.67 (1.02,2.73)	0.04	0.73 (0.32, 1.64)	0.44
Undecided	1.31 (0.86,2.01)	0.21	0.91 (0.33, 2.52)	0.85	1.41 (0.82,2.44)	0.22	0.90 (0.37, 2.18)	0.81
Diagnostic & Laboratory Medicine	1.0 (Reference)	-	1.0 (Reference)	-	1.0 (Reference)	-	1.0 (Reference)	-
Learner mistreatment
Belittlement	0.99 (0.94 1.05)	0.79	0.97 (0.87,1.08)	0.61	0.98 (0.91,1.05)	0.54	1.00 (0.87, 1.16)	0.95
Harassment	0.99 (0.89,1.09)	0.77	1.04 (0.83, 1.29)	0.75	1.04 (0.92,1.17)	0.52	0.94 (0.75, 1.18)	0.58
Perceived medical school support
“Overall, my medical school has encouraged me to lead a healthy life.”	1.19 (1.01, 1.40)	0.04	1.20 (0.82,1.76)	0.34	1.08 (0.89, 1.32)	0.43	0.98 (0.65, 1.48)	0.92
“My medical school tries to minimize student stress.”	0.89 (0.84,1.14)	0.77	1.12 (0.81,1.56)	0.49	0.98 (0.82, 1.18)	0.84	1.09 (0.76,1.59)	0.63
“My medical school discourages students from smoking.”	0.87 (0.74,1.03)	0.10	1.05 (0.73,1.49)	0.80	0.99 (0.82, 1.21)	0.95	0.78 (0.54, 1.13)	0.19
“My classmates discourage each other from smoking.”	1.12 (0.95,1.32)	0.17	1.03 (0.72,1.48)	0.87	0.98 (0.81, 1.19)	0.83	1.29 (0.87, 1.92)	0.21
“My medical school discourages students from binge drinking.”	0.89 (0.76,1.05)	0.18	0.83 (0.58,1.19)	0.31	0.96 (0.79, 1.16)	0.66	1.05 (0.71, 1.55)	0.81
“My classmates discourage each other from binge drinking.”	0.96 (0.73,1.25)	0.75	1.08 (0.59,1.97)	0.81	0.83 (0.59, 1.16)	0.27	0.47 (0.21, 1.06)	0.07

Abbreviation: CI, Confidence Interval; NPS, non-medical use of prescription stimulants; OR, odds ratio. Bold values indicate significant effects (P<.05).

^aNon-medical use of prescription stimulants is the use of pharmaceutical stimulant medicines (e.g., Ritalin, Dexedrine or Adderall) that are used outside of a doctor’s orders.³

^bMultivariable logistic regression was performed for the respective dependent variables. Regression analyses were conducted using weighted data based only on male and female respondents in years 1–4 (n = 4,438), as national enrollment demographic data were only available for these subgroups.⁴ Data were weighted for nonresponse using the raking ratio method⁵ to match students’ gender, year of study, medical school, and site of study with national enrollment data.

^cScoring of alcohol consumption was based on criteria from the National Institute on Alcohol Abuse and Alcoholism.^1,2 Alcohol consumption was classified as ‘excessive’ in the previous month if respondents met at least one of the following criteria: 1) reported at least one occasion on which they consumed five or more drinks (i.e., reported one or more episodes of binge drinking); or 2) men who drank more than two drinks per day on average or women who drank more than one drink per day on average. Alcohol consumption was classified as ‘non-excessive’ for respondents who consumed alcohol in the past month but did not meet the “excessive” alcohol use criteria. Alcohol consumption was classified as ‘none’ for respondents that reported zero alcohol consumption in the past month.

^dSmoking history is defined as having smoked at least 100 cigarettes during one’s entire life, based on CDC Behavioural Risk Factor Surveillance Survey definition.^6,7 Cigarette use is presented as a dichotomous categorical variable, with all those categorized as having a ‘current’ smoking history scored as ‘1’ and those with a “past” or “never” history scored as ‘0’.

^eLower OR is more desirable. A higher OR represents increased odds of alcohol, cigarette, marijuana, or NPS use in the categorical group relative to the referent group. For continuous predictor variables (age, year of study, belittlement, harassment), OR represents risk per increasing unit.

^fFor specialty of interest, respondents were asked, “Choose the one specialty you are now most interested in pursuing.” Specialties were grouped for analyses according to the table presented in Supplemental Appendix 1 & 2.

^kLearner mistreatment was scored using unweighted summative scoring, as described previously.⁸ Specifically, the authors collapsed the responses of ‘some’ and ‘severely’ into one category, with both ‘some’ or ‘severely’ responses receiving a score of 1 for each belittlement/harassment group (i.e., other students, residents or fellows, preclinical professors, clinical professors, and patients) and a response of ‘never’ receiving a score of 0. Then, scores were summed across all five different groups for both belittlement and harassment to give belittlement and harassment continuous scores of 0–5.

^lRespondents were asked about their medical school’s health promotion environment using previously described questions.⁹ Responses were scored on a five-point Likert scale, from ‘strongly disagree’ (0) to ‘strongly agree’ (4) and included in regression analyses as a continuous variable.

Table 2.

Results of Multivariable Logistic Regression of Professional Attitudes and Behaviours Associated with Substance Use

Dependent variable	Independent variable	OR (95% CI) a	P-value
Career Satisfaction b	Excessive alcohol used c	1.24 (1.04, 1.49)	0.02
	Current cigarette smoking d	0.80 (0.44, 1.46)	0.47
	Cannabis use in the past 12 months	1.16 (0.89, 1.39)	0.33
	NPS in past 12 months e	0.63 (0.42, 0.93)	0.02
Ability to handle workload due to physical health	Excessive alcohol used	1.26 (0.91, 1.75)	0.16
	Current cigarette smoking	0.97 (0.36, 2.61)	0.95
	Cannabis use in past 12 months	0.90 (0.60, 1.34)	0.59
	NPS in past 12 months	0.31 (0.18, 0.54)	<0.01
Ability to handle workload due to mental health	Excessive alcohol used	1.16 (0.96, 1.43)	0.17
	Current cigarette smoking	0.84 (0.43, 1.62)	0.60
	Cannabis use in the past 12 months	1.18 (0.89, 1.55)	0.24
	NPS in past 12 months	0.46 (0.30, 0.71)	<0.01
How often do you talk about smoking cessation with a typical patient?	Excessive alcohol used	0.97 (0.84, 1.13)	0.73
	Current cigarette smoking	1.32 (0.80, 2.17)	0.28
	Cannabis use in the past 12 months	1.12 (0.93, 1.35)	0.22
	NPS in past 12 months	0.86 (0.59, 1.24)	0.41
How relevant do you think talking to patients about smoking cessation will be?	Excessive alcohol used	0.98 (0.83,1.16)	0.84
	Current cigarette smoking	0.48 (0.29, 0.80)	<0.01
	Cannabis use in the past 12 months	1.08 (0.88, 1.32)	0.46
	NPS in past 12 months	0.71 (0.49, 1.03)	0.07
How often do you talk about alcohol use with a typical patient?	Excessive alcohol used	1.10 (0.93, 1.30)	0.27
	Current cigarette smoking	1.77 (1.05, 2.99)	0.03
	Cannabis use in the past 12 months	1.06 (0.86, 1.31)	0.56
	NPS in past 12 months	0.85 (0.56, 1.28)	0.44
How relevant do you think talking to patients about alcohol use will be?	Excessive alcohol used	1.00 (0.84, 1.19)	0.98
	Current cigarette smoking	0.42 (0.25, 0.70)	<0.01
	Cannabis use in the past 12 months	1.26 (1.01, 1.56)	0.04
	NPS in past 12 months	0.90 (0.60, 1.35)	0.60

Abbreviation: OR, odds ratio; CI, confidence interval; NPS, non-medical use of prescription stimulants. Bold values indicate significant effects (P<.05). Proportional odds logistic regression was performed for the respective dependent variables. Regression analyses were conducted using weighted data based only on male and female respondents in years 1-4 (n = 4,438), as national enrollment demographic data were only available for these subgroups.⁷⁹ Data were weighted for nonresponse using the raking ratio method⁷⁸ to match students’ gender, year of study, medical school, and site of study with national enrollment data.⁷⁹

^aHigher OR is more desirable. A higher OR represents increased odds of having higher career satisfaction, better ability to handle workload due to mental and physical health, and better patient counselling practices and attitudes.

^bRespondents were asked, “If you re-lived your life, would you still want to become a physician?” Responses were scored according to a Likert scale, from ‘definitely not’ (0) to ‘definitely’ (4) and subsequently dichotomized at the Likert score of 3 or higher.

^cScoring of alcohol consumption was based on criteria from the National Institute on Alcohol Abuse and Alcoholism.^1,2 Alcohol consumption was classified as ‘excessive’ in the previous month if respondents met at least one of the following criteria: 1) reported at least one occasion on which they consumed five or more drinks (i.e., reported one or more episodes of binge drinking); or 2) men who drank more than two drinks per day on average or women who drank more than one drink per day on average. Alcohol consumption was classified as ‘non-excessive’ for respondents who consumed alcohol in the past month but did not meet the “excessive” alcohol use criteria. Alcohol consumption was classified as ‘none’ for respondents that reported zero alcohol consumption in the past month.

^dSmoking history is defined as having smoked at least 100 cigarettes during one’s entire life, based on CDC Behavioural Risk Factor Surveillance Survey definition.^6,7 Cigarette use is presented as a dichotomous categorical variable, with all those categorized as having a ‘current’ smoking history scored as ‘1’ and those with a “past” or “never” history scored as ‘0’.

^eNon-medical use of prescription stimulants is the use of pharmaceutical stimulant medicines (e.g., Ritalin, Dexedrine or Adderall) that are used outside of a doctor’s orders.³

RESULTS

From the 11,469 surveyed students, 4,438 valid responses were analyzed after weighting (participation rate: 40.2%). Participant characteristics and substance use prevalence rates have been previously reported.^10,60 In brief, the mean age (SD) was 24.2 years (3.5), 64.1% were female, and 49.4% were single. Most participants (60.5%) were in the preclinical training stage.

Per our previous publication, nearly half (46.4%) of medical students reported excessive alcohol consumption in the past month. Lifetime, past-year, and past-month cannabis use prevalences were 45.6%, 20.7%, and 7.9%, respectively.¹⁰ Lifetime and past-year NPS use prevalences were 8.3% and 3.8%, respectively, while 2.2% and 6.8% reported current and lifetime cigarette smoking.¹⁰ Male medical students reported higher prevalence rates of all four substances than females.¹⁰ While alcohol, cigarettes, and NPS were associated with a higher risk of psychological distress, using NPS was associated with a risk of burnout.¹⁰

Table 1 outlines the results of multivariable logistic regression models for predictors of substance use. Individuals interested in “lifestyle” specialties (AOR, 1.81; 95% CI, 1.08-3.05) and surgical specialties (AOR, 1.69; 95% CI, 1.16-2.47) were more likely to report excessive alcohol use. In contrast, individuals interested in primary care specialties were more likely to report cannabis use in the past 12 months (AOR, 1.85; 95% CI, 1.14-3.00). No significant associations existed between current specialty of interest and cigarette use or NPS use in the past 12 months. Similarly, we did not identify any significant associations between learner mistreatment (measured by the degree of belittlement and harassment reported by survey participants) with any of the four categories of substance use. Finally, contrary to expectation, students reporting that they felt encouraged by their medical schools to lead a healthy lifestyle were more likely to report excessive alcohol use (AOR, 1.19; 95% CI, 1.01-1.40).

Table 2 outlines the results of multivariable logistic regression models for the association between substance use and professional attitudes and behaviours. While excessive alcohol use was associated with greater career satisfaction (AOR, 1.24; 95% CI: 1.04-1.49), non-excessive alcohol use was associated with lower career satisfaction (OR=0.92; p=0.035). In contrast, there was no significant association between no alcohol use and career satisfaction (OR=0.98; p=0.656). NPS in the past 12 months was associated with lower career satisfaction (AOR, 0.63; 95% CI: 0.42-0.93). In addition, there was a negative association between NPS use and the reported ability to handle workloads due to physical (AOR, 0.31; 95% CI, 0.18-0.54) or mental health issues (AOR, 0.46; 95% CI, 0.30-0.71), but not for other substances. We also found significant negative associations between current cigarette use and the perceived relevance of smoking cessation counselling (AOR, 0.48; 95% CI: 0.29-0.80) and alcohol cessation counselling (AOR, 0.42; 95% CI: 0.25-0.70).

DISCUSSION

This study uses data from the first comprehensive nationally representative survey of Canadian medical students’ health and health practices, including substance use, and one of the most extensive evaluations of substance use epidemiology in this population. The present analysis builds upon published baseline prevalence rates of excessive alcohol, cigarettes, cannabis, and NPS to examine the association between substance use and patient counselling attitudes and practices, career satisfaction, academic/clinical workload, and the medical school environment. Our analyses identified several important findings, for example, negative associations between current cigarette use and the perceived relevance of smoking cessation and alcohol cessation counselling. Ultimately, these findings suggest that encouraging medical students to practice healthy habits—including minimizing harmful substance use behaviours—could be important for improving their health and patient care.⁶⁵ However, our findings have other important implications, whether interventional or for further research. For example, one of the biggest remaining questions is whether the associations we found between substance use and professional outcomes are directly causative or (more likely) whether it is an indicator of a relationship between other personal-level factors and professional outcomes. Uncovering the actual determinants of the outcomes we have examined is important before targeted interventions can be designed.

Comparison with previous studies

Survey data indicate that alcohol is the most commonly reported substance consumed by Canadian post-secondary students, with 84% reporting using alcohol in the past year.^33,80 Previous studies have estimated that 7% to 56% of medical students report binge drinking^9,11,81, much higher than the estimated 1% of Canadian physicians who report binge drinking.⁶³ A 2002 study of US medical students found that the overall rate of binge drinking among medical students (34%) was lower than that seen in 18 to 24-year-olds (51%), 25 to 34-year-olds (40%), and US college seniors (45%).¹¹ A 2021 study of Italian medical students found that binge-drinking behaviour at least once in the last twelve months occurred in 65%, with higher prevalences in men, students living away from their parents, earlier age of onset of alcohol use, drinking outside meals and attending health courses.²⁰ A 2015 study of French medical students found that the prevalence of high-risk alcohol varied from 15% to 52%, with higher rates among preclinical trainees than clinical students (47% vs. 16%; p<0.05).²⁰ To put these findings into a Canadian context, the inaugural 2019/2020 Canadian Postsecondary Education Alcohol and Drug use Survey (CPEADS) indicated alcohol (84%), cannabis (48%), opioids (24%), non-medical prescription stimulants (12%), and tobacco (8%) were the most commonly used substances in the past year.³³

An unexpected finding in our study was that excessive alcohol use was associated with greater career satisfaction. We inferred career satisfaction based on our single question about whether one would re-train as a physician given a chance. Students who used alcohol excessively were likelier to report higher career satisfaction by our inferred measure. However, students who used alcohol in general (i.e., non-excessively) were likelier to have lower career satisfaction by our inferred measure, which suggests that there is something unique about those who use alcohol excessively versus those who use any alcohol, that may mediate this relationship between career satisfaction and excessive alcohol use. One possible mediating factor is social connectedness, which may be supported by prior research, which indicates alcohol use can be a maladaptive coping mechanism^82,83 associated with polysubstance use.^7,28,84,85 General post-secondary literature suggests that students who report binge drinking may rate their social connectedness higher, which can, in turn, influence career or educational satisfaction ratings due to the social context of alcohol use.⁸⁶ Consequently, excessive alcohol use may have been associated with the number and quality of social supports, thereby producing feelings of support and belonging, and could potentially mediate the relationship between alcohol use and career satisfaction.⁸⁷ In contrast, social integration in the medical student community may be more problematic for students who abstain from alcohol.⁸⁸ Our study also highlighted that excessive alcohol use was significantly associated with lower perceived relevance of counseling patients concerning alcohol use, consistent with a previous study of US medical students.¹¹ In this case, it may be possible that students who use alcohol excessively might normalize alcohol use behaviours in themselves and their patients and, consequently, are less likely to counsel them on alcohol cessation. Ultimately, more research is needed to explore this finding in more detail, including more explicit measures of career satisfaction and measures of hypothesized mediating factors, such as social factors. This could have important medical education implications: social connectedness and related experiences could be an important protective factor in medical students and, thus, an important target for interventions to improve student career satisfaction.

We did not find significant associations between perceptions of the medical school environment or learner mistreatment and smoking. However, a previous study found strong correlations between students’ and Deans’ perceptions of their respective schools’ health promotion environments.⁷¹ Consequently, future research will need to explore other personal and environmental factors that predict smoking behaviours or are associated with tobacco use among medical students. For example, while tobacco use often occurs in conjunction with other substance use (e.g., alcohol)¹³, we did not explore concurrent substance use in this study. Consistent with prior studies¹², we did find significant associations between current cigarette use and counselling practices. Unlike alcohol, however, tobacco use was associated with greater perceived relevance of providing patient counselling on tobacco cessation and alcohol use.

A 2004 study found no significant difference in lifetime smoking prevalence between Chinese medical and non-medical students (41% versus 45% for males and 4% versus 6% for females).⁸⁹ A 2020 study of Maltese medical students found that male and female medical students smoked significantly fewer cigarettes than the general Maltese population in 2016 (10.8% vs. 26.4%, p<0.0005, respectively).⁹⁰

We found that cannabis use in the past 12 months was associated with a current interest in primary care or surgical specialties. Previous studies have shown differences in medical student substance use by the current medical specialty of interest. For example, non-primary care students were more likely to drink excessively, even after adjustment for gender disparities across specialties.¹¹ Also, students interested in primary care were more likely to be non-smokers than those who were undecided or more interested in non-primary care specialties (83% vs. 79% vs. 77%; p=0.026).¹² Older studies found that emergency medicine physicians were twice as likely to use cannabis compared to other physicians, and that emergency medicine and psychiatry residents reported more current use of cannabis than their colleagues in other residency programs.^91,92 However, these findings have not been replicated with more recent data, and these studies explored specialty of practice as their independent variable, whereas we investigated students’ purported specialty of interest. These limitations affect the comparability of these prior studies with our findings. Regardless, it would be beneficial to further explore substance use in these subpopulations, as they may possess characteristics or be exposed to certain factors that predispose them to higher rates of cannabis use.

Previous literature has indicated that cannabis use among medical students might be associated with polysubstance use and personality traits associated with professional outcomes. For example, Schwarzbold et al. found that cannabis use was associated with extraversion, fun-seeking, and lower conscientiousness among medical students.⁹³ As lower conscientiousness may impact academic success⁹⁴, early identification of and psychological support for cannabis-using students could provide much-needed support and improve their chances of academic success.⁹³ Similarly, Fond et al. found that nearly 6% of medical students met the criteria for cannabis use disorder, which was associated with co-use of alcohol, tobacco, and hypnotics, as well as lower work hours, a history of trauma, and lower rates of psychological follow-up.⁹⁵ Although not examined in our study, previous surveys in the U.S. have shown that medical students’ attitudes towards cannabis (e.g., support for cannabis reform) are significantly correlated with previous cannabis use.^96–98 Given the medical profession’s role in informing cannabis policy, our findings have important professional implications in the wake of cannabis legalization in Canada. A 2021 Brazilian study indicated that the medical students in public universities in São Paulo have higher cannabis use rates than the general Brazilian population and other medical students worldwide, despite an awareness of the possible cannabis-related harms.⁹⁹

Although NPS use has not been widely studied among medical students, previous reports suggest that the desire for cognitive enhancement and academic performance are the primary drivers of NPS use among medical students.^100,101 These findings are important because NPS use could negatively impact medical students’ future performance as physicians.¹⁵ As our survey found associations between NPS use and greater burnout, higher distress, and reduced ability to cope with workload due to physical or mental health issues¹⁰, NPS may also represent a maladaptive coping strategy for distress. Bucher et al. found significant associations between NPS use and aggressive-hostility personality traits (e.g., tendency to express verbal aggression, quick temper, impulsivity), which can have wide-ranging impacts on their performance as future physicians and the development of professional physician-patient relationships.¹⁵ Taken together with the results from our survey, these findings demonstrate that it is important to identify medical students who use NPS within medical training environments to receive additional support to mitigate against the potential harms of such behaviour.

Strengths, Limitations, and Future Directions

The strengths of this study include the large sample size, the survey’s availability in both of Canada’s official languages, and the diverse array of measures, including four categories of substance use and a variety of covariates, including professional attitudes (e.g., career satisfaction, distress, patient counselling on alcohol or smoking cessation), the specialty of interest, learner mistreatment, and perceived medical school support. In addition, the statistical advantages of our study include the adjustment for non-response and sample weighting using national enrolment data, which improved the generalizability of our findings to Canadian medical students.

However, there are several important limitations. First, our study’s observational, cross-sectional nature prevents the assessment of causal relationships between substance use and the variables of interest. Second, our survey was primarily based on participant self-report; consequently, for some survey questions, social desirability and self-selection could have influenced the endorsement of more sensitive behaviours and topics. Third, our substance use categories were broad. For example, we did not consider more nuanced assessments (e.g., types of cannabis products, tobacco alternatives, doses), other categories (e.g., benzodiazepines, opioids, inhalants), or substance use disorder diagnoses.

Despite these limitations, our findings represent an important first step in quantifying patterns of substance use in Canadian medical students and their association with professional attitudes, specialty of interest, learner mistreatment, and perceived medical school support. Future research should prospectively evaluate these patterns, which can help identify further protective and risk factors for medical students who use substances.

Conclusions

The present survey builds upon our initial findings on substance use epidemiology among Canadian medical students.^10,60 Overall, our initial findings demonstrated that substance use is prevalent among Canadian medical students and appears to be associated with some adverse outcomes, including psychosocial distress. The present study demonstrates that specific patterns of substance use are significantly associated with various professional attitudes and behaviours. For example, NPS use was negatively associated with career satisfaction and students’ ability to handle their workload. It will be important to uncover potential mediating factors between these relationships and other predisposing factors for student substance use. Such research is needed so that interventions can be designed and targeted to those most in need to mitigate the impacts of student substance use on academic and professional outcomes.