The emotional wellness of caregiving professionals has been a topic of investigation for > 50 years, since burnout was first described as a negative consequence of work in health care and human services.1 Research has centered not only on the quality of life of the health-care professionals themselves, but also on the potential impacts of a caregiver's mental wellness on the recipients of their care.2 For veterinary professionals, recipients of care include both humans and animals, suggesting a broad range of potential impacts. As a result, veterinary organizations are increasingly cognizant that professional competency must comprise both technical skill and individual well-being.3
The veterinary community has also experienced an increased risk of adverse mental health outcomes among veterinary professionals.4 The proportional mortality ratio for suicide for veterinarians in the United States was first flagged in 1980 as being higher than that of the general population,5 and this trend has remained consistent until present day.6 Similar results have been found in Australia,7 Norway,8 Canada,9 and the United Kingdom,10 but not in Denmark11 or New Zealand.12 There are a variety of factors under investigation that may increase the occupational risk of suicide in the veterinary profession,13 including access to means,12 attitudes toward death and euthanasia,14,15 and job-related stress.13,16 Early work in the United Kingdom identified elevated rates of anxiety and depression among veterinarians relative to the general population,17 and as a result, much of the research surrounding suicide in the profession has focused on the prevalence of these and other indicators of poor mental health.18
Although identifying and describing occupational risks remains an important task, veterinary organizations and the veterinary community are also compelled to move forward with interventions to promote veterinarian wellness.19 This may involve identifying and mitigating risk factors that are associated with poor mental health outcomes, as well as bolstering and supporting positive predictors of psychological well-being.18,19 Resilience has been shown to be a key predictor of psychological well-being in both the general population and among medical professionals20,21 and highlighted as an area for further investigation in veterinary medicine.3,18 Previous work has shown that resilience is measurable through validated psychometric scales,22 contributes to positive mental health outcomes,23 and is responsive to training and educational efforts,20,24 highlighting it as a promising area for intervention and education.18
Resilience is a multidimensional construct that has been variously defined as “a capacity, a process, and an outcome.”18 All definitions describe a positive response to adversity,18,25–27 such that an individual is at least coping,22 if not thriving.28 The CD-RISC, the most widely used measure of resilience,29,30 defines it as “stress-coping ability.”22 With respect to occupational research, the ICWR-1 has proposed that resilience mediates the response to stressors through intrapersonal components including mindfulness, neuroticism, self-efficacy, and coping, with an end result of psychological adjustment.29 Psychological adjustment, or psychosocial adaptation, describes the dynamic process by which an individual uniquely perceives and defines elements of their world and their place within that world.31 Psychological adjustment may be positive as in the term “well adjusted,” or may be negative and include symptoms of stress, depression, anxiety, burnout, and compassion fatigue.29 The ICWR-1 model positions resilience as a key area for intervention to prevent or reduce these negative mental health outcomes.29 Although the association between resilience and mental health outcomes is complex and likely bidirectional in nature, the relative impact or association of individual characteristics may be estimated at a single point in time through the use of a DAG to reduce potential bias, as presented by Shrier and Platt.32 Similarly, as proposed in the ICWR-1 model,29 a DAG may be used to estimate the mediating effect of resilience on different mental health outcomes including burnout, secondary traumatic stress, perceived stress, and emotional distress.
The purpose of the study reported here was to investigate the association of demographic, career, and lifestyle factors with resilience and the association of resilience with mental health outcomes in a sample of Canadian veterinarians.
Materials and Methods
Data collection
The study protocol was approved by the Research Ethics Board at the University of Guelph (protocol No. 16SE032). An online survey was administered through the use of available softwarea from February 2017 through July 2017. The survey was available in French and English, and participation was restricted to veterinarians residing in Canada. Nonlicensing provincial and federal veterinary organizations assisted with recruitment by sending a notice of the survey via electronic mailing lists and newsletters. The Canadian Veterinary Medical Association, alumni organizations, species-specific or specialty veterinary associations, and a general distributor of veterinary products also assisted in advertising the survey. Links to the survey were also disseminated through social media. Participants were given an opportunity to be entered in a drawing to receive 1 of 5 prizes of $400 (odds of winning were given as approx 1/500). Participants were asked questions regarding demographics, lifestyle, career, and satisfaction with personal supports, such as relationship partners, family, friends, and workplace resources. The survey also contained 5 validated psychometric scales, including the CD-RISC,22 PSS,33 HADS,34 MBI,35 and ProQOL.36 All scales have previously been validated in both French and English and across a variety of populations.30–40 The full population of participants has been previously described41; only veterinarians who indicated they were currently in clinical practice were included in the analyses of the present study. The number of observations varied by question, as no questions were mandatory.
CD-RISC
For the purposes of the present study, resilience was defined as a participant's score on the CD-RISC, a 25-item questionnaire used to measure “successful stress-coping ability,”22 with each item scored on a 5-item rating scale. In the present study, total scores ranged from 0 to 100, with higher scores reflecting greater resilience. The scale incorporated global aspects of resilience, including “hardiness, which reflected control, commitment, and change viewed as challenge.”22
PSS
The PSS used in the present study was a 10-item questionnaire measuring the extent to which participants find their lives “unpredictable, uncontrolled, and overloading.”33 Total scores ranged from 0 to 40, with each question scored on a 5-point Likert scale and higher scores indicating a higher level of perceived stress.
HADS
The HADS, a 14-item questionnaire, has been used to screen for clinical anxiety and depression in both the hospital setting and general population.34 The scale was designed as 2 distinct subscales (ie, anxiety and depression), with each item measured on a 4-point Likert scale. Although the scale has demonstrated consistently reliable psychometric properties and case-finding ability, the capacity of the scale to differentiate between the constructs of anxiety and depression has come into question.42 As has been suggested in a recent systematic review, the scale was used in the present study as a single measure of emotional distress. Total scores ranged from 0 to 42.42
MBI–Human Services Survey
Burnout was assessed in the present study with the MBI–Human Service Survey, the most widely used, validated psychometric instrument for burnout. The MBI–Human Services Survey, a 22-item questionnaire, has been used to assess the feelings of caregiving professionals toward their job and the recipients of their care.35,38 A 7-point Likert scale was used to measure each question. Three distinct subscales were included in the questionnaire, 2 of which described negative job- and recipient-related feelings, with higher scores associated with burnout as follows: the Emotional Exhaustion subscale (9 items; scores ranged from 0 to 54)38 and the Depersonalization subscale (5 items; scores ranged from 0 to 30).38 The third subscale, Personal Accomplishment (8 items; scores ranged from 0 to 48),38 described positive job- and recipient-related feelings, with lower scores associated with burnout.
ProQOL
Version 5 of the ProQOL, a 30-item questionnaire, has been used to assess positive and negative emotional aspects of caregiving occupations.36 The ProQOL included 3 subscales. The Compassion Satisfaction subscale has been considered the positive emotional outcome, whereas Burnout and Secondary Traumatic Stress subscales have been considered negative emotional outcomes and the basis for “compassion fatigue.”43 Recent discussion around the construct of “compassion fatigue” has suggested that the research community move away from the term because of its potential inaccuracy, inconsistent usage, and lack of specificity,44 and “empathic distress” has been suggested as the correct description.45 The subscales of Burnout and Secondary Traumatic Stress were addressed individually in the present study. Burnout has been described as “feelings of hopelessness and difficulties dealing with work or in doing [one's] job effectively,”43 whereas Secondary Traumatic Stress has been defined as a product of “work-related, secondary exposure to extremely or traumatically stressful events.”43 Ten items were included in each subscale, and each question was measured on a 5-point Likert scale; subscale scores ranged from 10 to 50.
The ProQOL was designed for use by a variety of caregiving professionals with human patients or clients.43 As a result, the original scale referred to people and the test taker as a helper and their occupation as helping. The creators of the scale encouraged users to adapt it as needed for different caregiving professions.43 As veterinarians are caregivers for both nonhuman animal patients and human clients and are susceptible to secondary trauma in either relationship, references to people were replaced with client and patient. Helper was replaced with veterinarian, and helping was replaced with treating.
Statistical analysis
Statistical softwareb was used to conduct statistical analyses. Missing scale or subscale items for a given participant were imputed by use of the mean of the participant's remaining items, provided no more than 1 item was missing/scale or subscale, for the CD-RISC, PSS, HADS, and MBI.46 Missing items on the ProQOL were scored as zero in some instances, according to the 4 decision rules outlined in the scale manual.43 Cronbach α was calculated for each scale or subscale to assess reliability in this population.
Modeling of resilience as an outcome of demographic, lifestyle, and career factors—Potential demographic, lifestyle, and career predictors of resilience were selected on the basis of a DAG (ie, a causal diagram)32 constructed through a review of the existing literature. The mean, median, and range were calculated to describe continuous variables, and relative proportions were calculated as a percentage of total responses to each item for categorical variables.
Correlations between independent variables were calculated by use of the Spearman rank correlation coefficient (ρ), and variables were considered correlated if the absolute value was ρ > 0.9.47,48 In cases of correlation, the most biologically plausible variable was retained. Univariable linear regression models were created between the CD-RISC score and each independent variable, and each association was assessed for inclusion in the initial multivariable model by use of a liberal value of P < 0.2.48,49 In cases of a nonlinear relationship between a continuous independent variable and the outcome, transformations were attempted and assessed graphically. If transformation did not result in a linear association, the independent variable was categorized into 5 quantiles and the univariable relationship with CD-RISC score was assessed by use of the AIC.48
The multivariable linear regression model of the CD-RISC score was created by use of a manual backward elimination approach48 with purposeful selection of covariates as described by Bursac et al50 to prevent the rejection of potentially important predictors and confounders.48–50 Significance was defined as P < 0.05 by use of a t test. For instances in which no individual categories of a variable demonstrated significance, the whole variable was assessed with an F test prior to removal and retained if the F test indicated a significant impact on the model.49 The gender variable was of a priori interest given the well-recognized association between gender and mental health outcomes51 and thus forced into the model regardless of the P value. Age and “years since graduation from veterinary school” were also of a priori interest. On the basis of the causal diagram, the following interactions were hypothesized and tested: age by gender, marital status by gender, number of children by gender, practice role by gender, practice species by gender, satisfaction with relationship partner support by gender, satisfaction with work support by gender, and income by hours worked weekly. In the case of multiple interactions involving the same variable (gender), the AIC was used to determine the preferred model. Finally, all independent variables, including those eliminated during univariable analysis, were tested in the main-effects model for significance and confounding.48–50 A confounding variable was forced into the model if it caused a change of ≥ 30% to a continuous covariate coefficient or a significant category of a categorical covariate coefficient on removal from the model.48
The final model was evaluated for heteroscedasticity by use of the Cook-Weisberg test, and the standardized residuals were assessed graphically for normality. Outliers were assessed by graphical analysis of standardized residuals, high leverage observations, the Cook distance, and the difference in fits. An alternative model without outliers was compared with the full model to evaluate their effect. Outliers were not excluded from the final model unless sufficient justification, such as recording error, was discovered.48
Univariable modeling of resilience as a mediator of mental health outcomes—Subscale scores for the PSS, HADS, MBI, and ProQOL were each treated as an outcome in univariable linear regression analyses, with the CD-RISC score used as an independent variable, to investigate the potential for mediation by resilience on these outcomes. As key demographic characteristics of a priori interest, veterinarian gender and age were tested in each model to assess for confounding impact on the association.
Results
A total of 1,130 participants who met the inclusion criteria responded to the survey and completed the CD-RISC scale. Characteristics of participating veterinarians are described (Tables 1–5). Cronbach α predicted very good or excellent internal reliability for all scales and subscales evaluated.
Demographic, lifestyle, and career characteristic variables of Canadian veterinarians (n = 1,130) in clinical practice who responded to an online questionnaire that included 5 validated psychometric scales, CD-RISC, PSS, HADS, MBI, and ProQOL, to evaluate resilience, perceived stress, anxiety and depression, burnout, and secondary traumatic stress, respectively, from February 2017 through July 2017.
| Response | β coefficient* | |||||||
|---|---|---|---|---|---|---|---|---|
| Question | No. of respondents | Variable | Mean | Median | Range | Values | 95% CI | P value |
| What year were you born? | 1,123 | Age (y) | 44.2 | 43.0 | 25–73 | 0.18 | 0.11 to 0.25 | < 0.001 |
| On average, how many hours of sleep do you get per night? | 1,120 | Sleep/night (h) | 7.1 | 7.0 | 1.5–10.0 | 2.02 | 1.21 to 2.82 | < 0.001 |
| What year did you graduate from veterinary college? | 1,117 | Time since graduation (y) | 17.4 | 16.0 | 0–53 | 0.21 | 0.14 to 0.28 | < 0.001 |
| How many hours per week do you work as a veterinarian, on average (not including on call)? | 1,119 | Work/wk (h) | 39.8 | 40.0 | 0–100 | –0.035 | –0.096 to 0.026 | 0.261 |
| What is your yearly income from your work as a veterinarian? | 1,063 | Annual income ($) | 95,341 | 84,000 | 10,000–500,000 | < 0.001 | < 0.001 to < 0.001 | < 0.001 |
| How many veterinarians work at your practice? | 1,091 | Veterinarians (No.) | 4.3 | 3.0 | 0–85 | –0.002 | –0.14 to 0.13 | 0.98 |
| How many nonveterinary staff work at your practice? | 1,086 | Nonveterinary employees (No.) | 11.8 | 8.0 | 0–200 | –0.008 | –0.056 to 0.041 | 0.76 |
β coefficients represent the estimated impact of an independent variable on the outcome of interest in a linear regression model.
Results of univariable linear regression analysis of the association of demographic and lifestyle characteristics with CD-RISC scores of veterinarians who responded to an online questionnaire as described in Table 1.
| Response | β coefficient* | |||||
|---|---|---|---|---|---|---|
| Question | No. of respondents | Category | Proportion (%) | Value | 95% CI | P value |
| I identify my gender as: | 1,117 | Female | 78.4 | Referent | NA | NA |
| Male | 21.6 | 1.81 | –0.16 to 3.79 | 0.072 | ||
| Please select the province or territory in which you currently reside† | 1,130 | Atlantic provinces | 12.4 | Referent | NA | NA |
| Quebec | 23.0 | 0.59 | –2.26 to 3.45 | 0.69 | ||
| Ontario | 33.2 | 0.95 | –1.76 to 3.65 | 0.49 | ||
| Western provinces and the territories | 31.4 | –0.36 | –3.09 to 2.36 | 0.79 | ||
| Please indicate your relationship status | 1,118 | Single | 12.3 | Referent | NA | NA |
| Married | 59.9 | 4.42 | 1.89 to 6.94 | 0.001 | ||
| Divorced or separated | 5.5 | 6.93 | 2.77 to 11.09 | 0.001 | ||
| Committed relationship | 22.3 | 5.15 | 2.28 to 8.02 | < 0.001 | ||
| How many children do you have? | 1,127 | 0 | 43.6 | Referent | NA | NA |
| 1 | 12.3 | 1.56 | –1.03 to 4.15 | 0.24 | ||
| 2 | 28.0 | 4.02 | 2.08 to 5.96 | < 0.001 | ||
| 3 | 11.1 | 5.29 | 2.59 to 7.99 | < 0.001 | ||
| ≥ 4 | 5.1 | 6.50 | 2.74 to 10.27 | 0.001 | ||
| How much exercise do you get, on average, per week? | 1,114 | None | 8.9 | Referent | NA | NA |
| 1 h | 13.7 | 2.52 | –0.94 to 5.98 | 0.15 | ||
| 2 h | 18.5 | 3.44 | 0.16 to 6.73 | 0.040 | ||
| 3 h | 20.5 | 6.18 | 2.95 to 9.41 | < 0.001 | ||
| 4 h | 12.3 | 7.67 | 4.13 to 11.21 | <0.001 | ||
| ≥ 5 h | 26.1 | 8.63 | 5.51 to 11.75 | < 0.001 | ||
| Do you have a family history of mental illness (eg, depression, anxiety, schizophrenia, or bipolar disorder)? | 1,125 | Yes (vs no) | 34.6 | –4.59 | –6.28 to −2.91 | < 0.001 |
| Have you suffered from mental illness in the past? | 1,118 | Yes (vs no) | 28.6 | –6.53 | –8.29 to −4.77 | < 0.001 |
| Are you currently suffering from mental illness? | 1,116 | Yes (vs no) | 14.1 | –11.31 | –13.56 to −9.07 | < 0.001 |
| In general, I consider my overall health to be: | 1,126 | Poor | 2.4 | Referent | NA | NA |
| Fair | 15.8 | 6.55 | 1.53 to 11.56 | 0.011 | ||
| Good | 36.9 | 11.88 | 7.06 to 16.70 | < 0.001 | ||
| Very good | 32.4 | 18.74 | 13.90 to 23.58 | < 0.001 | ||
| Excellent | 12.5 | 26.12 | 21.01 to 31.22 | < 0.001 | ||
All responses are self-reported.
Categories were collapsed to avoid small cell size.
NA = Not applicable.
See Table 1 for remainder of key.
Results of univariable linear regression analysis of the association of career characteristics and CD-RISC scores of veterinarians who responded to an online questionnaire as described in Table 1.
| Response | β coefficient* | |||||
|---|---|---|---|---|---|---|
| Question | No. of respondents | Category | Proportion (%) | Value | 95% CI | P value |
| What type of practitioner are you?† | 1,099 | Mixed | 12.8 | Referent | NA | NA |
| Small animals only | 72.7 | –3.96 | –6.42 to −1.51 | 0.002 | ||
| Equine only | 3.4 | –3.74 | –8.70 to 1.23 | 0.14 | ||
| Large animals only | 11.1 | –1.37 | –4.69 to 1.95 | 0.42 | ||
| What type of practitioner are you? | 1,125 | Yes (vs no) | 8.1 | 1.31 | –1.67 to 4.29 | 0.39 |
| Responded: specialist | ||||||
| Any scheduled on-call hours?‡ | 1,115 | Yes (vs no) | 55.3 | 0.68 | –0.96 to 2.31 | 0.42 |
| What is your role at the practice?† | 1,089 | Practice owner | 41.9 | Referent | NA | NA |
| Practice associate | 51.1 | –3.62 | –5.31 to −1.93 | < 0.001 | ||
| Academic§ | 3.2 | –4.30 | –8.99 to 0.38 | 0.072 | ||
| Contract or locum | 3.9 | –1.46 | –5.77 to 2.85 | 0.51 | ||
| How much financial stress are you under? | 1,126 | None | 21.9 | Referent | NA | NA |
| A little | 35.4 | –3.77 | –5.94 to −1.60 | 0.001 | ||
| Some | 25.1 | –6.41 | –8.74 to −4.09 | < 0.001 | ||
| A lot | 17.7 | –8.43 | –10.98 to −5.88 | < 0.001 | ||
Respondents were originally asked how many scheduled on-call hours they had.
Including clinical faculty or other clinical teaching role, resident, or intern.
See Tables 1 and 2 for remainder of key.
Results of univariable linear regression analysis of the association of self-reported satisfaction with supports and CD-RISC scores of veterinarians who responded to an online questionnaire as described in Table 1.
| Response | β coefficient* | |||||
|---|---|---|---|---|---|---|
| Question | No. of respondents | Category | Proportion (%) | Value | 95% CI | P value |
| Please indicate how satisfied you are with support from relationship/partner | 980 | Not at all satisfied | 2.7 | Referent | NA | NA |
| Dissatisfied | 6.1 | 5.32 | –0.76 to 11.40 | 0.086 | ||
| Somewhat satisfied | 11.2 | 6.92 | 1.27 to 12.57 | 0.016 | ||
| Satisfied | 25.2 | 11.71 | 6.37 to 17.05 | < 0.001 | ||
| Very satisfied | 54.8 | 15.74 | 10.54 to 20.94 | < 0.001 | ||
| Please indicate how satisfied you are with support from friends | 1,086 | Not at all satisfied | 1.8 | Referent | NA | NA |
| Dissatisfied | 5.2 | 1.70 | –4.86 to 8.25 | 0.61 | ||
| Somewhat satisfied | 24.4 | 8.23 | 2.37 to 14.09 | 0.006 | ||
| Satisfied | 37.8 | 15.06 | 9.27 to 20.85 | < 0.001 | ||
| Very satisfied | 30.9 | 20.57 | 14.75 to 26.39 | < 0.001 | ||
| Please indicate how satisfied you are with support from family | 1,091 | Not at all satisfied | 2.8 | Referent | NA | NA |
| Dissatisfied | 6.1 | 1.94 | –3.64 to 7.51 | 0.50 | ||
| Somewhat satisfied | 20.4 | 6.93 | 2.01 to 7.51 | 0.006 | ||
| Satisfied | 32.9 | 10.86 | 6.05 to 15.66 | < 0.001 | ||
| Very satisfied | 37.8 | 16.58 | 11.81 to 21.36 | < 0.001 | ||
| Please indicate how satisfied you are with support from workplace resources (ie, colleagues, employee assistance program, supervisors) | 1,070 | Not at all satisfied | 7.7 | Referent | NA | NA |
| Dissatisfied | 10.7 | 3.83 | 0.20 to 7.46 | 0.039 | ||
| Somewhat satisfied | 32.6 | 7.58 | 4.50 to 10.66 | < 0.001 | ||
| Satisfied | 31.6 | 12.62 | 9.54 to 15.71 | < 0.001 | ||
| Very satisfied | 17.5 | 20.12 | 16.80 to 23.44 | < 0.001 | ||
See Tables 1 and 2 for remainder of key.
Description and internal consistency of 5 validated psychometric scales and univariable linear regression of the association between mental health outcome scores (as dependent variable) and the CD-RISC scores (as independent variable) of veterinarians who responded to an online questionnaire as described in Table 1.
| Psychometric scale | Score | β coefficient* | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Scale | Subscale | No. of respondents | Mean | Median | Range | Cronbach α | Value | 95% CI | P value |
| CD-RISC | NA | 1,130 | 69.9 | 70.0 | 20–99 | NA | NA | NA | NA |
| PSS | NA | 1,129 | 17.0 | 17.0 | 0–7 | 0.90 | –0.31 | –0.34 to −0.29 | < 0.001 |
| HADS§ | NA | 1,130 | 13.2 | 12.0 | 0–39 | 0.90 | –0.32 | –0.35 to −0.30 | < 0.001 |
| MBI | Emotional exhaustion | 1,130 | 26.1 | 26.0 | 0–54 | 0.94 | –0.52 | –0.57 to −0.47 | < 0.001 |
| Depersonalization | 1,129 | 8.9 | 8.0 | 0–8 | 0.80 | –0.20 | –0.22 to −0.17 | < 0.001 | |
| Personal accomplishment | 1,123 | 36.6 | 38.0 | 10–48 | 0.81 | 0.36 | 0.33 to 0.38 | < 0.001 | |
| ProQOL | Burnout | 1,129 | 25.2 | 25.0 | 10–45 | 0.86 | –0.33 | –0.36 to −0.31 | < 0.001 |
| Secondary traumatic stress | 1,129 | 23.6 | 23.0 | 10–46 | 0.84 | –0.23 | –0.25 to −0.21 | < 0.001 | |
| Compassion satisfaction | 1,129 | 37.8 | 39.9 | 14–50 | 0.91 | 0.35 | 0.33 to 0.37 | < 0.001 | |
Anxiety and depression subscales have been combined as a single measure of emotional distress.
See Tables 1 and 2 for remainder of key.
Modeling of resilience as an outcome of demographic, lifestyle, and career factors
Age and years since graduation from veterinary college were highly correlated (ρ = 0.95). Given that the CD-RISC measures global resilience rather than occupation-specific resilience, age was selected as the most biologically plausible predictor and was forced into the model. Once the model was constructed, the years since graduation variable was tested as a substitute for age, with minimal (< 10%) change to covariate coefficients. Neither variable was found to have a confounding effect on covariate coefficients.
The number of reported hours of sleep per night was found to have a quadratic relationship with the CD-RISC score through a graphical examination of the relationship, demonstrating a simple bend in only 1 direction, and through the significance of the quadratic term. The variable for annual income was found to have a nonlinear, nontransformable relationship with the CD-RISC score, and income was categorized into 5 quantiles ($10,000 to $65,000, $66,000 to $80,000, $81,000 to $92,000, $93,000 to $120,000, and $122,000 to $500,000). The generated income categories included all reported incomes (ie, $121,000 was not reported as an income by any participants). The AIC was preferable (lower) for the association between resilience and categorized income.
The final model (Figure 1) included 783 participants who provided responses for all of the variables of interest. The adjusted R2 for the final model was 37.2%. Only the following interactions were significant: income by reported number of hours worked per week (Figure 2) and satisfaction with workplace support by gender (Figure 3). The following variables were neither significant nor found to act as confounders in the final model: region of residence, hours of sleep per night, hours of exercise per week, family history of mental illness, past mental illness, satisfaction with support from family, role as a specialist, number of veterinarians in the clinic, number of nonveterinary staff in the clinic, and financial stress. Age was not significant in the final model but was forced in as previously described.
Multivariable linear regression model of predicted impact of the independent variables of demographic, lifestyle, and career characteristics on CD-RISC score, as a measure of resilience, for 783 Canadian veterinarians in clinical practice who responded to an online questionnaire from February 2017 through July 2017. Lines not crossing the null value of zero indicate significance. Referents or comparisons for each characteristic are as follows: overall health (referent, poor), satisfaction with support from friends (referent, not at all satisfied), satisfaction with support from a relationship or partner (referent, not at all satisfied), clinical role (referent, practice owner), any on-call hours (vs no on-call hours), number of children (referent, none), marital status (referent, single), age in years (no referent), current mental illness (vs none currently), species of practice (referent, mixed animal), hours of work per week by income category inter action (referent, $10k to $ 65k), and satisfaction with support from workplace by gender interaction (referent, not at all satisfied; female gender). k = 1,000.
Citation: Journal of the American Veterinary Medical Association 257, 10; 10.2460/javma.2020.257.10.1057
Predicted CD-RISC scores at different levels of the interaction between income categories and reported hours worked per week in a multivariable linear regression model of resilience, as measured by the CD-RISC score, among the veterinarians described for Figure 1.
Citation: Journal of the American Veterinary Medical Association 257, 10; 10.2460/javma.2020.257.10.1057
Predicted CD-RISC scores at different levels of the interaction between reported satisfaction with workplace supports and gender in a multivariable linear regression model of resilience, as measured by the CD-RISC score, among the veterinarians described for Figure 1.
Citation: Journal of the American Veterinary Medical Association 257, 10; 10.2460/javma.2020.257.10.1057
In the final model (Figure 1), participant-rated overall health (excellent vs referent, poor; β = 18.28 [95% CI, 11.89 to 24.67]; t = 5.61; P < 0.001), satisfaction with support from friends (very satisfied vs referent, not at all satisfied; β = 8.87 [95% CI, 1.81 to 15.92]; t = 2.47; P = 0.014), and satisfaction with support from relationship or partner (very satisfied vs referent, not at all satisfied; β = 6.21 [95% CI, 0.60 to 11.82]; t = 2.17; P < 0.030) had strong positive associations with resilience. Having 2 children (2 vs referent, none; β = 2.74 [95% CI, 0.58 to 4.90]; t = 2.49; P = 0.013), 3 children (3 vs referent, none; β = 3.09 [95% CI, 0.23 to 5.95]; t = 2.12; P = 0.034), or any scheduled on-call hours (vs no on-call hours; β = 1.91 [95% CI, 0.13 to 3.70]; t = 2.11; P = 0.035) was also positively associated with resilience. The self-reported presence of current mental illness had the strongest negative association with resilience (vs none currently; β = −5.03 [95% CI, −7.37 to −2.69]; t = −4.23; P < 0.001). Additionally, being married (married vs referent, single; β = −5.85 [95% CI, −10.88 to −0.81]; t = −2.28; P = 0.023) or practicing small animal medicine (small animals only vs referent, mixed; β = −2.53 [95% CI, −4.97 to −0.089]; t = −2.03; P = 0.042) was negatively associated with resilience. None of the categories for clinic role had a value of P < 0.05 (clinic associate vs referent, clinic owner; β = −1.85 [95% CI, −3.82 to 0.12]; t = −1.84; P = 0.066), but the variable itself was found to have a significant impact on the model (F value = 3.65; P = 0.012).
The final model was homoscedastic (Cook-Weisberg test was not significant at χ2 = 1.81; P = 0.18, and a histogram of the standardized residuals had an overall normal distribution). Four potential outliers were identified by graphical analysis; the model was then assessed with the outliers omitted. In this reduced model, the positive associations between resilience and levels of satisfaction with support from friends were reduced, such that the dissatisfied level had a modestly negative but nonsignificant impact on resilience. The positive impact of the contract or locum category was increased by approximately 50% and became significant at P = 0.012. As there was no justifiable reason for their exclusion, these outliers were retained in the final model.
Univariable modeling of resilience as a mediator of mental health outcomes
In univariable analyses, CD-RISC score had a significant and linear association with all mental health outcome scales and subscales assessed (Table 5). None of the associations were found to be susceptible to confounding by either veterinarian gender or age, with < 10% impact on the CD-RISC coefficient in each model.
Discussion
As anticipated on the basis of the literature from other populations,25 resilience among participants was associated with positive psychological outcomes, such as personal accomplishment and compassion satisfaction, and inversely associated with negative outcomes, such as stress, anxiety, depression, burnout, and secondary traumatic stress. These findings highlighted the importance of improving the psychological resilience of veterinary professionals as a valuable area for intervention to improve the overall mental well-being of this population. Additionally, predictors of resilience in the final multivariable linear regression model included both personal and workplace factors, presenting a variety of opportunities for intervention at both the individual and the institutional level.
In addition to the negative impact of the current mental illness variable on CD-RISC scores, participant-rated overall health, which was likely to be interpreted to include aspects of both mental and physical health, was a strong positive predictor of resilience. The association between physical health and psychological resilience has been described elsewhere,52 and both exercise and nutrition have been suggested to promote resilience in veterinarians18 and in the general population.53 Given that veterinarians reportedly experience occupational illness or injury at a rate nearly 3 times that for general practitioners of human medicine,54 and given the association of physical health with resilience generally, promotion of physical health may represent an important area of education and institutional support. Sleep is another well-recognized predictor and outcome of mental health55,56 and had a strong association with resilience in univariable analysis. However, the association was not significant in the full model, indicating that it may have been accounted for by the “overall health” variable or others. This may also be the case for the exercise variable, which initially demonstrated a strong univariable association.
Satisfaction with support from friends and from a relationship or partner were also important predictors of resilience in this population. This is in agreement with literature on resilience in veterinarians18 and in other health-care professionals20,25 regarding the high impact of social and organizational supports on psychological well-being. Satisfaction with support from family did not have a significant association in the final model, possibly implying that the self-selected social supports are more relevant to resilience.
An association between gender and resilience in the final model was anticipated because of the significant impact of gender on all of the mental health outcomes in this population41 and the univariable association with resilience, age, clinic role, marital status, number of children, and many of the other covariates. The finding of only 1 significant interaction with gender, that of satisfaction with workplace supports, was unexpected. A recent report57 by the British Veterinary Association also suggested that female veterinarians still face discrimination in the workplace, which may help explain why satisfaction with workplace supports was more strongly linked with resilience in female participants. Unique difficulties have been identified that are likely to impact women at work more than their male colleagues, including health considerations related to pregnancy.58 The strong association between satisfaction with workplace supports and resilience among female veterinarians is promising, particularly as the profession trends toward a greater proportion of women.58
For both genders, the association between satisfaction with workplace supports and resilience emphasizes the importance of employer leadership in relation to employee well-being. Indeed, in Canada, employers are legally required to provide for the occupational health and safety of their employees, including mental health.59 Although individual-level interventions such as mindfulness training24 and resilience building60 can be effective, they have been shown to be less effective than organization-level interventions, such as changes to workload or teamwork-focused meetings, among physicians.61 In the North American veterinarian population, there is evidence that the current workplace and organization-level mental health resources are not widely used.62 A cultural shift in the industry may be needed to build awareness of the importance of veterinarian well-being, and mentors, employers, and industry leaders are in a unique position to model well-being and provide a workplace environment in which individual professionals can thrive. Previous research has suggested that management skills training and increased employee participation in decision-making, in addition to decreased work hours and improved work-home boundaries,63 have the potential to improve the mental health of the veterinary community.
The interaction between income and hours worked per week demonstrated a strong positive association between resilience and work hours in the $66,000 to $92,000 range, and little or no association with work hours at higher or lower categories of income. The highest levels of resilience, given typical working hours (40 h/wk), were found in the highest and lowest income brackets, with lower levels of resilience in the second and third income brackets. Financial stress was suspected to have an impact on this association; had a strong, inverse relationship with both income and resilience in univariable analyses, but was not significant in the final model; and did not appear to confound the relationship between income and resilience. Unexpectedly, having scheduled on-call hours was positively associated with resilience. Follow-up studies employing qualitative methods would be useful in further exploration of these findings.
Age was not significant in the final resilience model. This was in line with previous studies64,65 and suggests that the reported increased risk of poor mental health outcomes in younger veterinarians63 is the result of other personal and career characteristics.
A third sociodemographic characteristic, marital status, had a significant negative association with resilience, although the implication is unclear. Prior reports of resilience have been inconsistent with regard to marital status.30 The literature suggests a protective psychological effect of marriage,66 but in this population, participants who identified themselves as single had a significantly higher CD-RISC score than married participants. A 2005 study67 of Belgian veterinarians may provide some insight; negative work-home interference was found to be significantly higher among veterinarians relative to other occupations, suggesting that the veterinary profession places a unique strain on spousal relationships and family life. Many of the stressors listed by veterinarians in the Hansez et al study67 were related to their availability to clients outside of office hours, suggesting that reduced on-call availability and firm work-home boundaries could improve psychological well-being. These restrictions are likely to be more feasible in larger, multiveterinarian practices, and also in regions with out-of-hours emergency facilities, than in practices or areas where out-of-hours coverage by other veterinarians is limited. Conversely, having 2 or 3 children appeared to have a positive association with resilience. Assuming a bidirectional association between resilience and lifestyle factors, it is possible that those individuals with relatively high levels of resilience may choose to have more children or remain unmarried.
Limitations of the present study included response bias, as the veterinarians who chose to respond to this survey may have been those who found the topic of mental wellness particularly salient. Although the exact number of veterinarians who received an invitation is unknown, the Canadian Veterinary Medical Association reports over 12,500 veterinarians in Canada, representing a conservative estimate of 9% response rate.68 The use of a multivariable analysis including sociodemographic characteristics reduces the effect of response rate bias.69
Many of the lifestyle and career factors investigated in this survey were assessed by use of single-item, self-reported measures. These items may represent complex constructs, and the questions used in this survey have not been previously validated except when noted. Future research may build on this exploratory analysis by delving into the predictive factors identified here. For example, self-reported items may be supported or refuted by information from external sources, and the impact of these constructs may be explored further by means of previously validated, gold-standard measures.
The present study was cross-sectional, and associations discussed here do not necessarily represent causal effects. A DAG has been used in this model to allow for assessment of the associations at a single point in time. However, as previously noted,23,29 the use of cross-sectional data limits the ability to evaluate how mental health predictors and outcomes are associated over time and whether the mediating role of resilience is stable. For example, resilience may have an indirect impact on career and lifestyle variables over time through individual behavioral changes according to the symptoms of mental health outcomes such as burnout. However, given the strong associations with all of the mental health outcomes measured here, it may be reasonable to consider resilience as a critical factor for overall psychological well-being. Indeed, resilience has been shown to partially or fully mediate aspects of burnout in critical care professionals.70 Many of the interventions designed to impact resilience, such as mindfulness,71 have been shown to have positive effects across many domains of mental health and wellness.72
Among physicians in human medicine, interventions to improve resilience have focused mainly at the individual level and have reported success.71 Strategies include mindfulness-based training,73 stress-management training,74 and individual or group counseling.75 At the institutional level, practice management style,76 effective communication,76 and physician-organization collaboration61 have been suggested to promote physician resilience. Other strategies, such as supportive professional relationships,76 conscious management of personal and professional boundaries,77 and limitation of working hours,77 require individual implementation but are unlikely to be achieved without institutional support and prioritization. Additionally, the literature suggests that professional competencies should be both modeled and explicitly taught,78 thus providing an opportunity for educational institutions and mentors to actively promote a culture of well-being and resilience among their students and mentees. These insights from the human medical field can likely inform potential interventions in the veterinary field, given the similarity of the occupations and relative sparsity of literature on veterinarians.
A variety of both individual and organizational factors appear to impact resilience in veterinarians and may result in a reduction of negative mental health outcomes such as stress, burnout, and secondary traumatic stress. Interventional strategies such as resilience building and mindfulness may be useful at the individual level, but workplace cultural changes, including improved boundaries between work and home life, are likely to be the most impactful for the occupation as a whole.
Acknowledgments
This manuscript represents a portion of a thesis submitted by Dr. Perret to the Department of Population Medicine, Ontario Veterinary College, University of Guelph, as partial fulfillment of the requirements for a PhD degree.
Funded by the Ontario Veterinary College Pet Trust and by a gift from Zoetis Canada. Funding sources did not have any involvement in the study design, data analysis and interpretation, or writing and publication of the manuscript.
The authors declare that there were no conflicts of interest.
The authors thank the members of the Advancing Wellness and Resilience in Research and Education (AWAR2E) Group for their assistance, José Denis-Robichaud and Tamara Best for assistance with translation, and David Pearl for assistance with multivariable modeling.
ABBREVIATIONS
| AIC | Akaike information criterion |
| CD-RISC | Connor-Davidson Resilience Scale |
| DAG | Directed acyclic graph |
| HADS | Hospital Anxiety and Depression Scale |
| ICWR-1 | International Collaboration on Workforce Resilience 1 |
| MBI | Maslach Burnout Inventory |
| ProQOL | Professional Quality of Life Scale |
| PSS | Perceived Stress Scale |
Footnotes
CoreXM, Qualtrics, Provo, Utah.
Stata/SE, version 15, StataCorp, College Station, Tex.
References
1. Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Rev Pyschol 2001;52:397–422.
2. Marmon LM, Heiss K. Improving surgeon wellness: the second victim syndrome and quality of care. Semin Pediatr Surg 2015;24:315–318.
3. McArthur M, Mansfield C, Matthew S, et al. Resilience in veterinary students and the predictive role of mindfulness and self-compassion. J Vet Med Educ 2017;44:106–115.
4. Platt B, Hawton K, Simkin S, et al. Suicidal behaviour and psychosocial problems in veterinary surgeons: a systematic review. Soc Psychiatry Psychiatr Epidemiol 2012;47:223–240.
5. Blair A, Hayes HM. Cancer and other causes of death among US veterinarians, 1966–1977. Int J Cancer 1980;25:181–185.
6. Tomasi SE, Fechter-Leggett ED, Edwards NT, et al. Suicide among veterinarians in the United States from 1979 through 2015. J Am Vet Med Assoc 2019;254:104–112.
7. Jones-Fairnie H, Ferroni P, Silburn S, et al. Suicide in Australian veterinarians. Aust Vet J 2008;86:114–116.
8. Hem E, Haldorsen T, Aasland OG, et al. Suicide rates according to education with a particular focus on physicians in Norway 1960–2000. Psychol Med 2005;35:873–880.
9. Champagne S. Vétérinaires en détresse: mal-ětre pour le bien-ětre animal. La Presse. Available at: mi.lapresse.ca/screens/70d0d375-28ba-47c5-9599-51e42a8d016a__7C___0.html. Accessed Oct 3, 2018.
10. Mellanby RJ. Incidence of suicide in the veterinary profession in England and Wales. Vet Rec 2005;157:415–417.
11. Hawton K, Agerbo E, Simkin S, et al. Risk of suicide in medical and related occupational groups: a national study based on Danish case population-based registers. J Affect Disord 2011;134:320–326.
12. Skegg K, Firth H, Gray A, et al. Suicide by occupation: does access to means increase the risk? Aust N Z J Psychiatry 2010;44:429–434.
13. Bartram DJ, Baldwin DS. Veterinary surgeons and suicide: influences, opportunities and research directions. Vet Rec 2008;162:36.
14. Ogden U, Kinnison T, May SA. Attitudes to animal euthanasia do not correlate with acceptance of human euthanasia or suicide. Vet Rec 2012;171:174.
15. Witte TK, Correia CJ, Angarano D, et al. Experience with euthanasia is associated with fearlessness about death in veterinary students. Suicide Life Threat Behav 2013;43:125–138.
16. Bartram DJ, Yadegarfar G, Baldwin DS. Psychosocial working conditions and work-related stressors among UK veterinary surgeons. Occup Med (Lond) 2009;59:334–341.
17. Bartram DJ, Yadegarfar G, Baldwin DS. A cross-sectional study of mental health and well-being and their associations in the UK veterinary profession. Soc Psychiatry Psychiatr Epidemiol 2009;44:1075–1085.
18. Cake MA, McArthur MM, Matthew SM, et al. Finding the balance: uncovering resilience in the veterinary literature. J Vet Med Educ 2017;44:95–105.
19. Cake MA, Bell MA, Bickley N, et al. The life of meaning: a model of the positive contributions to well-being from veterinary work. J Vet Med Educ 2015;42:184–193.
20. Jackson D, Firtko A, Edenborough M. Personal resilience as a strategy for surviving and thriving in the face of workplace adversity: a literature review. J Adv Nurs 2007;60:1–9.
21. Atkinson PA, Martin CR, Rankin J. Resilience revisited. J Psychiatr Ment Health Nurs 2009;16:137–145.
22. Connor KM, Davidson JRT. Development of a new resilience scale: the Connor-Davidson resilience scale (CD-RISC). Depress Anxiety 2003;18:76–82.
23. Hegney DG, Rees CS, Eley R, et al. The contribution of individual psychological resilience in determining the professional quality of life of Australian nurses. Front Psychol 2015;6(Oct 21):1613–1618.
24. Moffett JE, Bartram DJ. Veterinary students’ perspectives on resilience and resilience-building strategies. J Vet Med Educ 2017;44:116–124.
25. Robertson HD, Elliot AM, Burton C, et al. Resilience of primary healthcare professionals: a systematic review. Br J Gen Pract 2016;66:e423–e433.
26. Epstein RM. Realizing Engel's biopsychosocial vision: resilience, compassion, and quality of care. Int J Psychiatry Med 2014;47:275–287.
27. O'Dowd E, O'Connor P, Lydon S, et al. Stress, coping, and psychological resilience among physicians. BMC Health Serv Res 2018;18:730.
28. Beltman S, Mansfield CF, Price AE. Thriving not just surviving: a review of research on teacher resilience. Educ Res Rev 2011;6:185–207.
29. Rees CS, Breen LJ, Cusack L, et al. Understanding individual resilience in the workplace: the international collaboration of workforce resilience model. Front Psychol 2015;6:73.
30. Davidson J, Connor K. Connor-Davidson Resilience Scale (CD-RISC) manual. Available at: www.connordavidson-resiliencescale.com/CD-RISC%20Manual%2008-19-18.pdf. Accessed Oct 3, 2018.
31. Larsen PD. Psychosocial adjustment. In: Lubkin's chronic illness: impact and intervention. 10th ed. Burlington, Mass: Jones & Bartlett Learning LLC, 2019;43–62.
32. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008;8:70.
33. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24:385–396.
34. Zigmond AS, Snaith RP. Hospital anxiety and depression scale (HADS). Acta Psychiatr Scand 1983;67:361–370.
35. Maslach C, Jackson SE. The measurement of experienced burnout. J Organ Behav 1981;2:99–113.
36. Stamm BH. The ProQOL manual: the Professional Quality of Life Scale: compassion satisfaction, burnout & compassion fatigue/secondary trauma scales. Baltimore: Sidran Press, 2005;1–29.
37. Bjelland I, Dahl AA, Huag TT, et al. The validity of the Hospital Anxiety and Depression Scale. An updated literature review. J Psychosom Res 2002;52:69–77.
38. Maslach C, Jackson SE, Leiter M. Maslach Burnout Inventory manual. 4th ed. Menlo Park, Calif: Mind Garden Inc, 2016.
39. Lee EH. Review of the psychometric evidence of the perceived stress scale. Asian Nurs Res (Korean Soc Nurs Sci) 2012;6:121–127.
40. De La Rosa GM, Webb-Murphy JA, Fesperman SF, et al. Professional quality of life normative benchmarks. Psychol Trauma 2018;10:225–228.
41. Perret JL, Best CO, Coe JB, et al. Prevalence of mental health outcomes among Canadian veterinarians. J Am Vet Med Assoc 2020;256:365–375.
42. Cosco TD, Doyle F, Ward M, et al. Latent structure of the Hospital Anxiety and Depression Scale: a 10-year systematic review. J Psychosom Res 2012;72:180–184.
43. Stamm BH. The concise ProQOL manual. Pocatello, Idaho: ProQOL, 2010.
44. Sinclair S, Raffin-Bouchal S, Venturato L, et al. Compassion fatigue: a meta-narrative review of the healthcare literature. Int J Nurs Stud 2017;69:9–24.
45. Singer T, Klimecki OM. Empathy and compassion. Curr Biol 2014;24:R875–R878.
46. Morris J, Coyle D. Quality of life questionnaires in cancer clinical trials: imputing missing values. Psycho-Oncology 1994;3:215–222.
47. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989;79:340–349.
48. Dohoo IR, Martin SW, Stryhn H. Veterinary epidemiologic research. 2nd ed. Charlottetown, PEI, Canada: University of Prince Edward Island, 2009.
49. Sun GW, Shook TL, Kay GL. Inappropriate use of bivariable analysis to screen risk factors for use in multivariable analysis. J Clin Epidemiol 1996;49:907–916.
50. Bursac Z, Gauss CH, Williams DK, et al. Purposeful selection of variables in logistic regression. Source Code Biol Med 2008;3:17.
51. Seedat S, Scott KM, Angermeyer MC, et al. Cross-national associations between gender and mental disorders in the World Health Organization World Mental Health Surveys. Arch Gen Psychiatry 2009;66:785–795.
52. Gerber M, Jonsdottir IH, Lindwall M, et al. Physical activity in employees with differing occupational stress and mental health profiles: a latent profile analysis. Psychol Sport Exerc 2014;15:649–658.
53. Walsh R. Lifestyle and mental health. Am Psychol 2011;66:579–592.
54. Nienhaus A, Skudlik C, Seidler A. Work-related accidents and occupational diseases in veterinarians and their staff. Int Arch Occup Environ Health 2005;78:230–238.
55. Wolf MR, Rosenstock JB. Inadequate sleep and exercise associated with burnout and depression among medical students. Acad Psychiatry 2017;41:174–179.
56. Alvaro PK, Roberts RM, Harris JK. A systematic review assessing bidirectionality between sleep disturbances, anxiety, and depression. Sleep 2013;36:1059–1068.
57. Begeny C, Ryan M. Gender discrimination in the veterinary profession: a brief report of the BVA Employers’ Study 2018. Available at: www.bva.co.uk/media/2988/gender-discrimination-in-the-vet-profession-bva-workforce-report-nov-2018.pdf. Accessed May 5, 2020.
58. Slater MR, Slater M. Women in veterinary medicine. J Am Vet Med Assoc 2000;217:472–476.
59. Government of Canada. Canadian Centre for Occupational Health and Safety Act (R.S.C., 1985, c. C-13). Available at: laws-lois.justice.gc.ca/eng/acts/C-13/page-1.html#h-61686. Accessed Jun 11, 2019.
60. Swensen S, Kabcenell A, Shanafelt T. Physician-organization collaboration reduces physician burnout and promotes engagement: the Mayo Clinic experience. J Healthc Manag 2016;61:105–127.
61. Panagioti M, Panagopoulou E, Bower P, et al. Controlled interventions to reduce burnout in physicians. JAMA Intern Med 2017;177:195–205.
62. Volk JO, Schimmack U, Strand EB, et al. Executive summary of the Merck Animal Health Veterinary Wellbeing Study. J Am Vet Med Assoc 2018;252:1231–1238.
63. Bartram DJ, Sinclair JMA, Baldwin OS. Interventions with potential to improve the mental health and wellbeing of UK veterinary surgeons. Vet Rec 2010;166:518–523.
64. Campbell-Sills L, Stein MB. Psychometric analysis and refinement of the Connor-Davidson Resilience Scale (CD-RISC): validation of a 10-item measure of resilience. J Trauma Stress 2007;20:1019–1028.
65. Wagnild G. A review of the resilience scale. J Nurs Meas 2009;17:105–113.
66. Rendall MS, Weden MM, Favreault MM, et al. The protective effect of marriage for survival: a review and update. Demography 2011;48:481–506.
67. Hansez I, Schins F, Rollin F. Occupational stress, work-home interference and burnout among Belgian veterinary practitioners. Ir Vet J 2008;61:233–241.
68. Canadian Veterinary Medical Assocation. Statistics. Available at: www.canadianveterinarians.net/about/statistics. Accessed Oct 3, 2018.
69. Rindfuss RR, Choe MK, Tsuya NO, et al. Do low survey response rates bias results? Evidence from Japan. Demogr Res 2015;32:797–828.
70. Arrogante O, Aparicio-Zaldivar E. Burnout and health among critical care professionals: the mediational role of resilience. Intensive Crit Care Nurs 2017;42:110–115.
71. Fox S, Lydon S, Byrne D, et al. A systematic review of interventions to foster physician resilience. Postgrad Med J 2018;94:162–170.
72. Luken M, Sammons A. Systematic review of mindfulness practice for reducing job burnout. Am J Occup Ther 2016;70:1–10.
73. Goldhagen BE, Kingsolver K, Stinnett SS, et al. Stress and burnout in residents: impact of mindfulness-based resilience training. Adv Med Educ Pract 2015;6:525–532.
74. Sood A, Prasad K, Schroeder D, et al. Stress management and resilience training among department of medicine faculty: a pilot randomized clinical trial. J Gen Intern Med 2011;26:858–861.
75. Isaksson Ro KE, Tyssen R, Hoffart A, et al. A three-year cohort study of the relationships between coping, job stress and burnout after a counselling intervention for help-seeking physicians. BMC Public Health 2010;10:213.
76. Jensen PM, Trollope-Kumar K, Waters H, et al. Building physician resilience. Can Fam Physician 2008;54:722–729.
77. Zwack J, Schweitzer J. If every fifth physician is affected by burnout, what about the other four? Resilience strategies of experienced physicians. Acad Med 2013;88:382–389.
78. Paice E, Heard S, Moss F. How important are role models in making good doctors? BMJ 2002;325:707–710.
