Article Text

Download PDFPDF

A longitudinal and multicentre study of burnout and error in Irish junior doctors
  1. Paul O’Connor1,
  2. Sinéad Lydon1,
  3. Angela O’Dea2,
  4. Layla Hehir1,
  5. Gozie Offiah3,
  6. Akke Vellinga1,
  7. Dara Byrne1
  1. 1 School of Medicine, National University of Ireland Galway, Galway, Ireland
  2. 2 Department of Surgical Affairs, Royal College of Surgeons in Ireland, Dublin, Ireland
  3. 3 Royal College of Surgeons in Ireland, RCSI Education & Research Centre, Beaumont Hospital, Dublin, Ireland
  1. Correspondence to Dr Paul O’Connor, Discipline of General Practice, NUI Galway, Distillery Rd, Galway, Ireland; paul.oconnor{at}nuigalway.ie

Abstract

Background Junior doctors have been found to suffer from high levels of burnout.

Aims To measure burnout in a population of junior doctors in Ireland and identify if: levels of burnout are similar to US medical residents; there is a change in the pattern of burnout during the first year of postgraduate clinical practice; and burnout is associated with self-reported error.

Methods The Maslach Burnout Inventory—Human Services Survey was distributed to Irish junior doctors from five training networks in the last quarter of 2015 when they were approximately 4 months into their first year of clinical practice (time 1), and again 6 months later (time 2). The survey assessed burnout and whether they had made a medical error that had ‘played on (their) mind’.

Results A total of 172 respondents out of 601 (28.6%) completed the questionnaire on both occasions. Irish junior doctors at time 2 were more burned out than a sample of US medical residents (72.6% and 60.3% burned out, respectively; p=0.001). There was a significant increase in emotional exhaustion from time 1 to time 2 (p=0.007). The association between burnout and error was significant at time 2 only (p=0.03). At time 2, of those respondents who were burned out, 81/122 (66.4%) reported making an error. A total of 22/46 (47.8%) of the junior doctors who were not burned out at time 2 reported an error.

Conclusion Current levels of burnout are unsustainable and place the health of both junior doctors and their patients at risk.

  • Health & safety
  • Quality in health care
  • Risk management

Statistics from Altmetric.com

Introduction

Commencing in medical training, doctors report high levels of burnout throughout their career.1 Burnout is a multifactorial construct that is characterised by high levels of emotional exhaustion, high levels of depersonalisation (feeling detached from patients and others) and low levels of personal accomplishment.2

Internship, the first year of clinical practice in Ireland, is a particularly stressful part of postgraduate training for junior doctors as this is when they become responsible for the lives of patients.3 Junior doctors report that the job requirements and responsibilities often exceed their knowledge and experience.3 It is therefore unsurprising that junior doctors in their first year of clinical practice tend to experience high level of stress and burnout.3–6 Internationally, the prevalence of burnout among junior doctors has been shown to range from 18% to 82%.7 Moreover, burnout in junior doctor populations has been shown to be associated with poor health status, substance abuse, depression, anger, irritability, anxiety, suicidal thoughts and suboptimal patient care.7–10

Despite the prevalence of burnout, there have been few large multisite or longitudinal studies of burnout in junior doctors; most studies have been carried out with US residents and measure burnout at a single time point.1 7 9 11 Therefore, it is still unclear when junior doctors develop symptoms of burnout. It has been suggested that medical students enter postgraduate training with symptoms of burnout arising from medical school.12 There is also limited evidence to indicate whether levels of burnout change during the postgraduate training period. The small numbers of studies that have examined this issue find increases in burnout levels throughout the initial years of postgraduate training—particularly an increase in depersonalisation.1 13 14 High levels of burnout have been found to be significantly associated with a greater number of self-reported errors by healthcare providers.15 However, the number of studies that have investigated this relationship is small. Within a junior doctor population, self-perceived medical error has been correlated with burnout in a sample of internal medicine residents in the USA.16

Thus, although burnout in a junior doctor population is widely reported, there is a paucity of methodologically rigorous research examining its effects.7 9 11 This is particularly true within a European context. The purpose of this paper is to report the findings from a longitudinal multicentre survey of burnout during internship in Ireland. The goal of the research is to identify if (1) levels of burnout among Irish junior doctors are similar to US medical residents, (2) there is a change in the pattern of burnout during internship and (3) whether patterns of burnout are associated with self-reported error.

Method

Instrument

The questionnaire consisted of the Maslach Burnout Inventory—Human Services Survey (MBI-HSS)2 accompanied by items seeking specific demographic data (gender, age and nationality) and a series of self-generated and anonymous identification code questions so that the response from an individual at time one could be matched with their response at time two (eg, number of siblings living and deceased). The participants were also asked if they had made a medical error in the last 3 months that had ‘played on your mind’. The participants responded with either ‘yes’ or ‘no.’

The MBI-HSS is a 22-item measure of burnout intended for use with individuals working in human services that has been demonstrated to have acceptable psychometric properties.17 The survey consists of three subscales: emotional exhaustion (nine items), depersonalisation (five items) and personal accomplishment (eight items). Participants indicate the frequency with which they have certain feelings using a 7-point rating scale that ranges from 0 (never) to 6 (every day).

Within the medical literature, burnout in doctors is considered to be characterised by high levels of emotional exhaustion or depersonalisation.4–6 Therefore, consistent with convention, the respondents who were categorised as high on emotional exhaustion (score of 27 or higher) or high on depersonalisation (score of 10 or higher) were considered to have at least one manifestation of professional burnout.

Procedure

Following approval by five Irish Research Ethics boards, the anonymous questionnaire was distributed either electronically via e-mail or in paper form at teaching sessions to junior doctors from five national intern training networks in October and November 2015. All interns must complete a minimum of 3 months of general internal medicine and 3 months of general surgery in the 12-month period of training. There are some posts that have 3-month rotations in other specialties (eg, psychiatry or paediatrics). However, these posts are few in number.

The questionnaire was distributed a second time in April and May 2016 when the interns were nearing the end of their first clinical year. Reminder e-mails were sent 2 weeks and 4 weeks after the initial email request. The rationale for these two time points was to ensure that the junior doctors were settled in their job and to avoid capturing any temporary ‘spike’ in burnout at the beginning of the year and to coincide with the last teaching sessions of the year. Participation was incentivised by giving the participants the opportunity to enter a prize draw.

The comparison group of US medical residents1 4 was from a published report of burnout. In that multicentre sample survey, 1701 medical residents responded to a questionnaire at about the same time of year as the second measurement time for this study. That study reported the proportion of respondents with scores above or below cut-off values but did not report measures of variance in association with mean values.

Statistical analysis

Some respondents did not complete all elements of each subscale of the MBI-HSS and for these respondents a subscale score and whether it was above or below the cut-point for burnout could not be determined. The number of participants who could be analysed thus reflects only those with complete data.

χ2 tests of independence were used to compare the proportion of responders to non-responders, Irish doctors and US residents with values of the instrument scores above or below cut-off values and the proportion of Irish doctors with burnout who reported or did not report an error at each time point. Paired t-tests were used to compare the mean values of subscales for the two time points for Irish doctors for those with complete data, and McNemar’s test compares the paired proportions for respondents above or below cut-off values at the two time points. A t-test was used to compare the mean values of subscale scores for those reporting a medical error at each of the two time points.

At the second measurement point, respondents were categorised as: reporting no errors at either time point; reporting an error at the first time point only; reporting an error at the second time point only; and reporting errors at both time points. Analyses of variance (ANOVAs) were used to compare the mean values of the subscales by this categorisation with a Bonferroni adjustment for pairwise comparisons. For the t-tests; Q–Q plots and a Shapiro-Wilk test of normality; and for the ANOVA Levene’s test for homogeneity of variance were used to confirm that statistical assumptions were met. SPSS V.22 was used for the analysis.

Results

Participants

Of the junior doctors eligible to participate, 172/601 (28.6%) completed the survey on both occasions it was administered. The respondents’ demographics are shown in table 1. The respondents were found to be representative of the intern population in terms of gender and nationality. Distributions were not found to differ significantly from the intern population (p=0.50 for gender and p=0.10 for nationality).

Table 1

Respondent and non-respondent demographics

Table 2 summarises the questionnaire responses at time 1 and time 2, and the corresponding published data from the sample of US residents (where available).1 4 We used a high emotional exhaustion or depersonalisation score on the Maslach Burnout Inventory (indicating a frequency of weekly or more often) to categorise a respondent as ‘burned out’. Personal accomplishment scores did not contribute to the categorisation of participants as burned out/not burned out.

Table 2

Responses to the Maslach Burnout Questionnaire subscales

Comparison with US medical residents

As compared with US residents, a significantly larger proportion of the Irish sample of junior doctors at time 2 had high levels of emotional exhaustion (p<0.001) and high levels of depersonalisation (p<0.001; see table 2). There was no significant difference between the groups on the proportion of respondents with low levels of personal accomplishment (p=0.92). Overall, a significantly larger proportion of the Irish respondents at time 2 met the criteria for burnout as compared with the percentage of US residents (p=0.001; see table 2).

Comparison at time 1 and time 2

Emotional exhaustion increased between the two measurement points; mean difference (95% CI) −2.97 (−5.14 to −0.81), p=0.007; but there was no evidence of a difference in depersonalisation: −0.96 (−2.24 to 0.31), p=0.14; or personal accomplishment 1.07 (−0.47 to 2.62), p=0.17. There was no significant difference in levels of burnout at the two time points (p=0.72) (see table 2).

Association with errors

Summary data for burnout scale scores in relation to self-reported errors for both measurement points are shown in table 3.

Table 3

Summary MBI-HSS subscale scores by self-reported error

At the first measurement time, there was no evidence of a difference in subscale scores for those respondents who did or did not report an error:

  • Emotional exhaustion, p=0.35, mean difference =−1.67 (CI −5.22 to 1.87);

  • Depersonalisation, p=0.34, mean difference =−1.01 (CI −3.11 to 1.07) and

  • Personal accomplishment, p=0.09, mean difference =2.06 (CI −0.32 to 4.44).

Table 4 shows the proportion of those with burnout in relation to reporting an error at the first time point, and there was no evidence of a difference in these proportions at the first measurement point, p=0.35.

Table 4

Error versus burnout at time 1

For the ANOVA comparison of mean scores at the second measurement time in relation to the four-level categorisation of self-reported error there was strong evidence of a difference in depersonalisation, p<0.001; some evidence of a difference in emotional exhaustion, p=0.03; but no evidence of a difference in personal accomplishment, p=0.20. Pairwise comparisons identified the following important differences after the Bonferroni correction:

  • Respondents who made an error at both time 1 and time 2 had significantly higher level of emotional exhaustion than those respondents who reported an error at time 1 only (p=0.007, mean difference = 5.74 (CI 1.62 to 9.86);

  • Respondents who reported an error at time 1 and time 2 had significantly higher levels of depersonalisation than those who reported an error at time 1 only (p<0.001, mean difference = 5.49 (CI 3.08 to7.90)) or did not report any errors (p<0.001, mean difference = 5.94 (CI 3.13 to 8.74); and

  • Respondents who reported an error at time 2 only had significantly higher levels of depersonalisation than those who reported an error at time 1 only (p= 0.002, mean difference= 4.25 (CI 1.55 to 6.94)) or did not report any errors (p= 0.003, mean difference= 4.69 (CI 1.64 to 7.74)).

Further material is shown in the online supplementary material.

Supplementary Material

Supplementary material 1

Table 5 shows the proportion of those with burnout in relation to reporting an error at the second time point and there was some evidence that burnout was associated with self-reported error; 66.4% compared with 47.8%, p=0.03.

Table 5

Error versus burnout at time 2

Discussion

Doctors are significantly more likely to experience symptoms of burnout than the general public.4 5 Moreover, junior doctors have been shown to have higher levels of burnout than medical students or more senior doctors.4 Our results suggest that the symptoms of burnout are relatively common in junior doctors in Ireland.

The levels of burnout reported in this survey of Irish junior doctors are higher than those reported by recent surveys of US residents. However, an important caveat to this finding is that the US sample consists of responses from 4 years of training, with only 23% of respondents from the first year of postgraduate training.4 Another recent study found that junior doctors in Ireland also reported particularly high levels of psychological distress.3 These findings are in spite of the fact that Irish junior doctors report more sleep and spending more time on direct patient care and educational activities than junior doctors in the USA or Australia.18 Junior doctors in Ireland also work shorter hours, and generally have less debt than US junior doctors—both factors which have been found to be associated with burnout.16 19 20

The evidence presented here shows that although there was not a significant increase in the proportion of junior doctors that were burned out, there was a significant increase in the levels of emotional exhaustion reported as the year progresses. The small number of other longitudinal studies examining burnout show mixed results, and tend to be over a period of more than a year, so it is difficult to draw comparisons.1 However, the current study has shown that even within a 6-month period, there seems to be an increase in emotional exhaustion.

It was also found that burnout appears to be associated with self-reported errors that had ‘played on (the respondent’s) mind’. However, it seems that there may be a cumulative effect. At Time 2, significantly more of those reporting an error also met the criteria for burnout than those not reporting an error. This was not the case at the Time 1. It is not possible to draw a causal link (ie, does burnout lead to error or does error lead to burnout?). There could also be an intermediate effect21 in which error may be a consequence of acquiring greater responsibility, or burnout may be a result of the interns’ realisation of their limitations (or their realisation of the realities of hospital medicine). Nevertheless, similar to other research, it would appear that there is an association between self-reported error and burnout.15 16 It is conceivable that the increase in depersonalisation may be a psychological defence mechanism to protect one’s self from the feelings of guilt associated with making a medical error. However, this is speculative, and is an area that requires further research.

Limitations

There are a number of limitations to this study. Although the response rate is typical of this type of survey research,1 a large proportion of junior doctors did not respond at both time points. Similar to the majority of the research of this type, we relied on self-reported error rather than using objective measures of error. As such, we have no objective data as to whether an error was made or not. Similarly, the MBI-HSS1 survey is a subjective measure of burnout. However, the MBI-HSS is the most commonly used and well-validated measure of burnout in existence and is considered the ‘gold standard’ for measuring burnout among occupational groups.6 There is no normative data on burnout in an Irish population to draw a comparison with the levels of burnout found in the sample of interns. Finally, Type I error inflation might have occurred through multiple statistical testing.

Recommendations

Given the high levels of burnout reported in this study, it is suggested that this is an issue that must be prioritised by the health service and postgraduate training bodies. Moreover, there is a need to evaluate levels of burnout in more senior doctors, nurses and other allied health professionals. As a result of the systemic problem of high workload and low staffing across the Irish healthcare system,22 it seems very unlikely that junior doctors are not the only professional group that are experiencing high levels of burnout. Interventions for addressing burnout can be considered at three different levels.

Primary prevention is concerned with an organisation taking actions to modify or eliminate sources of burnout that are intrinsic to the work environment.23 This is likely to be most effective in bringing about long-term reductions in burnout as it would address the primary contributors to to burnout. There is a need to invest in the health service to address issues such as workload, low staffing and the pressures of a reduced a of beds. Working conditions are the main reason Irish junior doctors are leaving the healthcare system.22 A survey of doctors who had completed internship in Ireland in 2011 found that 45% were no longer working in the public health system in Ireland, with the majority having emigrated.22 24

Secondary prevention is concerned with improving the detection and management of burnout. It generally takes the form of educational programmes designed to teach individuals how to prioritise and safeguard their mental health. Self-care is not generally taught in medical schools or as part of postgraduate specialty training. This is beginning to change (eg, use of debriefing sessions),25 but more research is required to identify those interventions that can have a long-term impact on reducing burnout, and also fostering resilience (the ability to not just survive but flourish in a stressful environment).15

Tertiary prevention is concerned with the treatment, rehabilitation and recovery of individuals who have suffered, or are suffering, from high levels of burnout. Interventions at this level tend to involve the provision of counselling services. It has been found that doctors are resistant to seeking help.26 However, those responsible for the training and welfare of junior doctors have a duty to actively monitor the doctors for which they are responsible and encourage those experiencing high levels of burnout to seek help.

Conclusion

This multicentre study has demonstrated that a considerable proportion of junior doctors in Ireland are burned out, and more so than their counterparts in the USA. There is no ‘quick fix’ panacea for how to address the issue of burnout. It will require strategic thinking, resources and input from policymakers, healthcare organisations, undergraduate and postgraduate training bodies, researchers and doctors themselves. Medicine and medical training may always be stressful. However, the current levels of burnout are unsustainable and placing the health of both junior doctors and the patients for which they care at risk.

Main messages

  • Irish junior doctors are more burned out than US residents.

  • Emotional exhaustion increased during the first postgraduate training year as a doctor.

  • Burnout and error appear to be related, with an apparent cumulative.

Current research questions

  • What are effective long-term interventions for reducing burnout in medical professionals?

  • How do levels of burnout vary during a doctor’s career?

  • What is the relationship between burnout and medical error?

Acknowledgments

We thank the intern network coordinators and staff for facilitating the data collection.

References

Footnotes

  • Contributors POC, SL, AOD and DB were involved in the design and planning of the study. SL, AOD, LH and GO collected all data for the study. POC, SL and AV analysed the data. POC drafted the initial manuscript with all other authors assisted with redrafting it. All authors reviewed and approved the manuscript prior to submission.

  • Funding This research was partially supported by the Health Service Executive’s National Doctors Training and Planning.

  • Competing interests None declared.

  • Ethics approval Five ethics review boards associated with each of the participating intern training networks.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The authors are happy to consider sharing the data with any interested researchers. Please contact the corresponding author.

  • Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with ’BMJ Publishing Group'. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references. Affiliation for Gozie Offiah has been corrected.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.