Survey links doctors’ and nurses’ working conditions to mental health across Europe

Europe’s doctors and nurses are carrying a mental health burden that rivals the darkest days of the COVID-19 pandemic, and according to the largest survey of its kind ever conducted in the region, the culprit is not a virus but the everyday architecture of their working lives. A sweeping cross-sectional study published in The Lancet Regional Health – Europe has analysed responses from more than 90,000 physicians and nurses across all 27 European Union countries, Iceland, and Norway, revealing that almost one in three healthcare workers reports symptoms compatible with depression, roughly a quarter reports anxiety, and about one in six reports suicidal thoughts. The findings, drawn from the Mental Health of Nurses and Doctors (MeND) survey coordinated by the WHO Regional Office for Europe and funded by the European Commission, point to a sustained structural crisis rooted in excessive demands, workplace violence, and inadequate organisational support rather than a transient pandemic aftershock.

The scale of the undertaking is itself remarkable. Between 23 October 2024 and 10 April 2025, the research team distributed a 76-item online questionnaire through six European medical and nursing associations and, later, through the social media channels of professional bodies and the WHO Regional Office for Europe. Of 122,048 initial respondents, 90,171 valid responses remained after data cleaning, comprising 37,864 doctors and 52,307 nurses. The survey could be completed in every official language of the participating countries, took roughly ten minutes to finish, and was piloted before launch. The study protocol was registered on the Open Science Framework before data collection began, and the full analytic code has been released publicly on GitHub, an unusual level of transparency for research of this scale.

The questionnaire drew its working-condition items from established European assessment instruments, including the Copenhagen Psychosocial Questionnaire (Second Version), the European Working Conditions Telephone Survey 2021, and the European Survey of Enterprises on New and Emerging Risks 2019. Exposures were grouped into four categories: job demands such as long hours, night shifts, rotating shifts, tight deadlines, and dealing with angry patients or relatives; job hazards including sexual harassment, violent threats, physical violence, and bullying; job resources such as influence over work and support from colleagues and supervisors; and organisational support structures including unions, employee feedback mechanisms, stress action plans, and violence prevention protocols. Mental health was measured with validated screening tools: the nine-item Patient Health Questionnaire (PHQ-9) for depression, the seven-item Generalised Anxiety Disorder scale (GAD-7) for anxiety, the four-item CAGE questionnaire for alcohol dependence, and the ninth PHQ-9 item for suicidal thoughts. Both the depression and anxiety instruments showed excellent internal reliability in this sample, with Cronbach’s alpha values of 0.89 and 0.91 respectively.

The picture of daily working life that emerges is stark. Twenty-eight per cent of doctors and 9 per cent of nurses reported working more than 50 hours per week, a figure far above the roughly 2.7 per cent of the general EU workforce exceeding 45 hours recorded by Eurostat. More than a third of both professions worked nights at least weekly, 24 per cent of doctors and 56 per cent of nurses regularly worked rotating shifts, and 91 per cent of doctors and 88 per cent of nurses reported constant exposure to tight deadlines. Two-thirds of doctors and three-quarters of nurses said they regularly dealt with angry patients or relatives. Workplace violence was pervasive: a third of doctors and 37 per cent of nurses reported violent threats, 27 and 34 per cent respectively reported bullying, 11 and 14 per cent reported sexual harassment, and 10 and 18 per cent reported physical violence. By contrast, protective factors were thinner on the ground: only about 40 per cent reported support from superiors, only one in five to one in four workplaces had a stress action plan, and only about half reported formal protocols against bullying, harassment, or violence.

The statistical analysis was correspondingly sophisticated. The researchers used multilevel Poisson regression models with a random intercept for country to estimate prevalence ratios for each exposure–outcome pair, adjusting for age, gender, profession, household income, residency training, length of service, number of jobs, work setting, and sector. Because the pre-registered protocol assumed no data imputation but the observed missingness proved unlikely to be completely at random, the team adopted multiple imputation by chained equations as the primary approach, with complete-case analyses retained as sensitivity checks. They also conducted age- and gender-calibrated estimates in the 28 countries for doctors and 19 for nurses where disaggregated workforce data were available, and treated countries as separate studies in fixed- and random-effects meta-analyses, which revealed substantial between-country heterogeneity, with I-squared values exceeding 95 per cent for nearly all exposures and outcomes. Country-level prevalence of depression, for instance, ranged from 15 to 50 per cent.

The adjusted results form a coherent pattern. High job demands, particularly dealing with angry patients and relatives, were associated with 1 to 53 per cent higher prevalence of probable depression and anxiety and 2 to 33 per cent higher prevalence of suicidal thoughts. Workplace violence, especially bullying, showed the strongest links: exposed workers had 41 to 91 per cent higher prevalence of depression and anxiety and 43 to 94 per cent higher prevalence of suicidal thoughts compared with unexposed colleagues. Workplace violence was also the risk factor most consistently associated with probable alcohol dependence, raising prevalence by up to 41 per cent among those experiencing sexual harassment. In the opposite direction, job resources, particularly support from superiors, were associated with 12 to 44 per cent lower prevalence of adverse outcomes, and organisational structures, especially stress action plans, were linked to reductions of 13 to 49 per cent.

Perhaps most compelling was the evidence of dose–response relationships, which lends weight to a possible causal interpretation even though the cross-sectional design cannot prove causation. Mean PHQ-9 depression scores rose from about 6 among those working 11 to 20 hours per week to nearly 10 among those working 61 to 70 hours. Support from colleagues showed the largest protective gradient, with depression scores falling from 11 among those reporting no support to 6 among those who always received it. Similar graded patterns appeared for anxiety and suicidal thoughts, though not for alcohol dependence. The authors note that these monotonic relationships are difficult to explain entirely by reverse causation, in which depressed or anxious workers simply perceive their jobs more negatively.

The team also tested whether formal violence prevention protocols buffered the harm from workplace hazards. Statistical interactions were detected for several exposure–outcome combinations, but effect sizes were modest. The largest multiplicative interaction, a ratio of prevalence ratios of 1.28, was observed between violence protocols and sexual harassment in relation to depression. Predicted prevalences were generally lower in workplaces reporting formal protocols, but the exposure–response gradients ran largely parallel across strata, suggesting that protocols alone do little to blunt the psychological impact of violence once it occurs. The researchers caution that because organisational structures were self-reported, these measures likely capture perceived rather than objectively implemented support.

Writing in the paper, the authors argue that their findings challenge the notion that the mental health crisis among healthcare workers was primarily an acute pandemic phenomenon. The prevalence estimates they report are comparable to pooled figures from smaller single-country studies conducted between early 2020 and mid-2023, and dramatically higher than the roughly 6 per cent prevalence of probable depression in the general EU population estimated for 2013–2015. Pre-pandemic reports had already documented workforce shortages, demanding conditions, and persistent strain in the European health sector, suggesting that the patterns observed here reflect long-standing structural problems that the pandemic exposed rather than created. The authors also acknowledge important limitations: voluntary recruitment through professional associations and social media means response rates could not be calculated and selection bias cannot be excluded, notably the striking overrepresentation of night-shift workers, whose reported prevalence was roughly seven times the Eurostat benchmark. All measures were self-reported, brief screening tools were prioritised for survey brevity, and the CAGE questionnaire is less comprehensive than alternatives such as the AUDIT for problematic drinking.

Even so, the policy implications are hard to escape. The authors call for stronger adherence to the EU Working Time Directive, which limits the working week to 48 hours, and for implementation of International Labour Organisation recommendations on shift scheduling, including avoiding short intervals between shifts and increasing flexibility. They urge a zero-tolerance approach to workplace violence, in line with WHO and ILO guidance, alongside peer-support interventions such as those trialled during the pandemic, and expanded access to mental health and substance use services for healthcare workers, citing examples such as Spain’s Medical Council, which offers free psychological care to doctors. With Europe’s health systems struggling to retain staff and recruit new ones amid demographic transition and mounting demand, the authors conclude that improving working conditions is not merely a welfare measure but a structural necessity, and, echoing the WHO’s own framing, that it is time to act.