Academic research is built around long training periods, competitive funding, temporary contracts and pressure to publish. For people near the beginning of that career, those demands have increasingly raised questions about whether poor mental health is an individual problem or a structural feature of the research environment.
A large systematic review and meta-analysis now puts a clearer number on the problem. Across validated measures, 29.9% of early-career researchers met thresholds for elevated psychological distress. Rates of elevated depressive and anxiety symptoms were almost identical, at 29.8% and 29.7% respectively.
The analysis, published in Nature Human Behaviour, brought together evidence from 148 studies and four primary databases, covering 138,446 early-career researchers. The authors define this group as including doctoral researchers and early postdoctoral researchers, a period when academic careers are often least secure and professional expectations are changing rapidly.
A much larger evidence base
Previous surveys have repeatedly reported mental health concerns among PhD students and postdoctoral researchers, but estimates vary substantially depending on country, discipline, recruitment method and the questionnaire used. That makes isolated percentages difficult to interpret.
The new study addressed this by preregistering a systematic review and multilevel meta-analysis. Researchers searched Scopus, Web of Science, ERIC and PubMed through 28 October 2025. The final evidence base contained 148 studies, with 228 usable datasets across the included studies and primary databases.
Crucially, the team separated two related questions. The first was prevalence: how many researchers crossed validated thresholds indicating elevated symptoms or probable clinical concern? The second was average symptom severity across continuous scales. Keeping these outcomes separate prevents a population average from obscuring the share of individuals experiencing substantially greater difficulty.
The analysis used multilevel models to account for the fact that a single study could contribute several effect estimates. This is important because treating every estimate as fully independent can make a meta-analysis appear more precise than the underlying evidence allows.
Depression and anxiety affected about three in ten
Across all outcomes assessed with validated case-identification thresholds, the pooled prevalence of elevated psychological distress was 29.9%, with a 95% confidence interval from 26.1% to 33.9%.
Depressive symptoms were elevated in an estimated 29.8% of early-career researchers, with a 95% confidence interval from 26.1% to 33.8%. Anxiety symptoms produced an almost identical pooled estimate of 29.7%, ranging from 25.7% to 33.8%.
The review also quantified outcomes that receive less attention in discussions of academic wellbeing. An estimated 28.3% met elevated thresholds for eating-disorder symptoms, while 22.9% showed alcohol misuse. Non-suicidal self-injury was estimated at 18.6%, and suicidal ideation at 18.8%.
These figures should not be interpreted as formal diagnoses. Screening thresholds identify people with symptom levels sufficiently high to warrant concern or further assessment, but they are not equivalent to a clinical interview. The distinction matters when translating population research into claims about mental illness.
The comparison with similar-aged adults is striking
The researchers compared the pooled findings with age-matched general-population estimates. Elevated depressive symptoms among early-career researchers were approximately two to three times as common, while elevated anxiety symptoms were roughly three to five times as common.
That comparison does not prove academic employment caused the difference. People entering research are not randomly selected from the wider population, and the studies included different countries, disciplines and institutional environments. Even so, the size and consistency of the gap make it difficult to dismiss the pattern as simply reflecting the background mental health burden among young adults.
Average symptom severity added another layer. Across continuous measures, mean depression and anxiety severity generally fell in the mild range, while average stress was moderate. This can coexist with high prevalence because a population may have a modest average while a substantial minority experiences much more severe symptoms.
Who is at risk was harder to predict
The researchers tested whether demographic and individual characteristics explained the variation between studies. They found that outcome type and the measurement scale used accounted for the largest share of differences in reported prevalence and severity.
By comparison, demographic and individual characteristics showed limited associations with the estimates. This is an important result because workplace mental health programmes often focus on identifying vulnerable individuals. If distress is widespread across subgroups, interventions aimed only at personal resilience may leave major environmental pressures untouched.
The authors argue that the evidence supports structural reforms alongside evidence-based support services. In universities and research institutions, that could shift attention toward supervision quality, job security, workload, funding uncertainty, career expectations and access to professional mental health care rather than treating wellbeing primarily as an individual’s responsibility.
Why this matters for the research system
Early-career researchers occupy an unusual position. They are highly trained knowledge workers, but many remain on fixed-term funding while building publication records and competing for a relatively small pool of permanent academic roles. Doctoral candidates can simultaneously be students, employees and researchers, depending on the institution and country.
Mental health in this group therefore has consequences beyond individual wellbeing. Persistent distress can affect retention, productivity, collaboration and whether talented researchers remain in academia. Institutions also invest years of supervision, infrastructure and funding in developing researchers whose careers may become unsustainable before they reach more secure positions.
The issue is particularly relevant for research systems trying to expand postgraduate training and scientific capacity. Increasing the number of doctoral enrolments does not necessarily strengthen a research pipeline if institutions cannot also create environments in which researchers can complete training and remain productive without persistent psychological strain.
Important limitations remain
Meta-analysis can improve precision, but it cannot eliminate weaknesses in the underlying studies. Much of the literature relies on self-report questionnaires, and prevalence varies depending on the scale and threshold used. The researchers’ moderator analyses confirmed that measurement choices contributed substantially to variation between estimates.
The included evidence also spans different academic systems and employment conditions. A doctoral researcher with a secure stipend, structured supervision and clear working hours may face a very different environment from a researcher dependent on short-term contracts or uncertain funding.
Most importantly, the analysis estimates how common symptoms are. It does not establish that academic working conditions caused every reported mental health problem. Longitudinal and intervention research is still needed to determine which institutional changes produce meaningful improvements.
Even with those cautions, the scale of the evidence changes the discussion. A pooled estimate approaching 30% across psychological distress, depression and anxiety is difficult to reconcile with the idea that poor mental health among young researchers consists mainly of isolated individual cases.
The practical question for universities is increasingly not whether early-career researchers experience unusually high psychological strain, but which parts of the research environment can realistically be redesigned to reduce it.
Source Information
Study Title: Prevalence and severity of mental health problems in early-career researchers: a systematic review and meta-analysis
Journal: Nature Human Behaviour
Year: 2026
Study design: Preregistered systematic review and multilevel meta-analysis
Evidence base: 148 studies and four primary databases, k = 228, N = 138,446
Search period: Scopus, Web of Science, ERIC and PubMed searched through 28 October 2025
