Prevalence and severity of mental health problems in early-career researchers: a systematic review and meta-analysis

In the paper

“ECRs cite a range of structural and individual stressors, such as short-term employment contracts and related job insecurity, high competition, perceived isolation, feeling like an impostor, lack of feedback and power abuse as contributors to poor mental health12–18.”

What the sources say

• ?Not checked — no full text of the sourcecites [12] Jaremka, L. M. et al. Common academic experiences no one talks about: repeated rejection, impostor syndrome, and burnout. Perspect. Psychol. Sci. 15, 519–543 (2020). source
• Partly supported

  Section 2.3 discusses insecurity, a competitive atmosphere, isolation and supervision as possible sources of stress for PhD students; section 5.4 links career prospects and leadership to mental health problems. The cited paper does not establish the full list for ECRs generally: the passages do not identify short-term contracts, impostor feelings, lack of feedback or power abuse as contributors, and the paper studies PhD students rather than ECRs as a whole. (LLM, conf=0.86)

  cites [13] Levecque, K., Anseel, F., De Beuckelaer, A., Van Der Heyden, J. & Gisle, L. Work organization and mental health problems in PhD students. Res. Policy 46, 868–879 (2017). sourceExperimental — accuracy not yet measured
• Check pendingcites [14] Hazell, C. M. et al. Understanding the mental health of doctoral researchers: a mixed methods systematic review with meta-analysis and meta-synthesis. Syst. Rev. 9, 197 (2020). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [15] Keashly, L. & Neuman, J. H. Faculty experiences with bullying in higher education: causes, consequences, and management. Adm. Theory Prax. 32, 48–70 (2010).
• Check pendingcites [16] Abbott, A. Germany’s prestigious Max Planck Society conducts huge bullying survey. Nature 571, 14–15 (2019). source
• ?Not checked — no full text of the sourcecites [17] Else, H. Bullying in science: largest-ever national survey reveals bleak reality. Nature 607, 431 (2022). source
• ?Not checked — no full text of the sourcecites [18] Sverdlik, A. in Research Handbook of Academic Mental Health (eds Edwards, M. S. et al.) 307–324 (Edward Elgar Publishing, 2024). source

In the paper

“In contrast, sample-level moderators such as national income level, gender and calendar year showed no consistent effects.”

What the sources say

• ?Not checked — no full text of the sourcecites [10] Guo, L. et al. Prevalence and changes in depressive symptoms among postgraduate students: a systematic review and meta-analysis from 1980 to 2020. Stress Health 37, 835–847 (2021). source
• Partly supported

  The Results section, “Prevalence of anxiety mediated by moderators” (page unavailable), reports no significant gender difference. But “The trend of anxiety prevalence among graduate students” reports significant changes by year, and the paper compares countries rather than national income levels. Thus only the gender portion is supported. (LLM, conf=0.90)

  cites [9] Chi, T., Cheng, L. & Zhang, Z. Global prevalence and trend of anxiety among graduate students: a systematic review and meta-analysis. Brain Behav. 13, e2909 (2023). sourceExperimental — accuracy not yet measured

In the paper

“which has been proved to generate more adequate results than the double arcsine transformation when dealing with varying sample sizes228,229”

What the sources say

• Check pendingcites [228] Röver, C. & Friede, T. Double arcsine transform not appropriate for meta-analysis. Res. Synth. Methods 13, 645–648 (2022). source
• Partly supported

  The Introduction says that, unlike the single arcsine transformation, the double arcsine back-transformation depends on the sample size chosen when studies have different sample sizes. The Discussion reports misleading double-arcsine results in a case study with extremely skewed sample sizes and calls arcsine the best classic method. This supports the comparison, but not a claim of proof that single arcsine gives better results whenever sample sizes vary. (LLM, conf=0.84)

  cites [229] Schwarzer, G., Chemaitelly, H., Abu-Raddad, L. J. & Rücker, G. Seriously misleading results using inverse of Freeman–Tukey double arcsine transformation in meta-analysis of single proportions. Res. Synth. Methods 10, 476–483 (2019). sourceExperimental — accuracy not yet measured

In the paper

“Moderator analyses showed that outcome and scale type accounted for the largest share of variance, whereas demographic and individual characteristics had limited associations with prevalence and severity estimates.”

In the paper

“Specifically, pooled estimates indicated that 29.8% (26.1–33.8%) reported elevated depressive symptoms”

In the paper

“22.9% (18.4–27.9%) alcohol misuse”

In the paper

“Symptom severity estimates, capturing mean symptom levels on continuous scales, on average were ‘mild’ for depressive and anxiety symptoms and ‘moderate’ for stress.”

In the paper

“Symptom severity estimates, capturing mean symptom levels on continuous scales, on average were ‘mild’ for depressive and anxiety symptoms and ‘moderate’ for stress.”

In the paper

“28.3% (23.6–33.3%) eating disorder symptoms”

In the paper

“18.8% (15.2–22.6%) suicidal ideation”

In the paper

“Moderator analyses showed that outcome and scale type accounted for the largest share of variance, whereas demographic and individual characteristics had limited associations with prevalence and severity estimates.”

In the paper

“On the basis of validated case-identification thresholds, 29.9% (95% confidence interval, 26.1–33.9%) of ECRs report elevated psychological distress across all outcomes.”

In the paper

“18.6% (14.4–23.2%) non-suicidal self-injury”

In the paper

“29.7% (25.7–33.8%) anxiety symptoms”

In the paper

“In the severity model, R2 of the syndrome and symptom domain was 48.7%.”

In the paper

“According to the current data, a large proportion of ECRs reported elevated levels of psychological distress.”

In the paper

“Assessing the linear and quadratic association between prevalence and year of data collection indicated a similar effect of the time of the COVID-19 pandemic.”

In the paper

“Small but statistically significant associations were observed for loneliness and age (both were associated with lower prevalence) and history of mental health disorders (associated with higher prevalence).”

In the paper

“Likewise, suicidal ideation did not exceed the clinical cut-off.”

In the paper

“9.7% (95% CI, 8.2–11.4) severe anxiety.”

In the paper

“The pooled prevalence of anxiety symptoms in the most used scale, the GAD-7, was 27.6% (95% CI, 23.4–31.9)”

In the paper

“Small but statistically significant associations were observed for loneliness and age (both were associated with lower prevalence) and history of mental health disorders (associated with higher prevalence).”

In the paper

“with 30.9% (95% CI, 28.4–33.4) reporting mild”

In the paper

“From 228 samples drawn from 148 studies and four databases (N = 138,446), our results indicate that 29% of ECRs scored above screening thresholds for at least one syndrome or symptom.”

In the paper

“Study quality was a significant moderator, such that higher study quality was associated with lower prevalence (−0.14% per JBI score point; 95% CI, −0.40 to −0.01; R2 = 11.6%) and severity estimates (−3.06 POMP score points per JBI score point; 95% CI, −4.65 to −1.46; R2 = 7.7%).”

In the paper

“The multivariate moderator model explained 43.1% of the true effect-size heterogeneity in the prevalence model (F(55, 296) = 9.61, P < 0.001) and 52.4% in the severity model (F(100, 435) = 10.04, P < 0.001).”

In the paper

“Post-hoc sensitivity analyses showed that the pooled prevalence of depression in the PHQ-9 was slightly higher (by 0.37%; 95% CI, 0.07–0.90; explaining R2 = 5.7% of the true effect-size variance) than in all other scales measuring depression”

In the paper

“Beyond depressive and anxiety symptoms, elevated prevalence was also observed for other important syndromes and symptoms, such as poor sleep quality, eating disorder symptoms, alcohol abuse, suicidal ideation and non-suicidal self-injury.”

In the paper

“The scale used to assess mental health syndromes or symptom domains moderated both prevalence and severity estimates (R2 = 34.4% and 53.5%, respectively).”

In the paper

“In terms of severity, ECRs reported mild levels of depression and anxiety, moderate stress and generally poor sleep quality, while levels of severity of suicidal ideation and eating disorders fell below clinical cut-offs.”

In the paper

“For symptoms, average sleep quality was poor (with a wide estimate range across studies), and stress was consistently reported as moderate, underscoring the wide range of stressors experienced by ECRs62.”

In the paper

“The pooled prevalence of depressive symptoms in the most used scale, the Patient Health Questionnaire (PHQ-9 and PHQ-8), was 28.5% (95% CI, 24.3–32.9)”

In the paper

“A sensitivity analysis revealed that publication bias did not impact the significance of the already significant moderators (Supplementary Table 18).”

In the paper

“For example, the pooled severity for depressive symptoms as assessed with the PHQ-9 (44 studies, 67 effect sizes; n = 72.239) was 6.99 (95% CI, 6.12–7.86), indicating symptom levels in the mild range.”

In the paper

“Depending on the measurement scale, the average symptom severity across syndromes ranged from mild to severe for depressive symptoms and from mild to borderline abnormal for anxiety symptoms, and fell below screening-based severity cut-offs for eating disorder symptoms, substance abuse (for example, alcohol abuse) and suicidal ideation.”

In the paper

“4.7% (95% CI, 3.7–5.9) severe depressive symptoms.”

In the paper

“17.0% (95% CI, 15.2–18.9) moderate”

In the paper

“Across all outcomes, nearly one in three ECRs reported elevated psychological distress (pooled prevalence, 29.9%; 95% confidence interval (CI), 26.1–33.9%; k = 86).”

In the paper

“In terms of severity, ECRs reported mild levels of depression and anxiety, moderate stress and generally poor sleep quality, while levels of severity of suicidal ideation and eating disorders fell below clinical cut-offs.”

In the paper

“8.1% (95% CI, 6.8–9.5) moderately severe”

In the paper

“Sampling strategy was a significant moderator as well, with studies using full or random sampling reporting lower prevalence (−2.61%; 95% CI, −5.37 to −0.83; R2 = 11.6%) and lower severity estimates (−17.7 POMP score points; 95% CI, −23.1 to −12.29; R2 = 15.3%) than those using convenience sampling.”

In the paper

“The multivariate moderator model explained 43.1% of the true effect-size heterogeneity in the prevalence model (F(55, 296) = 9.61, P < 0.001) and 52.4% in the severity model (F(100, 435) = 10.04, P < 0.001).”

In the paper

“The pooled prevalence of suicidal ideation (item 9 of the PHQ-9) was 17.4% (95% CI, 13.8–21.4).”

In the paper

“In the prevalence model, COVID-19 significantly moderated the pooled prevalence of psychological distress (R2 = 8.7%), such that the prevalence was significantly higher during the COVID-19 period (34.2%; 95% CI, 29.4–39.2) than before (24.3%; 95% CI, 19.8–29.02).”

In the paper

“and loneliness decreased prevalence estimates (−0.002% per 1 POMP-score point increase of loneliness; 95% CI, −0.011 to −0.000)”

In the paper

“In terms of severity, ECRs reported mild levels of depression and anxiety, moderate stress and generally poor sleep quality, while levels of severity of suicidal ideation and eating disorders fell below clinical cut-offs.”

In the paper

“severity estimates indicated that 33.1% (95% CI, 30.1–35.5) of ECRs reported mild symptoms”

In the paper

“Additionally, a polynomial moderator analysis with publication year modelled as a linear and quadratic predictor revealed no significant time trends (Supplementary Table 18 and Supplementary Fig. 13).”

In the paper

“For severity, there was no evidence that sample characteristics acted as moderators.”

In the paper

“Post-pandemic prevalence remained significantly elevated (33.3%; 95% CI, 26.0–41.0; see also Fig. 4).”

In the paper

“while the pooled severity of anxiety in the GAD-7 was slightly higher (by 7.27 POMP scores; 95% CI, 3.70–10.84; R2 = 2.6%) than in all other scales measuring anxiety”

In the paper

“16.0% (95% CI, 14.1–18.0) moderate”

In the paper

“Sampling strategy was a significant moderator as well, with studies using full or random sampling reporting lower prevalence (−2.61%; 95% CI, −5.37 to −0.83; R2 = 11.6%) and lower severity estimates (−17.7 POMP score points; 95% CI, −23.1 to −12.29; R2 = 15.3%) than those using convenience sampling.”

In the paper

“For example, prevalence and severity estimates were comparable across disciplines, genders and career stages.”

In the paper

“In the prevalence model, COVID-19 significantly moderated the pooled prevalence of psychological distress (R2 = 8.7%), such that the prevalence was significantly higher during the COVID-19 period (34.2%; 95% CI, 29.4–39.2) than before (24.3%; 95% CI, 19.8–29.02).”

In the paper

“Prevalence was not consistently associated with most sample characteristics.”

In the paper

“Symptom severity across all outcomes was 33.9 (95% CI, 31.7–36.1; k = 151) on a normalized scale from 0 to 100, using per cent of maximum possible (POMP) scores”

In the paper

“Small but statistically significant associations were observed for loneliness and age (both were associated with lower prevalence) and history of mental health disorders (associated with higher prevalence).”

In the paper

“Study quality was a significant moderator, such that higher study quality was associated with lower prevalence (−0.14% per JBI score point; 95% CI, −0.40 to −0.01; R2 = 11.6%) and severity estimates (−3.06 POMP score points per JBI score point; 95% CI, −4.65 to −1.46; R2 = 7.7%).”

In the paper

“The qualitative assessments of asymmetry in the funnel plots of the prevalence and severity models (Supplementary Figs. 14 and 15) were both indicative of publication bias (via small-study effects—that is, larger effects among smaller and less precise studies).”

In the paper

“In the prevalence model, the presence of a mental health diagnosis prior to the study increased prevalence estimates (0.002% per 1% increase of prior mental health diagnosis; 95% CI, 0.000–0.006)”

In the paper

“Although this meta-analysis did not find any significant buffering or exacerbating effects of supervisor relationships, social support, loneliness, or life and work satisfaction, this does not mean that such effects do not exist.”

In the paper

“For example, only one of the studies conducted an external assessment via clinical-diagnostic interviews to establish diagnoses, rather than relying on self-rating scales49.”

What the sources say

• Check pendingcites [49] Nagy, G. A. et al. Burnout and mental health problems in biomedical doctoral students. CBE Life Sci. Educ. 18, ar27 (2019). source

In the paper

“Second, postdoctoral ECRs made up less than 3% of the overall sample5,52,57,66,68,87–93, yet they face slightly different challenges than doctoral students, and thus, more research should be dedicated to this group.”

What the sources say

• ?Not checked — no full text of the sourcecites [5] Trang, D. et al. A survey of the severity of mental health symptoms in the planetary science community. Nat. Astron. 8, 691–696 (2024). source
• Not checked yetcites [52] Healthy Minds Study among Colleges and Universities 2007–2024 (Healthy Minds Network, Univ. Michigan, Univ. California Los Angeles, Boston Univ. & Wayne State Univ., 2024).
• Check pendingcites [57] Thiagarajah, K., Sit, N. W. & Chee, H. P. Prevalence of sleep quality among academic staff of a private university in Malaysia. Malays. J. Med. Health Sci. 19, 28–34 (2023). source
• Check pendingcites [66] Van Der Weijden, I. & Teelken, C. Precarious careers: postdoctoral researchers and wellbeing at work. Stud. High. Educ. 48, 1595–1607 (2023). source
• Check pendingcites [68] Hanitzsch, T., Markiewitz, A. & Bødker, H. Publish and perish: mental health among communication and media scholars. J. Commun. 74, 429–442 (2024). source
• ?Not checked — no full text of the sourcecites [87] Bitter, A. N. & McCrea, S. M. “The ship is sinking, yet the band plays on”: COVID-19 and academia. J. Am. Coll. Health 72, 1876–1885 (2024). source
• ?Not checked — no full text of the sourcecites [88] Castelló, M., McAlpine, L. & Pyhältö, K. Spanish and UK post-PhD researchers: writing perceptions, well-being and productivity. High. Educ. Res. Dev. 36, 1108–1122 (2017). source
• Check pendingcites [89] Cilli, E., Ranieri, J., Guerra, F. & Di Giacomo, D. Early career researchers and mental health: observational study of challenge and wellbeing. Health Sci. Rep. 6, e1649 (2023). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [90] Dhont, J. et al. Conducting research in radiation oncology remotely during the COVID-19 pandemic: coping with isolation. Clin. Transl. Radiat. Oncol. 24, 53–59 (2020).
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [91] Gloria, C. T. & Steinhardt, M. A. Flourishing, languishing, and depressed postdoctoral fellows: differences in stress, anxiety, and depressive symptoms. J. Postdr. Aff. 3, 1–9 (2013).

In the paper

“Additionally, only a few studies (predominantly databases) used random (instead of convenience) sampling techniques50–53,56–58.”

What the sources say

• Check pendingcites [50] Charles, S. T., Karnaze, M. M. & Leslie, F. M. Positive factors related to graduate student mental health. J. Am. Coll. Health 70, 1858–1866 (2022). source
• Check pendingcites [51] Farrer, L. M., Gulliver, A., Bennett, K., Fassnacht, D. B. & Griffiths, K. M. Demographic and psychosocial predictors of major depression and generalised anxiety disorder in Australian university students. BMC Psychiatry 16, 241 (2016). source
• Not checked yetcites [52] Healthy Minds Study among Colleges and Universities 2007–2024 (Healthy Minds Network, Univ. Michigan, Univ. California Los Angeles, Boston Univ. & Wayne State Univ., 2024).
• ?Not checked — no full text of the sourcecites [53] Lt, L., Hense, S., Kodali, P. B. & Thankappan, K. R. Prevalence and underlying factors of depressive disorders among PhD students: a mixed-method study in the Indian context. J. Appl. Res. High. Educ. 14, 1704–1717 (2022). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [56] Talkhi, N., Ghavami, V., Jamali, M. R., Jamali, H. & Jamali, J. Relationship between minor psychiatric disorders and health literacy among students: a latent class regression. J. Health Lit. 5, 11–20 (2020).
• Check pendingcites [57] Thiagarajah, K., Sit, N. W. & Chee, H. P. Prevalence of sleep quality among academic staff of a private university in Malaysia. Malays. J. Med. Health Sci. 19, 28–34 (2023). source
• Check pendingcites [58] Ahalli, S., Fort, E., Bridai, Y., Baborier, N. & Charbotel, B. Mental health and working constraints of first-year PhD students in health and science in a French university: a cross-sectional study in the context of occupational health monitoring. BMJ Open 12, e057679 (2022). source

In the paper

“First, the quality of the primary studies eligible for this meta-analysis varied greatly.”

In the paper

“The prevalence of depressive symptoms was approximately two to three times higher and anxiety symptoms three to five times higher than in age-matched general-population samples.”

What the sources say

• Not checked yetcites [31] Terlizzi, E. P. & Villarroel, M. A. Symptoms of Generalized Anxiety Disorder among Adults: United States, 2019 NCHS Data Brief No. 378 (Centers for Disease Control and Prevention, 2020).
• Not checked yetcites [32] Adult Psychiatric Morbidity Survey: Survey of Mental Health and Wellbeing, England, 2023/4: Data Tables (NHS England, 2025); https://digital.nhs.uk/data-and-information/publications/statistical/adult-psychiatric-morbidity-survey/survey-of-mental-health-and-wellbeing-england-2023-24/data-sets
• ?Not checked — no full text of the sourcecites [33] Busch, M., Maske, U., Ryl, L., Schlack, R. & Hapke, U. Prevalence of depressive symptoms and diagnosed depression among adults in Germany. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 56, 733–739 (2013). source

In the paper

“Doctoral attrition remains at alarmingly high rates, with estimates between 30% and 50% of doctoral candidates leaving their programmes before graduating20–23”

What the sources say

• Check pendingcites [20] Wollast, R. et al. Facing the dropout crisis among PhD candidates: the role of supervisor support in emotional well-being and intended doctoral persistence among men and women. Stud. High. Educ. 48, 813–828 (2023). source
• ?Not checked — no full text of the sourcecites [21] Golde, C. M. The role of the department and discipline in doctoral student attrition: lessons from four departments. J. High. Educ. 76, 669–700 (2005). source
• Check pendingcites [22] Glorieux, A., Spruyt, B., Minnen, J. & van Tienoven, T. P. Calling it quits: a longitudinal study of factors associated with dropout among doctoral students. Stud. Contin. Educ. 47, 155–173 (2025). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [23] Hasgall, A., Saenen, B. & Borrell-Damian, L. Doctoral Education in Europe Today: Approaches and Institutional Structures (European University Association, 2019).

In the paper

“and in postdoctoral researchers (for example, short-term contracts, career uncertainty and pressure to publish or secure funding)12,19.”

What the sources say

• ?Not checked — no full text of the sourcecites [12] Jaremka, L. M. et al. Common academic experiences no one talks about: repeated rejection, impostor syndrome, and burnout. Perspect. Psychol. Sci. 15, 519–543 (2020). source
• ?Not checked — no full text of the sourcecites [19] Gewin, V. Postdoc depression and anxiety rates are rising, finds survey of 872 researchers. Nature https://doi.org/10.1038/d41586-025-02450-9 (2025). source

In the paper

“Average symptom severity typically falls in the mild range9–11.”

What the sources say

• Not checked yetcites [9] Chi, T., Cheng, L. & Zhang, Z. Global prevalence and trend of anxiety among graduate students: a systematic review and meta-analysis. Brain Behav. 13, e2909 (2023). source
• ?Not checked — no full text of the sourcecites [10] Guo, L. et al. Prevalence and changes in depressive symptoms among postgraduate students: a systematic review and meta-analysis from 1980 to 2020. Stress Health 37, 835–847 (2021). source
• Check pendingcites [11] Satinsky, E. N. et al. Systematic review and meta-analysis of depression, anxiety, and suicidal ideation among Ph.D. students. Sci. Rep. 11, 14370 (2021). source

In the paper

“The prevalence of depressive symptoms was approximately two to three times higher and anxiety symptoms three to five times higher than in age-matched general-population samples.”

What the sources say

• Not checked yetcites [31] Terlizzi, E. P. & Villarroel, M. A. Symptoms of Generalized Anxiety Disorder among Adults: United States, 2019 NCHS Data Brief No. 378 (Centers for Disease Control and Prevention, 2020).
• Not checked yetcites [32] Adult Psychiatric Morbidity Survey: Survey of Mental Health and Wellbeing, England, 2023/4: Data Tables (NHS England, 2025); https://digital.nhs.uk/data-and-information/publications/statistical/adult-psychiatric-morbidity-survey/survey-of-mental-health-and-wellbeing-england-2023-24/data-sets
• ?Not checked — no full text of the sourcecites [33] Busch, M., Maske, U., Ryl, L., Schlack, R. & Hapke, U. Prevalence of depressive symptoms and diagnosed depression among adults in Germany. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 56, 733–739 (2013). source

In the paper

“A broader systematic evaluation of other relevant mental health issues remains lacking, including clinically defined syndromes such as social phobia, obsessive-compulsive disorder (OCD), post-traumatic stress disorder (PTSD), attention deficit hyperactivity disorder (ADHD), personality disorders, eating disorders, alcohol and drug abuse, and burnout, as well as common symptoms of psychological”

What the sources say

• ?Not checked — no full text of the sourcecites [35] Inguaggiato, G. et al. The experience of women researchers during the COVID-19 pandemic: a scoping review. Res. Ethics 20, 780–811 (2024). source
• Check pendingcites [36] Jackman, P. C., Jacobs, L., Hawkins, R. M. & Sisson, K. Mental health and psychological wellbeing in the early stages of doctoral study: a systematic review. Eur. J. High. Educ. 12, 293–313 (2022). source
• ?Not checked — no full text of the sourcecites [37] Lokhtina, I. A. et al. The impact of the COVID-19 pandemic on early career researcher activity, development, career, and well-being: the state of the art. Stud. Grad. Postdr. Educ. 13, 245–265 (2022). source
• ?Not checked — no full text of the sourcecites [38] Mackie, S. A. & Bates, G. W. Contribution of the doctoral education environment to PhD candidates’ mental health problems: a scoping review. High. Educ. Res. Dev. 38, 565–578 (2019). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [39] Melián, E., Reyes, J. I. & Meneses, J. The online PhD experience: a qualitative systematic review. Int. Rev. Res. Open Distrib. Learn. 24, 137–158 (2023).
• Check pendingcites [40] Nicholls, H., Nicholls, M., Tekin, S., Lamb, D. & Billings, J. The impact of working in academia on researchers’ mental health and well-being: a systematic review and qualitative meta-synthesis. PLoS ONE 17, e0268890 (2022). source
• Check pendingcites [41] Schmidt, M. & Hansson, E. Doctoral students’ well-being: a literature review. Int. J. Qual. Stud. Health Well Being 13, 1508171 (2018). source
• Check pendingcites [42] Martínez-García, I., De Witte, H., García-Martínez, J. & Cano-García, F. J. A systematic review and a comprehensive approach to PhD students’ wellbeing. Appl. Psychol. Health Well Being 16, 1565–1583 (2024). source

In the paper

“One meta-analysis of doctoral students reported that 36% experienced mild, 12% moderate and 9% severe depressive symptoms10.”

What the sources say

• ?Not checked — no full text of the sourcecites [10] Guo, L. et al. Prevalence and changes in depressive symptoms among postgraduate students: a systematic review and meta-analysis from 1980 to 2020. Stress Health 37, 835–847 (2021). source

In the paper

“A recent meta-analysis on doctoral students and medical residents estimated that 9% reported suicidal ideation in the past year24.”

What the sources say

• Check pendingcites [24] Poli, M. et al. Suicide risk among residents and PhD students: a systematic review of the literature. J. Psychiatr. Res. 181, 433–462 (2025). source

In the paper

“Also, only a fraction (k = 23; 10.1%) of studies reported an adequate survey response rate (≥45%)”

In the paper

“The modest associations of individual-level moderators with psychological distress suggest that the primary drivers of poor mental health among ECRs are structural rather than personal.”

In the paper

“The most frequently cited systemic factor is the mismatch between the growing number of PhD graduates and the comparatively limited number of permanent positions in academia4,40,66–68.”

What the sources say

• Check pendingcites [4] Rahal, R.-M. et al. Quality research needs good working conditions. Nat. Hum. Behav. 7, 164–167 (2023). source
• Check pendingcites [40] Nicholls, H., Nicholls, M., Tekin, S., Lamb, D. & Billings, J. The impact of working in academia on researchers’ mental health and well-being: a systematic review and qualitative meta-synthesis. PLoS ONE 17, e0268890 (2022). source
• Check pendingcites [66] Van Der Weijden, I. & Teelken, C. Precarious careers: postdoctoral researchers and wellbeing at work. Stud. High. Educ. 48, 1595–1607 (2023). source
• Check pendingcites [67] Alberts, B., Kirschner, M. W., Tilghman, S. & Varmus, H. Rescuing US biomedical research from its systemic flaws. Proc. Natl Acad. Sci. USA 111, 5773–5777 (2014). source
• Check pendingcites [68] Hanitzsch, T., Markiewitz, A. & Bødker, H. Publish and perish: mental health among communication and media scholars. J. Commun. 74, 429–442 (2024). source

In the paper

“The consistency of these findings across sample characteristics further underscores the widespread extent of mental health issues among ECRs.”

In the paper

“Only one study used external assessments to measure psychological distress49”

What the sources say

• Check pendingcites [49] Nagy, G. A. et al. Burnout and mental health problems in biomedical doctoral students. CBE Life Sci. Educ. 18, ar27 (2019). source

In the paper

“Together, the consistently elevated prevalence across a wide range of mental health outcomes underscores the broad scope of psychological distress among ECRs.”

In the paper

“Moderation analyses indicated that prevalence and severity estimates were largely unaffected by demographic or individual factors, but rather due to study-level characteristics such as sampling strategy, study quality and type of measurement scale.”

In the paper

“These findings are in line with those of previous meta-analyses, which found that low-quality primary studies or those using non-random sampling techniques reported higher prevalence rates for depressive10 and anxiety symptoms11.”

What the sources say

• ?Not checked — no full text of the sourcecites [10] Guo, L. et al. Prevalence and changes in depressive symptoms among postgraduate students: a systematic review and meta-analysis from 1980 to 2020. Stress Health 37, 835–847 (2021). source
• Check pendingcites [11] Satinsky, E. N. et al. Systematic review and meta-analysis of depression, anxiety, and suicidal ideation among Ph.D. students. Sci. Rep. 11, 14370 (2021). source

In the paper

“These results suggest that ECRs most often reported feelings of being ineffective or lacking personal accomplishment.”

In the paper

“In other words, even mild or subclinical symptom severity can indicate cause for concern and elevated future risk60,63.”

What the sources say

• ?Not checked — no full text of the sourcecites [60] Fergusson, D. M., Horwood, L. J., Ridder, E. M. & Beautrais, A. L. Subthreshold depression in adolescence and mental health outcomes in adulthood. Arch. Gen. Psychiatry 62, 66–72 (2005). source
• ?Not checked — no full text of the sourcecites [63] van Os, J. The dynamics of subthreshold psychopathology: implications for diagnosis and treatment. Am. J. Psychiatry 170, 695–698 (2013). source

In the paper

“and prevalence estimates were almost never accompanied by confidence intervals”

In the paper

“random sampling was used rarely (for exceptions, see refs. 50–58).”

What the sources say

• Check pendingcites [50] Charles, S. T., Karnaze, M. M. & Leslie, F. M. Positive factors related to graduate student mental health. J. Am. Coll. Health 70, 1858–1866 (2022). source
• Check pendingcites [51] Farrer, L. M., Gulliver, A., Bennett, K., Fassnacht, D. B. & Griffiths, K. M. Demographic and psychosocial predictors of major depression and generalised anxiety disorder in Australian university students. BMC Psychiatry 16, 241 (2016). source
• Not checked yetcites [52] Healthy Minds Study among Colleges and Universities 2007–2024 (Healthy Minds Network, Univ. Michigan, Univ. California Los Angeles, Boston Univ. & Wayne State Univ., 2024).
• ?Not checked — no full text of the sourcecites [53] Lt, L., Hense, S., Kodali, P. B. & Thankappan, K. R. Prevalence and underlying factors of depressive disorders among PhD students: a mixed-method study in the Indian context. J. Appl. Res. High. Educ. 14, 1704–1717 (2022). source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [54] Panger, G., Tyron, J. & Smith, A. Graduate Student Happiness & Well-Being Report (University of California, Berkeley, 2014).
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [55] Sverdlik, A., Hall, N. C. & McAlpine, L. PhD imposter syndrome: exploring antecedents, consequences, and implications for doctoral well-being. Int. J. Dr. Stud. 15, 737–758 (2020).
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [56] Talkhi, N., Ghavami, V., Jamali, M. R., Jamali, H. & Jamali, J. Relationship between minor psychiatric disorders and health literacy among students: a latent class regression. J. Health Lit. 5, 11–20 (2020).
• Check pendingcites [57] Thiagarajah, K., Sit, N. W. & Chee, H. P. Prevalence of sleep quality among academic staff of a private university in Malaysia. Malays. J. Med. Health Sci. 19, 28–34 (2023). source
• Check pendingcites [58] Ahalli, S., Fort, E., Bridai, Y., Baborier, N. & Charbotel, B. Mental health and working constraints of first-year PhD students in health and science in a French university: a cross-sectional study in the context of occupational health monitoring. BMJ Open 12, e057679 (2022). source

In the paper

“Only 57.9% of studies had sufficient sample sizes (k = 132).”

In the paper

“This overestimation of the precision of individual prevalence rates entails the assignment of too-high weights in the meta-analytic model when estimating the pooled effect size.”

In the paper

“As almost all tests of these moderators lacked power, future research should direct more efforts to determine potential protective and risk factors in a congruent manner and thereby deduce effective and evidence-based interventions and implement systemic changes to reduce the psychological distress of ECRs.”

In the paper

“This causes the standard error to become artificially compressed for rates close to either 0% or 100%.”

In the paper

“Hence, prevalence rates were transformed prior to analysis, using the single (not double) arcsine transformation to stabilize the variance, which has been proved to generate more adequate results than the double arcsine transformation when dealing with varying sample sizes228,229:transformedES=ASIN(√ES),transformedvar=1/4n.”

What the sources say

• Check pendingcites [228] Röver, C. & Friede, T. Double arcsine transform not appropriate for meta-analysis. Res. Synth. Methods 13, 645–648 (2022). source
• Not checked yetcites [229] Schwarzer, G., Chemaitelly, H., Abu-Raddad, L. J. & Rücker, G. Seriously misleading results using inverse of Freeman–Tukey double arcsine transformation in meta-analysis of single proportions. Res. Synth. Methods 10, 476–483 (2019). source

In the paper

“As studies reported more than one effect size (for example, depression and suicidal ideation), we opted for a multilevel meta-analysis with three levels (for example, ref. 47), a random-effect model, which enables modelling the dependent effect sizes from the same study.”

What the sources say

• Not checked yetcites [47] Assink, M. & Wibbelink, C. J. M. Fitting three-level meta-analytic models in R: a step-by-step tutorial. Quant. Methods Psychol. 12, 154–174 (2016). source

In the paper

“Although different LLMs yielded converging classifications, further research is needed to develop and validate more robust frameworks for grouping countries on the basis of academic working conditions.”

In the paper

“Finally, for ECRs from underrepresented groups, additional barriers may exist, such as discrimination, limited mobility and unequal access to mentorship and resources40,78,79.”

What the sources say

• Check pendingcites [40] Nicholls, H., Nicholls, M., Tekin, S., Lamb, D. & Billings, J. The impact of working in academia on researchers’ mental health and well-being: a systematic review and qualitative meta-synthesis. PLoS ONE 17, e0268890 (2022). source
• Check pendingcites [78] Bowden, A. K. & Buie, C. R. Anti-Black racism in academia and what you can do about it. Nat. Rev. Mater. 6, 760–761 (2021). source
• Check pendingcites [79] Chen, C. Y. et al. Systemic racial disparities in funding rates at the National Science Foundation. Elife 11, e83071 (2022). source