Global Prevalence of Depression Among University Students: A Comprehensive Umbrella Review

Abstract

Depression has become a major concern for educational institutes due to its increasing incidence and detrimental impacts on students’ well-being. However, its global prevalence in the context of tertiary education remains unclear. This umbrella review summarised existing meta-analyses and systematic reviews to examine the worldwide prevalence of depression among university students. Several sociodemographic and methodological factors, including differences in depression prevalence before, during, and after COVID-19, were examined as moderators. Structured and exhaustive searches were conducted in five bibliographic databases and five journals, alongside two sources of grey literature. A total of 61 meta-analyses and systematic reviews were included in the final synthesis. Review-level sample sizes ranged from 823 to 8,238,954 (Mdn = 24,047), covering all inhabited continents and a broad range of countries across high-, middle-, and low-income settings. The global median prevalence of depression across the included reviews was 32.00%, with reported estimates ranging from 11.00% to 63.00%. Analysis of specific populations highlighted greater vulnerability among female students and those studying in Asia compared to Europe. The pandemic was shown to have contributed to the increase in depression prevalence among university students. Given this high prevalence of depression, there is an imperative need for educational institutions to prioritise mental well-being.

1. Introduction

University life introduces a host of new challenges as students transition into higher education. Navigating the shift to university often requires students to manage new educational demands (e.g., increased workload and expectations to perform well academically), financial concerns, and evolving social dynamics, which are challenges that may heighten stress and compromise emotional well-being (Andrews & Wilding, 2004; Mofatteh, 2021). Indeed, these stressors have been shown to contribute to the deterioration of mental health among university students (Hartanto et al., 2024; Lua et al., 2022; Mofatteh, 2021; Nakhostin-Ansari et al., 2020; Zhang et al., 2022), as well as an increased vulnerability to psychological disorders (Asif et al., 2020; Eisenberg et al., 2011; Lipson et al., 2022; Mohamad et al., 2021; Turner et al., 2007). The frequent comorbidity between depression and anxiety disorders may further exacerbate students’ mental health challenges (Hirschfeld, 2001). G. X. D. Tan et al. (2023) found that the global median prevalence of anxiety among university students was 32.00%, with reported estimates ranging from 7.40% to 55.00%, suggesting that depression may be similarly pervasive. Given that depression requires distinct assessment, treatment, and targeted support, establishing its prevalence independently is essential for guiding appropriate institutional interventions.

Research reflects a rise in depression rates among university students. A study of 155,026 university students from 196 campuses in the United States reported that the proportion of positive PHQ-2 scores increased from 24.80% in 2009 to 29.90% between 2016 and 2017 (Lipson et al., 2019). This finding was corroborated by a study of 3092 South African university students, which found that the proportion of moderate-severe BDI-II scores had increased from a low of 14.44% in 2016 to a high of 31.66% in 2019 (Rousseau et al., 2021). This upward trend in depression rates is a cause for concern, even more so as research on 560,000 college students from the United States found that suicidal ideation increased substantially alongside depressive symptoms between the years 2007 and 2022 (Vidal et al., 2026). Indeed, an increase in depression prevalence may lead to adverse outcomes such as poor academic performance, higher university dropout rates, increased vulnerability towards substance dependence as well as self-injurious or suicidal tendencies (Adams et al., 2021; Eisenberg et al., 2009; Garlow et al., 2008; Hjorth et al., 2016; Santana et al., 2023).

The unprecedented impact of COVID-19 has also exacerbated depression among university students (Azmi et al., 2022; Kaparounaki et al., 2020; Luo et al., 2021). Numerous studies report an increase in the prevalence of depressive symptoms during the pandemic (e.g., Kaparounaki et al., 2020; Lee et al., 2021; Luo et al., 2021; McLafferty et al., 2021). For example, McLafferty et al. (2021) examined 884 college students in Ireland and found that the prevalence rate of depressive symptoms increased from 14.90% before Autumn 2019 to 23.60% in Autumn 2020, corresponding to the pre-pandemic and early pandemic periods respectively. Similarly, Lee et al. (2021) found that 54.10% of college students reported an increase in depressive symptoms during the pandemic. This rise in depression prevalence may be attributed to factors such as the fear of contracting the virus, prolonged social isolation and the loss of loved ones (Azmi et al., 2022; Pietrabissa & Simpson, 2020). Emerging evidence suggests that the impact of COVID-19 on mental health may persist post-pandemic, particularly among young adults and student populations facing structural instability, socioeconomic insecurity, and uncertainty about their future direction (Catling, 2023; Essadek et al., 2025), highlighting a need to examine depression prevalence across different stages of the pandemic to inform preparedness for future public health crises.

Numerous meta-analyses and systematic reviews have examined the prevalence of depression among university students (Khan et al., 2021; Lei et al., 2016; Luo et al., 2021; Sarokhani et al., 2013). However, most of these reviews are limited in geographical variance (e.g., consisting of studies specific to a certain country such as China or continent such as Asia), study discipline (e.g., consisting of course-specific studies such as medicine, dental or nursing), degree type (i.e., consisting of studies which analysed either undergraduates or postgraduate students), and stages of the COVID-19 pandemic (i.e., consisting of studies made available before, during or after the pandemic). Most reviews are also limited in the type of assessment used, as they consist wholly of studies deriving prevalence estimates from self-reported depressive symptoms, which are not equivalent to clinician-administered diagnoses and may yield higher prevalence estimates. These methodological limitations obscure our understanding of the prevalence of depression among university students on a global scale.

To address these issues, the current umbrella review of meta-analyses and systematic reviews aims to examine the worldwide prevalence of depression among university students. Umbrella reviews are beneficial as they provide a comprehensive, macro-level overview of meta-analyses which themselves provide a narrower review of constituent studies (Belbasis et al., 2022), that is particularly useful in evidence-based decision-making (e.g., whether the risk factors associated with depression among university students warrant increased institutional investment in strengthening existing mental health support mechanisms; Choi & Kang, 2023; Hartanto et al., 2022). To distinguish the present review from umbrella reviews addressing broad mental health problems among university students, we focused specifically on depression to examine its sources of variation in greater depth. Specifically, we built on the subgroup analyses conducted by Paiva et al. (2025) by examining differences relating to geography and study discipline, while additionally investigating differences in degree and assessment types. The inclusion of these new moderators is supported by the different transitional and academic sources of stress faced by undergraduate and postgraduate students (Brooke et al., 2020; Guo et al., 2021; Wyatt & Oswalt, 2013), alongside evidence that self-report instruments produce higher prevalence estimates than clinician-administered diagnostic assessments (Lim et al., 2018). Consequently, the goals of this umbrella review are to (a) examine the global prevalence of depression in university students, (b) examine the trends in prevalence estimates from subgroup analyses, and (c) provide practical recommendations for schools and relevant institutions to consider.

2. Materials and Methods

2.1. Transparency and Openness

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021) following the Joanna Briggs Institute (JBI) Manual for Evidence Synthesis (Aromataris et al., 2024). The pre-registration, consisting of the design and synthesis plan, was first completed on 7 January 2023. Changes were made to the simplified search string and journal name (i.e., Journal of Abnormal Psychology was renamed Journal of Psychopathology and Clinical Science) on 9 January 2023. The original and updated versions of the pre-registration, the list of records screened across both searches, the data extraction and quality assessment files for the 61 included reviews, and the completed PRISMA 2020 checklist for this umbrella review were uploaded to OSF at https://osf.io/4nz6w/, accessed on 30 July 2026. Mendeley (v2.82.0) was utilised for the automatic removal of duplicate records.

2.2. Search Strategy

A detailed record of the sources searched and their search strings are provided in Table S1. Two comprehensive searches were conducted. The first was conducted for articles available up to 10 January 2023 by the second author. An updated search was conducted for articles available up to 18 February 2026 by the first author. To capture relevant peer-reviewed records, searches were conducted across five main databases (EBSCOhost ERIC, EBSCOhost PsycINFO, PubMed, Web of Science and Scopus) and five psychiatry journals with high Journal Citation Reports (JCR) rankings and impact factors (BMC Psychiatry, The American Journal of Psychiatry, Journal of Affective Disorders, Journal of Psychopathology and Clinical Science, and Depression and Anxiety). All records returned by the five main databases were retrieved for further screening. We supplemented our search by retrieving relevant records from the most recent issues of the five journals to minimise the risk of excluding records that had not yet been updated within the main databases. Additionally, two other sources (Google Scholar and ProQuest Dissertations & Theses Global) were searched to identify records not captured by the previous sources. These two sources support the inclusion of grey literature, which is shown to reduce publication bias and provide a more comprehensive understanding of the topic (Paez, 2017). The first 200 records sorted by relevance were retrieved from Google Scholar and ProQuest Dissertations & Theses Global respectively to balance search comprehensiveness with screening feasibility. No additional filters were applied to any source. All sources (i.e., five main databases, five journals, and two supplemental databases) were searched from the beginning of their available coverage until the relevant search date.

2.3. Selection Criteria

The eligibility of each record for inclusion in the current umbrella review was determined through a three-stage process following Hu et al. (2026), where (1) clearly irrelevant records were first removed based on information gathered from titles and abstracts, (2) relevant records were then identified for inclusion based on detailed readings of full texts, and (3) each relevant review was assessed for methodological quality and low-quality reviews were excluded. Records retrieved from the first search were independently screened by the second author and one of three authors (fourth, fifth or sixth) while records retrieved from the second search were independently screened by two authors (first and third). Agreement rates are presented in Table S2. Agreement across all criteria in each screening stage was generally good (see details in following sections) and all discrepancies were resolved via discussion between the two involved authors.

2.3.1. Assessment of Relevance

Firstly, titles and abstracts were screened based on the following exclusion criteria (mean inter-rater agreement = 95%, 88–99% raw agreement per criterion):

• Records were excluded if they were not reported in English.
• Records were excluded if they were neither meta-analyses nor systematic reviews.
• Records were excluded if they did not mention the specific term “depression”.

Secondly, the remaining records were screened based on their full texts against a detailed set of inclusion criteria (mean inter-rater agreement = 90%, 64–100% raw agreement per criterion):

• Records were deemed relevant if they were reported in English.
• Records were deemed relevant if they were meta-analyses or systematic reviews.
• Records were deemed relevant if they examined the prevalence of clinically diagnosed depression and/or depressive symptoms using at least one of the following assessment approaches:
  • A clinician-administered instrument and/or diagnostic interview (e.g., Structured Clinical Interview for DSM-5 [SCID-5]; First et al., 2016).
  • A clinician-administered or self-reported inventory (e.g., Hamilton Depression Scale; Hamilton, 1960), regardless of the psychometric validation.
• Records were deemed relevant if they reported depression prevalence as a percentage.
• Records were deemed relevant if they reported depression prevalence among university students, including those at the undergraduate and postgraduate levels. Records that conducted separate analyses for mixed samples of university and non-university students were eligible for inclusion.

2.3.2. Evaluation of Review Quality and Risk of Bias

We assessed the quality of each relevant review using the 11-item Joanna Briggs Institute checklist (JBI, 2017), which assesses key elements including how clearly the review question was formulated, the suitability of the inclusion criteria and search procedures, the thoroughness of quality appraisal processes, potential sources of bias (such as publication bias), and whether the recommendations and conclusions drawn are well supported by the evidence presented (mean inter-rater agreement = 86%, 38–100% raw agreement per criterion1). Reviews scoring at least five “Yes” out of 11 were considered sufficiently rigorous for inclusion.

2.4. Data Extraction

Two authors (i.e., second and fourth for the first search, first and third for the second search) independently extracted the following information from the included reviews: geographic coverage of the primary studies, type of constituent sample (i.e., undergraduate or postgraduate) and field of study (e.g., medical, dental, or nursing), total study count, total sample size, type of diagnosis (i.e., clinically diagnosed depression or depressive symptoms), type of measurement tools utilised (e.g., BDI, CES-D or DASS), overall depression prevalence, statistical model used to synthesise the overall prevalence (i.e., random-effects or fixed-effects model), and the reported results of any subgroup analyses (mean agreement across the two searches = 89%, 62–100% raw agreement per criterion; see Table S3 for breakdown). The authors of the corresponding reviews were contacted via email for clarification if any information was unclear. For consistency in reporting, each country was assigned to its corresponding geographic region (i.e., Africa, Asia-Pacific, Europe, Middle East, North America, or South America) based on the groupings provided by Wikimedia (Meta-Wiki, n.d.).

2.5. Data Synthesis

This umbrella review narratively synthesised the prevalence of depression from each included review. Beyond analysing the prevalence of depression among university students as a whole, we conducted subgroup analyses on several sociodemographic and methodological factors to further understand their moderating effects on depression. Subgroup synthesis at the umbrella level was conducted when conclusions could be drawn about the prevalence of depression across gender, degree type, study discipline, geographical location, COVID-19 phases (i.e., before, during, or after the pandemic), type of measurement tools, and type of diagnosis. A minimum of two reviews had to present the relevant sociodemographic or methodological information for conclusions to be drawn for each subgroup.

3. Results

3.1. Search Outcome and Eligibility

A total of 3790 unique records were retrieved across our first and second literature searches after automatic deduplication, from which 3322 records were removed during title and abstract screening, leaving 468 records for full-text screening. A subsequent 407 records were removed during full-text screening, after which all 61 reviews were critically appraised to be sufficiently rigorous for inclusion in our umbrella review. The full search and screening procedure is detailed in Figure 1.

All 61 reviews met the five-point cutoff criteria for the JBI critical appraisal tool (Mdn = 9, range = 5–11; see Table S4 for breakdown). Crucially, item four (“Were the sources and resources used to search for studies adequate?”) had the lowest number of “Yes” ratings (i.e., of the 61 reviews, only 23 or 38% were graded “Yes”), which highlights a need for methodological improvement in capturing grey literature when searching for studies (Figure 2). Nevertheless, none of the 61 reviews showed significant methodological biases overall, and thus all were considered sufficiently rigorous for inclusion in the final synthesis.

3.2. Review Characteristics

The characteristics of the 61 reviews included in the current umbrella review are shown in Table S5. Reviews were published or submitted between the years 2013 and 2025 (inclusive; Figure 3A) and consisted of studies that were made available between the years 1979 and 2024 (inclusive; Figure 3B). Overall sample sizes ranged from 823 to 8,238,954 (Mdn = 24,047; see Figure S1A), and the number of studies included in each meta-analysis or systematic review ranged from 5 to 188 (Mdn = 26; see Figure S1B). Among the 61 reviews, 19 reviews focused on medical students, 3 reviews focused on dental students, 1 review focused on both medical and dental students, 3 reviews focused on nursing students, 1 review focused on paramedic students, 1 review focused on health science students, 1 review focused on engineering students, 5 reviews focused specifically on undergraduate students across disciplines, 2 reviews focused specifically on postgraduate students across disciplines, and the remaining 25 reviews focused on college or university students in general (Figure 3C). The reviews covered at least 76 countries across all six regions, and China (35 reviews), India (22 reviews), and the United States (20 reviews) were the most frequently represented at the review level.

3.3. Prevalence of Depression

The overall prevalence of depression among university students ranged from 11.00% to 63.00% (Mdn = 32.00%) across the 61 reviews. Figure 4 summarizes the prevalence of depression reported in each review.

3.4. Subgroup Analyses

3.4.1. Demographic Factors

Gender. A total of 22 reviews examined the prevalence of depression in women (Mdn = 30.20%, Range = 14.00–56.00%) and in men (Mdn = 28.80%, Range = 15.00–51.30%). Of these 22 reviews, 15 (Batra et al., 2021; Chang et al., 2021; Deng et al., 2021; Heumann et al., 2024; Ibrahim et al., 2013; Jia et al., 2022; Kaur et al., 2024; Y. Li et al., 2021; Y.-K. Lin et al., 2024; Z.-Z. Lin et al., 2025; Luo et al., 2021; Melo et al., 2025; Pacheco et al., 2017; Puthran et al., 2016; Zatt et al., 2023) found the prevalence of depression to be higher in women (Mdn = 32.40%, Range = 18.73–56.00%) than in men (Mdn = 27.90%, Range = 16.43–50.00%), while 7 (Gao et al., 2020; Jaafari et al., 2021; Jiang et al., 2015; Jin et al., 2022; Lei et al., 2016; Sarokhani et al., 2013; Zeng et al., 2019) found the prevalence of depression to be higher in men (Mdn = 29.45%, Range = 15.00–51.30%) than in women (Mdn = 27.10%, Range = 14.00–48.90%). Most of these gender differences were statistically non-significant, except in two reviews where the prevalence rates of depression were significantly higher in women than in men (Chang et al., 2021; Ibrahim et al., 2013), and in one review where the prevalence rates of depression were significantly higher in men than in women (Sarokhani et al., 2013). Figure 5 provides an overview of these gender differences.

Degree Type. A total of eight reviews conducted subgroup analyses based on degree type (Akhtar et al., 2020; Deng et al., 2021; Jin et al., 2022; Y. Li et al., 2021; Luo et al., 2021; Puthran et al., 2016; J. Wang et al., 2023; Zhu et al., 2021). There was no evidence that the prevalence of depression differed between undergraduates (Mdn = 26.50%, Range = 22.80–32.50%) and postgraduates (Mdn = 28.60%, Range = 22.00–32.90%). Depression prevalence was higher among undergraduates and postgraduates in an equal number of reviews, and all differences were found to be statistically non-significant. Figure 6 provides an overview of these differences in degree type.

Study Discipline. We identified 14 reviews that conducted subgroup analyses based on study discipline (Akhtar et al., 2020; Cui et al., 2022; Demenech et al., 2021; Gao et al., 2020; Ibrahim et al., 2013; Khan et al., 2021; Lei et al., 2016; Y. Li et al., 2021; W. Li et al., 2022; Z.-Z. Lin et al., 2025; Luo et al., 2021; Niazi et al., 2025; Puthran et al., 2016; Zhu et al., 2021), with mixed findings (

Table 1. Comparison of Depression Prevalence by Study Discipline.

Author, Year	Subgroups	Number of Studies	Prevalence of Depression, 95% CI
Akhtar et al. (2020)	Medicine	15	22.40% [14.90%, 32.20%]
	Non-medicine	22	25.80% [18.70%, 34.50%]
Cui et al. (2022)	Health (Dental, Medicine, Pharmacy, Veterinary)	13	23.98% [14.68%, 36.63%]
	General	5	31.63% [13.08%, 58.72%]
Demenech et al. (2021)	Medicine	23	34.44% [8.10%, 70.00%]
	Health (Health area, Nursing, Odontology)	10	21.90% [7.34%, 47.70%]
	General	5	15.10% [4.61%, 25.70%]
Gao et al. (2020)	Medicine	6	25.70% [17.90%, 33.60%]
	Non-medicine	6	22.70% [16.70%, 28.60%]
Ibrahim et al. (2013)	Medicine	12	25.60% [23.20%, 26.60%]
	Non-medicine	11	35.60% [34.90%, 37.80%]
Khan et al. (2021)	Medicine	17	36.90% [27.14%, 47.86%]
	Non-medicine	9	53.59% [40.71%, 66.00%]
Lei et al. (2016)	Medicine	11	27.50% [19.80%, 38.30%]
	Non-medicine	28	22.40% [17.90%, 28.10%]
Y. Li et al. (2021)	Medicine	5	24.00% [19.00%, 30.00%]
	Non-medicine	2	18.00% [6.00%, 31.00%]
W. Li et al. (2022)	Medicine	10	39.40% [29.30%, 49.60%]
	Non-medicine	54	32.50% [28.00%, 37.00%]
Z.-Z. Lin et al. (2025)	Medicine	7	38.30% [28.30%, 48.50%]
	Non-medicine	25	33.70% [28.70%, 38.90%]
Luo et al. (2021)	Medicine	34	27.50% [21.00%, 34.60%]
	Non-medicine	20	27.50% [20.60%, 36.70%]
Niazi et al. (2025)	Medicine	24	48.72% [39.92%, 57.57%]
	Non-medicine	5	60.33% [49.04%, 71.08%]
Puthran et al. (2016)	Medicine	6	28.70% [15.70%, 46.50%]
	Non-medicine	6	30.60% [19.30%, 44.90%]
Zhu et al. (2021)	Healthcare	37	29.50% [23.80%, 36.00%]
	Non-healthcare	86	32.40% [28.40%, 36.60%]

). Of these 14 reviews, 11 examined the prevalence of depression, distinguishing between medical (Mdn = 27.50%, Range = 22.40–48.72%) and non-medical (Mdn = 30.60%, Range = 18.00–60.33%) students. A total of five reviews (Gao et al., 2020; Lei et al., 2016; Y. Li et al., 2021; W. Li et al., 2022; Z.-Z. Lin et al., 2025) suggested that medical students were more likely to be depressed (Mdn = 27.50%, Range = 24.00–39.40%) than non-medical students (Mdn = 22.70%, Range = 18.00–33.70%). These findings were corroborated by Demenech et al. (2021), who found that depression was most prevalent among medical (34.44%, 95% CI = [8.10%, 70.00%]) as compared to other health (21.90%, 95% CI = [7.34%, 47.70%]) or general (non-course specific) disciplines (15.10%, 95% CI = [4.61%, 25.70%]). In contrast, five reviews (Akhtar et al., 2020; Ibrahim et al., 2013; Khan et al., 2021; Niazi et al., 2025; Puthran et al., 2016) reported higher prevalence rates in non-medical students (Mdn = 35.60%, Range = 25.80–60.33%), as compared to medical students (Mdn = 28.70%, Range = 22.40–48.72%), while one review (Luo et al., 2021) found no differences between the two fields of study. Out of the 11 reviews that conducted subgroup analyses on medical and non-medical disciplines, 10 reviews did not have significant findings except for Ibrahim et al. (2013).

Regions. We identified 13 reviews that conducted subgroup analyses based on geographical regions (Akhtar et al., 2020; Ayoubi et al., 2023; Batra et al., 2021; Jia et al., 2022; W. Li et al., 2022; Y.-K. Lin et al., 2024; Peng et al., 2023; Puthran et al., 2016; Rotenstein et al., 2016; Santabárbara et al., 2021a, 2021b; Zatt et al., 2023; Zhu et al., 2021) with mixed findings (see Table S6). Of these 13 reviews, eight examined the prevalence of depression between Asia (Mdn = 33.15%, Range = 24.90–49.00%) and Europe (Mdn = 29.55%, Range = 12.21–52.30%). A total of seven reviews (Akhtar et al., 2020; W. Li et al., 2022; Y.-K. Lin et al., 2024; Peng et al., 2023; Puthran et al., 2016; Rotenstein et al., 2016; Zatt et al., 2023) found the prevalence of depression in Asia (Mdn = 30.10%, Range = 24.90–49.00%) to be higher than in Europe (Mdn = 23.20%, Range = 12.21–43.00%), while Jia et al. (2022) found the prevalence of depression in Europe (52.30%, 95% CI = [22.90%, 80.80%]) to be higher than Asia (36.20%, 95% CI = [27.60%, 45.30%]). Out of the eight reviews that conducted subgroup analyses on Asia and Europe, seven reviews did not have significant findings except for Rotenstein et al. (2016). Figure 7 provides an overview of the differences in prevalence rates in Asia and Europe.

COVID-19 Phases. A total of six reviews conducted subgroup analyses based on the phases of COVID-19 (Heumann et al., 2024; Y. Li et al., 2021; W. Li et al., 2022; Z.-Z. Lin et al., 2025; Luo et al., 2021; Melo et al., 2025)2. We observed a general trend where the prevalence of depression increased alongside the emergence of the COVID-19 pandemic (

Table 2. Comparison of Depression Prevalence by COVID-19 Phases.

Author, Year	Subgroups	Number of Studies	Prevalence of Depression, 95% CI
Heumann et al. (2024)	Before COVID-19	41	18.00% [14.70%, 21.20%]
	During COVID-19	15	30.60% [22.10%, 39.10%]
Y. Li et al. (2021)	Early stage of COVID-19	8	21.00% [16.00%, 25.00%]
	Late stage of COVID-19	10	54.00% [40.00%, 67.00%]
W. Li et al. (2022)	Before COVID-19	59	33.40% [28.90%, 37.90%]
	After COVID-19	5	35.90% [20.20%, 51.70%]
Z.-Z. Lin et al. (2025)	Before COVID-19	16	35.00% [26.90%, 43.40%]
	During/After COVID-19	13	38.70% [33.60%, 44.00%]
Luo et al. (2021)	Early stage of COVID-19	32	21.80% [18.30%, 25.50%]
	Late stage of COVID-19	34	31.00% [24.50%, 38.00%]
	Post COVID-19	6	28.90% [15.70%, 44.20%]
Melo et al. (2025)	Pre COVID-19 Onset	41	34.80% [29.10%, 40.70%]
	Post COVID-19 Onset	9	43.40% [31.00%, 56.30%]

). Of these six reviews, four reviews examined the prevalence of depression before (Mdn = 34.10%, Range = 18.00–35.00%) as compared to during or after the COVID-19 pandemic (Mdn = 37.30%, Range = 30.60–43.40%). These findings were corroborated by Y. Li et al. (2021), who reported higher prevalence rates after 1 March 2020 (54.00%, 95% CI = [40.00%, 67.00%]) compared to before 1 March 2020 (21.00%, 95% CI = [16.00%, 25.00%]), and Luo et al. (2021), who reported that prevalence rates were highest in the late stages of COVID-19 (31.00%, 95% CI = [24.50%, 38.00%]) as compared to the early stages of COVID-19 (21.80%, 95% CI = [18.30%, 25.50%]) or post-pandemic (28.90%, 95% CI = [15.70%, 44.20%]). Out of six reviews, four did not have significant findings except for Heumann et al. (2024) and Y. Li et al. (2021).

3.4.2. Method Factors

Measurement Tools. A total of 24 reviews conducted subgroup analyses on the types of measurement tools (Akhtar et al., 2020; Anbesaw et al., 2023; Ayoubi et al., 2023; Batra et al., 2021; Carvalho et al., 2022; Chang et al., 2021; Demenech et al., 2021; Dessauvagie et al., 2022; Duica et al., 2025; Guo et al., 2021; Heumann et al., 2024; Ibrahim et al., 2013; Jaafari et al., 2021; Jia et al., 2022; Khan et al., 2021; W. Li et al., 2022; Mekonnen et al., 2024; Moradi et al., 2024; Niazi et al., 2025; Peng et al., 2023; Santabárbara et al., 2021a, 2021b; Sheldon et al., 2021; Zatt et al., 2023). We identified 18 families of assessment tools reported across the reviews: AKUADS; BDI-I and BDI-II; CES-D and its revised version; the 42-item DASS and its 21-item abbreviation3; the 60-item GHQ and its 28-item abbreviation; HADS; HAMD; HRSRS; ICD-10; K10; KADS; PHQ and its 2-, 8- and 9-item derivatives; PRIME-MD; QIDS; RDSI; SCL-90; SRQ-20 and Zung-SDS (

Table 3. Comparison of Depression Prevalence by Measurement Tools.

Measure Family	Number of Reviews	Median	Range
AKUADS	1	53.67%	Included in only one review
BDI	18	26.07%	3.00–68.40%
CES-D	13	37.55%	19.00–74.70%
DASS	18	39.55%	9.10–74.00%
GHQ	3	28.00%	22.30–43.70%
HADS	9	33.33%	16.20–88.00%
HAMD	3	37.00%	19.00–53.00%
HRSRS	1	8.40%	Included in only one review
ICD-10	1	23.00%	Included in only one review
K10	1	28.00%	Included in only one review
KADS	1	2.90%	Included in only one review
PHQ	20	34.45%	4.50–69.00%
PRIME-MD	1	34.00%	Included in only one review
QIDS	1	51.50%	Included in only one review
RDSI	1	48.00%	Included in only one review
SCL-90	2	27.75%	7.00–48.50%
SRQ-20	1	28.44%	Included in only one review
Zung-SDS	8	29.50%	13.50–35.80%

).

Figure 8 provides an overview of differences in prevalence rates across the four most frequently used measurement tools in the included reviews. A full breakdown across all measurement tools is provided in Figure S2.

Of these 24 reviews, the PHQ was reported in 20 reviews, the BDI and DASS in 18 reviews, respectively, and the CES-D in 13 reviews. Median prevalence estimates generally fell between the upper 20.00% to 30.00% ranges for these four commonly utilised measurement tools, with the highest observed in the DASS family (Mdn = 39.55%, Range = 9.10%–74.00%). Estimates tended to be higher when depression was assessed using the CES-D family (Mdn = 37.55%, Range = 19.00–74.70%) or the PHQ family (Mdn = 34.45%, Range = 4.50–69.00%) than the BDI family (Mdn = 26.07%, Range = 3.00–68.40%). This finding was corroborated by Ibrahim et al. (2013), who reported a significantly higher rate of depression using the CES-D (36.80%, 95% CI = [35.20%, 38.40%]) and the PHQ (47.70%, 95% CI = [46.20%, 49.20%]) as compared to the BDI (24.00%, 95% CI = [23.10%, 24.90%]), as well as Duica et al. (2025), who reported a significantly higher rate of depression using the PHQ (63.97%, 95% CI = [51.88%, 74.51%]) than the BDI (32.82%, 95% CI = [20.38%, 48.25%]). Of note, the lowest estimate across the reviews assessing depression using the CES-D family (19.00%) was substantially higher than the lowest estimates across the reviews assessing depression using the PHQ (4.50%) or BDI families (3.00%).

The prevalence estimates drawn from lesser-known measurement tools varied greatly. For example, the prevalence rate assessed by the AKUADS was substantially higher than that of other measurement tools in our analyses (prevalence = 53.67%; included in only one review), while the prevalence rate assessed by the KADS was substantially lower than that of the other measurement tools in our analyses (prevalence = 2.90%; included in only one review). These findings should be cautiously interpreted as they may reflect the limited number of reviews available for estimates to be drawn, as well as differences in scale sensitivity and/or cut-off values across reviews.

Clinician-Administered Instrument versus Self-Reported Inventory. We initially aimed to conduct a comparison between clinician-diagnosed depression and self-reported depressive symptoms. However, we could not do so due to a lack of data (i.e., only Lei et al., 2016 provided data comparing clinician instruments and self-reported inventories).

4. Discussion

Educational institutions worldwide are increasingly concerned about the rising incidence of depression and its detrimental impacts on students’ well-being (American College Health Association, 2022; Andrews & Wilding, 2004; Lipson et al., 2022; Mofatteh, 2021; Rousseau et al., 2021). A large body of research has linked depression to a variety of adverse outcomes, including poor academic performance, higher university dropout rates, increased vulnerability towards substance dependence and suicidal ideation (Adams et al., 2021; Eisenberg et al., 2009; Garlow et al., 2008; Hjorth et al., 2016), which in conjunction pose a threat at the individual, institutional, and societal levels. Despite the evident challenges that depression poses to university students, the global prevalence of depression in the context of tertiary education remains underexamined. As such, this umbrella review was conducted to bridge this gap in the literature. We further refined our umbrella review through comprehensive subgroup analyses investigating the prevalence of depression in university students across gender, degree type, study discipline, geographical region, COVID-19 phases, and type of measurement tools used. Overall, our findings revealed a median depression prevalence of 32.00% (Range = 11.00–63.00%). This figure is a cause for concern as it suggests that one in three university students worldwide is affected by depression and at risk of its detrimental impacts.

Our subgroup analyses yielded several notable insights. In relation to gender, we observed a consistent trend where depression prevalence was higher among women than men for university students. This finding aligns with previous studies examining depression in the general population, which reported that depression is approximately twice as prevalent in women as in men (Kessler, 2003; Kuehner, 2017). We speculate that there are three possible explanations for this gender difference. Firstly, the variation between female and male university students may be attributed to biological factors such as genetics or hormones (Albert, 2015; Kuehner, 2003). Female students may be more susceptible to depression due to greater fluctuations in their hormone levels, such as rapid decreases in oestrogen that can affect serotonin function and contribute to mood instability (Albert, 2015; Kundakovic & Rocks, 2022). Secondly, women have been shown to ruminate more than men (Afifi, 2007; Robichaud et al., 2003). Female students who frequently engage in repetitive negative thinking may face difficulties concentrating on their academics, resulting in poorer grades and an increased vulnerability to depression (Hartanto & Yang, 2022; Lyubomirsky et al., 2003; Sun et al., 2014; Takagishi et al., 2013). Thirdly, female university students often face unique social pressures pertaining to their body image (Boggiano & Barrett, 1991; Murray et al., 2016). Research has shown that female students are often more dissatisfied with their weight than male students and are more likely to engage in unhealthy weight-loss behaviours (Grossbard et al., 2009; Lowery et al., 2005; Sepulveda et al., 2008). Female students may experience a disproportionate amount of pressure to attain a high standard of physical beauty characterised by a thin or lean body shape, resulting in stress, lower self-esteem, and a higher risk of depression (Boggiano & Barrett, 1991; Grossbard et al., 2009; Murray et al., 2016).

In relation to degree type, we did not find strong evidence that the prevalence of depression differed between undergraduates (Mdn = 26.50%, Range = 22.80–32.50%) and postgraduates (Mdn = 28.60%, Range = 22.00–32.90%), suggesting that students across both degree types may experience stressors of similar intensity. While existing studies suggest that undergraduate students experience heightened stress during the transition from pre-university to higher education as they adapt to new social and structural dynamics (Wyatt & Oswalt, 2013), postgraduate students may also face increased pressure due to the higher level of academic rigor and need to perform well for future career opportunities (Brooke et al., 2020; Guo et al., 2021). This similarity in depression prevalence may also reflect equal access to resources and support systems at both the undergraduate and postgraduate levels in most universities.

In relation to geographical region, we observed a consistent trend where the prevalence of depression in Asia was comparable with or higher than that of Europe. One review (Rotenstein et al., 2016) found a statistically significant result when comparing Asia (29.10%, 95% CI = [23.40%, 35.60%]) and Europe (16.90%, 95% CI = [12.80%, 21.90%]), consistent with research examining depression in the general population, which reported a higher predisposition to depression among Asians (Peng et al., 2023; Williams et al., 2015). These differences have been attributed to factors such as heightened parental expectations (Lee & Zhou, 2015) and culturally embedded beliefs about self-worth (Kwok & Tam, 2023). Notably, the education systems in Singapore, Hong Kong, and China place greater emphasis on academic excellence (Boman, 2022; C. Tan et al., 2016), a pressure often less pronounced in Western contexts. Many students in these Chinese-influenced Asian regions are influenced by Confucianism and feel a strong sense of duty to uphold their parents’ expectations in pursuing academic excellence (Chung & Walkey, 1989), which may contribute to an increased fear of failure and poor self-evaluations when their parents react poorly to their achievements.

Our research indicated that COVID-19 was associated with a rise in the prevalence of global depression rates among university students. For example, Y. Li et al. (2021) found a marked increase in the prevalence of depression after (54.00%, 95% CI = [40.00%, 67.00%]), as compared to before (21.00%, 95% CI = [16.00%, 25.00%]) the 1 March 2020. Indeed, COVID-19 has been described as a major stress event that precipitated a mental health crisis worldwide (American Psychological Association, 2020; Dong & Bouey, 2020). The pandemic disrupted the daily lives of university students, who experienced greater stress and poorer mental well-being as they adapted to online learning (Barbayannis et al., 2022; Lee et al., 2021; cf. Wong et al., 2024). Other factors such as prolonged isolation, a lack of in-person support from teachers, and increased workload due to online lessons may have contributed to the increased prevalence rates of depression during and after the pandemic (Deng et al., 2021; Giovenco et al., 2022; Grubic et al., 2020).

4.1. Practical Recommendations for Schools and Relevant Institutions

The high global prevalence of depression among university students highlights the importance of developing efficacious and scalable prevention and intervention programmes tailored to this population. One option for relevant institutions is to implement mindfulness- and resilience-based programmes, which have shown potential to improve students’ mental health and strengthen protection against depression (Akeman et al., 2020; Eun et al., 2026; González-Martín et al., 2023). Institutions could also explore technology-delivered ecological momentary interventions grounded in established cognitive and dialectical behavioural therapy principles (Liu et al., 2022; Kleinschmidt et al., 2025). Available evidence suggests that internet-assisted interventions may be as effective as in-person interventions (Kaltenthaler et al., 2006; Spence et al., 2011; Carlbring et al., 2018) while potentially offering greater cost-effectiveness (Ofoegbu et al., 2020; Donker et al., 2015; Rohrbach et al., 2023) and accessibility for technology-proficient student populations (Chen et al., 2026; Cuijpers et al., 2021). Greater attention could also be given to prevention and treatment strategies that respond to the needs of the vulnerable sex and regional subgroups identified in the present review. For example, depression literacy programmes could be adapted to account for gender differences in symptom recognition, help-seeking behaviour, and social stigmatisation (Swami, 2012; Townsend et al., 2019). Culturally adapted approaches that preserve the core content of the original treatment, such as interpersonal psychotherapy, might also be relevant in Asian settings, where higher prevalence estimates have been reported (Chowdhary et al., 2014; Kleinschmidt et al., 2025). Finally, incorporating asynchronous e-learning into regular lesson plans could be explored as one possible way of reducing transitional stress during future pandemics.

4.2. Future Directions and Limitations

Our umbrella review identified several gaps in the existing literature. Apart from Lei et al. (2016), all reviews assessed depression through self-report questionnaires rather than clinician-rated diagnoses, which may limit the accuracy of prevalence estimates in the student population. Future research should explore the use of clinician ratings to complement self-report measures of depression. The use of formal diagnoses would also enable researchers to differentiate between types of depressive disorders (e.g., major depressive disorder or dysthymia), which differ in chronicity and severity and may therefore have varying effects on students. This would also help to clarify whether university students are more vulnerable to particular depressive disorders and support more targeted prevention and intervention efforts.

We found in our quality appraisal that 38 of the 61 reviews did not use adequate sources when searching for studies. Crucially, most reviews overlooked grey literature sources, such as Google Scholar and ProQuest Dissertations and Theses Global. This is critical because unpublished studies can provide valuable data, offer a more balanced view of the subject matter, reduce the risk of publication bias, and strengthen the conclusions drawn from included studies (Hopewell et al., 2007). Future reviews should therefore include a broader range of evidence sources to contribute to a more holistic understanding of the topic and identify gaps for further research (Paez, 2017).

This umbrella review is not without limitations. Eligibility was restricted to English-language reviews, which may have resulted in the omission of relevant reviews published in other languages. We did not assess heterogeneity between the reviews reporting different ranges in their cut-off values across the same measurement tools. For example, the cut-off value for the Beck Depression Inventory (BDI) ranged from 10 to 14 points in Gao et al. (2020) and 14 to 17 points in Sarokhani et al. (2013), which may have affected the comparability of prevalence estimates across reviews. Future reviews should conduct subgroup analyses based on cut-off ranges within each measurement tool. At the same time, primary studies should aim to use psychometrically validated, instrument-specific cut-off values established against clinician-rated diagnoses to improve the accuracy of prevalence estimates.

5. Conclusions

In conclusion, the current umbrella review sought to provide a comprehensive, macro-level overview of meta-analyses and systematic reviews examining depression prevalence among university students, which were often limited in geographical coverage, study discipline, degree type, and stage of the COVID-19 pandemic. Our review underscored a high global prevalence of depression among university students, with a median rate of 32.00%. We identified the moderating effects of sex, country and stage of the COVID-19 pandemic on depression prevalence. Several potential sources of heterogeneity were found, including differences in prevalence estimates and cut-off values across families of measurement tools. Accordingly, schools and relevant institutions should allocate greater resources to prevention and treatment strategies tailored to the vulnerable gender and regional subgroups identified in the present review. The exacerbating effect of the COVID-19 pandemic on depression prevalence signals a need for preparedness during future health crises to prevent worsening mental health outcomes during periods of vulnerability. In summary, educational institutions worldwide must prioritise students’ mental well-being by strengthening existing support mechanisms and implementing evidence-based interventions to cultivate academic environments that are attuned to students’ mental health needs.