Psychiatric Symptoms Associated with Corticosteroid Use: A Systematic Review and Meta-analysis

Corticosteroids (CSs) are widely prescribed but can induce psychiatric adverse effects (depressive, manic, psychotic, and anxiety symptoms). This systematic review aimed to quantify the association between CS use and psychiatric symptoms and to explore potential moderators across clinical contexts.

We conducted a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-guided systematic review and meta-analysis. MEDLINE and Embase were searched through 10 October, 2025, and 19 hand-searched records were also screened. We included observational studies in medical CS users and excluded sex steroids, neonatal exposure, prescription-only outcomes, purely objective outcomes, sleep/cognition/delirium-focused studies, CS deficiency, withdrawal, and studies confounded by co-medications; randomized controlled trials were excluded by design. Two reviewers screened studies, extracted data, and assessed risk of bias with the Newcastle-Ottawa Scale. Certainty of evidence for each pooled outcome was additionally evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Pairwise meta-analyses pooled standardized mean differences and odds ratios using random-effects models; I² quantified heterogeneity. The acute phase was operationally defined as ≤ 8 weeks after CS initiation. An exploratory meta-regression across six studies (seven groups) examined dose, duration, cumulative dose, age, and sex.

Seventy-three studies (total N = 3,759,659) were included (22 cross-sectional, 31 cohort, 12 pre-post, 2 case-control, 2 registry, and 4 prospective patient-as-own-control investigations comparing outcomes during on-CS vs off-CS periods). Compared with non-users, CS use was associated with higher depressive symptom scores (standardized mean difference = 0.92, 95% confidence interval [CI] 0.40-1.45; p < 0.001; I² = 89%; seven studies; n = 1403; eight groups). During the operationally defined acute phase (≤ 8 weeks after CS initiation), manic symptoms were more frequent than depressive symptoms (six studies; n = 1086; odds ratio = 2.42, 95% CI 1.48-3.97; p < 0.001; I² = 0%). A single-arm meta-analysis estimated psychotic symptoms in 2.4% of CS users (95% CI 0.76-7.26; six studies; n = 1206). Anxiety showed no significant association with CS use (standardized mean difference = 0.13, 95% CI - 0.24 to 0.51; p = 0.26; I² = 7%; three studies; n = 1251). Overall study quality (Newcastle-Ottawa Scale) was generally moderate to high, whereas certainty of evidence according to GRADE was generally low to very low across pooled outcomes. The exploratory meta-regression suggested that higher daily dose and older age were associated with greater depressive severity; by contrast, cumulative dose, treatment duration, and sex showed no significant associations, and these moderator findings remain exploratory.

Corticosteroid use is associated with more severe depressive symptoms, and manic symptoms may predominate during acute treatment; psychotic reactions are uncommon. Dose and age may modulate depressive severity, but these moderator findings should be interpreted cautiously. Routine monitoring for mood and psychotic changes, especially early after initiation and in higher-dose or older patients, is warranted.