This meta-analysis was designed to identify risk factors associated with depression among individuals diagnosed with systemic lupus erythematosus (SLE). A comprehensive literature search was performed across eight electronic databases–Web of Science (WOS), the Cochrane Library, PubMed, Embase, China National Knowledge Infrastructure (CNKI), Wanfang Data, VIP Database, and Chinese Biomedical Literature Database (CBM)–with coverage up to May 2024. The search strategy combined Medical Subject Headings (MeSH) terms and relevant free-text keywords, including variations of “systemic lupus erythematosus” (e.g., “Lupus Erythematosus, Systemic,” “Disseminated Lupus Erythematosus”), along with terms related to depression (“depressive disorder,” “depression”) and risk assessment (“risk factors,” “risk indicators,” “health-related correlates”). An exemplar search string for PubMed is provided in Table 1. To further enhance retrieval completeness, reference lists of eligible and relevant publications were manually screened.

Studies were included if they met all of the following conditions: (a) participants were adults (≥18 years) with a confirmed diagnosis of systemic lupus erythematosus (SLE); (b) the primary aim was to examine risk factors for depression among individuals with SLE; (c) depression was assessed using internationally validated rating scales; (d) the design was observational–specifically cross-sectional, cohort, or case–control studies; (e) the full text was published in either English or Chinese; (f) appropriate statistical approaches (e.g., logistic regression) were used to estimate associations; and (g) effect estimates–such as odds ratios (ORs) with 95% confidence intervals (CIs)–were reported directly or could be derived from the provided data.

Exclusion criteria comprised: (a) duplicate or redundant publications; (b) unavailability of the full text despite reasonable retrieval efforts; and (c) insufficient or non-extractable outcome data. Disagreements regarding study eligibility were resolved through consensus between two reviewers, with input from a third reviewer when consensus could not be reached.

Two reviewers (LLZ and JJC) independently extracted data using a standardized form to ensure consistency and minimize error. Extracted information included the first author, publication year, study design, country or region, sample size, instruments used to assess depression in SLE patients, and reported risk factors. Any discrepancies were resolved through discussion, with adjudication by a third reviewer (YPS) when consensus could not be reached. The full set of extracted data is summarized in Table 2.

Study quality was assessed using instrument-specific criteria. For case–control and cohort studies, the Newcastle–Ottawa Scale (NOS) (21) was applied, with scores of 0–3, 4–6, and 7–9 classified as low, moderate, and high quality, respectively. Cross-sectional studies were evaluated according to the Agency for Healthcare Research and Quality (AHRQ) checklist (22), where scores of 0–3, 4–7, and 8–11 corresponded to low, moderate, and high methodological quality. Any disagreements between the two reviewers during study selection, data extraction, or quality appraisal were resolved through consensus; a third reviewer was consulted when necessary. Detailed quality ratings are provided in Supplementary Tables 1, 2.

All statistical procedures for this meta-analysis were performed exclusively using STATA software (version 18.0). For each analysis, effect sizes were expressed as odds ratios (ORs) with corresponding 95% confidence intervals (CIs). Given the anticipated clinical and methodological diversity among the included studies, we adopted a conservative approach by applying the random-effects model (DerSimonian-Laird method) across all analyses. Statistical heterogeneity was quantified using the I² statistic with its associated p-value; I² values exceeding 50% with p < 0.10 were interpreted as indicating substantial heterogeneity. We conducted comprehensive sensitivity analyses by systematically excluding individual studies to evaluate the robustness of our findings and identify potential outliers influencing the overall estimates. Publication bias was assessed through both visual inspection of funnel plots and Egger’s regression test, with statistical significance defined as p < 0.05.

Totally, 862 publications were found, among which 26 studies (12, 23–47) met the inclusion criteria and involved in the qualitative synthesis. The detailed flow chart of the study selection process is shown in Figure 1.

There were 26 qualified studies, including 2 case-control studies (31, 43), 5 cohort studies (33, 34, 39, 45, 46), and 19 cross-sectional studies (12, 23–30, 32, 35–38, 40–42, 44, 47), with a total of 8,960 SLE patients involved. Geographically, 14 studies were carried out in China (23–32, 35, 36, 42, 43), 5 in the US (33, 34, 38, 45, 46), 1 in Egypt (37), 1 in Portugal (39), 1 in South Korea (40), 1 in Greece (12), 1 in Brazil (41), 1 in Thailand (44) and 1 in Saudi Arabia (47). Study sample size varied from 70 to 1,609 people. The methodological quality was evaluated with the NOS scale for non-randomized studies and the AHRQ instrument for cross-sectional studies. In the cohort and case-control studies, 4 were rated as 8 stars (high quality) (33, 39, 43, 46) and 3 were rated as 7 stars (31, 34, 45). For cross-sectional studies, 6 scored 10 stars (high quality) (34, 40, 48–51), 6 scored 9 stars (24, 27, 29, 30, 35, 42), 4 scored 8 stars (25, 28, 37, 47), 1 scored 7 stars (32), and 1 scored 6 stars (26), which shows that the overall quality of the included literature is good. Full quality assessment result is shown in Table 2, see Supplementary Tables 3, 4 for detailed information. Nine different instruments were used to evaluate depression across the studies. The BDI was used in 4 studies (23, 28, 40, 47), the CES-D was used in 4 studies (33, 37, 38, 46), and the PHQ-9 was used in 2 studies (12, 44). HADS was used in 4 studies (12, 25, 34, 37), and SDS was used in 5 studies (25, 26, 30–32). Other tools include: DCS [depression clinical scale; 1 study (41)]; clinical diagnosis [2 studies (29, 34)]; HAMD [Hamilton Depression Rating Scale; 2 studies (24, 35)]; and PHQ-8 [Patient Health Questionnaire-8; 1 study (45)] (Table 2).

Among the 26 included studies, we identified over 20 potential correlates of depression in SLE patients across sociodemographic, socioeconomic, clinical, and patient-reported domains. Eleven factors met criteria for meta-analysis. As shown in Table 3, younger age, lower education, economic hardship, unemployment, smoking, higher disease activity, high-dose glucocorticoids, comorbidity, and greater fatigue were significantly associated with increased depression risk (all P < 0.05). Disease duration was not associated (OR = 0.90, P = 0.40), and low social support did not reach significance (P = 0.064). Remaining variables–including ethnicity, marital status, pain, sleep quality, body image, TCM-specific symptoms (e.g., menstrual irregularity), organ involvement, laboratory markers (e.g., low C3), and insurance type–were reported in ≤2 studies or measured inconsistently (e.g., “family burden” vs. “family dysfunction”) and were therefore summarized narratively (see Supplementary Table 5).

Our estimated prevalence (39.4%) is somewhat higher than those reported in earlier meta-analyses (typically 24%–39%) (7), possibly reflecting broader screening practices, heightened clinical awareness, or rising societal stressors in recent years. The markedly elevated burden among Chinese patients–nearly double that of other regions–is particularly concerning. While methodological differences (e.g., recruitment from tertiary referral centers where disease severity may be greater) may partially account for this gap, deeper contextual factors are likely at play. In many Asian societies, including China, mental health conditions remain heavily stigmatized, often leading individuals to delay seeking help until symptoms become severe (48). Our meta-regression analysis suggests that demographic features common in Chinese cohorts–namely, a higher proportion of women and younger average age–partially mediate this disparity, consistent with established epidemiological patterns showing elevated depression risk in women and young adults (49, 50).

Economic hardship emerged as the strongest correlate of depression (OR ≈ 6.05), with consistent effects across regions. This aligns with the social determinants of health framework: financial strain restricts access to medications and specialist care while simultaneously fueling chronic stress, sleep disruption, and social isolation–all of which have neurobiological links to mood dysregulation (51). Similarly, low education (OR = 2.21) and unemployment (OR = 2.46) independently increased depression risk, underscoring how structural inequities shape mental health outcomes in SLE.

Loss of employment not only reduces income but also erodes social identity and daily routine (52, 53); conversely, work participation can divert attention from illness, foster positive affect, and strengthen support networks through workplace interactions (54). These insights support integrating socioeconomic screening into routine rheumatology visits and advocate for policy-level interventions–such as income support, vocational rehabilitation, and reintegration programs–for patients with chronic autoimmune diseases.

Clinically, both high disease activity and glucocorticoid therapy were among the most potent predictors of depression. The robust association with disease activity supports the “inflammatory-depression” hypothesis: pro-inflammatory cytokines (e.g., IL-6, TNF-α) can cross the blood–brain barrier, disrupt monoamine metabolism, and dysregulate the hypothalamic–pituitary–adrenal axis, thereby directly promoting depressive symptoms (55). Meanwhile, high-dose glucocorticoid use showed an exceptionally strong link with depression (OR = 7.72), corroborating well-documented neuropsychiatric side effects ranging from mood lability and insomnia to steroid-induced psychosis (34, 56–58). Clinicians should therefore consider depression in active SLE not just as a comorbidity but as a potential manifestation of uncontrolled inflammation or iatrogenic drug effects.

Smoking, though assessed in fewer studies (OR = 3.17) (59), presents a modifiable risk factor with plausible biological mechanisms–namely, oxidative stress and neuroinflammation that heighten vulnerability to mood disorders (60, 61). Integrating smoking cessation support into routine SLE care could thus reduce psychiatric comorbidity.

Fatigue–one of the most disabling SLE symptoms–was strongly linked to depression. The FATILUP study (62) confirmed their tight interrelationship, and Da Costa et al. (63) identified depression as the strongest predictor of mental fatigue. Given the profound impact of fatigue on quality of life, evidence-based strategies such as supervised aerobic exercise have shown promise in alleviating both physical and emotional burden (64).