The searched returned a total of 2508 publications. After deduplication, screening, and full-text assessment, 17 met eligibility criteria (15 for prevalence, and 2 additionally for the impact of depression). A flowchart of the selection process is shown in Supplementary Figure S1.
Prevalence of depression in PSV
Details extracted from 15 studies that reported depression prevalence are shown in Table 1. Individual study sample sizes varied from 29 to 9978. Ten studies were of small vessel vasculitis; most of these were ANCA-associated vasculitis (AAV) [8, 13,14,15,16,17,18,19,20]. (Brezinova et al. reported results for small vessel vasculitis that were all AAV except one IgA vasculitis patient .) One study reported cerebral angiitis, which was included in the small vessel group . One study reported results on medium vessel vasculitis (polyarteritis nodosa) . Seven studies were of large vessel vasculitis, among which three were Takayasu arteritis and two GCA; we also included two PMR studies [14, 17, 18, 21,22,23,24]. (Brezinova et al. included polyarteritis nodosa in the large vessel group .)
Vasculitis was defined using ACR classification criteria in six studies [13, 15, 16, 19, 21, 22], diagnostic codes in five [14, 18, 23, 25, 26], self-report in two [8, 17], physician diagnosis in one  and unclear in one study .
Depression was defined by various means. Four studies used HADS [16, 20,21,22], 4 PHQ [13, 23, 24, 26], 2 Beck’s Depression Inventory (BDI) [18, 19] and 2 CESD [8, 15] were used with varying cut off values. One study used self-reported depression  and 2 diagnostic coding [14, 25].
The prevalence of depression in medium vessel vasculitis was available in only one study. Grayson et al. included 36 patients with polyarteritis nodosa, among whom 52% had self-reported depression .
Meta-analysis for small vessel vasculitis
The pooled prevalence of depression in small vessel vasculitis was 28% (95% CI 20–38%). Prevalence ranged from 14 to 55%. There was significant heterogeneity (93%) that was not improved by stratifying them into subgroups (Fig. 1). Studies using self-reported depression and CESD-R generally reported higher prevalence.
Meta-analysis for large vessel vasculitis
The pooled prevalence of depression in large vessel vasculitis was 24% (95% CI 17–34%). Prevalence ranged from 10 to 55%. Again, there was significant heterogeneity (96%) that was not improved by stratifying into subgroups (Fig. 2). Studies using self-reported depression generally reported higher prevalence.
Depression compared between PSV and controls
Compared to non-vasculitis patients in UK primary care data, Li et al. reported an increased risk of depression among granulomatosis with polyangiitis (GPA) patients, although this was limited to the first 3 years of follow-up (HR 1.77) but not 3 years after diagnosis (HR 1.01) . Hajj-Ali et al. also reported a higher prevalence of depression among GPA patients when compared to the general population (22 vs 7.6%; p < 0.001) .
Hinojosa-Azaola et al. reported a numerically higher prevalence of depression in AVV patients (9%) than RA (4%) and CKD (1%) controls, although this was not statistically significant (p = 0.29) . Basu et al. found no significant difference in depression (BDI ≥ 13; AAV 15% vs primary care controls 21%; p = 0.50) .
For large vessel vasculitides, Alibaz-Oner et al. reported similar HADS depression scores between patients with Takayasu and healthy controls (5.1 vs 3.8; p = 0.168), although their sample size was small (n = 55 vs 40) .
Associations between depression and disease outcomes
In a study of PSV (i.e., all vessel sizes), Brezinova et al. found BDI to be associated with poorer QoL (SF36 physical component score b = − 0.73, p < 0.001; mental component score b = − 0.56, p < 0.001) . Koutantji et al. reported higher pain in PSV patients with depression (HADS depression scores in those with pain vs those with none/little: 7.6 vs 5.2, p < 0.01) . Grayson et al. reported that a history of depression was associated with negative illness perceptions in PSV patients (OR 4.94; 95% CI 2.90–8.41); that is, organized beliefs that patients have about their illness that are important determinants of health-related behaviour . Among patients with all types of vasculitis, Carpenter et al. showed depressive symptoms (CESD-R) to independently predicted non-adherence to medication (b = 0.01, p = 0.02) .
Among GPA patients, Hajj-Ali et al. did not find associations between PHQ-9 depression score and disease activity (Birmingham Vasculitis Activity Score; median 4 vs 3, p = 0.77), damage (Vasculitis Damage Index, median 0 vs 0, p = 0.23), or glucocorticoid requirement . They did, however, report a significant association between PHQ-9 and fatigue (r = 0.73, p < 0.05). Hinojosa‑Azaola et al. also found an association between HADS and fatigue (r = 0.48, p = 0.01) among AAV patients . When they dichotomised HADS scores (threshold > 7), those with depression or anxiety had more steroid use (prednisolone ≥ 10 mg/day; OR 6.65, 95% CI 1.37–32) than those without. These patients also had poorer quality of life (SF36 score 51.4 vs 66.8, p = 0.003) and more frequent sleep impairment (65 vs 33%, p = 0.01) than those without anxiety or depression. Similarly, Basu et al. found AVV patients with depression (HADS > 8) to have higher odds of having poor quality of life (OR 5.6; 95% CI 2.0–15.8) . In a separate AAV study, Basu et al. also showed depression to be independently associated with unemployment (OR 4.4, 95% CI 1.8–10.8) .
There were no studies comparing depression and disease outcomes in large vessel vasculitides.