Real World Mortality and Specific Causes of Death in Chronic Oral Glucocorticoid Use: A Systematic Review, Meta-Analysis and Meta-Regression

Background: Glucocorticoids (GCs) are widely used as therapeutic agents with prevalence 0.9–3.7%, but they are associated with significant side effects. Understanding of mortality ratios and causes of death from GC use is poorly appreciated and likely to help shape future stratification of clinical practice. Aims: To perform a meta-analysis of all-cause and specific cause -mortality amongst chronic GC users. Methods: The protocol was registered in PROSPERO (CRD42017067530). Searches were undertaken of PubMed, EMBASE, CINHAL, web of science and Cochrane Central from 1966 to April 2019. The primary outcomes were proportion of death and SMR in chronic GC use patients. The meta-analysis was performed with STATA version 16.1. The I2, subgroup analysis and meta-regression were used to assess heterogeneity among included studies. Results: A total of 109,511 articles, were screened. One hundred eighteen articles with 128 patient cohorts containing 51,374 patients reporting mortality fulfilled the eligibility criteria and were included in the meta-analysis. SMR from seven autoimmune/inflammatory disease studies was 1.84 (95%CI 1.27,2.41) with I2 70.2 6%. The proportion of overall death was 0.12 (95% CI 0.1, .014) with I2 89.3%. The proportion of death was 0.18 (95% CI 0.13,0.24) with I2 92.0% in vasculitis diseases (40 studies), 0.10 (95% CI 0.08, 0.13) with I2 86.2% in connective tissue diseases (67 studies), 0.07 (95% CI 0.03, 0.13) with I2 88.7% in inflammatory diseases (15 studies), 0.28 (95% CI 0.21–0.37) with I2 0.0% in haematologic diseases (2 studies), and 0.06 (95% CI 0.05, 0.09) with I2 0.0% in respiratory diseases (3 studies). GC prescription reports were different across studies and led to different prediction of mortality with high heterogeneity. Proportion of death amongst a GC cumulative dose of 0.3 to 3.9 gram, 4.0 to 7.3 gram and 7.4 to 36.7 gram were 0.11 (95% CI 0.06, 0.20), 0.04 (95% CI 0.02, 0.08) and 0.16 (95% CI 0.08, 0.28), respectively. The proportion of deaths predicted by average mean dose of ≥ 5mg/d, >5–7.5 mg/d, >7.5–10 mg/d and >10–30 mg/d were 0.02 (95% CI 0.01, 0.10), 0.15 (95% CI 0.15, 0.16), 0.08 (95% CI 0.03, 0.19) and 0.14 (95% CI 0.11, 0.19), respectively. The proportion of death predicted by a maintenance dose of ≥5mg/d, >5–7.5 mg/d, >7.5–10 mg/d and >10–30 mg/d were 0.08 (95% CI 0.05, 0.13), 0.12(95% CI 0.05, 0.23), 0.11(95% CI 0.06, 0.210) and 0.12(95% CI 0.05, 0.24) respectively. The causes of death (77 studies) were cardiac (25.3%), infection (13.2%), malignancy (15.6%), respiratory failure (10.6%), active underlying disease (4.4%), cerebrovascular disease (1.1%) and thromboembolism (0.9%). Conclusion: This is the first meta-analysis of oral GC use and mortality from real-world clinical practice publications. Multiple factors contribute to mortality, including GC dose, duration of exposure, route, preparation, together with patient and disease-specific factors.