Systemic corticosteroids in acute exacerbation of COPD: a meta-analysis of controlled studies with emphasis on ICU patients

Guidelines on systemic corticosteroids in chronic obstructive pulmonary disease (COPD) exacerbation rely on studies that excluded patients requiring ventilatory support. Recent publication of studies including ICU patients allows estimation of the level of evidence overall and in patients admitted to the ICU. We included RCTs evaluating the efficacy and safety of systemic corticosteroids in COPD exacerbation, compared to placebo or standard treatment. The effect size on treatment success was computed by a random effects model overall and in subgroups of non-ICU and ICU patients. Effects on mortality and on the rate of adverse effects of corticosteroids were also computed. Twelve RCTs (including 1,331 patients) were included. Pooled analysis showed a statistically significant increase in the treatment success rate when using systemic corticosteroids: odds ratio (OR) = 1.72, 95% confidence interval (CI) = 1.15 to 2.57; p = 0.01. Subgroup analysis showed different patterns of effect in ICU and non-ICU subpopulations: a non-significant difference of effect in the subgroup of ICU patients (OR = 1.34, 95% CI = 0.61 to 2.95; p = 0.46), whereas in the non-ICU patients, the effect was significant (OR = 1.87, 95% CI = 1.18 to 2.99; p = 0.01; p for interaction = 0.72). Among ICU patients, there was no difference in the success whether patients were ventilated with tracheal intubation (OR = 1.85, 95% CI = 0.14 to 23.34; p = 0.63) or with non-invasive ventilation (OR = 4.88, 95% CI = 0.31 to 75.81; p = 0.25). Overall, there was no difference in the mortality rate between the steroid-treated group and controls: OR = 1.07, 95% CI = 0.67 to 1.71; p = 0.77. The rate of adverse events increased significantly with corticosteroid administration (OR = 2.36, 95% CI = 1.67 to 3.33; p < 0.0001). In particular, treatment with systemic corticosteroids significantly increased the risk of hyperglycemic episodes requiring initiation or alteration of insulin therapy (OR = 2.96, 95% CI = 1.69 to 5; p < 0.0001). We found corticosteroids to be beneficial in the whole population (non-critically ill and critically ill patients) in terms of treatment success rate. However, subgroup analysis showed that this effect of corticosteroids was only observed in non-critically ill patients whereas critically ill patients derived no benefit from systemic corticosteroids regardless of the chosen ventilatory mode (invasive or non-invasive ventilation). Further analyses showed no effect on mortality of corticosteroids, but higher side effects, such as hyperglycemic episodes requiring the initiation or alteration of insulin therapy.