Fourteen studies were included: 9 randomised trials (406 participants) and 5 cohort studies (393 participants).
The studies assessing corticosteroids alone were not pooled, owing to statistical heterogeneity, and no further analysis of these studies was undertaken.
No evidence of statistical heterogeneity was found between the studies assessing oral corticosteroids and radiotherapy (Q test, P=0.59), or between the studies assessing IV corticosteroids (Q test, P=0.98). Treatment with oral corticosteroids plus radiotherapy was associated with a statistically significant reduction in the number of treatment failures (RR 0.3, 95% confidence interval, CI: 0.15, 0.6), compared with treatment with oral corticosteroids or radiotherapy alone. Treatment with IV corticosteroids, with or without radiotherapy, was also associated with a statistically significant reduction in the number of treatment failures (RR 0.36, 95% CI: 0.19, 0.66). The results from the random-effects analyses were similar.
When the 7 oral corticosteroids plus radiotherapy studies were pooled with the 3 IV corticosteroid studies, the overall RR was 0.3 (CI not reported). From this, the number of patients who needed to be treated with oral corticosteroids plus radiotherapy, or IV corticosteroids, to prevent one more failure in comparison with other treatments was estimated to be between three and eight. The results were similar across subgroups by sample size, publication year, study design and risk of failure in the control group (these results were not presented). A regression analysis investigating publication bias suggested that some publication bias may have been present.