Table 3.
Results of meta-analyses and sensitivity analyses.
| Outcome | Review results | Number of studies suspected of ORB | Number of participants missing from the meta-analysis (%) | Sensitivity analysis results | |
|---|---|---|---|---|---|
| ORB alone | Study publication bias * | ||||
| Intravenous Magnesium: children | |||||
| Hospital admission | RR 0.69 (95% CI 0.53, 0.90) from 3 studies, I2 = 17.7%, RE. Favours intervention | 3 [18,23,24] | 117 (50%) | RR 0.76 (95% CI: 0.58, 0.99) for one study as trialists confirmed that for the other two it was not measured. | 1 |
| Pulmonary function | SMD 1.94 (95% CI: 0.80, 3.08) from 4 studies, I2 = 84.4%, RE. Favours intervention | 2 [18,20] | 104 (45%) | NA: both studies given H classifications and this was confirmed by trialists | 4 |
| Intravenous Magnesium: adults | |||||
| Hospital admission | RR 0.87 (95% CI 0.70, 1.08) from 8 studies, I2- = 30%, RE. Favours intervention | 2 [22,26] | 129 (14%) | NA: confirmed by trialists that this was not measured | 4 |
| Pulmonary function | SMD 0.25 (95% CI -0.01, 0.51) from 9 studies, I2 = 70.6%, RE. Favours intervention | 1 [19] | 33 (3%) | NA: results from one study obtained and included in an updated meta-analysis | |
| Updated SMD 0.24, (95% CI 0, 0.48) from 10 studies, RE. I2 = 67% | 7 | ||||
| Nebulised Magnesium: children | |||||
| Hospital admission | RR 2 (95% CI 0.19, 20.93) from 1 study, RE. Favours control | 1 [25] | 40 (39%) | NA: confirmed by the trialists that this was not measured | 1 |
| Pulmonary function | SMD -0.26 (95% CI -1.49, 0.98) from 2 studies, I2 = 88.9%, RE. Favours control | 0 | 0 | NA: all eligible studies reported on this outcome | 1 |
| Nebulised Magnesium: adults | |||||
| Hospital admission | RR 0.68 (95% CI 0.46, 1.02) from 6 studies, I2 = 0, RE. Favours intervention | 1 [21] | 74 (17%) | RR 0.76 (95% CI 0.51, 1.13) | 8 |
| Pulmonary function | SMD 0.17 (95% CI -0.02, 0.36) 7 studies, I squared 1.2%, RE. Favours intervention | 0 | 0 | NA: all eligible studies reported on this outcome | 5 |
*: number of unpublished studies required to overturn review conclusions after allowing for ORB. When the result is statistically significant this implies the result becoming non-significant. When the result is not statistically significant, the reported value is the number required to change the direction of treatment effect.
NA: Not Applicable; RE: random effects