Nine trials with 792 patients (range 31 to 238) were included. Six trials included patients with systematic inflammatory response syndrome, four trials included patients with sepsis and three trials included patients with septic shock. Two trials were judged at low risk of bias, three at high risk and four at unclear risk.
Selenium use led to a statistically significant reduction in mortality compared to control (OR 0.73, 95% CI 0.54 to 0.98; NNT 13; Ι²=0%). There was no evidence that selenium altered length of ICU stay (MD 2.03, 95% CI -0.51 to 4.56; three trials; Ι²=0%) or changed the incidence of nosocomial pneumonia (OR 0.83, 95% CI 0.28 to 2.49; three trials; Ι²=56%).
The effect of selenium on mortality was statistically significant in the high dose (>500 mcg) trials, but not quite statistically significant in the low dose (<500 mcg) trials, however there was no statistically significant difference in the effect of selenium between low and high dose trials. The two high quality trials showed no benefit of selenium on mortality; the observed benefit arose primarily from the seven high or unclear risk of bias trials. However, the difference between the two quality groups was not statistically significant.
There was no evidence of publication bias.