Allogeneic red blood cell transfusion
All five trials reported on the number of patients who received an allogeneic RBC transfusion. In meta-analysis there was no evidence that iron supplementation to treat anaemia in critical care patients reduced allogeneic RBC transfusion requirements (RR 0.88, 95 % CI 0.74 to 1.06; p = 0.18, I
2 = 46 %) (Fig. 3a). In subgroup analysis by route of administration there was no evidence of an effect of iron supplementation in patients who received oral iron or those who received intravenous iron (Fig. 4).
Mean number of RBC units
Only two trials reported the number of RBC units received in both groups [19, 24]. After transforming to a log scale, meta-analysis of these two trials showed no evidence that iron supplementation reduced the mean number of allogeneic RBC units transfused (MD -0.45, 95 % CI -1.34 to 0.43, p = 0.32, I
2 = 79 %) (Fig. 3b).
Mean difference in haemoglobin concentration
Mean (and standard deviation (SD)) changes in haemoglobin concentration at short-term and mid-term follow-up time points was reported in three trials [19, 21, 24]. There was no significant difference in the mean haemoglobin concentration between patients who received iron and those who did not, either at short-term (MD -0.25, 95 % CI -0.79 to 0.28, p = 0.35, I
2 = 57 %) (Fig. 3c) or mid-term follow-up (MD 0.21, 95 % CI -0.13 to 0.55, p = 0.23, I
2 = 0 %) (Fig. 3d). In the two trials that did not report mean and SD, one trial reported no change in haematocrit concentration over a period of 28 days in both groups [22] and the other trial found no significant differences in haemoglobin concentration between groups at any follow-up time point over 14 days [23].
In-hospital mortality
Four trials reported in-hospital mortality as an outcome [21–24]. In meta-analysis there was no difference in the risk of mortality between patients who received iron supplementation and those who did not (RR 1.04, 95 % CI 0.43 to 2.52; p = 0.92, I
2 = 52 %) (Fig. 5a). Causes of mortality were not reported in any trials.
In-hospital infection
Two studies provided data on infection and antibiotic usage [22, 23]. There was no evidence of a difference in the risk of infection in patients who received iron supplementation and those who did not (RR 0.89, 95 % CI 0.74 to 1.08, p = 0.23, I
2 = 0 %) (Fig. 5b). Neither study found a significant difference in mean number of antibiotic days between the intervention and control groups (14 vs. 16 days, p = 0.45; 14 vs. 16 days, p = 0.64).
Length of stay
Three trials reported ICU LOS [22–24] and two trials additionally reported hospital LOS [22, 23]. One trial reported a significantly longer ICU LOS in the control group compared to the intervention group (58 +/- 31 vs. 29 +/- 18 days, p < 0.05) [24], whereas two trials found no difference in ICU LOS (12 vs. 14 days, p = 0.69; 10 vs. 11 days, p = 0.53) [22, 23]. Similarly, these trials found no differences in hospital LOS (14 vs. 16 days, p = 0.24; 14 vs. 16 days, p = 0.50) [22, 23]. Meta-analysis was not performed as only one trial provided mean and SD values.
Biomarkers of iron status
All trials reported serum ferritin measurements. Three trials reported mean and SD data [19, 21, 24] which were log-transformed due to the likely skewing of this outcome. Iron supplementation was associated with higher log serum ferritin levels at both short-term (MD 0.41 ng/mL, 95 % CI 0.06 to 0.77, p = 0.02, I
2 = 78 %) (Fig. 5c) and mid-term follow up (MD 0.68 ng/mL, 95 % CI 0.53 to 0.83, p < 0.00001, I
2 = 0 %) (Fig. 5d). In two trials that did not report mean and SD values, in one trial of oral iron supplementation there was no difference in serum ferritin between iron and placebo groups throughout the study period [22] and in one trial of intravenous iron supplementation there were significantly higher serum ferritin levels in the iron group compared to the placebo group at day 7 (median 808 ng/mL vs. 457 ng/mL, p < 0.01) and day 14 (median 1,046 ng/mL vs. 551.5 ng/mL, p < 0.01) [23].
Three trials reported serum iron and eZPP measurements [22–24], all of which found no significant difference in concentrations between iron and placebo groups at any time point. Meta-analysis was precluded by a lack of reporting of mean and SD values. Two trials reported Tsat levels – in one trial there was a statistically significant increase in Tsat levels at day 7 in the intravenous iron group when compared to placebo (15 % vs. 11 %, respectively, p = 0.02) but this was below the normal range and did not persist through to the end of the study period [21], whereas in another trial there were no differences in Tsat levels throughout the study period [24].
Other secondary outcomes
No trials reported data on morbidity scales, HrQoL scores or serious adverse events as per the definition above.
Adverse events
Only one trial [22] comparing oral iron to placebo reported adverse effects - namely gastrointestinal upset defined as “nausea, vomiting, abdominal cramping, or constipation, necessitating discontinuation of the study drug for any period of time”. The authors found no difference between the iron and placebo groups (12.4 % vs. 8.7 % respectively, p = 0.62).