1. Objectives

To assess the effect of prophylactic enteral iron supplementation on growth, neurodevelopment, haematological parameters, morbidity, and mortality in preterm and low birth weight infants

2. How studies were identified

The following databases were searched in August 2011:

• CENTRAL (The Cochrane Library 2011, Issue 8)
• MEDLINE
• CINAHL
• Oxford Database of Perinatal Trials
• ClinicalTrials.gov
• WHO International Trials Registry and Platform
• The ISRCTN Registry

Reference lists and conference proceedings were also searched and the authors directly contacted researchers

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, including quasi-randomized trials

(Crossover trials were excluded)

3.2 Study participants

Preterm (<37 weeks’ gestation) or low birth weight (<2500 g) infants

3.3 Interventions

i) Enteral iron supplementation (≥1 mg iron/kg/day) versus no supplementation (<1 mg iron/kg/day)

ii) Comparisons of different regimens of enteral iron supplementation (varying doses, duration, and timing of commencement of supplementation)

(Studies using erythropoietin in conjunction with iron were excluded)

3.4 Primary outcomes

Standardized measures of neurodevelopment, for example:

• Bayley Mental Development Index (at ≤12 months, ≤ two years, ≤ five years)
• Bayley Psychomotor Development Index (at ≤12 months, ≤ two years, ≤ five years)

Growth

• Length (at ≤12 months, ≤ two years, ≤ five years)
• Weight (at ≤12 months, ≤ two years, ≤ five years)
• Head circumference (at ≤12 months, ≤ two years, ≤ five years)

Secondary outcomes included haemoglobin and mean corpuscular volume (MCV); serum ferritin concentrations, transferrin saturation and total iron binding capacity; severe anaemia (haemoglobin <8 g/dL or hematocrit <0.25); mortality; chronic lung disease (oxygen requirement at 36 weeks post-menstrual age); retinopathy of prematurity; necrotising enterocolitis; invasive infection; feed intolerance requiring cessation of enteral feeds; duration of hospitalization; and readmission to hospital in first year of life

4. Main results

4.1 Included studies

Twenty-six controlled trials, enrolling 2726 infants, were included in this review

• Twenty-one trials compared enteral iron supplementation to no or minimal iron supplementation; four trials compared early (two to three weeks postnatal) versus late (around eight weeks postnatal) initiation of enteral iron supplementation and iron supplementation versus no or minimal iron supplementation; three trials compared high dose (3.4 to 7.1 mg/kg/day) with low dose (2 to 3.6 mg/kg/day) iron supplementation and iron supplementation versus no or minimal iron supplementation; and one trial compared a low dose of iron (2 mg/kg/day) with a very low dose (1 mg/kg/day), and with a placebo
• Most trials compared iron supplementation (approximately 2 mg/kg/day to 4 mg/kg/day of elemental iron) to placebo or no supplementation; some earlier trials conducted prior to the mid-1960’s used higher doses (>10 mg/kg/day)
• Iron supplementation began between four to six weeks postnatal age in most trials
• Most trials enrolled only preterm infants, some trials used low birth weight as an inclusion criterion
• In five trials infants were formula-fed, in 17 trials infants were either fully breast milk-fed, or given a combination of breast milk and formula, and in four trials the feeding method was not reported

4.2 Study settings

• Brazil, Canada (1 trial), Finland (2 trials), Germany (2 trials), India (2 trials), Israel, Sweden (2 trials), the United Kingdom of Great Britain and Northern Ireland (4 trials), and the United States of America (11 trials)
• Publication dates ranged from 1952 to 2010

4.3 Study settings

How the data were analysed
Four comparisons were made: i) enteral iron supplementation versus no iron supplementation; ii) early versus late commencement of iron supplementation; iii) high versus low dose iron supplementation; and iv) short duration versus long duration iron supplementation. Dichotomous data were summarized using risk ratios (RR) with corresponding 95% confidence intervals (CI), while continuous data were summarized using mean differences (MD) and 95% CI. Fixed-effects models were used for all meta-analyses, and the following subgroup analyses were planned to investigate potential sources of heterogeneity:

• By type of feeding: trials involving exclusively formula-fed infants versus trials involving exclusively or partially breast-fed infants
• By age of iron supplementation initiation: early commencement (<28 days postnatal) versus late commencement (≥28 days postnatal)
• By daily dose of supplemental iron: low dose (≤2 mg/kg/day) versus high dose (>2 mg/kg/day)
• By duration of iron supplementation: short duration (≤ six months) versus long duration (> six months)
• By gestational age and birth weight: ≤33 completed weeks’ gestation or <1500 g versus >33 completed weeks’ gestation or ≥1500 g

Results
Enteral iron supplementation versus no iron supplementation
Neurodevelopment and growth
Numerical data were not supplied in many trials and thus results were not pooled in meta-analysis. One of the thirteen trials reporting on weight gain found a statistically significant treatment effect among infants weighing 1000 to 1500 g at birth at 12 months follow-up (MD 1400 g, no measure of statistical significance provided), but did not find statistically significant effects for those weighing 1500 to 2000 g at birth (MD 794 g) or those weighing 2000 to 2250 g at birth (MD 1400 g). These data should be interpreted with caution, however, as only six infants were included in the 12-month follow-up. In one of the six trials reporting on length at ≤12 months the treatment group was on average 1.6 cm longer (95% CI [0.06 to 3.14 cm], 23 infants). Six trials measuring change in head circumference did not report statistically significant differences between groups. No trials reported on neurodevelopment.

Additional outcomes
Overall, haemoglobin concentrations at six to eight weeks postnatal were not different between groups, but were significantly improved in subgroup analyses of those fully or partially breastfed or feeding method not stated (MD 4.14 g/L, 95% CI [0.59 to 7.70], 4 trials/137 infants), and in infants receiving low dose iron at ≤2 mg/kg/day (MD 3.92, 95% CI [1.53 to 6.32], 3 trials/209 infants). At three to four months postnatal, no overall difference between groups was found in haemoglobin concentrations, but a significant improvement was noted in subgroup analyses of infants receiving late commencement of iron supplementation (MD 5.59 g/L, 95% CI [1.39 to 9.79], 2 trials/89 infants) and in those receiving low dose iron (MD 4.00 g/L, 95% CI [0.85 to 7.15], 2 trials/165 infants). In pooled analysis of three trials, haemoglobin concentrations at six to nine months postnatal were improved among infants who received iron supplementation (MD 5.91 g/L, 95% CI [4.25 to 7.58], p<0.00001; 458 infants). In one trial of 81 infants, haemoglobin concentrations were statistically significantly improved at 12 months (MD 16.0 g/L, 95% CI [10.66 to 21.34], p<0.00001), while another trial reported no effect (data not provided). A further trial reported a difference favouring the treatment group of 59 g/L in haemoglobin among those born at 1000 to 1500 g (n=6 infants, no measure of statistical significance provided). Iron supplementation improved serum ferritin concentrations at six to eight weeks (MD 6.70 μg/L, 95% CI [0.13 to 13.27], p=0.046; 3 trials/124 infants) and at three to four months (MD 11.13 μg/L, 95% CI [0.74 to 21.52], p=0.036; 3 trials/104 infants). MCV was significantly greater by 2.10 fL among the treatment group at six to nine months in one trial of 326 infants (95% CI [1.24 to 2.96 fL], p<0.00001). MCV was also reported to be significantly greater at six to eight weeks in one trial, at three to four months in two trials, and at six to nine months in one other trial (data not supplied). In one trial reporting on transferrin saturation at six to eight weeks, saturation was greater by seven percentage points in the treatment group (95% CI [0.05 to 13.95%], 43 infants), and a similar increase was noted at six to nine months in another trial of 326 infants (MD 7.14%, 95% CI [5.19 to 9.08], p<0.00001). In one trial, none of the 22 infants receiving iron supplementation developed anaemia (defined as haemoglobin <7.5 g/dL for two consecutive months), while three of the 29 control infants did (data on significance not supplied). Mortality and duration of hospitalization were not reported on in any trials. For other outcomes, results were not indicative of an effect.

Early versus late iron supplementation
Neurodevelopment and growth
In one trial of 164 infants, the Mental Processing Composite scale at 5 years of age was not statistically significantly different between early and late iron supplementation groups (MD 3.0 points, 95% CI [-2.06 to 8.06]). In the same trial, no significant differences in weight, length, and head circumference at five years of age were observed between treatment groups. A greater proportion of children in the late iron group had an abnormal clinical neurological examination (35% versus 17%; p=0.02).

Additional outcomes
In pooled analysis of two trials involving 144 infants, haemoglobin concentrations at six to eight weeks were significantly higher in the early iron group (MD 15.49 g/L, 95% CI [12.74 to 18.24], p<0.00001). In one trial 116 infants, serum ferritin at six to eight weeks postpartum was greater by 25 μg/L in the early iron group (95% CI [16.94 to 33.06 μg/L], p<0.00001); however, in two other trials, no statistically significant effect was found (numerical data not available for meta-analysis). Severe anaemia, mortality, enteral feed intolerance, requirement for readmission, and duration of hospitalization were not reported on in any trials. For other outcomes, results were not indicative of an effect.

High versus low iron supplementation
Neurodevelopment and growth
The Griffiths Developmental Assessment score at 12 months of age was not different between infants who received high or low iron supplementation (MD 0.0 points, 95% CI [-6.38 to 6.38], 1 trial/42 infants). Only one trial reported numerical data for growth, in which no difference was noted between treatment and control groups at 12 months postnatal (MD 0 Z-scores for height and MD 0.1 Z-scores for weight; 42 infants).

Additional outcomes
Haemoglobin concentrations at three to four months were greater with high iron supplementation in pooled analysis of three trials involving 117 infants (MD 4.49 g/L, 95% CI [0.84 to 8.13], p=0.016), and MCV was also statistically significantly higher by 3.30 fL (95% CI [0.06 to 6.54 fL]) in one trial of 36 infants. Serum ferritin was 4 μg/L (95% CI [0.12 to 7.88]) greater in the high iron group at six to nine months postnatal in one trial involving 42 infants. In the same trial, transferrin saturation was lower by 10 percentage points in the high dose group at six to eight weeks (95% CI [-17.73 to -2.27%], 52 infants). Severe anaemia, mortality, chronic lung disease, retinopathy of prematurity, necrotising enterocolitis, enteral feed intolerance, requirement for readmission, and duration of hospitalization were not reported on in any trials. For other outcomes, no significant difference between treatment groups was found.

Duration of iron supplementation
Neurodevelopment and growth
No trials reported on the primary outcomes of neurodevelopment and growth.

Additional outcomes
Haemoglobin was greater at six to nine months follow-up in one trial of 44 infants (MD 5.7 g/L, 95% CI [0.5 to 10.9], p=0.032). No clear evidence of a difference in haemoglobin between groups was found in one trial with a high rate of attrition, with benefits in favour of both short duration (MD 3 to 5 g/L) and long duration (MD 2 to 28 g/L) iron supplementation depending on the birth weight and time period (no data on statistical significance provided). No other pre-specified outcomes were reported on.

5. Additional author observations*

The overall methodological quality of the included trials was fair to poor, with older studies in particular being less rigorous and only eight of the 26 trials reporting adequate allocation concealment methods. Four trials conducted between 1954 and 1977 removed anaemic infants from the trial. As anaemic infants were overrepresented in the control groups, the effect of iron supplementation on haemoglobin concentrations was likely attenuated by their removal. It has been argued that infants who have received a prior blood transfusion may not require iron supplementation. Included studies differed on their blood transfusion protocols, with some permitting transfusion, some excluding infants who had received a transfusion, and some not stating any transfusion policy. Few trials investigated potential adverse effects of iron supplementation, such as necrotising enterocolitis, retinopathy of prematurity, and chronic neonatal lung disease.

In comparison to no iron supplementation, enteral iron supplementation appears to improve haemoglobin and serum ferritin concentrations in preterm and low birth weight infants after eight weeks postnatal age, and may reduce the risk of anaemia. No benefit of providing more than 2 to 3 mg/kg/day was found. There is currently insufficient evidence to assess the effect of enteral iron supplementation on long-term neurodevelopmental and growth outcomes. No randomized controlled trials comparing enteral iron supplementation with no iron supplementation have evaluated the effects on neurodevelopmental outcomes.

Further trials evaluating the long-term neurodevelopmental effects of enteral iron supplementation in preterm and low birth weight infants are required. In addition, further trials that stratify infants according to birth weight and gestation, feeding method and prior blood transfusion are needed to determine whether iron supplementation is of benefit.