Table 1

 Characteristics and results of studies

First author, location, and study typenStudy periodDefinition of human milk (HM) intakeStatistical analysis and outcomeConfounding factorsMajor problems*
*No precalculation of sample size in any study.
†T, total number in study; I, number in intervention group; C, number in control group.
‡OR, odds ratio.
§CI, 95% confidence interval.
¶RR, relative risk.
**All control infants were exclusively fed formula milk (FM), except in Narayanan et al19 where the control group was given pasteurised HM + formula milk.
El Mohandes,20 USA, cohortT 173† I 59 C 1140–38 daysHM only as enteral nutrient ⩾1 week or as 40% of enteral calorie intake + FMSurvival analysis using time to infection: lower incidence of sepsis in HM group; OR = 0.38‡, CI = 0.05–0.95§, p = 0.04None mentioned or controlledFlawed definition of HM feeding No exclusively HM fed group Infection severity not assessed
Schanler,21 USA, cohortT 108 I 62 C 460–9 weeks>50 ml/kg/day HM averaging through hospital stay + FMLogistic regression: lower proportion of sepsis episodes in HM group; OR = 0.46, CI = 0.24–0.87, p = 0.016Controlled: antenatal steroid exposure Not controlled: maternal education, maternal contact and holding, milk intakesNo subgroup analysis of exclusive HM fed group Severity of infection not assessed Important confounders not controlled
Furman,22 USA, cohortT 119 I 79 C 400–6 weeksGraded doses 1–24, 25–49, ⩾50 ml/kg HM through week 4 + FMPoisson regression analysis: lower number of sepsis episodes in ⩾50 ml/kg HM group; RR = 0.27, CI = 0.08–0.95, p<0.05Controlled: birth weight, gender, ethnicity Not controlled: dexamethasoneNo exclusive HM fed group Severity of infection not assessed Important confounder not controlled
Hylander,23 USA, cohortT 212 I 123 C 89Hospital stayAny amount of HM+FM Duration unknownLogistic regression on selected group of measured variables: lower odds of infection in HM groups; OR = 0.46, CI = 0.24–0.87, p = 0.016Controlled: maternal sociodemographic factors, birth weight, 5 minute Apgar, days of mechanical ventilationFlawed definition of HM feeding No exclusive HM fed group Severity of infection not assessed, duration of feeding unknown
Blaymore-Bier,25 Australia, cohortT 39 I 24 C 151, 3, 7, 12 monthsAny amount of HM+FM up to 1 yearANCOVA: lesser days of URTI symptoms at: 1 month, p = 0.02, and 7 months, p = 0.006Controlled: socioeconomic statusSmall numbers with only 5 infants exclusively HM fed Bias in detecting outcome
Contreras-Lemus,24 Mexico, cohortT 118 I 59 C 59Hospital stayPreterm HM only Duration unknownχ2 Lower incidence of diarrhoea; RR¶ = 9, and urinary infection; RR5 in HM group, p<0.01None mentioned or accounted forNo details of diagnosis of outcome measures. Duration of feeding unknown
Narayanan,17 India, RCTT 70 I 32 C 38Hospital stayEBM (mother’s own or mature donor) + FM Duration unknownχ2 test Lower infection rate in HM group (n = 9), FM (n = 24), p<0.01None mentioned or accounted forSmall numbers with only 5 infants <1500 g Flawed definition of HM feeding, duration unknown Lack of exclusive HM fed group
Narayanan,18 India, RCTT 66 I 33 C 33Hospital stayColostrum 10 ml TDS (mother’s own or mature donor) + FM Duration unknownχ2 test Lower infection rate in HM group (n = 7), FM (n = 18), p<0.01None mentioned or accounted forSmall numbers with only 5 infants <1500 g Flawed definition of HM feeding, duration unknown Lack of exclusive HM fed group
Narayanan,19 India, RCT NarayananT 226 I 169Hospital stayRaw or pasteurised EHM (mother’s own or mature donor) Duration unknownχ2 test Greater infection rate pasteurised HM + FM group 33.3%, raw HM group: 10.5%, p<0.05None mentioned or accounted forInadequate numbers with only 20 infants <1500 g No exclusively formula fed group** Duration of HM feeding unknown