Table 2.

Human milk (HM) and developmental outcome in very low birth weight (VLBW) infants: non-nutritive methodological parameters

Reference	Pinelli et al.  [29]	Furman et al.  [30]	Tanaka et al.  [31]	Horwood et al.  [32]	Smith et al.  [33]	Johnson et al.  [34]	Vohr et al.  [35]
Primary study for the subjects	Structured breast feeding counseling to sustain lactation	Self selected provision of breast milk vs. preterm formula	RBC DHA in breastfed VLBW infants at 4 weeks	Retinopathy of prematurity	Epidemiology of brain injuries in VLBW infants and Epidemiologic study of multiple births	Early predictors of education attainment at 11 y	Parenteral glutamine trial
Subject source	One center, Canada	One center, USA	One center, Japan	All VLBW births, New Zealand	5 centers, USA	All extremely preterm births in United Kingdom and Ireland	12 of 15 NICHD NRN sites, USA
Exclusions	Multiple birth, severe congenital, surgical, chromosome abnormality, non-English speaking parents	Positive drug screen, major congenital anomalies, intrauterine infection, overwhelming maternal social concerns	Cerebral palsy, no RBC DHA data, severe chronic lung disease, minor anomalies, hearing problem	Sensorineural deficit and no breast milk data	NS	None	Unable to test including those with sensorineural deficit
Assessment age	6 and 12 m corrected	20 m corrected	5y	91 m	6 to 8y	Median 10y 11 m	30 m corrected
Sample size (assessed/eligible)*	138/148	98/119	18/26	298/338	439/770	219/307; and 153 term “Control”	704/939
Observers blinded to feeding protocol	Yes	Yes	NS	NS	NS	Yes	NS
Neurodevelopment tests	BSID II: MDI and PDI	BSID II: MDI and PDI	KABC	WISC-R: verbal and performance IQ	CELF; CCVL; KABC; PPVT; WRAVMA	KABC; TAAS; WIAT-II	BSID-II
Effect size for human milk†	See below	See below	See below	See below	See below	See below	See below

CI – confidence interval; NICHD NRN – National Institute of Child Health and Human Development Neonatal Research Network; NS – not stated or specified or not significant; RBC DHS red blood cell - Docosahexaenoic acid; SD – standard deviation.

Neurodevelopment tests: BSID II Bayley Scales of Infant Development (2nd edition), MDI Mental Developmental Index, PDI Psychomotor Developmental Index; CCVL California Children’s Verbal Learning Test; CELF Clinical Evaluation of Language Fundamentals 3rd ed; KABC - Kaufman Assessment Battery for Children; PPVT - Peabody picture vocabulary test 3rd ed; TAAS Teachers Academic Attainment Scale; WIAT-II Wechsler Individual Achievement Test – II; WISC-R - Revised Wechsler intelligence scale for children; WRAVMA - Wide Range Assessment of Visual Motor Abilities.

*None had stated a priori power calculation to measure the neurodevelopment effect of HM. For Smith et al. [33], total sample size included 4 gestation matched VLBW controls for each VLBW subject with abnormal cranial ultrasound. For Johnson et al. [34], control subjects included one subject selected randomly from 3 classmates born at term with same sex and ethnicity.

†Effect size for human milk (maximum amount or as specified) vs none or limited human milk. Mean (SD or CI if available) scores after adjustment for covariates and confounders:

Pinelli et al. [29]. Dichotomous groups based on 80% HM intake as cut point: No significant difference in MDI 92 (15) vs. 91 (12) or PDI 78 (15, SD) vs. 77 (14).

Furman et al. [30]. No significant difference in MDI 85 +/−21 vs. 80 +/−16 or PDI 76 +/−16 vs. 80 +/−16.

Tanaka et al. [31]. Dichotomous groups based on 80% HM intake as cut point: higher raw score for sequential 106.7 (14.5) vs. 94.7 (11.6) but not simultaneous or composite mental processing. No adjusted scores.

Horwood et al. [32]. Higher mean verbal IQ (102.1 vs 96.1 p < 0.05) and performance IQ (103.3 vs. 99.6, p > 0.15) after >8 m HM.

Smith et al. [33]. Significantly different in visual motor skills only: WRAVMA drawing 97.7 (14.6) vs. 90.6 (13.5), 95% CI = 1.0-9.2; and K-ABC triangle completion 10.6 (3.0) vs. 9.1 (2.5), 95% CI = 0.1-1.7. No HM effect from increased HM duration (>4 m, in infants with abnormal cranial ultrasound or <28 weeks gestation).

Johnson et al. [34]. Regardless of HM intake, significantly lower composite scores in extremely preterm children without serious neurosensory or cognitive impairment for reading 91 (13.4) vs. 98.7 (11.5); for mathematics 84.0 (15.6) vs. 98.8 (14.8). Data analyzed for children with and without serious functional or cognitive impairment while attending mainstream or special schools. Multivariate model show breast milk is one of the independent predictors of reading scores but not mathematic scores at 11 years. Other independent predictors included BSID-II MDI, and head circumference at 30 m, and perinatal and social factors. All independent predictors accounted for 31% of the variance for reading scores at 11 years.

Vohr et al. [35]. MDI and PDI in the highest 3 quintiles of human milk groups were higher (p <0.05) than no human milk group. Mean values for highest vs. no human milk for MDI 89.7 vs. 76.5, and for PDI 90.2 vs 78.4. HM intake as a continuous measure show that each 10 mL/kg/d increase in human milk ingestion, the estimated increase of 0.59 points in MDI and 0.56 points in PDI.