See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/9081998

Growth and Development of Premature Infants

Fed Predominantly Human Milk, Predominantly
Premature Infant Formula, or a Combination of
Human Milk and Premature Formula
Article in Journal of Pediatric Gastroenterology and Nutrition November 2003
Impact Factor: 2.63 DOI: 10.1097/00005176-200310000-00008 Source: PubMed

CITATIONS

READS

89

318

21 authors, including:
Sonja L Connor

Kathleen M. Gustafson

Oregon Health and Science University

University of Kansas Medical Center

81 PUBLICATIONS 3,341 CITATIONS

55 PUBLICATIONS 1,627 CITATIONS

SEE PROFILE

SEE PROFILE

Lynn T Singer

Vance Zemon

Case Western Reserve University

Yeshiva University

158 PUBLICATIONS 5,844 CITATIONS

102 PUBLICATIONS 2,314 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

SEE PROFILE

Available from: Sharon Groh-Wargo

Retrieved on: 30 May 2016

Journal of Pediatric Gastroenterology and Nutrition

37:437446 October 2003 Lippincott Williams & Wilkins, Inc., Philadelphia

Growth and Development of Premature Infants Fed

Predominantly Human Milk, Predominantly Premature Infant
Formula, or a Combination of Human Milk and
Premature Formula
*Deborah L. OConnor, Joan Jacobs, Robert Hall, David Adamkin, Nancy Auestad,

Marcella Castillo, William E. Connor, Sonja L. Connor, Kathleen Fitzgerald,

#Sharon Groh-Wargo, **E. Eugenie Hartmann, Jeri Janowsky, Alan Lucas, Dean Margeson,

Patricia Mena, Martha Neuringer, Gail Ross, Lynn Singer, Terence Stephenson,
Joanne Szabo, and ***Vance Zemon
*University of Toronto and The Hospital for Sick Children, Toronto, Canada; Ross Products Div, Abbott Labs, Columbus,
Ohio; The Childrens Mercy Hospital, Kansas City, Missouri; University of Louisville & Kosair Childrens Hospital,
Louisville, Kentucky; INTA Univ de Chile, Santiago, Chile; Oregon Health & Sciences University, Portland, Oregon;
#MetroHealth Medical Center, Cleveland, Ohio; **Hunter College, New York, New York; Institute of Child Health, London,
United Kingdom; Weill Medical College, Cornell University, New York, New York; Rainbow Babies & Childrens Hospital,
Cleveland, Ohio; University of Nottingham, Nottingham, United Kingdom; Arkansas Childrens Hospital, Little Rock,
Arkansas; ***Yeshiva University, Bronx, New York; on behalf of The Ross Preterm Lipid Study Group: R. Carroll and B. Meyer
(The Childrens Mercy Hospital); P. Radmacher and S. Rafail (Kosair Childrens Hospital); A. Blanco Gomez (INTA Univ de
Chile); P. Fisher and S. Escoe (Oregon Health & Sciences University); R. Arendt and M. Davillier (Rainbow Babies &
Childrens Hospital); K. Kennedy (Institute of Child Health); J. Putis (Leeds General Infirmary); S. Newell (St. James Hospital,
Leeds); S. Carlisle (Arkansas Childrens Hospital); C. Broestl, C. Downs, Q. Liang, P. Pollack, W. Qiu, and D. Smart (Ross
Products Division); J. Deeks, S. Sullivan, R. Tressler (Abbott Labs); S. Buckley (Yeshiva University); J. Gordon and L.
Garcia-Quispe (Hunter College); and D. Pinchasik and M. Nesin (Weill Medical College).

human milk was fed before hospital discharge, it was fortified

(3,0503,300 kJ/L, 2224 kcal/oz). As infants were weaned
from human milk, they were fed nutrient-enriched formula with
or without arachidonic and docosahexaenoic acids (3,300 kJ/L
before term, 3,050 kJ/L thereafter) until 12 months CA. Formula fed infants were given nutrient-enriched formula with or
without added arachidonic and docosahexaenoic acids (3,300
kJ/L to term, 3,050 kJ/L thereafter) until 12 months CA. For the
purposes of this evaluation, infants were categorized into four
mutually exclusive feeding groups: 1) predominantly human
milk fed until term CA (PHM-T, n 43); 2) 50% energy
from human milk before hospital discharge ( 50% HM, n
98); 3) < 50% of energy from human milk before hospital
discharge (< 50% HM, n 203); or 4) predominantly formula
fed until term CA (PFF-T, n 119).
Results: PFF-T infants weighed approximately 500 g more at
term CA than did PHM-T infants. This absolute difference

ABSTRACT
Background: In a recent meta-analysis, human milk feeding of
low birth-weight (LBW) infants was associated with a 5.2 point
improvement in IQ tests. However, in the studies in this metaanalysis, feeding regimens were used (unfortified human milk,
term formula) that no longer represent recommended practice.
Objective: To compare the growth, in-hospital feeding tolerance, morbidity, and development (cognitive, motor, visual,
and language) of LBW infants fed different amounts of human
milk until term chronologic age (CA) with those of LBW infants fed nutrient-enriched formulas from first enteral feeding.
Methods: The data in this study were collected in a previous
randomized controlled trial assessing the benefit of supplementing nutrient-enriched formulas for LBW infants with arachidonic acid and docosahexaenoic acid. Infants (n 463,
birth weight, 7501,800 g) were enrolled from nurseries located in Chile, the United Kingdom, and the United States. If

Received: September 5, 2002; revised: March 14, 2003; accepted: April 9, 2003.
Address correspondence and reprint requests to Dr. Deborah L. OConnor, Associate Professor, Department of Nutritional Sciences, University of
Toronto, Senior Associate Scientist, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada Canada, M5G 1X8 (e-mail:
Deborah_l.oconnor@sickkids.ca).

437

438

OCONNOR ET AL.

persisted until 6 months CA. PFF-T infants were also longer

(1.01.5 cm) and had larger head circumferences (0.31.1 cm)
than both PHM-T and 50% HM infants at term CA. There
was a positive association between duration of human milk
feeding and the Bayley Mental Index at 12 months CA (P
0.032 full and P 0.073 reduced, statistical models) after
controlling for the confounding variables of home environment
and maternal intelligence. Infants with chronic lung disease fed
50% HM until term CA (n 22) had a mean Bayley Motor
Index about 11 points higher at 12 months CA compared with
infants PFF-T (n 24, P 0.033 full model).

Conclusion: Our data suggest that, despite a slower early

growth rate, human milk fed LBW infants have development at
least comparable to that of infants fed nutrient-enriched formula. Exploratory analysis suggests that some subgroups of
human milk fed LBW infants may have enhanced development, although this needs to be confirmed in future studies.
JPGN 37:437446, 2003. Key Words: Low birth weight
Infant nutritionHuman milkFormula feedingGrowth
Development.

In a recently published meta-analysis, Anderson et al.

(1) conclude that breast feeding is associated with higher
scores for cognitive development than is formula feeding, even after adjustment for confounding factors. Interestingly, this benefit was greatest for low birth weight
(LBW) infants, among whom those fed human milk
showed a 5.2 point advantage in IQ. The conclusions of
Anderson et al. are consistent with the individual observations of Lucas et al. (2,3) and others (4,5). Lucas et al.
reported that LBW infants fed unfortified human milk
for 4 weeks after initiation of enteral feeding performed
better on standardized tests of development at 18 months
corrected age (CA) than did LBW infants fed a significant proportion of total energy via standard term formula. Lucas et al. also reported that LBW infants fed a
specially designed nutrient-enriched preterm formula for
4 weeks postnatally had more advanced motor and mental development at 18 months CA than did infants fed
term formula. They also observed higher IQ, most notably verbal IQ, in boys 712 to 8 years of age who were fed
nutrient-enriched formula compared with those fed term
formula in infancy (6).
In contrast to these comparisons between unfortified
human milk and term formula, Lucas et al. (2) found no
difference in the developmental outcomes of LBW infants fed unfortified human milk for 4 weeks compared
with those fed a specially designed nutrient-enriched preterm formula. These comparisons suggest that human
milk feeding and nutrient enrichment may have independent, perhaps additive, effects on the development of
infants born prematurely.
In contemplating how these observations might be applied to current clinical practice, the following points are
worthy of consideration. First, the American Academy of
Pediatrics (7) now recommends that, under most circumstances, fortified human milk is the feeding of choice for
hospitalized LBW infants. Neither unfortified human
milk nor term formulas are currently recommended during the initial in-hospital course because they do not
meet the nutritional requirements of LBW infants, especially those of very LBW infants(<1,500 g) (7,8). Thus,
the aforementioned studies compared feeding regimens
that no longer represent ideal or recommended feeding
practices for LBW infants. Few published data exist to

date that directly examine the development of LBW infants fed fortified human milk compared with specially
designed nutrient-enriched formulas during and after
hospital discharge.
The purpose of this article is to compare the growth,
in-hospital feeding tolerance, morbidity, and development at 14 months CA of LBW infants fed different
amounts of human milk until term-corrected age (CA)
and then fed nutrient-enriched formulas until 12 months
CA when weaned with those of LBW infants fed
nutrient-enriched formulas from first enteral feeding until 12 months CA.

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

SUBJECTS AND METHODS

Subjects
The data presented here represent a further analysis of LBW
infants from a large prospective, randomized controlled trial
designed to assess possible benefits of supplementing nutrientenriched formulas for premature infants with oils containing
arachidonic acid (AA) and docosahexaenoic acid (DHA) (9).
Statistical comparisons of growth and development outcomes
in the original study were made among study formula groups as
randomized, but not according to the quantity of human milk
consumed.
Four hundred seventy preterm infants (< 33 wks gestational
age) with birth weights of 750 to 1,805 g were enrolled between
October 1996 and January 1998 from Neonatal Intensive Care
Units in the United States (n 334), United Kingdom (n
85), and Chile (n 51). Of these infants, 463 consumed either
human milk and/or formula before exiting the study and were
included in this data analysis. The remaining seven infants
exited the study before their first enteral feeding because of
medical circumstances or parental withdrawal of consent.
Singleton, twin, appropriate-for-gestational age, and small-forgestational age infants were eligible to participate, providing
they could be enrolled within 72 hours of first enteral feeding
(including trophic feeds or water) initiated by the 28th day of
life. Infants with congenital abnormalities that could affect
growth and development or who had undergone major surgery
before randomization were not eligible to participate. Other
exclusion criteria included periventricular/intraventricular
hemorrhage > Grade II, maternal incapacity (including maternal cocaine or alcohol abuse), liquid ventilation, asphyxia resulting in neurologic damage, or uncontrolled systemic infection at the time of enrollment. A human ethics review board at
each participating institution approved the study procedures.

GROWTH AND DEVELOPMENT OF PREMATURE INFANTS

Study design
After written informed consent was provided by at least one
parent or guardian, infants were randomized to one of three
study formula regimens (one control, two containing
AA+DHA). The study was designed to reflect actual feeding
practices at the participating nurseries. Thus, on Study Day 1,
infants could be exclusively human milk fed, exclusively formula fed, or fed a combination of human milk and formula.
Study Day 1 was defined as the day on which infants were
randomized and began enteral feeding. There were no restrictions in the study protocol limiting or discouraging either the
exclusivity or duration of human milk feeding. The protocol did
specify that from birth until hospital discharge, human milk
was to be fortified with a powdered or liquid fortifier to contain
3,050 to 3,300 kJ/L (2224 kcal/oz) and to deliver a minimum
of 2.8 g/kg/d of protein. No guidelines were provided with
regard to fortification of human milk after discharge. The energy density of unfortified human milk was estimated to be
2,800 kJ/L (20 kcal/oz). Formula-fed infants were fed a specially designed in-hospital preterm formula (modified Similac
Special Care ready-to-feed [3,300 kJ/L] Ross Products Division, Columbus, Ohio, USA) with or without (control) AA- and
DHA-enriched oils from the time of first enteral feeding until
term CA as described previously (9). Corrected age (CA) is
equal to the menstrual age plus chronological age of an infant.
At term CA, formula-fed infants were transitioned to their assigned postdischarge nutrient-enriched formula (modified NeoSure powder [3,050 kJ/L] Ross Products Division, Columbus,
Ohio, USA) with or without the same sources of AA+DHA as
in the assigned preterm formula to 12 months CA. In the
AA+DHA-containing formulas, AA and DHA provided 0.42%
and 0.26% of fatty acids, respectively, in the premature formula
and 0.44% and 0.16%, respectively, in the enriched formula. As
infants were weaned from human milk, they were transitioned
onto their randomized study formula (3,300 kJ/L until term,
3,050 kJ/L until 12 months CA).

Human milk feeding groups

Infants were classified into four mutually exclusive feeding
groups based on their relative human-milk-to-formula intake
from the first enteral feeding until their term CA birth date.
Infants were categorized as follows: 1) predominantly human
milk until term CA (PHM-T) if they consumed < 100 mL/kg
birth weight of formula for the total duration of their initial
hospital stay and > 80% of all feedings provided as human milk
at term CA ; and 4) predominantly formula fed until term CA
(PFF-T) if they consumed < 100 mL/kg birth weight of human milk for the total duration of their initial hospital stay and
> 80% of all feedings provided as formula at term CA. Infants
failing to meet the criteria for PHM-T or PFF-T were categorized as: 2) 50% of total energy as human milk ( 50%
HM); and 3) < 50% total energy as human milk (< 50% HM).
Because of the significant dilution of human milk, liquid fortifiers (e.g., Similac Natural Care, liquid formula concentrates,
etc) were classified as infant formula for the purposes of categorizing infants into human milk feeding groups.

439

records before the term, 2, 4, 6, 9, and 12 months CA study

visits. In addition, breast-feeding mothers were asked at each
post-term study visit if they were still breast-feeding. If they
had discontinued breast-feeding, the date of discontinuation
was recorded.

Demographic data
Neonatal, perinatal, and family characteristics of enrolled
infants were obtained from medical records or parental report
as described previously (9). The HOME Inventory (10) and the
vocabulary subtest of the WAIS-R (11) administered to the
biologic mother (if living in the home) were obtained to assess
the quality and quantity of cognitive, social, and emotional
support available to each infant in the home environment and
maternal intelligence, respectively.

Growth, in-hospital feeding tolerance,

and morbidity
Weight, length, and head circumference of infants were measured according to standardized procedures on Study Day 1 and
at term, 2, 4, 6, 9, and 12 months CA as described in detail
previously (9). Visit windows of 7 days were allowed up to
9 months CA and 10 days at 12 months CA. The percentages
of infants who had enteral feedings withheld for at least 1 day,
necrotizing enterocolitis (NEC), systemic infection, or chronic
lung disease were determined from medical records, as were
the number of days infants took to reach full enteral feeding
(420 kJ/kg/d or 100 kcal/kg/d) (9). Chronic lung disease was
defined as supplemental oxygen required beyond 1 month
chronological age or 36 weeks postconception.
The number of serious adverse events (SAEs) was determined from study day 1 until each infants 12-month CA birth
date. This definition of SAE excluded events during initial
hospitalization that might be expected in the natural course of
the preterm infant (except death) (9).

Visual, cognitive, motor, and language development

Visual acuity of each infant was assessed using the Teller
Acuity Card Procedure (Vistech Inc, Dayton, OH) (12) at 2, 4,
and 6 months CA. Each tester and his or her backup was trained
and certified. The Bayley Scales of Infant Development II (13)
were administered primarily by a single certified tester at each
site at 12 months CA to assess cognitive (Mental Index) and
motor development (Motor Index). The vocabulary checklist
from the infant version of the MacArthur Communicative Development Inventories (14), a standardized parent-report instrument, was completed to provide information about each childs
vocabulary comprehension at 9 and 14 months CA and vocabulary production at 14 months CA. Details of the training of
developmental testers and quality control procedures to reduce
variability among testers have been described previously (9).

Statistical methods
Duration and exclusivity of human milk feeding
Human milk and formula intake data were obtained from
flow sheets during initial hospitalization and from 3-day dietary

All available data from enrolled infants who received either

human milk or formula before exiting the study (463 of 470
infants) were used in these analyses. Once enrolled, no infants

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

440

OCONNOR ET AL.

were dropped from follow-up for any reason other than parental
withdrawal of consent, infant death, or change in family residence precluding follow-up (intent-to-treat analysis). Categorical variables were analyzed using 2 or Cochran-MantelHaenszel tests. Continuous variables were evaluated using
analysis of covariance (ANCOVA). Data obtained at more than
one time point were analyzed using repeated-measures analyses
(SAS PROC MIXED, Cary, NC) (15). In addition, analyses
of continuous outcome variables considered site, randomization
strata (gender and birth weight group [7501,250 g or 1,251
1,800 g]), and the formula assignment to which the infant was
originally randomized as preplanned covariates. For the growth
analyses, the infants size for gestational age (either
appropriate-for-gestational age [AGA] or small-for-gestational
age [SGA]) were included as additional preplanned covariates.
For developmental outcomes, additional preplanned covariates
included size for gestation, gestational age, HOME Inventory,
the vocabulary subtest of the maternal WAIS-R, prenatal smoking, and in-home smoking at hospital discharge. Because of its
previously reported influence on development, chronic lung
disease (yes/no) was included as a covariate in developmental
analyses (1618). Weight at 2 months corrected age was added
to these analyses to serve as a proxy for growth after hospital
discharge (only three infants remained in the hospital at this
time). Because of the small numbers of infants, the Little Rock
(n 24) and New York sites (n 16) were treated as a single
site in the statistical analysis.
Because of the large number of preplanned covariates for
development outcomes, a backward step-wise approach was
used to produce a reduced statistical model. In this approach,
covariates or factors, except the patients feeding group, with a
P value 0.15 were excluded sequentially until a final step
was reached, with only feeding group and covariates with a P
value < 0.15 remaining. Unless there were differences between
the full and reduced models with respect to the presence or
absence of a feeding group effect on developmental outcomes,
results for only the reduced models are presented. All statistical
tests of hypotheses were two-tailed with P 0.05 for main
effects and P 0.10 for interaction effects.

human milk to their infants after hospital discharge, 26

(11%) supplemented human milk feedings with a fortifier (primarily a liquid) or other nonstudy formula.
The duration of human milk feeding was highly related to the exclusivity of human milk feeding to term
CA (Fig. 1). Infants no longer on the assigned study
feeding were generally transitioned to a standard term
formula, unless they were transitioned very close to the
12-month CA birth date. In this case, they were switched
to cows milk. The time at which solids were introduced
did not differ among human milk feeding groups. Solids
were introduced at a median age (1st, 3rd quartile) of 4
months CA (25 months). Study formula assignments
(one control, two containing AA + DHA) were equally
distributed among the four human milk feeding groups.
Demographic data
Human milk feeding groups differed statistically in a
number of infant and family baseline characteristics
(Tables 1 and 2). Birth weights for male infants were
similar in the four human milk feeding groups, but
among female infants, those in the 50% HM group had
lower mean birth weights than did those in the PFF-T
group (LS means SE, 1,188 24 g vs. 1,314 20 g).
Infants in the < 50% HM group had a slightly lower, but
significant, gestational age at birth than did infants in the
PFF-T group. The ethnicity of feeding groups was different. Predominantly human milk fed infants were
mostly Caucasian (95%), whereas 54% to 65% of infants
in the other groups were Caucasian, with the remainder
being of African, Hispanic, or Latino.
The mothers of infants in the < 50% HM group were
2.5 to 3.1 years younger than were mothers of infants in

RESULTS
Duration and exclusivity of human milk feeding
Eighty-one percent of the 463 infants completed the
study to 12 months CA. Ninety percent of all infants
enrolled in the study remained on human milk and/or the
assigned study formula feeding until term CA. At 12
months CA, 68% of infants remained on human milk or
the assigned study formula. Forty-three (9%) infants
were predominantly human milk fed until term CA
(PHM-T, human milk feeding group 1). Ninety-eight
(21%) infants were fed 50% of total energy during
initial hospitalization from human milk ( 50% HM,
human milk feeding group 2). Two hundred three (44%)
infants were fed < 50% of total energy during initial
hospitalization from human milk (< 50% HM, human
milk feeding group 3), and 119 (26%) infants were predominantly formula fed until term CA (PFF-T, human
milk feeding group 4). Of the 232 mothers providing any

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

FIG. 1. Duration of human milk feeding (mean SD) by feeding

group. Unlike superscript letters denote statistically significant
differences. Horizontal hatched lines represent the first and the
third quartile within each feeding group.

GROWTH AND DEVELOPMENT OF PREMATURE INFANTS

441

TABLE 1. Neonatal and perinatal characteristics*

Characteristics

PHM-T

50% HM

<50% HM

PFF-T

Birth weight (g)

Gestational age at birth (wks)
Size at birth, n (%)
Small for gestational age
Appropriate for gestational age
Gender, n (%)
Male
Female
Multiple birth status, n (%)
Singleton
Twin
Apgar score
Ethnicity, n (%)
Caucasian
African
Hispanic/Latino
Other
Postnatal age at study day 1 (d)

1,275 312 (43)

29.7 2.0ab (43)

1,287 279 (98)

29.6 1.9ab (98)

1,288 287 (203)

29.5 2.1a (203)

1,332 279 (119)

29.9 2.0b (119)

7 (16)
36 (84)

11 (11)
87 (89)

13 (6)
190 (94)

13 (11)
106 (89)

20 (47)
23 (53)

58 (59)
40 (41)

114 (56)
89 (44)

59 (50)
60 (50)

39 (91)
4 (9)
8.3 1.2 (43)

81 (83)
17 (17)
8.2 1.0 (97)

160 (79)
43 (21)
8.0 1.4 (202)

87 (73)
32 (27)
7.9 1.5 (119)

41 (95)
2 (5)
0
0
5.5 2.5 (43)

64 (65)
10 (10)
9 (9)
15 (15)
5.1 3.3 (98)

110 (54)
31 (15)
44 (22)
18 (9)
5.0 3.3 (203)

71 (60)
28 (24)
7 (6)
13 (11)
5.0 3.0 (119)

* Values are mean SD (number of participants) unless otherwise noted. Unlike superscript letters within each row denote statistically significant
differences.
A statistically significant human milk feeding group*gender effect was found (P 0.042). Birth weights for male infants were similar among
the four feeding groups, but among female infants, those in the 50% HM group had a lower mean birth weight than did female infants in the PFF-T
group (LS, means SE; 1,188 24 g versus 1,314 20 g).
A statistically significant feeding group effect was found (P 0.015).

the other feeding groups. At the United States sites,

mothers of infants fed PHM-T or 50% HM had about
2 more years of education than did mothers of infants fed
< 50% HM or PFF-T. Similarly, about 40% of mothers in

the PHM-T or 50% HM groups from the United Kingdom had a university degree, compared with fewer than
10% in the < 50% HM or PFF-T groups. Maternal smoking during pregnancy and the proportion of infants ex-

TABLE 2. Family characteristics*

Characteristics

PHM-T

50% HM

<50% HM

PFF-T

Maternal age (yrs)

Maternal education
US (yrs)
Chile (yrs)
UK, highest qualification obtained, n (%)
None
<3 O-level
>3 O-level
A levels
Degree+
Maternal smoking during pregnancy, n (%)
Yes
No
Postnatal smoking in the home, n (%)
Yes
No
Prenatal care, n (%)
1st trimester
2nd trimester
3rd trimester or none
HOME inventory score
Maternal WAIS-R raw vocabulary score

29.7 5.1b (43)

29.0 6.5b (98)

25.6 6.3a (203)

28.0 6.4b (119)

15.1 2.0b (34)

N/A

14.0 2.6b (72)

11.8 1.5 (8)

12.7 2.2a (126)

9.3 2.4 (39)

12.4 1.9a (96)

11.3 1.0 (4)

1 (11)
0 (0)
2 (22)
2 (22)
4 (44)

4 (22)
2 (11)
4 (22)
1 (6)
7 (39)

9 (26)
8 (23)
16 (46)
1 (3)
1 (3)

6 (35)
3 (18)
7 (41)
0 (0)
1 (6)

2 (5)
41 (95)

18 (19)
79 (81)

57 (28)
146 (72)

41 (34)
78 (66)

4 (10)
38 (90)

15 (17)
73 (83)

70 (35)
132 (65)

35 (30)
80 (70)

42 (98)
0 (0)
1 (2)
39.4 3.4b (40)
53.2 8.9 (40)

85 (87)
11 (11)
2 (2)
36.9 5.8b (81)
45.0 13.5 (79)

163 (81)
32 (16)
3 (1)
35.0 5.8ab (175)
35.7 12.7 (174)

98 (83)
13 (11)
7 (6)
35.2 5.6a (103)
36.3 15.6 (100)

* Values are mean SD (number of participants) unless otherwise noted. Unlike superscript letters within each row denote statistically significant
differences.
A statistically significant human milk feeding group effect was found (P < 0.01).
UK education equivalents: <3, certificate of education or general CSE below C grade.
UK education equivalents: >3, certificates of education or any O level or general certificates of education grade AC.

A statistically significant human milk feeding group*gender effect was found (P 0.032).

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

442

OCONNOR ET AL.

posed to cigarette smoke in the home after hospital discharge were greater in the primarily formula-fed groups.
HOME Inventory scores were higher (better) in the
families of infants in the PHM-T and 50% HM groups
than those of infants in the PFF-T group but not the
< 50% HM group. Scores on the verbal subtest of the
WAIS-R were higher in mothers of infants in the PHM-T
group than in those of the < 50% HM or PFF-T groups.
These scores were also higher for mothers of male infants in the 50% HM-T group than for mothers of male
infants predominantly fed formula, but these differences
were not seen among female infants.

head circumferences at study entry. By term CA, PFF-T

infants had significantly greater mean length (1.01.5
cm) and head circumference (0.31.1 cm) than did infants fed either PHM-T or 50% HM. Statistically significant differences disappeared by 9 months CA for
length and by 4 months CA for head circumference. No
statistically significant interaction was found between
the human milk feeding group or birth weight (750
1,250 g or 1,2511,800 g) for weight, length, and head
circumference measurements, suggesting that feeding
type did not differentially influence the growth of
smaller versus larger infants.
In-hospital feeding tolerance and morbidity

Growth
There were consistent differences in the growth of
infants depending on human milk feeding group (Table
3). Predominantly formula-fed infants weighed more
(3548 g) at study entry than did infants in the other
feeding groups. By term CA, a dose-dependent relationship existed between infant weight and the proportion of
energy consumed as infant formula. On average, infants
fed predominantly formula weighed about 500 g or
18.5% more at term CA than did infants fed predominantly human milk. This absolute difference in weight
persisted until 6 months CA, after which the < 50% HM
group, but not the PFF-T group, continued to weigh more
than the PHM-T group until 12 months CA. PFF-T infants also tended to be slightly longer and had larger

Infants in the four feeding groups were similar in most

measures of early feeding tolerance and in-hospital morbidity (Table 4). The slightly shorter interval for the
PFF-T group to reach full enteral feeding may have resulted from differences in caloric density between preterm formula (3,500 kJ/L) and unfortified human milk
(2,800 kJ/L). It was standard practice in the participating
nurseries not to fortify human milk until full enteral feeding (100 kcal/kg/d) was established. The mean age of
infants at study entry (5 days) did not differ among feeding groups, nor did the number of days between study
day 1 and hospital discharge (about 40 days).
The percent of infants with at least one SAE or rehospitalization was lower in the groups that received more
human milk (P < 0.05, Table 4). Thirteen infants died (8

TABLE 3. Weight, length, and head circumference*

Characteristic
Weight (g)
Study day 1
Term-corrected age (CA)
2 months CA
4 months CA
6 months CA
9 months CA
12 months CA
Length (cm)
Study day 1
Term-corrected age (CA)
2 months CA
4 months CA
6 months CA
9 months CA
12 months CA
Head circumference (cm)
Study day 1
Term-corrected age (CA)
2 months CA
4 months CA
6 months CA
9 months CA
12 months CA

PHM-T

50% HM

<50% HM

PFF-T

1,198 326a (43)

2,740 374a (42)
4,389 656a (41)
5,658 845a (41)
6,755 958a (40)
7,948 1,072a (41)
8,814 1,151a (41)

1,192 270a (98)

2,952 524b (90)
4,793 741b (85)
6,065 878b (79)
7,101 964ab (79)
8,083 1,080ab (80)
8,871 1,021ab (79)

1,205 278a (203)

3,152 491c (199)
4,920 723b (191)
6,286 861b (189)
7,326 955b (181)
8,376 1,048b (174)
9,161 1,112b (172)

1,240 278b (119)

3,249 496d (115)
4,898b 692 (112)
6,163 871b (106)
7,228 957b (104)
8,342 991ab (103)
9,157 1,076ab (96)

38.6 3.1ab (41)

46.9 2.2a (42)
53.5 2.6a (41)
59.3 2.9a (41)
63.7 3.0a (40)
68.8 2.9a (41)
73.1 2.8a (41)

38.4 2.9a (94)

47.4 2.4a (88)
54.9 2.3b (84)
60.6 2.5b (79)
64.9 2.4ab (79)
69.3 2.6a (80)
73.1 2.7a (79)

38.9 3.1a (190)

48.2 2.4b (197)
55.3 2.6b (189)
61.0 2.6b (188)
65.5 2.6b (180)
69.9 2.9a (173)
73.8 3.0a (172)

39.3 2.9b (109)

48.5 2.2c (112)
55.2 2.3b (111)
60.7 2.1b (106)
64.9 2.6b (103)
69.5 2.5a (102)
73.0 2.6a (96)

27.0 2.2ab (40)

34.4 1.1a (42)
38.4 1.2a (41)
41.3 1.3a (41)
43.3 1.5a (40)
45.1 1.4a (41)
46.1 1.4a (41)

27.1 2.0a (95)

35.2 1.5ab (89)
39.4 1.5b (84)
41.9 1.6a (79)
43.8 1.7a (79)
45.3 1.7a (80)
46.4 1.7a (79)

27.3 2.0ab (191)

35.4 1.5bc (198)
39.2 1.5ab (189)
41.8 1.5a (189)
43.6 1.5a (179)
45.3 1.7a (174)
46.4 1.7a (172)

27.4 1.8b (111)

35.5 1.5c (113)
39.2 1.5b (112)
41.7 1.7a (106)
43.5 1.6a (103)
45.3 1.7a (103)
46.3 1.7a (96)

* Values are mean SD (number of participants) unless otherwise noted. Unlike superscript letters within each row (corrected age of child or visit)
denote statistically significant differences (P < 0.05).
A statistically significant feeding group*visit effect (P < 0.0001) was found for all anthropometric measures.

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

GROWTH AND DEVELOPMENT OF PREMATURE INFANTS

443

TABLE 4. In-hospital feeding tolerance and clinical problems*

Characteristics

PHM-T

50% HM-T

<50% HM-T

PFF-T

Number of subjects
Feedings withheld for at least 1 days (% of infants)
Feedings withheld because of gastric residuals (% of infants)
Days to reach full enteral feeds (100 kcal/kg/d) mean SD
(1st, 3rd quartile)
Suspected cases of NEC (% of infants)
Confirmed cases of NEC (% of infants)
Suspected systemic infection (% of infants)
Confirmed cases of systemic infection (% of infants)
Chronic lung disease (% of infants)
Days from study day 1 to hospital discharge mean SD
(1st, 3rd quartile)
Serious adverse events (SAEs), n (%)
Hospital readmissions n (%)

43
26
14

98
38
18

203
28
18

119
28
18

16 19 (4, 21)a
21
0
33
7
28

15 16 (5, 18)a
21
5
33
11
22

15 15 (4, 21)b
22
1
39
16
27

8 8 (3, 10)b
30
7
33
14
20

42 23 (20, 54)
14 (33)
14 (33)

37 19 (25, 49)
38 (39)
28 (29)

45 25 (28, 58)
93 (46)
85 (42)

37 23 (21, 43)
61 (51)
54 (45)

* Unlike superscript letters within each row denote statistically significant differences.
A statistically significant human milk feeding group effect was found (days to first enteral feeding P < 0.0001; SAEs and hospital readmissions,
P < 0.05).

before discharge), of whom 0, 9, 2, and 2 died in the

PHM-T, 50% HM, < 50% HM, and PFF-T groups,
respectively.
Visual, cognitive, motor, and language development
A statistically significant main effect of human milk
feeding on visual acuity as assessed by the Teller Acuity
Card Procedure was found (P 0.003), although absolute differences were not large (Table 5). Infants in the
feeding groups consuming any appreciable amount of
human milk had higher Teller Acuity scores during the
2- to 6-month period than did infants in groups predomi-

nantly formula fed. Site (P < 0.0001), absence of maternal prenatal smoking (P 0.028), greater weight at 2
months CA (P < 0.001), higher scores on the HOME
Inventory (P 0.004), and higher maternal scores on
the vocabulary subtest of the WAIS-R (P 0.025) were
statistically significant factors in predicting Teller Acuity
scores (reduced statistical model).
No differences in the Bayley Mental Index or Motor
Index were found among feeding groups. However, a
positive association between duration of human milk
feeding and Bayley Mental Index was found in the full
(P 0.032), but not the reduced, statistical model (P

TABLE 5. Visual, cognitive, motor, and language development*

Characteristics
Visual acuity, mean (cycles/degree) SD (octaves)
Teller Acuity Card Procedure
2 months CA
4 months CA
6 months CA
Bayley Scales of Infant Development II at 12 months CA
Mental Index
Motor Index
MacArthur Communicative Development Inventories
Vocabulary comprehension scores
9 months CA
14 months CA
Vocabulary production score**
14 months CA

PHM-T

50% HM-T

<50% HM-T

PFF-T

1.8 0.8 (40)

3.8 0.6 (41)
7.4 0.8 (40)

1.8 0.7 (84)

3.6 0.6 (76)
7.2 0.5 (75)

2.0 0.6 (179)

3.7 0.5 (183)
7.0 0.5 (174)

1.7 0.7 (110)

3.4 0.6 (102)
6.8 0.8 (101)

93.1 14.5 (41)

86.8 15.2 (41)

95.0 13.4 (79)

84.6 14.8 (79)

91.6 10.5 (172)

86.5 15.1 (172)

92.9 13.4 (96)

88.1 14.6 (95)

96.2 17.6 (39)

97.0 14.5 (38)

100.3 21.5 (78)

101.3 16.1 (68)

105.0 20.5 (168)

100.8 18.3 (145)

102.6 18.4 (100)

100.7 17.0 (79)

96.6 18.9 (38)

98.0 18.5 (68)

97.7 17.6 (146)

96.7 18.0 (79)

* Values are mean SD (number of participants).

A statistically significant human milk feeding group effect was observed for behavioral acuity (PHM-T, 50% HM, <50% HM, >PFF-T,
P 0.003).
Infants with a Bayley Mental or Motor score of <50 were excluded from the Mental Index statistical analyses. Likewise, infants with a Motor
Score <50 were excluded from the Motor Index statistical analysis.
No statistically significant human milk feeding group effect was found; however, in the full (P 0.032) but not reduced statistical model (P
0.0731) duration of human milk feeding was positively associated with the Mental Index.
A statistically significant human milk feeding group*chronic lung disease effect was found in the full (P 0.080) but not reduced model (P
0.152). Among children with chronic lung disease, 50% HM > PFF-T.

A statistically significant human milk feeding group*visit effect was found in the full (P 0.033) but not reduced statistical model (P 0.058).
At 9 months CA, PFF-T > 50 50% HM.
** A statistically significant human milk feeding group*birth weight stratum effect was found in the full (P 0.019) and reduced models (P
0.0632). Among infants born >1,250 g, 50% HM > PFF-T.

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

444

OCONNOR ET AL.

0.073). Site (P < 0.0001), gender (female > male, P

0.001), gestational age at birth (+ association, P
0.005), and birth size (AGA > SGA, P 0.034) were all
statistically significant factors in the final reduced model.
A relationship was found between chronic lung disease and feeding group in the full model (P 0.08) for
the Bayley Motor Index but was not sustained in the
reduced statistical model. Both the WAIS ( association,
P < 0.0001) and HOME Inventory (+ association, P
0.004) were statistically significant factors in the full
statistical model. In the full model, there was an indication that the subset of infants with chronic lung disease in
the 50% HM-T group (n 22, LS mean SE, 87.9
1.7) had greater Motor Index scores than did the same
subset of infants in the PFF-T group (n 24, 76.3 3.5,
P 0.033). Infants in the < 50% HM-T (n 54) group
did not differ from infants in either the 50% HM-T or
the PFF-T groups. Because of the large number of covariates included in the full model and the small number
of PHM-T infants (n 9) with chronic lung disease, we
were unable to produce estimates for this group.
A statistically significant relationship between feeding
group by age was observed for vocabulary comprehension (P 0.033 full, P 0.058 reduced). At 9 months
CA, the vocabulary comprehension score of PFF-T infants (LS mean SE, 105.6 1.8) was greater than that
of PHM-T infants (97.7 2.9, P 0.023), although
neither differed from infants in the mixed feeding
groups. This difference was not sustained at 14 months
CA. Site (P < 0.0001), lower maternal age (P < 0.0001),
greater gestational age (P < 0.001), and higher scores on
the HOME Inventory (P < 0.001) were statistically significant predictors of vocabulary comprehension in the
reduced model.
A significant interaction between feeding and birth
weight group was found for vocabulary production at 14
months CA using both full (P 0.019) and reduced
(P 0.063) models. Infants born weighing more than
1,250 g in the 50% HM-T group (LS mean SE, 96.8
3.5) had greater vocabulary production scores than did
infants weighing more than 1,250 g in the PFF-T group
(88.3 3.7). Site (P < 0.041), chronic lung disease (No
> Yes, P 0.008), weight at 2 months CA (+ association, P 0.006), and maternal scores on the HOME
Inventory (+ association, P < 0.0001) and vocabulary
subtest of the WAIS-R ( association, P 0.006) were
significant factors in the reduced statistical model.
DISCUSSION
This study suggests that the growth of LBW infants is
inversely related to the percentage of total energy consumed as human milk until term CA. These results, particularly the post-discharge growth comparisons between
human milk-fed and formula-fed preterm infants, are
consistent with previous reports in the literature (1921).

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

Our data suggest that, despite slower early growth,

human milk-fed LBW infants have development at least
comparable to that of infants fed nutrient-enriched formula from the time of first enteral feeding to 12 months
CA. Our exploratory analyses, which require confirmation in future studies, suggest that some subgroups of
human milk-fed LBW infants may have improved scores
on developmental tests. Specifically, infants in the feeding groups that consumed any appreciable amount of
human milk had higher Teller visual acuity scores than
did predominantly formula-fed infants, although the
quantitative difference was not clinically obvious. This is
of particular interest because two-thirds of formula-fed
infants in the current study consumed nutrient-enriched
feeding containing AA+DHA, supplements previously
shown to improve the visual acuity of LBW formula-fed
infants (9).
No differences in the Bayley Mental Index or Motor
Index were found among feeding groups. However, a
positive relationship between duration of human milk
feeding and the Bayley Mental Index was observed, even
after controlling for maternal and family characteristics
known to influence development. In addition, infants
with chronic lung disease who were fed 50% HM had
a mean Bayley Motor Index score about 11 points higher
than that of corresponding infants in the PFF-T group. It
is important to note that mean Bayley Mental (9295)
and Motor (8588) Index scores of LBW infants in this
study, regardless of feeding group, were below those of
term infants, suggesting that additional strategies, including nutritional ones, should be considered to improve the
development of LBW infants.
The relationship between human milk intake and measures of language development was mixed. It is conceivable that growth differences among feeding groups may
be confounding the relationship between human milk
feeding and development. A positive association between infant weight at 2 months CA and Teller visual
acuity scores was likewise observed (P 0.001). The
observed growth and development relationship is consistent with that reported by Mattia and deRegnier (22), in
which slower growth during the first month of life was
associated with poorer receptive and expressive language
development at 2 and 3 years of age. Hack et al. also
found that very LBW (VLBW) infants (n 249) whose
head circumference had not reached a normal percentile
by 8 months CA had poorer cognitive, educational, and
psychosocial development at 8 years of age (23). Several
studies, although not all, suggest that nutritional interventions that support growth and, in particular, brain
growth will lead to an improved outcome (6,2427). For
example, Lucas et al. followed the test results of cognitive function of premature infants fed a term or an enriched preterm formula for about 4 weeks after birth
(6,24). Those fed nutrient-enriched formula demonstrated not only better growth but also improved motor

GROWTH AND DEVELOPMENT OF PREMATURE INFANTS

and mental development at 18 months CA (Bayley) and
improved IQ, notably verbal IQ in boys, at 712 to 8 years
when compared with those who were fed term formula.
In addition to possible developmental advantages, human milk feeding was associated with fewer serious adverse events, specifically a reduction in the number of
hospitalizations after initial discharge. Several studies,
although not all, have shown an inverse relationship between breast-feeding and incidence of infection in term
infants (2837). There are a number of compounds in
human milk that might account for these observations,
including immunoglobulin A, lactoferrin, lysozyme, oligosaccharides, growth factors, and cellular components.
Nonetheless, differences across feeding groups in maternal education, maternal intelligence, in utero/postnatal
smoke exposure, and quality of the home environment
may also have influenced morbidity, independent of the
anti-infective and other unique components of human
milk (38).
The strengths of this study are the large sample size,
the detailed accounting of milk feedings, and the inclusion of a comprehensive set of growth, tolerance, morbidity, and developmental tests. Unlike in most comparisons of human milk-fed and formula-fed LBW infants, in
this study we attempted to control for genetic and environmental differences between human milk-feeding and
formula-feeding families. In addition, infants in this
study were fed according to current recommendations
(7,8). We acknowledge that the self-selection of families
to human milk-feeding groups, as opposed to random
assignment, allows for the possibility that study findings
could be attributable to unknown, and thus uncontrolled,
genetic and environmental differences among feeding
groups (e.g., quality of medical care). In addition, as a
consequence of the exploratory nature of this retrospective study, a large number of comparisons were made
without statistical adjustment for multiple comparisons;
thus, it is possible that some of our observations could
have occurred by chance. Finally, the relatively small
PHM-T group compared with the other feeding groups
and the variability in feeding after term CA diminishes
the power of these analyses.
In summary, LBW infants fed predominantly human
milk from first enteral feeding until term CA (fortified
in-hospital) and weaned to nutrient-enriched formula until 12 months CA are physically smaller for some time
after hospital discharge compared with LBW infants fed
a nutrient-enriched formula. Nonetheless, their development is at least comparable. In addition, human milk-fed
LBW infants appear to have reduced postdischarge morbidity, and in some subgroups perhaps even enhanced
development compared with infants fed nutrientenriched formula from the time of first enteral feeding to
12 months CA. Given the exploratory nature of this
study, future studies are needed to confirm these latter
observations.

445

REFERENCES
1. Anderson JW, Johnstone BM, Remley DT. Breast-feeding and
cognitive development: a meta-analysis. Am J Clin Nutr 1999;70:
52535.
2. Lucas A, Morley R, Cole TJ, et al. A randomised multicentre study
of human milk versus formula and later development in preterm
infants. Arch Dis Child 1994;70:F1416.
3. Lucas A, Morley R, Cole TJ, et al. Breast milk and subsequent
intelligence quotient in children born preterm. Lancet 1992;339:
2614.
4. Pollock JI. Mothers choice to provide breast milk and developmental outcome. Arch Dis Child 1989;64:7634.
5. Doyle LW, Rickards AL, Kelly EA, et al. Breastfeeding and intelligence. Lancet 1992;339:7445.
6. Lucas A, Morley R, Cole TJ. Randomised trial of early diet in
preterm babies and later intelligence quotient. BMJ 1998;317:
14817.
7. American Academy of Pediatrics, Committee on Nutrition. Pediatric Nutrition Handbook. Elk Grove Village, IL: American Academy of Pediatrics; 1998.
8. Nutrition Committee of the Canadian Paediatric Society. Nutrient
needs and feeding of preterm infants. Can Med Assoc J 1995;
152:176585.
9. OConnor DL, Hall R, Adamkin D, et al. Growth and development
in preterm infants fed long-chain polyunsaturated fatty acids: a
prospective, randomized controlled trial. Pediatrics 2001;108:
35971.
10. Caldwell BM, Bradley RH. Home Observation for the Measurement of the Environment. Little Rock, AR: University of Arkansas;
1984.
11. Wechsler D. Wechsler Adult Intelligence Scale, rev. San Antonio,
TX: The Psychological Corporation; 1981.
12. Teller DY, McDonald MA, Preston K, et al. Assessment of visual
acuity in infants and children: the acuity card procedure. Dev Med
Child Neurol 1986;28:77989.
13. Bayley N. Bayley Scales of Infant Development. San Antonio, TX:
The Psychological Corporation; 1993.
14. Fenson L, Dale P, Reznick J, et al. MacArthur Communicative
Development Inventories: Users Guide and Technical Manual.
San Diego: Singular Publishing Group; 1993.
15. SAS Institute Inc. SAS/STAT Users Guide. 4 ed. Cary, NC: SAS
Institute, Inc; 1989.
16. Carlson SE, Werkman SH, Tolley EA. Effect of long-chain n-3
fatty acid supplementation on visual acuity and growth of preterm
infants with and without bronchopulmonary dysplasia. Am J Clin
Nutr 1996;63:68797.
17. Singer LT, Siegel AC, Lewis B, et al. Preschool language outcomes of children with history of bronchopulmonary dysplasia and
very low birth weight. J Dev Behav Pediatr 2001;22:1926.
18. Singer L, Yamashita T, Lilien L, et al. A longitudinal study of
developmental outcome of infants with bronchopulmonary dysplasia and very low birth weight. Pediatrics 1997;100:98793.
19. Chan GM. Growth and bone mineral status of discharged very low
birth weight infants fed different formulas or human milk. J Pediatr 1993;123:43943.
20. Wauben IP, Atkinson SA, Shah JK, et al. Growth and body composition of preterm infants: influence of nutrient fortification of
mothers milk in hospital and breastfeeding post-hospital discharge. Acta Paediatr 1998;87:7805.
21. Wheeler RE, Hall RT. Feeding of premature infant formula after
hospital discharge of infants weighing less than 1800 g at birth.
J Perinatol 1996;16:1116.
22. Mattia FR, deRegnier RA. Chronic physiologic instability is associated with neurodevelopmental morbidity at one and two years in
extremely premature infants. Pediatrics 1998;102:E35.
23. Hack M, Breslau N, Weissman B, et al. Effect of very low birth
weight and subnormal head size on cognitive abilities at school
age. N Engl J Med 1991;325:2317.

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

446

OCONNOR ET AL.

24. Lucas A, Morley R, Cole TJ, et al. Early diet in preterm babies and
developmental status at 18 months. Lancet 1990;335:147781.
25. Cooke RJ, Embleton ND, Griffin IJ, et al. Feeding preterm infants
after hospital discharge: growth and development at 18 months of
age. Pediatr Res 2001;49:71922.
26. Morley R. The influence of early diet on later development. J
Biosoc Sci 1996;28:4817.
27. Morley R, Lucas A. Influence of early diet on outcome in preterm
infants. Acta Paediatr Suppl 1994;405:1236.
28. Cunningham AS, Jelliffe DB, Jelliffe EF. Breast-feeding and
health in the 1980s: a global epidemiologic review. J Pediatr 1991;
118:65966.
29. Dewey KG, Heinig MJ, Nommsen-Rivers LA. Differences in morbidity between breast-fed and formula-fed infants. J Pediatr 1995;
126:696702.
30. Woodward A, Douglas RM, Graham NM, et al. Acute respiratory
illness in Adelaide children: breast feeding modifies the effect of
passive smoking. J Epidemiol Community Health 1990;44:22430.
31. Wright AL, Holberg CJ, Martinez FD, et al. Breast feeding and
lower respiratory tract illness in the first year of life. Group Health
Medical Associates. BMJ 1989;299:9469.

J Pediatr Gastroenterol Nutr, Vol. 37, No. 4, October 2003

32. Frank AL, Taber LH, Glezen WP, et al. Breast-feeding and respiratory virus infection. Pediatrics 1982;70:23945.
33. Savilahti E, Jarvenpaa AL, Raiha NC. Serum immunoglobulins in
preterm infants: comparison of human milk and formula feeding.
Pediatrics 1983;72:3126.
34. Hanson LA, Ahlstedt S, Andersson B, et al. Protective factors in
milk and the development of the immune system Pediatrics 1985;
75:1726.
35. Adebonojo FO. Artificial vs breast feeding: relation to infant
health in a middle class American community. Clin Pediatr (Phila)
1972;11:259.
36. Cushing AH, Anderson L. Diarrhea in breast-fed and non-breastfed infants. Pediatrics 1982;70:9215.
37. Rubin DH, Leventhal JM, Krasilnikoff PA, et al. Relationship
between infant feeding and infectious illness: a prospective study
of infants during the first year of life. Pediatrics 1990;85:46471.
38. Sauls HS. Potential effect of demographic and other variables in
studies comparing morbidity of breast-fed and bottle-fed infants.
Pediatrics 1979;64:5237.