Growth and Nutrient Intakes of Human Milk–Fed Preterm Infants Provided With Extra Energy and Nutrients After Hospital Discharge

Growth and Nutrient Intakes of Human Milk Fed Preterm Infants Provided

With Extra Energy and Nutrients After Hospital Discharge Deborah L. O'Connor, Sobia Khan, Karen Weishuhn, Jennifer Vaughan, Ann Jefferies, Douglas M. Campbell, Elizabeth Asztalos, Mark Feldman, Joanne Rovet, Carol Westall, Hilary Whyte and on behalf of the Postdischarge Feeding Study Group

Pediatrics 2008;121;766-776 DOI: 10.1542/peds.2007-0054

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.pediatrics.org/cgi/content/full/121/4/766

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2008 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

ARTICLE

Growth and Nutrient Intakes of Human Milk–Fed Preterm Infants Provided With Extra Energy and Nutrients After Hospital Discharge

Deborah L. O’Connor, PhD, RD a,b,c , Sobia Khan, MSc, RD a,b,c , Karen Weishuhn, RD a,b,c , Jennifer Vaughan, RN, IBCLC a,c , Ann Jefferies, MD d,e , Douglas M. Campbell, MD d,f , Elizabeth Asztalos, MD d,g , Mark Feldman, MD d,h , Joanne Rovet, PhD d , Carol Westall, PhD i , Hilary Whyte, MD d,j , on behalf of the Postdischarge Feeding Study Group

a Physiology and Experimental Medicine Program, b Department of Clinical Dietetics, and j Department of Neonatology, Hospital for Sick Children, Toronto, Ontario, Canada; Departments of c Nutritional Sciences, d Pediatrics, and i Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada; e Department of

Pediatrics, Mount Sinai Hospital, Toronto, Ontario, Canada; f Department of Pediatrics, St Michael’s Hospital, Toronto, Ontario, Canada; g Department of Newborn and Developmental Pediatrics, Sunnybrook Hospital, Toronto, Ontario, Canada; h Department of Paediatrics, St Joseph’s Health Centre, Toronto, Ontario, Canada

The authors have indicated they have no financial relationships relevant to this article to disclose.

What’s Known on This Subject

What This Study Adds

Although more could be done to improve the nutritional status of low birth weight Adding a multi-nutrient fortifier to ⬃half of the milk provided to predominantly human (LBW) infants during their initial hospitalization, it is apparent that correction of acquired

milk-fed infants for 12 weeks after discharge may be an effective strategy in addressing nutrient-deficits for many very LBW infants is difficult to accomplish before discharge.

early discharge nutrient-deficits without unduly influencing human milk feeding when lactation support is provided.

ABSTRACT

OBJECTIVES. The purpose of this pilot study was to determine whether mixing a multi- nutrient fortifier to approximately one half of the human milk fed each day for a finite period after discharge improves the nutrient intake and growth of predomi-

www.pediatrics.org/cgi/doi/10.1542/

nantly human milk–fed low birth weight infants. We also assessed the impact of this

peds.2007-0054

intervention on the exclusivity of human milk feeding.

doi:10.1542/peds.2007-0054 This trial has been registered at www.

METHODS. Human milk–fed (ⱖ80% feeding per day) low birth weight (750 –1800 g)

clinicaltrials.gov (identifier NCT00413985).

infants (n ⫽ 39) were randomly assigned at hospital discharge to either a control or

Key Words

an intervention group. Infants in the control group were discharged from the

low birth weight, premature infant,

hospital on unfortified human milk. Nutrient enrichment of human milk in the

breastfeeding, human milk, growth

intervention group was achieved by mixing approximately one half of the human

Abbreviations

milk provided each day with a powdered multinutrient human milk fortifier for 12

LBW—low birth weight GTA— greater Toronto area

weeks after discharge. Milk with added nutrients was estimated to contain ⬃80 kcal

CA— corrected age

(336 kJ) and 2.2 g protein/100 mL plus other nutrients. Intensive lactation support

SGA—small for gestational age

was provided to both groups.

SAE—serious adverse event VLBW—very low birth weight

RESULTS. Infants in the intervention group were longer during the study period, and those

Accepted for publication Aug 27, 2007

born ⱕ1250 g had larger head circumferences than infants in the control group. There

Address correspondence to Deborah L.

was a trend toward infants in the intervention group to be heavier at the end of the O’Connor, PhD, RD, Room 8511C, The

Hospital for Sick Children, 555 University

intervention compared with those in the control group. Mean protein, zinc, calcium,

Ave, Toronto, Ontario, Canada M5G 1X8.

phosphorus, and vitamins A and D intakes were higher in the intervention group.

E-mail: deborah_l.o’connor@sickkids.ca PEDIATRICS (ISSN Numbers: Print, 0031-4005;

CONCLUSIONS. Results from this study suggest that adding a multinutrient fortifier to

Online, 1098-4275). Copyright © 2008 by the

approximately one half of the milk provided to predominantly human milk–fed infants American Academy of Pediatrics for 12 weeks after hospital discharge may be an effective strategy in addressing early

discharge nutrient deficits and poor growth without unduly influencing human milk feeding when intensive lactation support is provided.

N GENERAL, THE I earlier an infant is born before his or her expected term delivery date, the greater is their risk

for morbidity and malnutrition. 1,2 The reasons for undernutrition are widely known to clinicians and include that these low birth weight (LBW) infants are born with compromised nutrient reserves. Preterm infants often acquire nutrient deficits because initiation of parenteral nutrition and intralipids is frequently delayed; paren- teral glucose and lipid solutions are poorly tolerated; and feedings are frequently withheld for clinical proce-

dures, sepsis, or suspicion of necrotizing enterocolitis. 3 Furthermore, comorbidities such as chronic lung disease limit the volume of feeding and/or route that nutrients are supplied, further complicating the provision of adequate nutrition.

O’CONNOR et al

Although more could be done to improve the nutri- tional status and growth of LBW infants during their initial hospitalization, it is becoming increasingly appar- ent that correction of acquired nutrient deficits is diffi- cult to accomplish before discharge. In fact, many LBW infants leave the hospital with poorer nutritional status than when they started their postnatal life. 3–8 For exam- ple, using weight-for-age centiles (⬍10th centile) as a

crude indicator of nutritional status, Lemons et al 6 dem-

onstrated that whereas 22% of the 4500 VLBW infants who were born at 1 of the US Neonatal Network sites were at risk for undernutrition at birth, ⬃96% were so around the time of discharge. Furthermore, human milk–fed infants often accrue the greatest nutritional

deficits by discharge. 9 A plethora of review articles and

position statements from authoritative bodies under- score the concern about the nutritional status of human milk–fed LBW infants after hospital discharge. 2–4,10–13 In fact, the European Society for Paediatric Gastroenterol- ogy, Hepatology and Nutrition Committee on Nutrition recently recommended that human milk–fed preterm infants who are discharged from the hospital with sub- normal weight for postconceptional age be routinely supplemented to provide an adequate nutrient supply. 11 Despite the aforementioned concerns, recommendation, and the widely known advantages of human milk over

formula feeding, 14 no randomized, controlled trials have been conducted to ascertain whether multinutrient for- tification of human milk after hospital would be benefi- cial. The objectives of this pilot study, then, were to determine whether mixing a multinutrient fortifier to approximately one half of mother’s milk for 12 weeks after discharge would improve the nutrient intakes and growth of LBW (750 –1800 g) infants. Because few data specific to the LBW infant exist suggesting what impact this intervention might have on human milk feeding, we also assessed the duration and exclusivity of human milk use.

METHODS Study Population

Primarily and exclusively human milk–fed infants and their mothers were enrolled between April 2004 and November 2005 from NICUs located in the greater To- ronto area (GTA). These nurseries included Mount Sinai Hospital, Sunnybrook Hospital, The Credit Valley Hospi- tal, St Joseph’s Health Center, St Michael’s Hospital, Scarborough Hospitals (Grace and General Divisions), Scarborough Centenary (Rouge Valley Health System), and Toronto East General Hospital. Singleton or twin infants (born ⬍33 weeks’ gestational age and between 750 and 1800 g) who received ⱖ80% of their total feedings as human milk (fortified or unfortified) 3 days before hospital discharge were eligible to participate. Families agreed to feed their infants primarily human milk after hospital discharge and to supplement a pre- determined volume of the human milk fed with a pow- dered multinutrient fortifier for 12 weeks if so randomly assigned. Infants with serious congenital or chromo- somal anomalies that could affect growth were not eli-

gible to participate. Other exclusion criteria were grade 3 or 4 periventricular or intraventricular hemorrhage, oral steroids within 14 days of randomization, severe as- phyxia (hypoxia or ischemia characterized by an Apgar score of ⬍5 at 5 minutes), and known maternal alcohol or drug abuse. Families were also ineligible when their principal residence was outside the GTA or when the mother was unable to communicate verbally in English. Last, infants were excluded from participation when any single feeding at hospital discharge needed to be nutrient enriched or concentrated to ⬎24 kcal/fl oz (3800 kJ/L) or when ⬎50% of the daily feeds required nutrient enrichment.

Experimental Design

The Human Ethics Committees at the Hospital for Sick Children and each of the aforementioned recruiting hos- pitals approved the study, which was conducted accord- ing to the policies and procedures of each institution and the Canadian Tri-council policy statement on ethical

conduct of research involving human subjects. 15 A com- puter-generated randomization schedule stratified for gender and birth weight group (ⱕ1250 g, ⬎1250 g) was prepared by 1 of the authors (Dr O’Connor), who was not involved in enrolling study participants. A separate randomization schedule was prepared for twins. For avoidance of biasing in-hospital feeding practices or which individuals were approached about the study, randomization assignments were placed in individually sealed envelopes that were opened the day before hos- pital discharge (study day 1).

Infants who were randomly assigned to the control group were discharged from the hospital on unfortified human milk (from breast or expressed), as is routine clinical practice in the GTA. For infants who were ran- domly assigned to the intervention group, our goal was to supply roughly half the volume of human milk as nutrient-enriched feedings after hospital discharge. Nu- trient enrichment of human milk provided to infants in the intervention group was achieved by having caregiv- ers mix a predetermined volume (150 mL ⫻ infant weight [kg]/2) of human milk (fresh or thawed) with a powdered multinutrient human milk fortifier (4 single- use packets [0.9 g each] per 100 mL of human milk; Table 1). Remaining feedings were to be provided as unfortified milk (from the breast or expressed). Using the milk intake data collected from term-corrected age (CA) breastfed preterm infants published by Wauben et

al, 16 we estimated that infants in this study would con- sume ⬃150 mL/kg per day around the time of hospital discharge. The exact volume (150 mL ⫻ infant weight [kg]/2) of nutrient-enriched feeding to be fed was de- termined by the study coordinator (Ms Vaughan) at discharge and recalculated at 4 and 8 weeks after dis- charge to accommodate changes in body weight. In com- bination with unfortified human milk feedings, the av- erage daily energy and nutrient density provided to infants in the intervention group was predicted to be similar to those of infants who were fed commercially available nutrient-enriched formula designed for post- discharge feeding of preterm infants (eg, 22 kcal/fl oz

PEDIATRICS Volume 121, Number 4, April 2008

TABLE 1 Approximate Energy and Select Nutrient Composition of

discretion of the infant’s pediatrician how nutrient en-

Mature Human Milk With and Without a Multinutrient

richment was to be accomplished, but, in general, pow-

Fortifier

dered postdischarge formula (eg, Similac Neosure) was

added to human milk. The study coordinator, a certified lactation consultant, offered lactation support to moth-

Per 100 mL

Nutrients

Mature Preterm

Human Milk With

ers in both feeding groups. Mothers in both study groups

Human Milk

a Multinutrient

were provided free of charge the vitamin/iron drops,

feeding devices, and use of a double breast-expression Energy, kcal (kJ)

Fortifier (4 Packets)

electric breast pump (Purely Yours Breast Pump; Ameda, Protein, g

1.30 2.20 Mississauga, Ontario, Canada). Mothers in the interven- Fat, g

3.90 4.20 tion group were also provided the powdered human Carbohydrate, g

milk fortifier free of charge.

Minerals

Calcium, mg

Phosphorus, mg

14.00 78.80 Demographic Data, In-hospital Tolerance, and Morbidity

Magnesium, mg

3.00 9.70 Family characteristics of enrolled infants were obtained Sodium, mg

20.50 34.50 from a personal interview with 1 or both parents. This Potassium, mg

included mother’s gravidity and parity as well as paren- Chloride, mg

0.27 1.20 tal age, height, ethnicity, and years of education. Each Copper, mg

Zinc, mg

0.06 0.22 infant’s weight and gestational age at birth, size at birth Iron, mg

0.10 0.44 (ie, small for gestational age [SGA]or appropriate for Choline, mg

18.20 19.80 gestational age), and in-hospital course were obtained Selenium, ␮g

1.80 2.20 from medical charts. The last included the number of Vitamins

days that parental nutrition was provided, days to reach Vitamin A, ␮g

full enteral feeds (100 kcal/kg per day [420 kJ/kg per Vitamin D, IU

day]), and days the infant received nothing by mouth Vitamin E(dl-␣ tocopherol), mg

0.50 2.00 after reaching full enteral feeds. Morbidity outcomes Folic acid, ␮g

8.50 30.70 such as the number of infants with a confirmed case of

Milk nutrient composition values were obtained from the literature.

Where possible, milk 44 – 46

necrotizing enterocolitis (more than Bell stage II), sys-

composition values were used from studies that reported data on mature milk samples col- lected from mothers who delivered preterm. The powdered human milk fortifier used in this

temic infection (positive blood culture), or chronic lung

study was Similac Human Milk Fortifier (Abbott Nutrition, Montreal, Quebec, Canada)

disease before study day 1 were also recorded. Chronic lung disease was defined as the need for supplemental oxygen beyond 1 month chronological age or 36 weeks’

[3100 kJ/L] and 18 g/L protein). 2 Families could choose

postconception.

when during the day they wished to provide the nutri- The number of serious adverse events (SAEs) or un- ent-enriched feedings and whether they would use a

expected adverse events was determined from study day bottle or supplemental nursing system.* Most families

1 until 12 weeks after hospital discharge. An SAE was elected to use bottles. As is routine clinical practice in the

defined as any event that occurred during the clinical GTA, infants in the control group were provided with

trial that resulted in death or was life-threatening or vitamin drops consisting of vitamins A (1500 IU), D (400

disabling, required hospital admission, or required inter- IU), and C (30 mg) after discharge. For reduction of the

vention to prevent permanent impairment. possibility of inappropriately high intakes of fat soluble

vitamins, infants who were randomly assigned to the

Growth

intervention group were not provided with vitamins A The weight, length, and head circumference of infants and C and only 200 IU of vitamin D (ie, half the man-

was determined according to standardized procedures by ufacturer recommended dosage of D-Vi-Sol [Mead

the study coordinator at study day 1 and at 4 (⫾3 days), Johnson Nutritionals, Ottawa, Ontario, Canada]). A

8 (⫾3 days), and 12 (⫾3 days) weeks after hospital daily iron supplement (15 mg/day) was prescribed for

discharge during home visits. 19 Infants were weighed infants in both feeding groups.

twice in the nude using a precision scale (⫾2 g; Medela An algorithm was created before study initiation to

BabyWeigh; Medela, Mississauga, Ontario). Recumbent assess whether nutrient enrichment should be initiated

length and head circumference were measured twice to for an infant in the control group who demonstrated

the nearest 0.1 cm with a lengthboard (Ellard Instru- poor growth. In general, when an infant’s growth, as

mentation, Munroe, WA) and nonstretchable tape mea- assessed by weight gain over ⬎7-day period or 2 con-

sure (InserTape, Ross Canada, Abbott Laboratories, secutive home visits, dropped by 2 percentile curves on

Montreal, Quebec, Canada), respectively. the Infant Health and Development growth charts, 17,18 a

feeding intervention was initiated. The same was true

Enteral Intake

when an infant demonstrated clinical signs (eg, low Dietary intake diaries were mailed to parents 3 days tone, lethargy) of poor intake and growth. It was at the before each home visit (4, 8, and 12 weeks after dis-

*Usually made up of a plastic container serving as a reservoir for supplemental milk and a

charge). Information collected in these diaries included

capillary tube extending from the reservoir and placed adjacent to the mother’s nipple.

intake of human milk consumed at the breast, unforti-

O’CONNOR et al O’CONNOR et al

procedure. 20 The test-weighing protocol required a care- giver to weigh infants before and after every feeding for

3 consecutive days before the 4-, 8-, and 12-weeks post- discharge home visit, using the Medela BabyWeigh scale, which was specifically designed for this purpose. The increase in weight after each feeding (in grams) provides an estimate of the amount of milk (in millili- ters) consumed. Parents were instructed not to change the infant or add or remove blankets or clothing be- tween the pre- and postfeeding weights. Use of vitamin and mineral drops was also recorded in the food diaries. Energy and select nutrient intakes (protein, calcium, phosphorus, zinc, iron, vitamin A, and vitamin D) were estimated using human milk composition values from the literature, manufacturer label claims for infant for- mulas and the powdered human milk fortifier, and the Canadian Nutrient File. 21

To capture changes in human milk feeding frequency between 3-day food diaries, the study coordinator at the 4-, 8-, and 12-week postdischarge home visits asked mothers to estimate the number of times each day that her infant was fed since hospital discharge or the last visit and the number of these feeding that were provided as human milk. When the mother was no longer pro- viding human milk, she was asked to recall the date when human milk feeding was discontinued.

Statistical Analysis

We estimated a priori that 34 infants who completed the feeding intervention of this pilot study would allow us to detect a 1-SD difference in the mean weight of infants in the 2 feeding groups at 12 weeks after hospital discharge with 80% power at an ␣ level of .05. All data were analyzed using SAS 9.1 for Windows (SAS Institute, Cary, NC). All statistical tests were 2-tailed using an ␣ -level of .05. Data were checked to ensure that they were normally distributed (PROC UNIVARIATE) and, as appropriate, transformed. Demographic data were ana-

lyzed using t tests for continuous variables and ␹ 2 tests

for categorical variables. Continuous outcome data col- lected at ⬎1 time point were analyzed using mixed repeated-measures analysis of variance, which accom- modates missing observations controlling for gender and birth weight strata (750 –1250 g or 1251–1800 g [PROC MIXED]). Growth rates from study day 1 to 4 weeks, study day 1 to 8 weeks, and study day 1 to 12 weeks were individually analyzed using analysis of variance (PROC GLM) controlling for gender and birth weight strata. Volume of human milk consumed at the breast and the intake of fortified human milk were not nor- mally distributed; hence, nonparametric pairwise com- parisons were completed for these 2 variables at each time point (Wilcoxon rank-sum test).

Given that infants who are SGA account for 25% of the preterm population in some nurseries, we elected to in- clude these infants in this study. However, preterm infants

who are SGA often exhibit a different growth pattern than preterm infants who are appropriate for gestational age, so all statistical analyses were rerun excluding infants who were SGA to ensure that their inclusion did not influence

the study findings. 22 In these latter statistical analyses, 1 infant who developed hydrocephalus during the study in the intervention group was also excluded.

RESULTS Study Sample

Families of 143 infants who were known to be primarily human milk fed were provided an information letter about the study. Of these infants, 51 did not meet the eligibility criteria and 53 refused to participate. Of the infants who met the eligibility criteria, 29 were excluded because their primary residence was outside the GTA and/or their mother decided to formula feed after hos- pital discharge. Mothers of 22 infants did not want to use bottles, a breast pump, or a human milk fortifier. Of the

53 eligible infants, the main reasons for refusing to par- ticipate were (1) the study was too much for them or they were busy with other children at home, (2) it was too far to travel to clinic visits planned for after the feeding intervention phase, and (3) they were already participating in other research studies (Fig 1).

For the remaining 39 infants, 20 were randomly as- signed to the control group and 19 to the intervention group. We had sufficient data on 34 infants at 12 weeks after discharge to conduct an intention-to-treat analysis. There were 5 withdrawals in the study: 3 in the control group before study day 1 because of insufficient milk supply and 2 in the intervention group after study day 1 but before the 4-week home visit because of a sick infant and an overwhelmed mother who chose to withdraw from the study.

Infant and Family Demographics

Most baseline infant and family demographics, including infant weight, length, and head circumference, at study day 1 did not differ statistically between the 2 feeding groups (Table 2). There was a trend toward older gesta- tional age at birth (29.8 ⫾ 1.7 weeks) in the control versus intervention group (28.9 ⫾ 1.2 weeks; P ⫽ .06) and fewer male infants in the control (11 of 20) versus intervention group (14 of 19; P ⫽ .07). The potential impact of these trends on outcomes was addressed by including the randomization strata (birth weight and gender) in all of the statistical analyses.

Infants and their families were primarily white, and parents were well educated (Table 3). Domperidone (Motilium; Loudwater, High Wycombe, United King- dom), a galactagogue prescribed as part of usual clinical practice in the GTA, was used by a significant number of women in both feeding groups (control: n ⫽ 12; inter- vention: n ⫽ 14) after hospital discharge. 23

Growth

By the end of the study period, infants in the interven- tion group tended to be heavier (P ⫽ .07), and during the study period they were longer (P ⫽ .02) than infants

PEDIATRICS Volume 121, Number 4, April 2008

Approached for participation (n = 143)

Excluded (n = 104) Did not meet inclusion

Total No. of infants assigned

criteria (n = 51)

randomly (n = 39)

Refused to participate (n = 53)

Control group

Intervention group

(n = 20)

(n = 19)

FIGURE 1 Study sample recruitment flow chart.

Withdrawals before

Withdrawals after study

study day 1 (n = 3)

day 1 (n = 2)

Insufficient milk

Infant sick (n = 1)

supply (n = 3)

Mom overwhelmed (n = 1)

12 wk after discharge home visit

Control group

Intervention group

(n = 17)

(n = 17)

in the control group (Table 4). Among infants who were weeks after discharge, respectively. No difference in ⱕ 1250 g, those in the intervention group had a larger

mean head circumference existed between feeding mean head circumference than those in the control

groups for infants born ⬎1250 g. group (P ⫽ .0009). The adjusted mean head circumfer-

The aforementioned statistical analyses included all ences of infants who were ⱕ1250 g in the control group

infants as randomly assigned (intention-to-treat) includ- were 32.5 ⫾ 0.4, 34.7 ⫾ 0.4, 36.6 ⫾ 0.4, and 38.0 ⫾ 0.4

ing 2 infants in the control group who were fed human cm at study day 1 and at 4, 8, and 12 weeks after

milk that contained powdered discharge formula to in- discharge, respectively. Likewise, the adjusted mean

crease the nutrient concentration of the human milk fed head circumferences of infants who were ⱕ1250 g in the

to address slow growth. All statistical analyses controlled intervention group were 34.0 ⫾ 0.3, 37.0 ⫾ 0.3, 39.0 ⫾

for gender and birth weight strata (750 –1250 g or 1251–

0.3, and 40.6 ⫾ 0.3 cm at study day 1 and at 4, 8, and 12 1800 g). Re-analysis of these growth data without the

TABLE 2 Neonatal Characteristics of the Study Population of Preterm Infants

Intervention Birth weight, mean ⫾ SD (n), g

Gestational age at birth, mean ⫾ SD (n), wk a 29.8 ⫾ 1.7 (20)

14 (73.6) Birth weight strata (ⱕ1250g), n (%)

Male gender, n (%) b 11 (46.7)

8 (42.0) Weight at study day 1, mean ⫾ SD (n), g

2785 ⫾ 465 (18) Postconceptional age at study day 1, mean ⫾ SD (n), wk

37.8 ⫾ 3.3 (18) Days on parenteral nutrition, mean ⫾ SD (n)

12 ⫾ 8 (19) Days to full enteral feeds (100 kcal/kg per day), mean ⫾ SD (n)

12 ⫾ 7 (19) Days feedings withheld, mean ⫾ SD (n)

0.1 ⫾ 0.2 (19) Chronic lung disease, n (%)

8 (42.0) Confirmed cases of NEC, n (%)

0 (0.0) Confirmed cases of systemic infection, n (%)

1 (5.3) Size at birth, n (%) SGA

1 (5.3) Appropriate for gestational age

18 (94.7) Multiple birth status, n (%) Singleton

4 (21.1) Differences between feeding groups for continuous variables were assessed by 2-sided t tests and for categorical variables using ␹ 2 analyses.

Mean ⫾ SD values presented are unadjusted. NEC indicates necrotizing enterocolitis. a P ⫽ .06.

b P ⫽ .07.

O’CONNOR et al

TABLE 3 Family Characteristics a vention group having more rapid head circumference gains Characteristics

versus the control group from study day 1 to 8 weeks after Mother

Control

Intervention

discharge (P ⫽ .09). This difference in head circumference gain was highly significant for infants born ⱕ1250 g up to

Age, mean ⫾ SD (n), y

4 weeks after discharge (P ⬍ .0001). No differences in Parity, mean ⫾ SD (n)

Height, mean ⫾ SD (n), cm

weight gains were observed between the feeding groups at Gravidity, mean ⫾ SD (n)

any interval after discharge. When study day 1 anthropo- Ethnicity, n (%)

metrics (weight length or head circumference as appropri- White

ate) were added as a covariate to the statistical models, the Other

results for length did not change, but a statistically signifi- Education, n (%)

cant difference in head circumference gain between feed- No university degree

ing groups was observed from study day 1 to 12 weeks after University degree

Father discharge for infants born ⱕ1250 g (P ⫽ .03). There was a

trend toward infants in the intervention group having Age, mean ⫾ SD (n), y

Height, mean ⫾ SD (n), cm

greater weight gains to 4 weeks after discharge compared Ethnicity, n (%)

with infants in the control group (P ⫽ .05). These observed White

differences in length and head circumference gains early Other

after hospital discharge are consistent with previous obser- Education, n (%) a vations that infants are at greatest nutritional risk at this

time. 9,24 Re-analysis of the aforementioned growth gains University degree

No university degree

without the infants who were SGA and a child who devel-

Mean ⫾ SD values presented are unadjusted. Differences between feeding groups for contin-

oped hydrocephalus after randomization did not apprecia-

uous variables were assessed by 2-sided t tests and for categorical variables using ␹ 2 analyses.

bly change the results, and, if anything, P values were

P ⫽ .04.

strengthened.

infants who were SGA (n ⫽ 3) and a child who devel-

Enteral Intake

oped hydrocephalus after randomization did not appre- Most infants consumed at least 1 human milk feeding ciably change the results for length and head circumfer-

per day at the 12-week postdischarge visit; only 1 infant ence, but the trend for a difference in weight between

in the control group and a set of twins in the interven- the 2 feeding groups disappeared.

tion group did not. Likewise, the percentage of daily

A statistically significant difference was found between feedings provided as human milk at the 4-, 8-, and feeding groups among infants ⱕ1250 g in length gains

12-week postdischarge home visits did not differ be- (cm/wk) between study day 1 and 12 weeks after discharge

tween the control (81 ⫾ 37%, 71 ⫾ 40%, and 71 ⫾ (P ⫽ .009). There was a trend toward infants in the inter-

38%) and intervention (82 ⫾ 27%, 93 ⫾ 10%, and 88 ⫾

TABLE 4 Weight, Length and Head Circumference Measurements from Study Day 1 to 12 Weeks After Hospital Discharge

Mean ⫾ SD (n)

Mean ⫾ SD (n)

Treatment

Visit

Treatment and Birth

Effect

Effect

Weight Interaction

NS Study day 1

Weight, g a .13

Length, cm

NS Study day 1

Head circumference, cm

⬍.001 b Study day 1

Mean ⫾ SD values presented are unadjusted. Differences between feeding groups were assessed by repeated measures analysis of variance (ANOVA) controlling or gender and birth weight strata (ⱕ1250 g, ⬎1250 g). NS indicates not significant.

a An interaction between feeding group and time was found (P ⫽ .06). Intervention ⬎control at 8 (P ⫽ .10) and 12 weeks (P ⫽ .07) postdischarge. b Infants born ⱕ1250 g in the intervention group had larger heads than those in the control group during the study (P ⫽ .0009). Unadjusted

mean ⫾ SD head circumference; control (n ⫽ 6): 32.4 ⫾ 1.2, 34.6 ⫾ 0.6, 36.5 ⫾ 0.7, and 37.9 ⫾ 1.4 cm; intervention (n ⫽ 9): 34.1 ⫾ 1.2, 37.1 ⫾

1.0, 39.1 ⫾ 0.8, and 40.7 ⫾ 0.8 cm.

PEDIATRICS Volume 121, Number 4, April 2008

TABLE 5 Intake Volumes (mL/kg per D) by Feeding Type at 4, 8, and 12 Weeks After Hospital Discharge

12 wk, Mean ⫾ SD (n) At breast a Control

Parameter

4 wk, Mean ⫾ SD (n)

8 wk, Mean ⫾ SD (n)

38 ⫾ 26 (15) Unfortified expressed human milk b

13 ⫾ 21 (15) Fortified human milk c

Control d 9 ⫾ 34 (16)

49 ⫾ 21 (15) Total human milk (all sources)

Intervention e 51 ⫾ 18 (17)

11 ⫾ 17 (15) Total intake (all sources, human

milk, formula, other) f Control

Mean ⫾ SD values presented are unadjusted. Differences between feeding groups were assessed by repeated measures ANOVA controlling for gender and birth weight strata (ⱕ1250 g, ⬎1250 g) except for milk consumed at the breast and fortified human milk, which were analyzed at each time point by a Wilcoxon rank-sum test. Unless otherwise indicated, no statistically significant differences were noted between feeding groups.

a 4 weeks only: P ⫽ .06. b Control ⬎intervention, P ⫽ .03. c Intervention ⬎control, P ⬍ .0001. d Powdered postdischarge formula (eg, Neosure) was added to expressed human milk to a concentration of 24 kcal/oz (3300 kJ/L). e Similac Human Milk Fortifier was added to expressed human milk to a concentration of 24 kcal/oz (3300 kJ/L).

f Control ⬎intervention, P ⫽ .02.

15%) groups. Furthermore, the total volume of human For confirmation that intakes of vitamin A and D re- milk consumed (all sources) did not differ between feed-

mained within a safe intake range among infants who ing groups (Table 5). As planned, approximately half of

were consuming nutrient-enriched human milk, intakes the total human milk consumed each day in the inter-

were compared with adverse effect levels as published by vention group had human milk fortifier added. None of

the Institute of Medicine Dietary Reference Intakes for the aforementioned enteral intake results differed when

term-born infants. 26,27 This comparison demonstrated that the infants who were SGA and the infant with hydro-

mean vitamin D intakes (diet ⫹ vitamin and mineral cephalus were removed from the statistical analyses.

drops) remained below the no observed adverse effect level The mean volume of feedings (mL/kg per day) from

of 1800 IU/day for infants 0 to 12 months. 26 Likewise, any source (human milk ⫹ formula ⫹ other) was greater

mean vitamin A remained below the lowest observed ad- in the control than the intervention group at the 4-, 8-,

verse effect level of 6000 ␮g/day preformed vitamin A for and 12-week home visits (P ⫽ .02). Except for energy

infants who are younger than 1 year. 27 and iron intake, mean nutrient intakes of infants in the intervention group were greater than that of infants in

Serious Adverse Events

the control group (Table 6). Because the iron content of Only 1 infant who was randomly assigned to the inter- both human milk and the human milk fortifier used in

vention group had an SAE, which consisted of 2 hospital this study was low, the primary source of iron came from

admissions as a result of gastroesophageal disease. He iron drops. These results did not change appreciably

was withdrawn soon after study day 1 at the request of when the statistical analyses were rerun without the

his parents. This infant had not yet received the study infants who were SGA and the infant with hydroceph-

human milk fortifier after hospital discharge. alus; however, the difference in protein intake was no

longer statistically significant (P ⫽ .06). The nutrient

DISCUSSION

intakes of study infants were also compared with the recommended intakes for postdischarge feeding of pre-

Growth

term infants published by the Canadian Paediatric Soci- Data from this pilot study suggest that LBW (⬍1800 g) ety. 25 Mean energy, protein, zinc, and iron intakes gen-

infants who were fed human milk that contained extra erally were just at or below recommended levels for both

nutrients for 12 weeks after hospital discharge were feeding groups (Table 6). Mean vitamin D and calcium

longer during the study period (P ⫽ .02), and those born intakes generally exceeded the recommended levels in

ⱕ 1250 g had larger head circumferences than infants the intervention but not in the control group.

who were sent home on human milk alone (P ⫽ .0009).

O’CONNOR et al

TABLE 6 Daily Energy and Select Nutrient Intake at 4, 8, and 12 Weeks After Hospital Discharge

P-RNI 25 P Treatment

Visit

Effect Energy intake, kcal/kg

Protein intake, g/kg

Vitamin A, ␮g/d

NS Control

Vitamin D, IU

NS Control

Calcium, mmol

Phosphorus, mmol

Zinc, ␮mol/kg

Iron, ␮mol/kg

Data are for 15 to 17 infants and include vitamin and mineral drops. Differences between feeding groups were assessed by repeated measures ANOVA controlling for gender and birth weight strata (ⱕ1250 g, ⬎1250 g). P-RNI indicates recommended nutrient intakes for premature infants.

There was a trend toward infants in the intervention imens most consistently occurred within the first few group to be heavier at the end of the study period

weeks of hospital discharge and among very low birth (5535 ⫾ 766 g) compared with those in the control

weight (VLBW) infants. 11,24,32,38 This likely reflects the group (5042 ⫾ 967 g; P ⫽ .07). All statistical models

time frame and population at highest risk for malnutri- controlled for gender and birth weight strata (750 –1250

tion and that feeding strategies designed for term-born

g or 1251–1800 g). As far as we are aware, this is the first infants are probably insufficient to address nutrient def- randomized, controlled trial conducted to evaluate the

icits for VLBW formula-fed infants early after discharge. impact of adding energy and multiple nutrients to hu-

We know from our previous work with a large cohort of man milk fed to LBW infants after discharge.

LBW preterm infants (n ⫽ 463) that human milk–fed The growth results for human milk–fed infants re-

infants may, in fact, be at increased nutritional risk at ported herein are consistent with observations of higher

hospital discharge compared with their formula-fed growth among LBW infants fed nutrient-enriched as

counterparts. 9

opposed to a standard term formula after hospital dis- charge in most 24,28–35 but not all studies. 36,37 For example,

Estimated Energy and Nutrient Intakes

Carver et al 24 reported that infants (born ⬍1800 g) who The estimated energy intakes (kcal/kg per day [kJ/kg per were fed a nutrient-enriched postdischarge formula (eg,

day]) of infants in our study did not differ between

22 kcal/fl oz [3100 kJ/L], 19 g/L protein) weighed more feeding groups, suggesting that human milk–fed LBW and gained more weight until 2 months’ CA and were

infants are able to compensate to some degree for the longer at 3 months’ CA compared with infants who were

energy and/or nutrient density of their feeding. These fed a standard term formula (20 kcal/fl oz [2800 kJ/L],

observations are consistent with previously published

14 g/L protein). As was the case with human milk–fed comparisons among LBW infants who were fed standard infants in this study, infants in the study by Carver et al

term or nutrient-enriched formula after hospital dis- who had birth weights ⱕ1250 g and were fed nutrient-

charge in which no differences in energy intake were enriched feedings had larger head circumferences than

observed. 24,28,31 Furthermore, these data are consistent LBW infants who were fed a standard term formula after

with that of a well-controlled study in which term-born hospital discharge.

infants exhibited an early capacity to upregulate intake Although published comparisons of infants who were

in response to the energy density of formula provided. 39 fed standard term versus nutrient-enriched formula af-

Although energy intakes of infants in both feeding ter hospital discharge differed considerably in experi-

groups in this study were generally just at or below those mental design, growth differences between feeding reg-

recommended, they are consistent with those reported

PEDIATRICS Volume 121, Number 4, April 2008

Mean intakes of protein, calcium, phosphorus, vita- min A, vitamin D, and zinc differed between infants who were fed human milk that contained extra nutrients and those who were fed human milk alone. These data sug- gest that although infants fed human milk alone may be able to upregulate their intake to match the energy intake of infants in the intervention group, they were unable to compensate for the disproportionate concen- tration of several nutrients. Given that the intakes of protein, calcium, and zinc not only differed between feeding groups but also were consumed below recom- mended levels in the control group, we speculate that these nutrients in particular may be responsible, in part, for our observations of improved length and head cir- cumference gains (cm/wk) among infants ⱕ1250 g in the intervention group immediately after hospital dis- charge. The lack of difference in weight gains (g/kg per day) may reflect the lack of a difference in energy intake.

Our intake data are consistent with the results from studies that demonstrated improved protein, vitamin D, calcium, phosphorus, and zinc intakes among infants who consumed nutrient-enriched formula compared with those who consumed standard term formula or human milk alone after hospital discharge. 24,28–31,35 Lucas

et al 39 suggested that the higher calcium and phosphorus intakes among infants who consumed nutrient-enriched formula may have contributed to the observed increase in linear growth up to 9 months’ CA in their study.

Brunton et al 29 attributed the greater linear growth ve- locity and lean body mass accretion among infants who had bronchopulmonary dysplasia and were fed nutrient- enriched formula to higher protein and zinc intakes.

Finally Wauben et al 16 speculated that the higher per-

centage of body fat among LBW infants who were fed human milk after hospital discharge compared with in- fant formula (by parental choice) may have been attrib- utable to lower calcium, phosphorus, and protein in- takes.

As we extensively describe elsewhere, there are a number of approaches that could be used to provide extra nutrients to human milk–fed infants after dis- charge, each with its own set of strengths and limita-

tions. 2 One advantage of using a powdered human milk fortifier is that it minimizes the dilution of human milk. Limitations of using a powdered human milk fortifier include that they are not sterile and that these products were never designed or studied for use after hospital discharge. Unless carefully planned, using a powdered human milk fortifier to nutrient-enrich human milk after hospital discharge could result in inappropriately high intakes of certain nutrients, such as iron, vitamin A, and vitamin D. It is very important to note that different human milk fortifiers vary considerably in their nutrient content and are not interchangeable. To address some of the potential shortcomings of using the human milk fortifier that was used in this study (Similac Human Milk Fortifier), we altered the type and quantity of vitamin and mineral drops typically provided to human milk–fed

LBW infants at discharge. Because the human milk for- tifier that was used in this study was low in iron, infants were provided with iron drops (15 mg/day elemental iron) after hospital discharge, regardless of their feeding assignment. Instead of providing drops that contained vitamins A, D (400 IU), and C as we did in the control group, infants in the intervention group were given half the dosage of vitamin D (200 IU) and no vitamins A and

C. Nutrient intakes from all sources in the intervention group were well below the lowest observed adverse effect level for vitamin A (6000 ␮g/day) and iron (40 mg/day) and below the no observed adverse effect level for vitamin D (1800 IU/day). 26,27 Although the afore- mentioned adverse effect level cutoffs were established for healthy term-born infants, these data do suggest that adding this particular powdered human milk fortifier to approximately half of expressed human milk for a finite period of time after hospital discharge can be done in such a manner so as not to provide inappropriately high levels of select nutrients.

Impact of the Intervention on Human Milk Feeding

During the feeding intervention, we saw no differences between groups with respect to the number of infants who were being human milk fed, the total volume of human milk provided each day, or the percentage of daily feedings provided as human milk. At 12 weeks after discharge, 71 ⫾ 38% and 88 ⫾ 15.4% of daily feedings in the control and intervention groups, respec- tively, were provided as human milk. The percentage of infants who were still being fed human milk 12 weeks after discharge in this study is much higher than that reported in the literature and is likely because they were predominantly human milk fed at discharge and because of the significant amount of lactation support that they received at home. 41–44

Clinical Application

We acknowledge that there our limitations with this study that need to be considered before applying it to practice. First, this is a pilot study and, hence, the sample size was small. Second, mothers in this study had access to an unusually high level of lactation support after hospital discharge: a lactation consultant, breast pumps, scales to measure milk intake, and the galactagogue domperidone to stimulate milk production. Confirma- tion of the findings reported herein in larger studies with lactation support levels more typically available are war- ranted. Given competing priorities of families at hospital discharge and the resultant low rate of willingness to participate in research studies, future studies require multicenter collaboration. In addition to the short-term growth and nutrient intake outcomes reported herein, the success or failure of this feeding intervention needs to include evaluation of longer term human milk feeding and developmental and body composition outcomes. Fi- nally, whereas in this study the volume of human milk was directly assessed, literature values were used to es- timate the energy and nutrient content of breast milk; therefore, ascribing a specific nutrient intake level to precise growth outcomes needs to be done with caution.

O’CONNOR et al

CONCLUSIONS

7. Lucas A. Nutrition, growth and development of postdischarge Results from this pilot study suggest that adding a multi-

preterm infants. In: Silverman E, ed. Posthospital Nutrition in the nutrient fortifier to approximately half of the milk that

Preterm Infant. Columbus, OH: Ross Products Division, Abbott was fed to predominantly human milk–fed infants for 12

Laboratories; 1996:81– 89

weeks after hospital discharge may be an effective strat- 8. Merko S, Shah PS, Wong KY, et al. Nutrient intakes and egy in addressing early discharge nutrient deficits and

growth of very preterm infants born ⬍28 weeks gestation poor growth without unduly influencing human milk

[abstract]. Can J Diet Pract Res. 2002;63(2 Suppl):105 feeding when intensive lactation support is provided. In

9. O’Connor DL, Jacobs J, Hall R, et al. Growth and development addition to the short-term growth and nutrient intake

of premature infants fed predominantly human milk, predom- outcomes reported herein, future larger studies to eval- inantly premature infant formula, or a combination of human uate the success or failure of this feeding intervention milk and premature formula. J Pediatr Gastroenterol Nutr. 2003; need to include an assessment of human milk feeding 37(4):437– 446 rates at 1 year and longer term developmental and body 10. American Academy of Pediatrics. Nutritional needs of the pre-

term infant. In: Klienman R, ed. Pediatric Nutrition Handbook. composition outcomes.

5th ed. Elk Grove Village, IL: American Academy of Pediatrics; 2004:23–54

ACKNOWLEDGMENTS

11. Aggett PJ, Agostoni C, Axelsson I, et al. Feeding preterm in- This study was supported by the Institute of Musculo-

fants after hospital discharge: a commentary by the ESPGHAN skeletal Health and Arthritis, Canadian Institute of

Committee on Nutrition. J Pediatr Gastroenterol Nutr. 2006; Health Research. Ms Khan and Ms Weishuhn received 42(5):596 – 603 graduate student stipends from the Canadian Institute of 12. Greer FR. Feeding the preterm infant after hospital discharge. Health Research Training Program in Clinical Nutrition Pediatr Ann. 2001;30(11):658 – 665

13. Schanler RJ. Post-discharge nutrition for the preterm infant. Research.

Acta Paediatr Suppl. 2005;94(449):68 –73 The Postdischarge Feeding Study Group also included

14. Gartner LM, Morton J, Lawrence RA, et al. Breastfeeding and Kirsten Kotsopoulos (Mount Sinai Hospital), Kirsten

the use of human milk. Pediatrics. 2005;115(2):496 –506 McFadden and Pauline Darling (St Michael’s Hospital),

15. Canadian Institute of Health Research, Natural Sciences and En- Andrea Nash (Sunnybrook Hospital), Debby Arts-Rodas

gineering Research Council of Canada, Social Sciences and Hu- (St Joseph’s Health Care), Sandra Gabriele and Jaimie

manities Research Council of Canada. Tri-council Policy Statement: MacKinnon (Credit Valley Hospital), Peter Azzopardi

Ethical Conduct for Research Involving Humans. Ottawa, Ontario, (Scarborough Hospitals), and Jelena Popovic (Toronto

Canada: Public Works and Government Services; 1998 East General).

16. Wauben IP, Atkinson SA, Shah JK, Paes B. Growth and body We grateful to all of the families who participated in

composition of preterm infants: influence of nutrient fortifica- this study and especially thank the dietitians, nurses, and

tion of mother’s milk in hospital and breastfeeding post- physicians who assisted with study recruitment at each

hospital discharge. Acta Paediatr. 1998;87(7):780 –785 site. We also thank Abbott Nutrition (Montreal, Quebec,

17. Casey PH, Kraemer HC, Bernbaum J, et al. Growth patterns of Canada) and Mead Johnson Nutritionals (Ottawa, On-

low birth weight preterm infants: a longitudinal analysis of a tario, Canada) for providing, at our request, the human

large, varied sample. J Pediatr. 1990;117(2 pt 1):298 –307 milk fortifier and many of the disposable supplies used to

18. Casey PH, Kraemer HC, Bernbaum J, Yogman MW, Sells JC. collect and store human milk in this study.

Growth status and growth rates of a varied sample of low birth weight, preterm infants: a longitudinal cohort from birth to three years of age. J Pediatr. 1991;119(4):599 – 605

REFERENCES

19. Gibson R, ed. Principles of Nutritional Assessment. 2nd ed. New 1. Ehrenkranz RA, Younes N, Lemons JA, et al. Longitudinal

York, NY: Oxford University Press; 2005 growth of hospitalized very low birth weight infants. Pediatrics.

20. Woolridge M, Butte N, Dewey K, et al. Methods for the measure- 1999;104(2 pt 1):280 –289

ment of milk volume intake of the breast-fed infant. In: Jensen G, 2. O’Connor DL, Merko S, Brennan J. Human milk feeding of