Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full...

12
Am J C/in Nutr l988;48: 1375-86. Printed in USA. © 1988 American Society for Clinical Nutrition 1375 Studies in human lactation: milk volumes in lactating women during the onset of lactation and full lactation3 Margaret C Neville, PhD; Ronald Keller, BA; Joy Seacat, RN, MA; Valerie Lutes; Marianne Neifert, MD; Clare Casey, PhD; Jonathan Allen, PhD; andPhilip Archer, PhD Introduction ABSTRACT After validation of test-weighing procedures milk volumes produced by 13 multiparous Caucasian women were followed longitudinally through the first year of lactation. All practiced exclusive breast-feeding for at least 5 mo. Milk transfer to the infant was low on days 1 and 2 and increased rapidly to 498 ± 129 g/d (1± SD) on day 5 and then more slowly to 753 ± 89 g/d during months 3-5. There was a characteristic milk volume for each mother- infant pair that was significantly related neither to milk yield on days 4-6 nor to birth weight. It was, however, strongly related to infant weight at 1 mo, suggesting that infant and/or mater- nal factors coming into play during the first month of life are strong determinants of subse- quent milk transfer to the infant. Am J C/in Nutr 1988;48: 1375-86. KEY WORDS Lactation, milk intake, breast-feeding, human milk yield, test weighing Methods Time is an inescapable variable in all studies of lacta- tion. A number ofcross-sectional studies ofmilk output broken down by duration of lactation are available (1, 2). Although these studies provide a series ofviews of the lactation performance ofa sample from a population, in the absence of longitudinal data on individuals, the cx- tent to which such data predict the temporal course of lactation in a particular mother-infant pair is not clear. A better understanding of the temporal course of lactation may help clarify the factors that govern milk transfer to the infant. Ofparticular interest in this regard is the rela- tion between milk transfer during the initiation of lacta- tion and later lactational performance. These considerations led us to perform a longitudinal study in 13 highly motivated lactating women, focusing particularly on the first 14 d postpartum. Because the amount of data required for each subject in a longitudi- nal study restricts the subject number, we chose as homo- geneous a sample as possible thus reducing the number of maternal variables that might have confounded inter- pretation ofthe results. The study was therefore confined to multiparous, nonsmoking Caucasian women of mid- dle-to-upper socioeconomic status living in Denver, CO. In this paper we report volume data from these subjects. Some data on milk composition and nutritional status ofthe mothers and infants have appeared in brief reports elsewhere (3, 4). Before reporting the data from these studies we examine the validity oftest weighing as a mea- sure ofbreast-milk production. Subjects All subjects in the longitudinal study were multiparous (mean parity 2.9 ± 1 .2, SD), nonsmoking Caucasian women who planned to continue breast-feeding for at least 1 y. None used hormonal contraception during the study (barrier meth- ods, abstinence, or sterilization were used). Mean age at the birth ofthe study infant was 3 1.9 ± 4.4 (SD) y (range 25-39 y). Median family income was > $35 000/y. All mothers were high school graduates and all but two had college degrees. Measure- ment of lactational performance began within 12 h of giving birth in all but one subject. The infant ofthis subject, who had gestational diabetes, was hospitalized for 2 wk after birth for the respiratory effects of meconium aspiration and evaluation of a possible ventricular septal defect. During this period the mother pumped her breasts to obtain milk for the infant, who was breast-fed only during maternal visits. Data from this sub- ject were analyzed separately and combined with the other data only where stated. A second subject was involved in an auto- mobile accident 2.5 mo postpartum resulting in hospitalization for 1 wk. She continued to pump her breasts during this time I From the Departments of Physiology, Pediatrics, and Preventive Medicine and Biometrics, University ofColorado School of Medicine, Denver, CO. 2 Supported by contract 0 1 HD 22801 and grant HD-19547 from the National Institutes of Health and grant RR-69 from the General Clinical Research Centers Program of the Division of Research Re- sources. 3 Address reprint requests to M C Neville, Department of Physiol- ogy, Box C240, UCHSC, Denver, CO 80262. Received June 22, 1987. Accepted for publication January 5, 1988. by on August 22, 2009 www.ajcn.org Downloaded from

Transcript of Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full...

Page 1: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

Am J C/in Nutr l988;48: 1375-86. Printed in USA. © 1988 American Society for Clinical Nutrition 1375

Studies in human lactation: milk volumes in lactating womenduring the onset of lactation and full lactation�3

Margaret C Neville, PhD; Ronald Keller, BA; Joy Seacat, RN, MA; Valerie Lutes;

Marianne Neifert, MD; Clare Casey, PhD; Jonathan Allen, PhD; andPhilip Archer, PhD

Introduction

ABSTRACT After validation of test-weighing procedures milk volumes produced by 13

multiparous Caucasian women were followed longitudinally through the first year of lactation.

All practiced exclusive breast-feeding for at least 5 mo. Milk transfer to the infant was low on

days 1 and 2 and increased rapidly to 498 ± 129 g/d (1± SD) on day 5 and then more slowlyto 753 ± 89 g/d during months 3-5. There was a characteristic milk volume for each mother-

infant pair that was significantly related neither to milk yield on days 4-6 nor to birth weight.

It was, however, strongly related to infant weight at 1 mo, suggesting that infant and/or mater-

nal factors coming into play during the first month of life are strong determinants of subse-

quent milk transfer to the infant. Am J C/in Nutr 1988;48: 1375-86.

KEY WORDS Lactation, milk intake, breast-feeding, human milk yield, test weighing

Methods

Time is an inescapable variable in all studies of lacta-

tion. A number ofcross-sectional studies ofmilk outputbroken down by duration of lactation are available (1,2). Although these studies provide a series ofviews of thelactation performance ofa sample from a population, in

the absence of longitudinal data on individuals, the cx-tent to which such data predict the temporal course oflactation in a particular mother-infant pair is not clear. A

better understanding of the temporal course of lactationmay help clarify the factors that govern milk transfer tothe infant. Ofparticular interest in this regard is the rela-

tion between milk transfer during the initiation of lacta-tion and later lactational performance.

These considerations led us to perform a longitudinal

study in 1 3 highly motivated lactating women, focusingparticularly on the first 14 d postpartum. Because the

amount of data required for each subject in a longitudi-nal study restricts the subject number, we chose as homo-

geneous a sample as possible thus reducing the numberof maternal variables that might have confounded inter-pretation ofthe results. The study was therefore confinedto multiparous, nonsmoking Caucasian women of mid-

dle-to-upper socioeconomic status living in Denver, CO.In this paper we report volume data from these subjects.Some data on milk composition and nutritional statusofthe mothers and infants have appeared in brief reportselsewhere (3, 4). Before reporting the data from these

studies we examine the validity oftest weighing as a mea-

sure ofbreast-milk production.

Subjects

All subjects in the longitudinal study were multiparous(mean parity 2.9 ± 1 .2, SD), nonsmoking Caucasian womenwho planned to continue breast-feeding for at least 1 y. None

used hormonal contraception during the study (barrier meth-ods, abstinence, or sterilization were used). Mean age at thebirth ofthe study infant was 3 1 .9 ± 4.4 (SD) y (range 25-39 y).Median family income was > $35 000/y. All mothers were high

school graduates and all but two had college degrees. Measure-ment of lactational performance began within 12 h of giving

birth in all but one subject. The infant ofthis subject, who hadgestational diabetes, was hospitalized for 2 wk after birth for

the respiratory effects of meconium aspiration and evaluationof a possible ventricular septal defect. During this period themother pumped her breasts to obtain milk for the infant, whowas breast-fed only during maternal visits. Data from this sub-

ject were analyzed separately and combined with the other dataonly where stated. A second subject was involved in an auto-mobile accident 2.5 mo postpartum resulting in hospitalization

for 1 wk. She continued to pump her breasts during this time

I From the Departments of Physiology, Pediatrics, and PreventiveMedicine and Biometrics, University ofColorado School of Medicine,Denver, CO.

2 Supported by contract 0 1 HD 22801 and grant HD-19547 from

the National Institutes of Health and grant RR-69 from the GeneralClinical Research Centers Program of the Division of Research Re-sources.

3 Address reprint requests to M C Neville, Department of Physiol-

ogy, Box C240, UCHSC, Denver, CO 80262.

Received June 22, 1987.Accepted for publication January 5, 1988.

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 2: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

Bottle Weight Difference per Feed (9)

1376 NEVILLE ET AL

Q.

C.,

C

w

0

a’4)

U,4)

I-

20

FIG 1. Bottle-fed infants: comparison between milk intake obtainedby test weighing of infants with milk intake obtained by weighing the

bottle before and after the feed. A. Three infants weighing 1.7-3.8 kghad 77 test weighings. B. Fifteen infants were test weighed once each.

The line represents the best-fitting linear regression. For A, the y inter-

cept was 1. 1 ± 4.2, the slope, 0.97 ± 0.03, and r, 0.93; for B the y inter-

cept was -0.01 ± 5.1, the slope, 0.99 ± 0.048, and r, 0.96.

and the infant returned to breast-feeding as soon as the motherreturned from the hospital. Three-month data from thismother were not included in the analysis. Where mastitis wasconcurrent with test weighing, as shown by high levels of so-

dium and chloride in the milk (5), the data were excluded fromthe analysis.

Six study infants were male (mean birth weight 3529 ± 337

[SDI g; mean gestational age 40.0 ± 1 . 1 wk [SD]). Seven studyinfants were female (mean birth weight 3054 ± 3 17 g [SD];mean gestational age, 39. 1 ± 1 .9 wk [SD]). Solids were intro-

duced between 4 and 9 mo (mean 7.0 ± 1.5 mo, SD) with for-mula being used occasionally (� 240 mL/wk) after 4 mo inthree subjects. Most other infants went directly to cow milk at

12 mo. One infant was weaned at 8 mo; all others wereweaned at � 12 mo. The study continued through weaningfor nine subjects; four others discontinued the study beforeweaning.

Mothers participating in the validation of the test-weighingstudy had characteristics similar to those in the longitudinalstudy. Infants used in validation oftest weighing were all hospi-talized in the pediatric intensive care unit of University Hospi-tal, mostly for low birth weight. All were growing well at the

time of measurement.Data on infant and maternal morbidity; estimated duration

of feeds; intake of formula, water, and beikost; and menstrual

status were recorded at monthly intervals. All subjects (for hos-pitalized infant, parents) gave informed consent for these pro-cedures, which had been previously approved by the Human

Subjects Committee ofthe University ofColorado Health Sci-ences Center.

Measurement ofmilk volume production

All milk amounts produced during any given measurementperiod were recorded. Milk received by the infant was mea-

sured by test weighing as outlined below. Pumped volumeswere measured when milk was expressed. Nine subjects weretest weighed at all feeds from birth to 14 d. The other threewere test weighed for at least 9 ofthe first 14 d. All 12 were testweighed weekly from weeks 3 through 8 and then monthly un-til the end ofthe study. One subject undertook a modified, lon-gitudinal protocol, making test weighings on days 6 and 14 andat monthly intervals thereafter. Five-milliliter samples were

taken from both breasts at midfeed at regular intervals through-out this study (6); compositional findings will be reported else-where.

Test weighing. Mothers or research nurses were instructedin the use of the electronic balance for test weighing to obtain

a measure of milk transfer to the infant. They estimated anyspit-up or spilled milk and those who leaked substantial quanti-

ties of milk wore breast pads that were weighed at each feed.Infants were weighed before and after every feed for a giventime period, usually 24 h plus two feeds into the next 24-h pe-riod but up to every feed during the first 14 d oflactation. Frombirth to 4 mo, infants were weighed on a Sartorius electronic

balance (model 3862MP6, Westbury, NY; accurate to 1 g) setto integrate the values of2O successive weighings.

Test weighing ofolder infants. Because movement artifactswere found to be highly significant with older infants, an dcc-

tronic balance with a swing was devised for test weighing: aninfant swing was suspended from a Mettler PC24 electronicbalance (Mettler Instrument Corp, Hightstown, NJ; range 24kg with a readability of 1 g, integration time 1 s, stability detec-tor update speed of 0.2 s, ME 41648 data output module).Weights and a stability signal were transmitted to a computerwith printer. Even with this equipment repetitive weights on amoving infant could vary by � 300 g. We found it possible to

overcome the variation in weights by performing a variation ofthe statistical techniques described by Ansombe (7) for rejec-

tion of outliers. We called this method iterative trimming (IT,see Appendix A). To use IT, 50-100 repetitive weights wereobtained, the mean and SD ofthe weights were calculated, andthose weights lying > 1 SD from the mean were dropped. Themean and SD were then recalculated; the procedure was re-

peated until the SD was < 3 g.Validation oftest weighing. The limitations of test-weighing

were examined in a number of ways: 1) We evaluated IT forweighing active infants. 2) We compared milk intake volumesin hospitalized, growing, premature infants receiving eitherbottles or gavage feedings. Research nurses weighed both the

bottle and the infant before and after feeding. 3) We comparedthe results of 48-h test weighing with hourly milk productionvalues obtained by using a special pumping technique with syn-thetic oxytocin. 4) We examined the day-to-day variation inmeasured volumes. The detailed techniques used for analysis

are outlined below.Comparison oftest weighing with hourly pumped volumes.

Five Caucasian women who had at least some college educa-tion and who were fully breast-feeding an infant aged 1-6 mo

test weighed their infants before and after each feed for a 48-h

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 3: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

HUMAN MILK VOLUME 1377

a)

>-

0.-’.���0-C,

C

0a)

70-

60-

50-

40-

30

20-

I0-

C-

30Test

35We I g hi

40Yield

I I I

-C

E

a,

E0

>

4,Q.

E

0 3 6 9

Time (hours)

FIG 2. Hourly volumes ofpumped milk in five lactating women. The points represent the mean across individuals

with twice the SEM indicated by the distance between bars. The dotted line represents average milk yield obtained

by test weighing and measurement ofall pumped and expressed volumes for 48 h. Inset: Mean and SEM of pumpedvolumes for hours 2-8 for each woman as a function ofmean hourly milk production obtained by 48-h test weighing.The line represents equality between the two measures.

period. They also recorded the volume of all milk pumped

from the breasts during this period. Within the succeeding 2wk these women were admitted to the Clinical Research Centerof the University Hospital where their breasts were pumpedevery hour for 8 h according to the following schedule: An dcc-

tric pump (Medela, Inc, Crystal Lake, IL) fitted with dual headswas used to pump both breasts simultaneously for a minimumof 10 mm or until milk ceased to flow freely. One drop (0.05mL) ofsynthetic oxytocin (Syntocinon#{174}, Sandoz Pharmaceuti-cal, East Hanover, NJ) was administered intranasally and thepumping was continued for another 5 mm. Milk volumes were

obtained by weighing the collection vials before and after

pumping using a Sartorius electronic balance.

Data analysis

Determination ofdaily milk intake. To overcome the problemof deciding what feeds to include in a day (8), we analyzed thedata three ways: we calculated 1) the total milk produced duringthe first 24 h of the total interval, 2) the milk produced during

the last 24 h of the total interval, and 3) a value derived for aninterval from the beginning ofthe first feed to the beginning of

the last feed. For this derived value the summed weight ofall butthe first feed was divided by the number of hours in the totalinterval and multiplied by 24 to obtain the daily yield. The dailyyields obtained by the three methods were compared and the

median value, usually the value obtained by method 3, was re-ported. Unusable data were obtained for 54 of the total 3484

feeds for which test weights were collected in the longitudinalstudy; this was due to problems with the balance, mother fallingasleep before the second weighing, etc. When this occurred,method 3 was used for the intervals before and after the missedfeeds if the total time of test weighing was > 24 h; if the total

time was < 24 h the data for that day were discarded. Twenty-five (7%) of373 test-weigh days were discarded for these reasons.Expressed and leaked milk was weighed and added to the total.

Corrections to test-weighing data. Two types of correctionwere applied to data obtained by test weighing, a correctionfor the insensible weight loss by the infant during the feeds,amounting to 2 g. kg� . h’ (9, 10), and a correction for specificgravity (1.032) when the data were to be expressed in volumerather than weight units. Because these two corrections by andlarge offset each other, infant data are not corrected and arestated in grams/day. This facilitates comparison with data fromother studies (8).

Standard statistical methods from the program ABSTAT(Anderson-Bell, Canon City, CO) were used to obtain meansand variances across subjects. Multiple-linear-regression tech-niques and Student’s I test were used in the analysis of individ-ual temporal trends and other relations among variables.

Results

Validation oftest weighing

Test weighing ofbottle-fed infants. Test weights werecompared with the weight of milk taken from the bottle

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 4: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

Time (Days Post Partum)

1378 NEVILLE ET AL

a�0

E

C0

C.,

.a0

1,

E

0>

FIG 3. Rates ofmilk production in exclusively breast-feeding women. A. The corrected daily milk yield is plotted

for days 1 through 6; thereafter lumped values derived from regression analysis as described in the text are plottedfor several segments of the postpartum interval. B. Means (small circles), SD, and range of values from A. Large

circles represent a subject with gestational diabetes.

as shown in Figure 1 . For the three infants measured atevery feed for a period of 3-5 d each (A), test weighinggave significantly different values (1 ± SEM, 50.4 ± 1.9

g/feed) from bottle weighing (52.9 ± 2 g/feed; one-tailedpaired t test, p < 0.00 1). When the data were correctedfor insensible weight loss using an average feed length of

30 mm, the mean value from test weighing was 52.7± 1 .9 g/feed, not significantly different from the valueobtained from weighing the bottle (one-tailed paired t

test, p > 0.2). Figure lB shows milk intake per feed plot-ted as in Figure lA for 15 different infants measured at asingle feed each. For these data the test weight was lessthan the bottle weight by a mean of 2.5 ± 1 .3 g (SEM).This value is not significantly different from the pre-

dicted insensible weight loss of 2.8 g/feed. The randomerror ofthe daily difference between test weighing of theinfant and weighing of the bottle before and after feedsamounted to ±3% of the weight of formula taken forboth groups, suggesting that the process oftest weighingitselfis associated with an acceptably small error.

Comparison oftest weighing with hourlypumped milkvolumes. The question of whether test weighing in thehome can produce representative data was addressed by

comparing results from 48-h test weighing with milkyields obtained by hourly pumping with synthetic oxyto-cm used to obtain a complete letdown. Figure 2 shows

the mean hourly pumped milk volumes obtained fromfive women over an 8-h period in a hospital. For compar-ison, the horizontal dashed line gives the mean 48-h testweighing volume obtained at home expressed as an

hourly rate of milk production. Volumes from the firsttwo pumpings were higher than the mean volumes ob-

tamed by test weighing because of the presence of resid-ual milk; thereafter, the hourly pumped volumes did notdiffer significantly from the test-weighing mean. The vol-umes obtained from hours 3 to 8 were averaged for eachsubject and compared to the mean volume obtained by48-h test weighing by that subject in her home, again cx-pressed as grams/hour(Inset, Fig 2). The mean difference

between the two techniques across individuals was 0.7 g/h (�-� 3%), suggesting that test weighing in the home bymotivated, educated subjects carries an acceptably lowerror and does not interfere with lactation performance.

Milk yield and transftr in thefirst week of lactation

The upper portion (A) of Figure 3 presents individual

patterns of milk yield during the period of exclusivebreast-feeding in our 12 normal subjects. After the first 6d postpartum the data are grouped to reduce the varianceofthe individual yields (see below). The lower portion of

the figure (B) shows the mean, SD, and range of valuesfor the same periods. These average data indicate a grad-

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 5: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

1400

1200

400

200

0 I 2 3 4 5 6 7 8 9 10 4Days Postpartum

HUMAN MILK VOLUME 1379

U,

Ea

4,

E

0

>

a

FIG 4. Milk production during the onset of lactation. Milk volume production from the present study (open

circles) is compared with infant intakes from the present study (closed circles) and from references 1 1 (open squares,n = 9) and 12 (open triangles, n = 9). Data from references 13 and 14, obtained by manual expression ofall milk for

the first 10 d oflactation, is depicted by closed triangles (n = 7). Distance between bars represents 2 SEM.

ual increase in milk yield for the first 36 h followed by anabrupt increase during hours 49-96 (mean slope 200mL/d) after which yield levels off.

During the first 6 d only four subjects actually followedthe mean pattern shown in Figure 3B. Three subjectsshowed a more gradual increase over the first 1 3-30 dand four produced large volumes of milk on day 4 fol-lowed by a decreased yield as lactation progressed. Thedifference between total milk yield and milk transferredto the infant during the onset of lactation is shown inFigure 4. The difference is greatest on day 4 (fl’.- 140 g),

primarily reflecting the last group ofmothers who tendedto produce more milk than their infants took in and useda breast pump to reduce engorgement. By 10 d the yieldin all but one of these subjects had leveled off to within30 mL of infant intake. One subject (shown by the + inFig 3A) continued to overproduce, leaking considerablequantities of milk into breast cups, until weaning wasbegun after 5 mo.

The data from one subject, the hospitalization ofwhose infant forced her to remove most milk by breastpump during the first 14 d postpartum, are shown aslarge circles in Figure 3B. Despite use ofthe breast pump,her volumes fell within the range of the breast-feedingsubjects.

Figure 4 also shows the data from the three other de-tailed studies (1 1-14) of milk volume in the first 2 wk

postpartum. Data in two of these studies (1 1 , 1 2) repre-sent infant intakes obtained by test weighing the infantas in the present study. Data from the third study (13,14) were obtained by breast pump or hand expression.

Milk yield and transfer duringfulllactation

The mean daily intakes are tabulated along with theyields in Table 1 . Milk yield increased slowly from a

mean of �‘.-550 mL/d on day 5 to “.-750 mL/d at 1 mo to“.-850 mL/d at 5 mo. The differences between the meanvalues at 2 mo (yield, 744 ± 3 1 mL/d [SE]; transfer 694± 27 g/d [SE]) and 5 mo (yield, 849 ± 33 mL/d [SE];transfer 838 ± 37 g/d [SE]) were significant (p < 0.02)indicating that both variables continue to increasethrough the period ofexclusive breast-feeding. The rangeof individual daily yields was large, � 800 mL/d duringweek 2 decreasing to � 400 mL/d by 2 mo.

Figure 5 shows a comparison between our data andinfant intake volumes from literature studies that fit the

following criteria: data were obtained by test weighingthe infant, exclusive breast-feeding was validated, threeor more subjects were studied, and milk transfer was bro-

ken down by monthly interval. The mean values fromthe 1 5 qualifying studies are similar to those observed inour study, suggesting that milk transfer to the infant in

these Denver mother-infant pairs is representative ofmilk transfer in most populations throughout the world.

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 6: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

1380 NEVILLE ET AL

TABLE 1

Milk yields and intakes

Days

postpartum Corrected milk yield

Numberof

subjects Infant milk intakeNumber of

subjects

n (range) mL/d g/d

1

2

3

4

5

6

7

8

910

11

14

21 (19-23)

28(24-32)

35(33-39)

42(40-46)

49(47-53)

56(54-63)

90(84-98)

120(117-126)

150(145-162)

180(172-185)

l801

210 (208-221)

2lOf

240(230-256)

240t

270(275-287)

270j

300(295-316)

300:�

330(317-344)

360(346-375)

56± 65(-ll-155)t

185± 103(12-379)

393±158(226-745)

580±250(306-1010)

563± 145(354-929)

558±156(360-888)

610± 187(421-1008)

657 ± 236 (442-1223)

606 ± 105 (485-872)682±254(395-1283)

654±168(410-1001)

668 ± 163 (423-1024)

703 ± 108 (601-935)

794± 178(550-1134)

726± 135(506-1054)

743 ± 113(601-926)

749± 124(586-986)

744±111(560-914)

772± 127(616-943)

754± 103(603-889)

849± 119(736-1170)

790±134(523-1027)

848± 99(719-1027)

738 ± 161 (486-963)

872± 67(796-945)

645± 195(288-1006)804± 135(720-1002)639 ± 221 (242-889)

853± 46(787-889)

578 ± 248 (143-896)

838 - -

578±233(132-861)

427±250(73-772)

7

10

11

11

12

10

8

9

1010

8

10

10

13

12

12

10

13

12

13

13

13

9

12

6

104

12

5

11

1

9

9

44± 7l(-31-149)t

182± 86(44-355)

371±153(209-688)

451 ± 176(164-694)

498± 129(323-736)

508±167(315-861)

573± 167(406-842)

581 ± 159 (410-923)

580 ± 76 (470-720)

589± 132(366-866)

615±168(398-934)

653 ± 154(416-922)

65 1 ± 84 (554-786)

770± 179(495-1144)

668± 117(465-930)

71 1 ± 111 (554-896)

709± 115(559-922)

694± 98(556-859)

734± 114(613-942)

711 ± 100(570-847)

838± 134(688-1173)

766±121(508-936)

820± 79(710-936)

721 ± 154 (486-963)

848± 63(796-945)

622±210(288-1002)818± 158(720-1002)618 ± 220(223-871)

817± 91(681-871)

551 ± 234(129-894)

720 - -

554±240(120-860)

403±250(65-770)

6

9

10

10

11

9

7

8

9

9

8

9

10

13

12

12

10

12

10

12

12

13

9

12

6

93

12

4

11

1

8

8

a 1± SD. Range given in parentheses.

t Negative values are due to insensible weight loss.t Includes only subjects giving < 100 kcal/d beikost.

Longitudinal correlations in milk output

There is no simple mathematical relation to describethe longitudinal course of milk-volume production. Wedivided the postpartum interval into six segments corn-prising days 2-6, days 7-14, days 15-28, days 29-56,days 60- 150, and days 154-240. Days 2-6 represent theonset of lactation. We found the mean volume for eachindividual from days 4 through 6 to be highly correlated(r > 0.9) with daily volumes from days 2 to 6. This meanvolume was used as representative ofthe pattern of lacto-genesis in each individual. The rate of change of yieldand transfer over the remaining time segments as well asthe mean milk yields and transfers were determined bylinear regression for each individual. The SD about theregression line (S�.�) was used to assess the day-to-dayvariability within subjects. These quantities were thenaveraged across subjects to give the values for the mean

yield, intake, slope, and day-to-day variation (Table 2).Data for months 6-9 are not included in Table 2 becauseonly five subjects (four shown in Fig 3A and one in Fig3B) continued exclusive breast-feeding during this inter-val (Table 1). Because the slopes and variations were sim-ilar for yields and intakes, only the data for intake areshown. It is the mean yields from this analysis that areplotted for each individual in Figure 3B.

Milk volume intake increased rapidly from days 7 to14 (slope 12 mL/d) and more slowly from days 14 to 28(mean slope 7 mL/d). The mean rate ofincrease thereaf-ter was 1 mL/d from the individual slopes correspondingto the increase in mean milk intake described above. Inthe five subjects continuing full breast-feeding frommonths 5 to 8, there was a slight, statistically insignificantdecrease in milk output over this final interval.

The mean day-to-day variability within subjects was

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 7: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

ll00�

1000

900

800

700

600

500

400

0

0

S

�18

I

2-I-

3 4 5 6 7

Months Postpartum

FIG 5. Milk intakes during established lactation. The lines show the smoothed mean from this study and ±1 SD.Points are data from the literature obtained by test-weighing offully breast-fed infants (see Appendix B). C, Pao et al

(15); N, Hofvander et al (16); 0, Butte and Calloway (17); #{149},Butte et ai (18); �, Steenberger et al (19); A, Dewey et al(20); i�, Dewey et al (21); D, Salmenpera et al (22); �9, Walgren (23); *, Janus et al (24, 25); U, Whitehead and Paul

(26); Q, Stuffet al (27); 0, Chandra (28); V, Carvalho (29); �, Prentice et al (2); and 5, present study.

HUMAN MILK VOLUME 1381

C.�

�2)0.

C,

E

a)

C

C

H

fairly constant throughout lactation (SD 60-80 mL/d,giving a maximum upper 95% confidence limit of 120g/d; Table 2). The interindividual variation in intake de-

creased from a high of 1 50 mL/d (CV ‘-�.-22%) at 1 mo to80 mL/d (CV 10%) after 6 mo. Two types of evidence

suggest that there is a characteristic rate of milk transferfor each mother-infant pair after the first month of lacta-tion: 1) During months 2-5, when all subjects were fullybreast-feeding, there was a highly significant difference(p < 0.00 1) between the highest (878 ± 26 mL/d, SEM)

and the lowest (59 1 ± 33 mL/d, SEM) daily milk yields.2) The correlation coefficient for milk transfer in an mdi-vidual mother-infant pair during month 2 comparedwith months 3-5 was 0.63 (p < 0.02, n = 13).

Sources ofvariability in milk transfer to the breast -ftd

infant

To determine whether the yield at the onset of lacta-tion, presumably controlled by such maternal character-istics as breast development and hormonal milieu (30),was related to milk production in full lactation, we exam-

med the correlation of mean milk yield and intake fordays 4-6 with milk intake in full lactation (Fig 6). Therewas a highly significant relation (p < 0.01) between milk

intake on days 4-6 and milk intake through day 28 (Fig

6A). Thereafter, there was no significant relation. A sim-ilar pattern of correlation between milk yield days 4-6and milk intake was observed (Fig 6A). When milk yield

and intake on days 4-6 were correlated with milk yieldthrough lactation, a similar pattern emerged (data notshown). Results obtained using the Spearman rank cor-relation were similar. The conclusion is that milk trans-

fer in full lactation cannot be predicted from milk yieldor intake during the first week postpartum.

The relation between milk intake and the weight of theinfant at the beginning and end of each time interval is

shown in Figure 6B. The correlation coefficient betweenmean milk intake and infant weight at the beginning ofthe segment was “.�0.4 (p > 0.1, n = 12) for the firstmonth increasing to -‘.--0.6 (p < 0.05, n = 13) after 1 mo(Fig 6B). The correlation coefficient between milk intakeand infant weight at the end of the interval was signifi-cantly higher. By 150 d there was a very strong relation

between infant weight and milk intake at that time (p< 0.0 1) as has been found by others (2, 3 1). This increasein correlation with time implies, but certainly does notprove, that milk intake is the independent variable in therelationship between infant weight and milk intake.

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 8: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

1382 NEVILLE ET AL

TABLE 2

Characteristics ofmilk intake in exclusively breast-feeding mother-infant pairs

Time interval (days postpartum)

7-14 15-28 30-59 60-150

Subjects(n)Meandailyproduction(mL/d±SD)

Mean daily intake (g/d ± SD)Meanslopeforintake(g/d±SEM)Mean day-to-day variation in intake (g/d ± SEM)Meanfeedfrequency(feeds/d±SEM)Mean feed duration (mm ± SEM)

12661 ± 175

6 15 ± I 30

12.4± 3.456.2 ± 12

7.5± 1.117 ± 2

12 13739 ± 164 751 ± 107

689 ± 148 707 ± 104

7.0± 2.3 0.8± 1.155 ± 6 63 ± 12

8.2± 0.8 8.1± 0.618 ± 2 17 ± 2

13

787 ±88753 ± 89

1.3± 0.679 ± 18

7.3± 0.516 ± 2

r

Days0-28

p

Days3O-59

r p r

Days6O-l50

0

Correlation between mean daily intake andInfant birth weight

Infant weight at 1 moInfant weight gain 1 mo postpartum

Total milk intake 1 mo postpartum

0.30

0.69

0.79I .0

NS

<0.02

<0.01-

0.06 NS0.57 <0.05

0.84 <0.010.64 <0.05

0.440.66

0.6 10.49

>0.10<0.02

<0.05<0.10

a Subject 3 excluded from analysis up to day 30 because ofgestational diabetes. There were at least three times per subject within each interval.

r is correlation coefficient, p is probability that there is no correlation using a two-tailed test. A one-tailed test would give higher levels of significance;

statistical measures obtained from linear regression analysis with n = 12.

Although birth weight is not a strong predictor of milk

intake throughout lactation, infant weight at 1 mo is (Ta-ble 2). Weight gain during month 1 is, in turn, stronglycorrelated with milk intake over this interval (p < 0.01).Thus, lactation performance during weeks 2-4 postpar-tum is a strong predictor ofmilk output during the subse-quent period of full lactation as shown by the significantrelation between total milk intake during month 1 post-partum and subsequent milk intake (Table 2).

The mean feed frequency and duration (Table 2) var-ied little during lactation. There was no significant rela-tion between milk intake and feed frequency, feed dura-tion, the product of feed frequency and duration, or in-fant sex (p > 0.1) at any time between days 5 and 150.Because ofthe limited subject number and the homoge-neous sample, we did not examine the effects of maternalvariables such as age, parity, or weight gain during preg-nancy.

Discussion

Test weighing as a measure ofmilk volume

Test weighing was evaluated in a large number of labo-ratones with a wide variety of methodologies (8, 16, 18,27, 32-38). In assessing the available comparisons be-tween formula weight and test weighing, it becomes clearthat good results were achieved where an integratingelectronic balance was used (1 8, 37) and errors in estima-tion of intake from the bottle were avoided by carefulweighing before and after the feed. Recording errors werereduced ifthe balance had a print-out. With some excep-tions (32, 36), good results were also achieved in the fieldwith mothers carrying out their own test weighing. Hof-

vander et al (I 6) found no significant differences between24-h test weighings performed in the home and thoseperformed by the same mothers in a hospital. Butte et alfound (33) that deuterium oxide dilution and test weigh-ing in the home give remarkably similar values for milkvolume production. Our comparison ofmilk productionvalues obtained by test weighing with those obtained by

consecutive pumping suggests that intelligent, well-moti-vated women are able to obtain accurate data in the field.These studies, as well as the broad agreement betweenmilk volume intakes recently measured throughout theworld (Fig 5), indicate that when carried out with appro-priate equipment by reliable mothers or research work-ers, test weighing is a consistent measure of milk in-take (9).

A serious problem in determining mean milk transferfor an individual mother-infant pair is the day-to-dayvariability in volumes obtained by test weighing, a prob-1cm that has been extensively discussed, particularly byButte and Garza ( 1 8) and by Woolridge et al (37). In the

present study, regression analysis oftest weights obtainedduring several 24-h periods showed that mean intraindi-vidual variation averaged 60-80 g/d throughout lacta-tion. Thus, an individual 24-h test weigh result can differfrom the true mean milk transfer by � 1 5% (95% confi-dence interval). The precision of the mean estimate canbe increased ifseveral days oftest weighing are averaged(8, 27, 38). Three to 4 d oftest weighing will usually re-duce the coefficient ofvariation (CV) to � 5%.

Onset of lactation

Including this study four investigations of milk vol-ume production during the onset of lactation are avail-

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 9: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

C

a)

U

a)0

UC

0

C

a)

0

U

a)

0

C

H

A1.0 -

0.8-P’(o.oI

\::\-_\

0.6 :?��P� ‘�

B

4-6 7-14 5-2830-60 61-150I I I I I I

4-6 7-14 15-2830-60 61-150

HUMAN MILK VOLUME

Days Post Partum

FIG 6. Relation of milk intake at various times postpartum with milk yield and intake days 4-6 postpartum (A)

and infant weight (B). Correlation coefficients were determined by regression analysis on 12 subjects who practicedexclusive breast-feeding from the onset oflactation. The dashed horizontal lines give the critical values for two-sidedlevels ofsignificance for sample correlation coefficients under the null hypothesis ofno correlation. A. Closed trianglesdepict correlation coefficients between mean milk yields days 4-6 postpartum and milk transferred to the infant

during the stated interval. Open triangles represent the correlation between milk transfer days 4-6 and subsequentmilk transfer. B. Closed circles represent the correlation between average milk intakes for that interval and infantweights at the beginning ofeach ofthe intervals. Open circles represent the correlation between average milk intakes

for that interval and infant weights at the end ofeach ofthe intervals shown on the abscissa.

1383

able (Fig 4) ( 1 1 , 1 2, 14). In all studies volumes were lowfor the first 2 d, increased markedly on days 3 and 4, andleveled offafter day 5. The actual volumes obtained weresimilar in Denver, Perth (1 1), and Edinburgh (12) wheredata were obtained by test weighing the infant. In thestudy by Macy et al (13, 14), all milk was obtained bymanual expression and milk volume production rosemore sharply starting on day 3 and reached values nearlydouble those of the present study by day 5. AlthoughMacy used wet nurses in earlier studies, these data wereobtained from Detroit multiparas ostensibly selected bymuch the same criteria as the subjects in our study. Thevolume differences suggest that the potential for milkproduction in early lactation is larger than infant needand can be fully realized by complete emptying of thebreasts on a consistent basis (3, 3 1). Several subjects inthis study did produce more milk than the infant con-sumed in early lactation (Fig 4), providing additional cvi-dence that infant demand rather than maternal capacitylimits milk intake after day 3, at least in some subjects.

Milk volumes obtained on days 1 and 2 were similarin all four studies and were low, the mean for day 2 being

175 mL/d. At this stage maternal steroid hormone 1ev-els probably inhibit milk production, the onset of copi-

ous milk secretion after day 2 resulting from falling ma-ternal progesterone and possibly estrogen levels (30).Healthy full-term infants obviously possess nutrientstores that allow them to cope with low milk suppliesduring this period; however, milk output from the breastmay not be adequate for premature or other infants withan atypical requirement for fluids or glucose during thefirst 2 d after birth (39).

Milk output duringfull lactation

We found that milk transfer increased rapidly to ‘�‘.-700g/d during the first month of full breast-feeding, levelingoff to a steady increase of < 1 g/d thereafter to give amean intake of -� 800 mL/d at 6 mo. Volumes in this

study were similar to those obtained elsewhere by test-weighing the infant (Fig 5). (A set of early values fromPicciano et al [40] are systematically lower than the morerecent values from this same group shown in Figure 5.The discrepancy may be because test weighing was per-formed in the early group of infants 3 d after routineDPT vaccination [MF Picciano, personal communica-lion].) Isotope dilution gives similar values. Coward etal (41) reported data obtained by this technique in The

Gambia and in Papua New Guinea; these fall neatly

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 10: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

1384 NEVILLE ET AL

within the ranges shown in Figure 3B, ie, 795 mL/d dur-ing months 1 to 4 in The Gambia and 670 mL/d inPapua New Guinea during the same postpartum inter-val. Mean values currently in the literature obtained bytest weighing the mother were not corrected for insensi-ble weight loss and were elevated (42): although the tech-nique can be used with accuracy (43), correction for in-sensible weight loss is essential (9).

The mean intraindividual CV in all the studies shownin Figure 5, weighted for subject number, was 1 7% com-pared with a value of 14% in our study. The consistencyamong the milk-volume data derived from 16 studiesfrom nine countries suggests that there is now sufficientdata in the literature on exclusively breast-feedingwomen during the first 8 mo postpartum so that refer-ence data can be established for the temporal depen-dence ofhuman-milk-volume transfer. Further, analysisofour longitudinal data suggests that the increase in milktransfer from months 2 to 5 (�- 1 g/d) reflects a generalpattern observed in most individuals. During the firstmonth of lactation, the individual temporal patternstend to deviate rather widely from the population patternand must be studied on an individual basis.

Sources ofvariability among individuals

We conclude from our longitudinal data that there isa characteristic milk-volume transfer for each mother-infant pair after the first month oflactation. The volumetransferred was not related to yields or intakes at the on-set oflactation, suggesting that maternal variables associ-ated with lactogenesis do not regulate milk productionduring full lactation. This conclusion is strengthened bythe observation that mean yield in mothers who overpro-duced during early lactation (eg, those who removed sig-nificant volumes by breast pump or manual expression)was not different from mean yield in nonoverproducersat any time during lactation (data not shown). Volumetransfer after the first month was significantly related toinfant weight (p < 0.02). Because the correlation be-tween milk transfer and infant weight was stronger at theend ofeach time interval examined that at the beginning(Fig 6B), we suggest that some factor or factors other thaninfant weight per se are the primaiy determinants of

transfer from mother to infAnt. Future studies might wellfocus on the relation between milk output in early lacta-tion and such factors as milk composition, the infant’sgrowth potential as expressed in parental and grandpa-rental size, the strength of the infant’s suck, and infantbasal metabolic rate.

Conclusion

The major objective ofthis study was to provide longi-tudinal data on milk-volume production and infant in-take in a small, relatively homogeneous group of exclu-sively breast-feeding women. Such data are necessary toprovide a temporal basis for the design of cross-sectionalstudies with a large enough number of subjects to allowreliable statistical analysis of maternal and infant vari-

ables influencing milk production. It is gratifying, there-fore, that our data differ in no significant respect frommilk-volume data obtained in larger, cross-sectionalstudies from a variety of populations throughout theworld. Mean milk transfer is low during the first 2 d post-partum, rises rapidly on days 3 and 4, and then increases

more slowly to reach maximum levels of ‘�-800 mL/d at6 mo postpartum in exclusively breast-feeding women.The pattern of milk transfer for a given mother-infantpair appears to be largely set by the end of the firstmonth. The interindividual CV is ‘� 16% in most studies;

this variation is related most closely to infant size andmay be more a function ofinfant demand than any ma-

ternal variable (2, 3, 32). Future studies focusing on in-fant factors that influence the demand for breast milkin the first month postpartum may be of considerableimportance in understanding the growth of the breast-fedinfant. 13

We acknowledge the helpful discussions of Mary Frances Picciano

and an unknown referee in the interpretation of these studies. Most

important, however, was the unstinting enthusiasm and cooperation of

the mothers who served as subjects through a tedious year oftest weigh-ing, milk samples, and diet records. Without the dedication of these

subjects to furthering our knowledge ofthe physiology and nutritional

significance oflactation, these studies could not have been done.

References

1. Morrison SD. Human Milk. Yield, proximate principles and inor-

ganic constituents. Farnham Royal, UK: Commonwealth Agricul-

turalBureaux, 1952.

2. Prentice AM, Paul A, Prentice A, Black A, Cole T, Whitehead RG.Cross-cultural differences in lactational performance. In: HamoshM, Goldman A eds. Human lactation 2: maternal and environ-mental factors. New York: Plenum Press, 1986:13-44.

3. Neville MC, Oliva-Rasbach J. Is maternal milk production limit-ing for infant growth during the first year of life in breast-fed in-

fants? In: Goldman A, Atkinson 5, eds. Human lactation 3: theeffect of human milk on the recipient infant. New York: PlenumPress, 1987:123-33.

4. Neville MC, Seacat J, Lutes V, Keller RP, Allen J, Casey C. Lacto-

genesis in women: milk volumes for the first two months postpar-tum. Fed Proc l985;43:668(abstr).

5. Neville MC, Allen JC, Watters C. The mechanisms of milk secre-tion. In: Neville MC, Neifert MR, eds. Lactation: physiology, flu-trition and breast-feeding. New York: Plenum Press, 1983:49-104.

6. Neville MC, Keller RP, Seacat J, Casey CE, Allen JC, Archer P.

Studies on human lactation. I Within-feed and between-breastvariation in selected components of human milk. Am J Clin Nutr

l984;40:635-46.

7. Anscombe FJ. Rejection of outliers. Technometrics 1960;2:123-147.

8. Woolridge MW, Butte N, Dewey KG, Ferris AM, Garza C, KellerRP. Methods for the measurement of milk volume intake of the

breast-fed infant. In: Jensen RG, Neville MC, eds. Human lacta-

tion: milk components and methodologies. New York: PlenumPress, 1985:5-20.

9. Neville MC. The measurement of milk transfer from mother to

breast-feeding infant. J Pediatr Gastroenterol Nutr l987;6:659-662 (editorial).

10. Hendrikson EC, Seacat JM, Neville MC. Insensible weight loss inchildren under one year ofage. Acta Paediatr 1985;74:678-80.

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 11: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

HUMAN MILK VOLUME 1385

1 1. Saint L, Smith M, Hartmann PE. The yield and nutrient content

ofcolostrum and milk ofwomen from giving birth to month post-

partum. BrJ Nutr 1984; 52:87-95.12. McClelland DBL, McGrath J, Samson RR. Antimicrobial factors

in human milk. Studies ofconcentration and transfer to the infant

during the early stages of lactation. Acta Paediatr Scand [Suppl]1978;271:3-20.

13. Kaucher M, Moyer EZ, Richards AJ, Williams HH, Wertz AL,Macy 1G. Human milk studies. XX The diet oflactating women

and the collection and proparation of food and human milk for

analysis. Am J DisChild l945;70:l42-7.

14. Roderuck C, Coryell MN, Williams HH, Macy 1G. Metabolism ofwomen during the reproductive cycle. IX The utilization of ribo-

flavin during lactation. J Nutr 1946; 32:267-83.15. Pao EM, Himes JM, Roche AF. Milk intakes and feeding patterns

ofbreast-fed infants. J Am Diet Assoc l980;77:540-5.

16. Hofvander Y, Hagman U, Hillervik C, Sjolin S. The amount ofmilk consumed by 1-3 months old breast- or bottle-fed infants.Acta Paediatr Scand l982;7 1:953-8.

17. Butte NF, Calloway DH. Evaluation oflactational performance of

Navajo women. Am J Clin Nutr 198 l;34:22l0-5.

18. Butte NF, Garza C. Energy and protein intakes of exclusively

breast-fed infants during the first four months oflife. In: Gracey M,Falkner F eds. Nutritional needs and assessment ofnormal growth.Nestle nutrition series. New York: Raven Press, 1985:63-83.

19. Van Steenbergen WM, Kusin JA, Van Rens MM. Lactation per-

formance of Akamba mothers, Kenya. Breast feeding behaviour,breast milk yield and composition. J Trop Pediatr 1981:27:155-61.

20. Dewey KG, Finley DA, Lonnerdal B. Breast milk volume and

composition during late lactation (7-20 months). J Ped Gastroent-

erol Nutr 1984; 3:713-20.

21. Dewey KG, Lonnerdal B. Milk and nutrient intake of breast-fedinfants from 1 to 6 months: relation to growth and fatness. J Ped

Gastroenterol Nutr l983;2:497-506.

22. Salmenpera L, Perheentupa J, Siimes M. Exclusively breast-fedhealthy infants grow slower than reference infants. Pediatr Res1985; 19:307-12.

23. Wallgren A. Breast-milk consumption ofhealthy full-term infants.

Acta Paediatr 1944; 32:778-90.

24. Janas LM, Picciano MF, Hatch TF. Indices ofprotein metabolismin term infants fed human milk, whey-predominant formula, or

cow’s milk formula. Pediatrics 1985;75:775-84.

25. Janas MJ, Picciano MF. Quantities ofamino acids ingested by hu-man milk-fed infants. J Pediatr 1986; 109:802-7.

26. Whitehead RG, Paul AA. Infant growth and human milk require-ments: a fresh approach. Lancet 1981;2: 16 1-3.

27. StuffJE, Garza C, Boutte C, et al. Sources ofvariance in milk and

caloric intakes in breast-fed infants: implications for lactation

study design and interpretation. Am J Clin Nutr 1986;43:361-6.

28. Chandra RK. Breast feeding, growth and morbidity. Nutr Res1981; 1:25-3 1.

29. Carvalho MD, Robertson 5, Friedman A, Klaus M. Effect of fre-quent breast-feeding on early milk production and infant weight

gain. Pediatrics l983;72:307-1 1.

30. Neville MC. Regulation ofmammary development and lactation.In: Neville MC, Neifert MA, eds. Lactation: physiology, nutrition

and breast-feeding. New York: Plenum Press, 1983:103-40.

31. Dewey KG, Lonnerdal B. Infant self-regulation ofbreast milk in-take. Acta Paediatr Scand 1986; 75:893-8.

32. Borschel MW, Kirksey A, Hanneman RE. Evaluation of test-weighing for the assessment of milk volume intake of formula-fed

infants and its application to breast-fed infants. Am J Clin Nutr1986;43:367-73.

33. Butte NF, Wong WW, Patterson BW, Garza C, Klein PD. Com-parison ofthe deuterium dilution and test-weighing techniques for

the determination of human milk intake. In: Goldman A, Atkin-son SA, eds. Human lactation III: the effects ofhuman milk upon

the recipient infant. New York: Plenum Press, 1987:374.

34. Vio FR, Infante CB, Lara WC, Mardones-Santander F, Rosso PR.Validation of the deuterium dilution technique for the measure-

ment offluid intake in infants. Human Nutr Clin Nutr l986;40C:

327-32.

35. Whitfield MF, Kay R, Stevens S. Validity of routine clinical testweighing as a measure ofthe intake ofbreast-fed infants. Arch Dis

Child 198l;56:9l9-2l.36. Brown KH, Black RE, Robertson AD, Akhtar NA, Ahmed G,

Becker S. Clinical and field trials ofhuman lactation: methodologi-cal considerations. Am J Clin Nutr 1982; 35:745-56.

37. Woolridge MW, Baum JD, Drewert RF. Effect ofa traditional andof a new nipple shield on sucking patterns and milk flow. Early

Hum Dev l980;4:357-64.38. Garza C, Butte NF, StuffJE, Motil K!, Montandon CM, Schanler

RJ. Estimating intakes of breastfed infants. In: The breastfed in-

fant: a model for performance. Columbus, OH: Ross Laboratories,1986:6- 12.

39. Casey CE, Neifert MR. Seacat JM, Neville MC. The intake of nu-

trients by breast-fed infants during the fIrst five days after birth.Am J Dis Child 1986: 140:933-6.

40. Picciano MF, Calkins EJ, Gamck JR. Deering RH. Milk and mm-eral intakes ofbreastfed infants. Acta PaediatrScand l981;70: 189-94.

41. Coward WA, Cole TJ, Sawyer MB, Prentice AM, Orr-Ewing AK.Breast-milk intake measurement in mixed-fed infants by adminis-

tration ofdeuterium oxide to their mothers. Hum Nutr Clin Nutr

l982;36C: 14 1-8.

42. Rattigan 5, Ghisalberti AV, Hartmann PE. Breast-milk production

in Australian women. BrJ Nutr 198 l;45:243-9.43. Arthur PG, Hartmann PE, Smith M. Measurement of the milk

intake of breast-fed infants. J Pediatr Gastroenterol Nutr l987;6:

243-9.

Appendix A

Evaluation ofswing balancefor older infants

We tested the swing-type balance and the technique of itera-

tive trimming (IT) on 20 infants weighing 8-18 kg who at-tended a pediatrics clinic and compared the results obtainedwhen the child was moving actively with the stable-weight re-sult from when the child was relatively motionless. No stableweight or true weight was obtained for comparison in three ofthe infants. For the remainder, we found that when the coeffi-cient of variation (CV) ofthe repetitive weights was < 1%; themean, median, the mean of The Middle 50% of values, Tri-mean, and IT all gave a value within 1 g of the true weight(Table 1A). We, therefore, used the eight cases in six differentsubjects where the CV was > 1% as a test of the method. The

data are shown in Table lA where we compare the mean of fiveor more true weights obtained when the balance returned areading that had been stable for 1 s to the mean, median, mode,and three other values statistically derived from 100 successiveweights. The stable weights for a single infant never differed by> 10 g and had a CV ofO.03% (< 3.5 g). When obtainable, the

mode agreed most closely with the true weight. However, nomodal value was found in four of the eight cases. The valuereturned by IT differed from the true value by � 2 g in seven

ofthe eight cases. In the eighth case, the infant was practicallyclimbing out of the infant swing using violent movements. IT

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from

Page 12: Studies In Human Lactation Milk Volumes In Lactating Women During The Onset Of Lactation And Full Lactation

1386 NEVILLE ET AL

returned a value differing from the true weight by only 17 g. Over- mean of 6.5 g from the true weight. Under conditions when theall, the use of the swing-balance apparatus and IT resulted in a infant waslessactive, the weighing procedure was terminated aftermeasurement on an active infant that differed no more than a 5 stable weights were returned and the error was < � g.

TABLE lA

Comparison ofiterative trimming with other statistical methods for estimating the weight ofa moving infant

Subject True weight

Difference fr om true weight

Mean Median Mode iMid 50% Trimean IT

g

2 8385 -18 -2 -2 -2 -19 -2

3 11051 17 6 - 7 16 4

13 8768 -10 36 0 -42 -22 2

14 8555 5 2 0 2 5 1

15 8785 125 5 -1 58 16 -1

l5At 8788 -5 0 - -10 1 2

19 14658 47 -6 - 0 48 -2

19A� 14658 253 54 - -104 233 17

SRMSD.� 101 23 - 37 82 6

Number ofweights differingfromtrueweightby>5g 6 4 - 5 6 1

a iMid 50% isthe mean ofthe middle 50% ofvalues. The Trimean weights the points > 1 SD from the mean by one half. IT is iterative trimming.

t Infant IS reweighed during a slightly quieter period.:1:Infant 19 reweighed while crying and thrashing about.§ Square root ofmean square difference.

Appendix B

Fifteen studies cited in Figure 5

Symbol Author LocationMonths

postpartumNumber of

subjects SD

C Pao et al United States 1 1 1 159U Hofvander et al Sweden 1-3 25 Range 4000 Butte and Calloway United Statesa 1 10 113. Butte et al United States 1-4 37-41 1 14-1310 Steenbergeretal Kenya 1-3 7-13 180-197A Dewey et al United States 7-8 8 99-1421� Deweyetai UnitedStates 1-6 11-19 122-1920 Salmenpera et al Finland 4-9 12-3 1 138-140

�32 Walgren Sweden 5-6 26-72 60-133a Janusetal UnitedStates 1-3 10 135

J ? Whitehead and Paul United Kingdom 1-6 4-23 87-187Q Stuffet al United States 5 9 850 Chandra Canada 3-8 24-33 71-126

V Carvaiho United States 1 28 237Prenticeetal TheGambia 1-3 6-8 113-183

S Presentstudy UnitedStates 0.5-8 5-13 98-158

a Navajo women.

by on August 22, 2009

ww

w.ajcn.org

Dow

nloaded from