A Comparison of the Safety of Cupfeedings and Bottlefeedings in
Premature Infants Whose Mothers Intend to Breastfeed

"The Foley Cup Feeder was not used in this study and therefore is not endorsed by it.
We are posting this study for reference material only."

Kathleen A. Marinelli, MD, IBCLC
Georgine S. Burke, PhD
Virginia L. Dodd, MS, RNC


To compare the safety of cupfeeding, an alternative feeding method, to bottlefeeding, the current standard of artificial feeding in the
United States, in preterm infants whose mothers intend to breastfeed.


In prospective, randomized crossover study, 56 infants &Mac178;34 weeks at birth, whose mothers indicated a desire to
breastfeed, were studied. Skin-to-skin care and attempts at breast were encouraged frequently when babies were physiologically
stable. Then infants were &Mac179;34 weeks' corrected gestatiotial age, the order of the first two non-breast oral feedings was
randomized by coin toss to one cupfeeding and one bottlefeeding. Trained Neonatal Intensive Care Unit nurses provided the
feeding. Heart rate, respiratory rate, and oxygen saturation were recorded at 1 minute intervals for 10 minutes before and during
the feeding. Volume taken, time required to complete the feed, and any apnea, bradycardia, choking, or spitting episodes were


Heart rate (p < 0.0001) and respiratory rate (p < 0.0001) increased and oxygen saturation decreased (p = 0.0002) during both cup
and bottefeedings compared to pre-feeding baselines. The amount of change in these three parameters from baseline to feeding
period was similar for both feeding methods. The fraction of O 2 saturation <90% during baseline compared to the study period
was different between these two feeding methods (p = 0.02). There was a 10-fold increase in desaturations < 90% during
bottlefeeds compared to no change during cupfeeds. When comparing cupfeeding periods to bottlefeeding periods, heart rates re
higher (p = 0.009) and oxygen saturations lower (p = 0.02) during bottlefeeds. There were no differences between methods in
respiratory rate, choking, spilling or apnea, and bradycardia. Volumes taken were lower (p =0.001) and duration of feeds longer (p
= 0.002) during cupfeedings.


During cuptfeedings, premature infants are more physiologically stable, with lower heart rates, higher oxygen saturations, and less
desaturations, than during bottlefeedings. However, cupfed infants took less volume, over more time, than bottlefed for these initial
feedings. Based on better physiologic stability and no difference in untoward effects, cupfeeding is at least as safe, if not safer,
than bottlefeeding in this population. This study supports the use of cupfeeding as a safe alternative feeding method for premature
infants learning to breastfeed.

Journal of Perinatology 2001; 21:350-355.


There is no question that human milk is uniquely superior to other forms of nutrition for infants. The most recent American
Academy of Pediatrics policy statement on breastfeeding summarizes the research supporting the benefits to the infant, the mother,
and society. Further, this statement specifically recommends human milk as the preferred feeding, not only for healthy term
infants, but for sick and premature infants as well.

The increasing research-based evidence of the positive effects, both short- and long-term, conferred on premature babies by
breastmilk has recently been reviewed. Some of these effects include easier digestion, with less vomiting, gastric aspirates, and
time to achieve full enteral feeds, decreased incidence and severity of infections, including sepsis and meningitis, and less frequent
and less severe necrotizing enterocolitis. Lucas et al. documented improved neurodevelopmental outcomes at 7.5 to 8 years of age
in former premature infants receiving breastmilk, which is supported by an Australian study ' showing a reduced prevalence of low
IQ in preterm breastfed babies. There is also emerging evidence that those preterm babies who receive human milk have decreased
incidence and severity of Retinopathy of Prematurity, and improved retinal function.

Based on this increasing body of data, breastmilk is arguably the standard of care in enteral nutrition for sick or premature infants.
There are, however, a number of challenges. One is providing this milk to these babies in a manner that supports breastfeeding
when their mothers are not present, or before they are strong or mature enough to complete a feed at breast. Commonly, mothers
of these infants initiate lactation with the use of a breast pump, and feedings of expressed human milk are begun by nasogastric or
orogastric gavage. In the United States, when oral feedings are introduced, breastfeeding is frequently supplemented by the
introduction of bottles. The use of artificial nipples has not been shown to foster breastfeeding success in a term population, and
there are data to show that it is detrimental in a preterm population as well.

The oro-motor skills used to suckle at breast and suck from a bottle differ. Babies who use bottlefeeding techniques at breast can
cause maternal sore nipples and decreased milk supply, and are often unsuccessful in transitioning to feeding at breast. This may
contribute to the relatively short breastfeeding duration another of the challenges in this high-risk population. A recent survey
shows the overall breastfeeding initiation rate in the United States to be 59.7%, with breastfeeding prevalence at 6 months
decreased to 21.6%. For babies < 2500 g, breastfeeding initiation is only 47.7%, with prevalence at 6 months falling to 12.6%.36
Alternative feeding methods to the bottle, which do not adversely affect breastfeeding duration in this population, are highly

Cupfeeding is one such method. It has been used successfully in special care nurseries and pediatric units in developing countries,
and in several European countries, to feed premature and lowbirth-weight infants until they are strong and mature enough to be
exclusively breastfed. These reports describe the technique and the transition from gavage to cupfeedings to breastfeeding. There
are no previous controlled studies reporting the safety of cupfeeding in a preterm population. Attempts to introduce cupfeeding in
Neonatal Intensive Care Units (NICUs) in the United States are often met with strong opposition, based on the lack of
research-based evidence for its safety and efficacy. This study was designed to compare the safety of cupfeeding to the current
standard of artificial feeding in this country (bottles with an artificial nipple) in preterm infants whose mothers were providing
expressed breastmilk.

Study Design

Infants less than or equal to 34 weeks' gestation at birth, admitted to the NICU of Connecticut Children's Medical Center and
whose mothers indicated a desire to breastfeed, were candidates for this study. Babies were excluded from participation if they had
concurrent diagnoses or conditions that would interfere with oral feedings. These included chromosomal defects, congenital
anomalies, or neurological conditions affecting feedings, and oxygen dependence or receiving feeds thickened with cereal at the
time of study entry. Infants could be enrolled any time after birth prior to receiving their first non-breastfeeding oral feed. The
study sample was estimated at 50 subjects. The power of this sample size is adequate to detect a mean difference from baseline to
feeding of 8% in oxygen saturation, with 80% power and 5% error. Assuming that baseline oxygen saturations will be in the
mid-90%, a decrease of 8% represents a value of 88% or less, which was considered to be clinically significant for the purposes of
this study.

The study was approved by the Institutional Review Board of Connecticut Children's Medical Center. Informed written consent
was obtained from the mothers prior to enrollment.

A randomized crossover design was used. Prior to entry into the study, infants were fed expressed breastmilk or preterm formula
by gavage. Mothers were encouraged to practice skinto-skin care and place infants at breast as frequently as mother was able and
baby tolerated. Randomization occurred when a study infant reached at least 34 weeks' corrected gestational age, and was deemed
by the medical team caring for that infant ready to begin oral feedings other than at breast. Indicators in this unit for oral readiness
include eagerness to suck, sustained alertness while receiving care, and maintenance of physiologic stability during attempted
nutritive breastfeeding. Thirty-four weeks was chosen for randomization, as this is the time in this unit that attempts with
bottlefeeds have traditionally been made if infants otherwise meet indicators for oral readiness. Although breastfeeding is
encouraged at a much younger gestational age based on physiologic stability data, the NICU Health Care Team did not feel it was
appropriate to attempt cupfeedings at an earlier age without previous evidence to support safety. The order of the first two oral,
non-breastfeeding feeds were determined by coin toss to one cupfeeding followed by one bottlefeeding or one bottlefeeding
followed by one cupfeeding. There was a minimum of one gavage feeding between these two oral feedings. Study feedings were
given at scheduled feeding times when the infants were awake and alert.

Nine NICU nurses were trained to give cupfeedings, and provided all the study feedings. Whenever possible, babies had each of
their two study feeds administered by the same nurse. Cupfeedings were given with a 30-ml medicine cup, using the procedure
previously published. 48 Infants were swaddled, with head and neck supported in a semi-upright position. The rim of the cup,
containing the prescribed amount of expressed breastmilk for the feed, up to 15 to 20 ml, was placed gently against the infant's
lower lip. The cup was tipped so that milk just touched the lower lip; it was not poured into the infant's mouth. The cup was
refilled as necessary. Bottlefeedings were given from a graduated feeder with a standard yellow nipple. Total volume offered at
each feed was that volume ordered by the physician caring for the infant. For both methods, the baby's cues were used to
determine when to pause or when to terminate the feeding session. The feeding was paused whenever the baby needed to be
burped. If an adverse physiological event occurred, such as sustained oxygen saturation below 88%, the feeding was paused until
recovery. If the adverse event continued, or the baby would not take further volume, the feeding was terminated.

Data Collection

All infants were monitored by Hewlett- Packard Merlin Series monitors for a baseline period of 10 minutes prior to the feed, and
during the feed. Heart rate (HR), respiratory rate (RR), and oxygen saturation (02 sat) were recorded continuously. Data were
stored each minute and recovered for analysis. Uncorrelated oxygen saturation data were eliminated from the analysis. The volume
taken at the feed, time required to complete the feed, and any apnea (respiratory pause &Mac179; 15 seconds), bradycardia (heart
rate 100 beats/minute), choking, or spitting episodes were recorded.

Data Analysis

Descriptive statistics were calculated for all measures. Changes in physiologic parameters from baseline to feed and between
feeding methods were evaluated with multivariate analysis of variance (MANOVA) with repeated measures. Effects of potential
confounders, such as nurse administering the feed, were tested. If no effect was found, the F and p values for intergroup
differences from the repeated measures MANOVA are reported. Differences in physiologic measures during a feed, as well as
characteristics of the feed (duration, amount), were evaluated with the paired t -test. The frequency of any adverse event (choking,
spitting, apnea, and bradycardia) between feeding methods was examined using McNemar procedure for paired categorical
observations. An alpha of 0.05 or less was considered evidence of statistical significance. All p values are two - tailed. The
statistical package used was JMP from the SAS Institute (Cary, NC).


In the 22 months from November 1996 through September 1998, 56 infants were enrolled in this study. Their characteristics are
presented in Table 1. Birthweight ranged from 550 to 2528 g. Gestational age at birth was 24 to 34 weeks, with corrected
gestational age at time of study at 33.7 to 36.4 weeks. Fifty-four percent were male. All infants had nasogastric tubes present at the
time of the study, and all but three infants had documented previous experience at breast.

In comparing each feeding method to its own baseline (Table 2), changes from baseline were significant for both cupfeeds and
bottlefeeds, with increases in heart rate (F= 20.3, p < 0.0001) and in respiratory rate (F= 29.1, p < 0.0001), and decreases in
oxygen saturation (F= 15.8, p = 0.0002). However, the feeding methods did not differ in the amount of change from baseline
(Table 2). We also evaluated the fraction of measured oxygen saturation less than 90% (number of 02 saturation measurements <
90% divided by the total number of 02 saturation measurements) for each feeding method compared to baseline. Overall, there was
an increase in the fraction of 02 saturations < 90% (F= 10.01, p = 0.003) over baseline during feeding periods. But different from
the results reported above of heart rate and respiratory rate, there is a statistically significant difference from baseline between the
two groups (F= 6.20, p = 0.02), with the fraction of 02 saturations <90% increasing 10-fold from baseline during bottlefeeds, but
remaining unchanged during cupfeeds.

Average heart rate during the feed (Table 2 ) was higher for bottlefeeds ( 171.8 ± 10.9) than for cupfeeds (168.4 ± 11. 1; p =
0.009), while average respiratory rate was no different. Mean oxygen saturation was lower for bottlefeeds (94.5 ± 5.3) than for
cupfeeds (96.5 ± 2.6; p = 0.02 ), with the lower end of the range as low as 74.4% during a bottlefeed compared to 90% during a
cupfeed. There was also a higher proportion of recorded oxygen saturations less than 90% during the bottlefeeds (0.13 ± 0.22)
than the cupfeeds (0.05 ± 0.09; p = 0.02).

The fraction of the feed taken (volume taken/volume prescribed in milliliters) was higher (Table 3) for bottlefeeds (bottle 0.76 ±
0.3, cup 0.61 ± 0.3;p = 0.003 ), while cupfeedings were of longer duration (16.3 ± 5.5 minutes bottle, 20.1 ± 5.9 minutes cup; p =
0.002). There were no significant differences in episodes of choking, spitting or apnea and bradycardia between the two methods,
although of note is the high percentage of feeds of both types associated with a choking episode (40% of bottlefeeds, 43% of

Nine trained study nurses gave the feeds. Four nurses administered 85% of the study feeds. Eighty percent of the babies had each
of their two study feeds done by the same nurse. The results were not affected by differences among nurses administering the feed.


This is the first prospective randomized study in premature babies of the safety of cupfeeding compared to bottlefeeding as an
alternative feeding method while learning to breastfeed. Over 50 years ago in the United States, Fredeen published a 10-year
descriptive experience with cupfeeding at the University of Kansas Medical Center, finding this method "particularly suited to the
premature and feeble infant and those with oral defects." Until recently, there has been no further published work in this area in the

Our data show that cupfeeding is at least as safe as bottlefeeding in this population. In defining safety, we looked at both measures
of physiologic stability and untoward effects during a feeding compared to the baseline pre-feeding period. There are similar
increases in heart rate and respiratory rate for both feeding methods when compared to baseline, with no differences in respiratory
rate between them. There were no differences in the incidence of choking, spitting or apnea and bradycardia during feeds between
the two methods. Further, it appears from our study that, in fact, during cupfeeding, premature infants may be more
physiologically stable than during bottlefeeding sessions. This is shown by both higher average heart rates and lower oxygen
saturation during bottlefeeds, whether looking at mean saturation during the feeding session, lower end of the recorded range
during the feed, or the fraction of saturations that fell below 90%.

We demonstrated a high incidence of choking associated with both feeding methods. Although we can see appropriate feeding
behaviors and coordinated suckswallow-breathe at 34 weeks' gestation and even younger, these data corroborate that these babies
are not developmentally mature feeders. Previous work showing that premature babies are more physiologically stable during
breastfeeding than during supports early breastfeeding attempts in this population. Our study demonstrates that cupfeeding is at
least as safe as bottles. But in our efforts to discontinue the gavage feedings as soon as possible, using an alternative oral feeding
method to enable discharge home from the hospital before exclusive breastfeeding is established, our data also show that we must
keep this developmental immaturity in mind.

These infants took less volume and required more time for a cupfeeding than for a bottlefeeding. At face value, this may appear to
make the use of cupfeeding less desirable in premature infants. The study by Howard et al. showed that in a full-term baby
population, physiologic parameters, administration times, and amounts ingested were not different during cupfeedings and
bottlefeedings. They concluded that in the term population, cupfeeding is a safe and efficient alternative to bottlefeeding. We have
shown the same conclusion with regard to safety. However, the differences in feeding duration and amount taken by this less
developmentally mature population could be viewed as reduced efficiency. An alternative interpretation might suggest that longer
duration and less volume are more developmentally appropriate for these premature infants.

While breastfeeding, the infant controls the feeding session, regulating when there is flow of milk. While bottlefeeding, milk flows
constantly out of the artificial nipple, forcing the baby to swallow. With manipulation of that nipple in the infant's mouth, the
feeding session is controlled by the feeder, not the infant. In effect, an infant can be force-fed. From our results, cupfeeding may
be more similar to breastfeeding in these respects. The infant is in control of his own intake, able to pace his feeding similar to
attempts at breast, provided the cupfeeding is being offered so that the milk just touches the infant's lips and is not poured into his
mouth .4, This difference in control of the feeding may explain why cupfeeding sessions took longer, with less volume taken. Both
breastfeeding and cupfeeding then are consistent with developmentally supportive care as described by Als and Gilkerson
(Newborn Individualized Care and Assessment Program, NIDCAP), which has been embraced by many neonatal units. Rather than
looking at the smaller volumes consumed and longer times for cupfeeding as a negative effect, when viewed in this light, they are
in fact indicative of a feeding method that permits the infant to determine his needs and competencies, allowing for infant initiation
and control.

These were also the first two oral, non-breastfeeding feeds these infants received. It is likely that with time and experience, these
feeds would become more proficient as indicated in previous descriptive reports. 37394'49 The lower heart rates and better oxygen
saturations compared to bottlefeeds suggest more physiologic stability similar again to that seen during breastfeeding. During
cupfeeding, extension of the tongue forward over the lower alveolar ridge to lap the milk into the mouth supports correct
breastfeeding tongue movement. For these reasons, cupfeeding is a safe - and may prove to be the preferred alternative feeding
method in the breastfed premature infant.

An important limitation of this study is that we were unable to quantify the number of times each study infant has had experience at
breast, to determine if that oral experience influenced how well a baby did with these study feeds. Our unit encourages early and
frequent skin-to-skin contact, with early experiences at breast, based on the known physiologic and psychological benefits to
premature infants and their mothers, and the observation that these practices facilitate transition to breastfeeding. These encounters
are not always clearly documented. We do know, however, that virtually all of these babies have had oral experience at breast prior
to study feeds. Another limitation of the study is that only the first two artificial feedings were included, an intentional approach
designed to limit learning bias. In addition, safety issues should have been more likely to manifest during these inexperienced
alternative oral feeds. This study does not address whether the use of cupfeedings is more efficacious in promoting eventual
success at breastfeeding in this population. That is an area that remains in need of investigation.


In a population of premature infants whose mothers intend to breastfeed, cupfeeding is at least as safe as bottlefeeding as an
alternative feeding method. With less increase in heart rate, better oxygenation, and the ability to pace their own feeding during
cupfeeds as compared to bottlefeeds, premature infants demonstrate better physiologic stability during cupfeedings, reminiscent of
the differences previously shown between breastfeeding and bottlefeeding. For these reasons, cupfeeding may prove to be the
preferred alternative feeding method for the breastfed premature infant.


We thank Susan Zaremba for help with data and chart review Linda Kaczmarczyk, research librarian, Mark Bamberger for his
invaluable assistance with data management, and Victor Herson for his thoughtful review of this manuscript. We are indebted to the
staff of the NICU for their patience and assistance in carrying out this study, and especially to the study nurses whose devotion to
supporting breastfeeding mothers enabled us to introduce cupfeeding to our NICU.

Journal of Perinatology 2001; 21:350-355

© 2001 Nature Publishing All rights reserved
Cup Feeding Study