Objective. Approximately 90% of infants who develop necrotizing enterocolitis (NEC) do so after being fed. Previous prospective studies have shown that infants given small enteral feedings for the first 7 to 10 days of feeding do not have an increased risk for NEC compared with those given no feedings. Although neonatologists now commonly increase feeding volumes, no study has compared the risk for NEC between infants fed these small volumes and those fed volumes that are increased slowly. The purpose of this study was to compare the risks and benefits of small and increasing feeding volume.
Methods. In a randomized, controlled trial, we randomly assigned 141 preterm infants in the newborn intensive care unit to be fed 10 days using 1 of 2 schedules. One group was fed 20 mL/kg/d for the first 10 study days (minimal). The other group (advancing) was fed 20 mL/kg/d on study day 1; feeding volume was increased by 20 mL/kg/d up to 140 mL/kg/d, which was maintained until study day 10. The main outcome measure was incidence of NEC; secondary outcomes were maturation of intestinal motor patterns, time to reach full enteral feedings, and incidence of late sepsis.
Results. The study was closed early because 7 infants who were assigned to advancing feeding volumes developed NEC, whereas only 1 infant fed minimal feeding volumes did, or 10% versus 1.4%. Although infants who were fed minimal volumes established full enteral feeding volumes later than infants who were fed advancing volumes, maturation of intestinal motor patterns and the incidence of late sepsis and feeding intolerance was similar in the 2 groups.
Conclusion. Given that advancing feeding volumes increase the risk of NEC without providing benefits for motor function or feeding tolerance, neonatologists should consider using minimal feeding volumes until future trials assess the safety of advancing feeding volumes.
Although the cause of necrotizing enterocolitis (NEC) is considered to be multifactorial, 90% of infants develop this disease after being fed, whereas only 10% develop NEC before being fed.1 Because of concerns for precipitating NEC and the widespread availability of parenteral nutrition during the 1970s and 1980s, preterm infants were commonly not fed.2 This practice was challenged when a series of prospective trials failed to show that the incidence of NEC was higher in infants given small enteral feedings (2–24 mL/kg/d) for the first 7 to 10 days compared with those who were unfed.3–7 Unfortunately, it seems that the neonatal community interpreted the findings from these studies to show that it is safe to increase feeding volumes after initiating early enteral feedings, and feeding volumes are commonly increased—although slowly—after these minimal feeding volumes are tolerated.8,9 The incidence of NEC was low in some of these previous studies, ranging from 0% to 5%,3,4,10 but none of these studies was designed to compare the incidence of NEC between treatment groups. A recent study compared 2 rates of advancement and found no difference in the incidence of NEC,11 but with either regimen the incidence of NEC was high. No study to date has answered the question of whether using minimal feeding volumes for a prolonged time, as was the case in previous studies,3–6,10 results in a lower incidence of NEC compared with using increasing feeding volumes from the start. Although it is not known, one could speculate that infants who are fed small volumes would not achieve full enteral feedings until some time after 10 days and, hence, would have a greater risk for late sepsis and cholestatic jaundice.12 For this reason, the study reported here examines not only the incidence of NEC but also the incidence of late sepsis (infection occurring after the initiation of feeding) and cholestatic jaundice (direct bilirubin >2 mg/dL) among study infants. Small enteral feedings accelerate maturation of gastrointestinal function3–6,10; however, it is not known whether larger feeding volumes accelerate maturation of gastrointestinal function even more. In an effort to assess both risk and benefits of using larger feeding volumes, we also assessed whether larger feeding volumes would accelerate maturation of gastrointestinal function even more than minimal feeding volumes do.
Infants in this study were admitted to the newborn intensive care unit (NICU) at Ben Taub General Hospital in Houston, TX, between January 1996 and January 2000. Infants who were considered for inclusion in this study were born at gestational ages <32 weeks with appropriate weights for gestational age. Gestational age was determined by ultrasound and the last menstrual period. When discrepancies were present using these 2 assessment tools, the neonatologist’s physical examination to determine gestational age was used. Infants were excluded when they had congenital anomalies or had birth weight not appropriate for gestational age. They were also excluded when they developed NEC or isolated ileal perforation before feedings were begun. The Institutional Review Board at Baylor College of Medicine approved this protocol for use, and parents provided informed consent for infants to participate in this study. Infants entered the study on the day that the attending neonatologist decided to initiate enteral feedings.
On the day that the attending neonatologist chose to begin feedings, study infants were randomly assigned to be fed using minimal or advancing volumes, as described below. By NICU routine, patients were not fed until umbilical vascular catheters had been withdrawn and infusions of pressor agents had been discontinued. Patients were randomized by requesting the ward secretary of the intermediate care nursery or a nurse in the NICU who was not involved in the patient’s care to draw 1 of 4 lots. Two of the lots were marked minimal, and 2 were marked advancing, and the same 4 lots were used throughout the study. Care providers were not blinded to an infant’s assignment.
On study day 10, feedings were withheld after midnight, and an assessment of gastrointestinal function was performed at 8 AM on study day 11. At completion of this assessment, feeding volumes for infants who were initially fed minimal volumes were increased by 20 mL/kg/d until a volume of 140 mL/kg/d was reached and maintained; infants who had received advancing volumes resumed their full enteral feeding volumes. Bedside records were reviewed daily for feeding characteristics and outcomes as described below.
Infants were initially fed unfortified expressed breast milk or preterm formula (Enfamil Premature Formula 24; Mead Johnson, Evansville IN). As shown in Fig 1, infants who were randomized to minimal volumes were fed 20 mL/kg/d for 10 days in 4-hour cycles consisting of a 2-hour infusion of milk or formula followed by 2 hours of fasting. These infants were given full parenteral nutritional support throughout these 10 days. Infants who were randomized to advancing volumes were fed 20 mL/kg/d using the same strategy as for infants fed minimal volumes. On study day 2, however, feeding volumes were increased to 40 mL/kg/d; on study day 3, volumes were increased to 60 mL/kg/d, and so forth, until a volume of 140 mL/kg/d was achieved. Parenteral nutrition was gradually tapered as enteral feeding volumes were increased. Feeding volume was then maintained at 140 mL/kg/d until study day 10. Two packs of breast milk fortifier (Enfamil Human Milk Fortifier; Mead Johnson) was added to 100 mL of expressed breast milk on the day that feeding volumes reached 120 mL/kg/d and 4 packs when feeding volumes reached 140 mL/kg/d or more. Decisions concerning feeding intolerance were made uniformly in both groups using a previously published algorithm.9,13 In short, this algorithm provides a series of uniform clinical actions based on the presence of and volume of milk remaining in the stomach 2 hours after the completion of feeding and physical findings, such as abdominal distention.
Feeding and Clinical Outcomes
Feeding characteristics, such as the presence of milk in the stomach 2 hours after completion of a feeding (ie, gastric residuals), postnatal ages when full enteral feedings were tolerated and when oral feedings were established, and the incidence of late sepsis and cholestatic jaundice were recorded daily. By NICU routine, plasma concentrations of bilirubin were determined every other week while infants were receiving parenteral nutrition, and these values were also recorded.
Antroduodenal manometry was performed using a neonatal manometry system, which has been previously validated for use in preterm infants.14 The manometric probe, which is made of polyvinyl extrusion tubing, has an outer diameter of 3.5 mm and contains 5 recording ports, located 2.5 cm apart, is perfused by a low-compliance, continuous-perfusion system that provides a response rate of 57 mmHg/sec at 10 pounds per square inch.14 The probe was positioned such that 3 or more ports were located in the duodenum and 1 or more in the antrum. Fasting motor contractions were recorded for 4 hours. Then the infant was given a feeding consisting of 5 mL/kg/15 min via the antral port, and motor contractions were recorded throughout this feeding and for an additional 2 hours after its completion.
Plasma Concentrations of Hormones
On the day of the manometric study, a single 2-mL blood sample was drawn via venipuncture while the infant was still fasting. The blood was placed on ice and centrifuged at 2000 rpm for 15 minutes. The resulting plasma was removed and stored at −70°C. Plasma concentrations of gastrin and motilin were determined in duplicate by radioimmunoassay.15,16 Concentrations were expressed in picograms per milliliter (pg/mL).
At the completion of the manometric recording, the infant was given 2 mL of carmine red via the mid-duodenal recording port. The manometric probe was removed and feedings were then resumed. The time for the passage of the first red stool after the administration of the dye was recorded in hours.
The primary outcome was the diagnosis of NEC. At the time that this study was designed, the incidence of NEC in this NICU was 12%. We determined that studies in 250 infants would permit a power of 0.8 to detect a decrease in the incidence of NEC to 4% at P < .05. This low incidence was chosen, as the incidence of NEC was low in studies that compared minimal enteral feedings to no feedings,3,4 and the incidence was higher in a study that used larger feeding volumes.8 The Data Safety Monitoring Board was concerned that infants given advancing feeding volumes would have a higher incidence of NEC than infants given minimal feeding volumes and wanted to close the study as soon as a statistically significant adverse outcome was detected. Thus, before the study began, the Data Safety Monitoring Board established preset times that data would be analyzed. They adjusted P values for multiple analyses in the protocol before the study opened and set P values at which the study would be closed, using a 1-tailed analysis. The Data Safety Monitoring Board planned that they would review data when 60, 100, 140, 180, and 220 patients had completed participation in the protocol, using predetermined P values that ranged from .01 for the first interim analysis, .02 for the second, .03 for the third, .04 for the fourth, and .05 for the fifth.
When the diagnosis of NEC was suspected, feedings were stopped and abdominal radiographs were performed. NEC was defined, using Bell’s criteria of stage 2 or greater,17 to be the presence of pneumatosis intestinalis or pneumoperitoneum, and the diagnosis was confirmed by unanimous decision by the attending neonatologist as well as a radiologist and a pediatric surgeon not involved in the care of the infant or the study. When an infant with pneumoperitoneum was found to have an isolated ileal perforation at surgery, that infant was excluded as a case of NEC. On 1 occasion when a unanimous decision was not present for an infant assigned to the advancing feeding group, the records were further reviewed by the director of the NICU, who agreed with 2 of the reviewers that the infant had isolated ileal perforation and not NEC.
To limit the number of manometric evaluations that were performed, manometric recordings were made in the first 110 infants who participated in the study. This number was chosen on the basis of the variance of the presence of the migrating motor complex (MMC) in preterm infants and the success of obtaining adequate recordings to assess the presence of the MMC.3,10 Similarly, venipunctures for determination of plasma concentrations of motilin and gastrin were performed in the first 65 infants enrolled into the study on the basis of the variance of plasma concentrations of gastrin in previous studies.3,10
Recordings of antral and duodenal motor patterns were analyzed by hand in 30-minute intervals by a laboratory technician who was not aware of the infant’s assignment (MZ, see acknowledgments), as previously described for preterm infants.3,12 We determined the presence of MMCs during fasting and the presence of a mature motor response.18 With the exception of the incidence of NEC described above, all other data were analyzed using Fisher exact test and χ2 analysis using 2-tailed analysis. The postnatal days to reach full enteral feedings, full oral feedings, and hospital discharge were not normally distributed, and these data were compared using the Mann-Whitney U test.
Of the 226 infants who were eligible for the study, parents of 148 consented to their infant’s enrollment in the study and 78 declined or died, as detailed in Fig 2. Of the 148 patients whose parents provided consent for participation, 4 failed to enter the study: 2 developed isolated ileal perforations, and 2 died before being fed. Of the 144 infants randomized, 3 more were removed from the study by physician request (n = 2) or by parental request (n = 1). These 3 patients had been assigned to the minimal feeding group and were fed using the advancing feeding schedule, and none developed NEC. Because all 3 were removed from the study on study day 1 and were fed using the advancing schedule, they were not considered to be protocol violations and instead were removed from the study. As a result, 141 patients were randomized and completed the study, and the analysis of their outcome is considered to be an intent-to-treat analysis. Demographic and clinical characteristics were similar between the 2 groups (Table 1).
The study was limited to 141 infants when the incidence of adverse outcome differed significantly between the 2 feeding interventions. Only 1 infant (1.4%) who was assigned to minimal volumes developed NEC, whereas 7 infants (10%) who were fed advancing volumes did (P < .03, Fisher exact test; Table 2). Characteristics of infants who developed NEC are shown in Table 2. Gestational ages of these infants ranged from 24 to 32 weeks, and birth weights ranged from 675 to 1540 g. Five of the 8 infants had received expressed breast milk. The time from the initiation to feedings to the time of diagnosis of NEC ranged from 7 to 29 days.
Infants who were fed advancing volumes reached full enteral feeding volumes significantly sooner than infants who were fed minimal volumes (P < .001; Table 3), and they required less time to reach full oral feeding and hospital discharge (P = .023 and P < .001, respectively; Table 2). Weights at discharge were similar (P = .83; Table 3). More central venous lines were placed in infants who were fed minimal volumes (P < .001; Table 3), and these infants required parenteral nutrition support longer (P < .0001; Table 3). However, they did not experience a higher incidence of late sepsis or cholestatic jaundice (P = .91 and .93, respectively; Table 3).
Antroduodenal motor patterns were recorded in the first 110 patients who entered the study. Sixty-one recordings were adequate for assessing the presence of the MMC, and 81 were adequate for assessing the motor response to feeding. Of 24 infants who were fed minimal volumes 14 (58%) had MMCs, whereas 18 (49%) of 37 infants who were fed advancing volumes did (P = .59). Of 34 infants who were fed minimal volumes, 27 79%) had mature motor responses to feeding, whereas 31 (66%) of 47 of infants who were fed advancing volumes did (P = .19). Thus, occurrence of mature gastrointestinal motor patterns was similar in the 2 groups of infants.
Blood was drawn for determination of plasma concentrations of gastrin and motilin in the first 65 infants; an inadequate volume was obtained in 1 infant. As a result, plasma concentrations of these hormones were determined in 64 infants; 30 of these infants had been assigned to advancing feeding volumes, and 34 had been assigned to minimal feeding volumes. Infants who were fed advancing volumes had higher plasma concentrations of gastrin and motilin than infants who were fed minimal volumes (226 ± 136 [mean ± standard deviation] vs 135 ± 66 pg/mL [P < .001] and 358 ± 170 vs 244 ± 177 pg/mL [P < .01], respectively). Duodeno-anal transit times were faster in infants who were fed advancing volumes compared with those who were fed minimal volumes (33 ± 41 vs 54 ± 36 hours; P < .03).
Although infants who were fed advancing volumes reached larger feeding volumes by study day 10 than infants who were assigned to minimal volumes (P < .001; Table 4), the number of hours that feedings were withheld (P > .99; Table 4) and the number of gastric residuals during the intervention (21 ± 9 vs 19 ± 9 residuals; P = .33) did not differ between the 2 groups. Infants who were fed advancing volumes, however, had more stools than infants who were fed minimal volumes (22 ± 10 vs 16 ± 9 stools during study days 1–10 and 27 ± 915 vs 20 ± 911 stools during study days 11–20; both P < .001). By study day 20, ie, 10 days after feeding volumes were increased in infants initially fed minimal volumes, the feeding volumes were similar in the 2 groups (P = .91; Table 4), as were the time feedings were withheld (P = .24; Table 4) and the number of gastric residuals (17 ± 8 vs 18 ± 9 residuals; P = .71). Infants in both groups required similar days to reach full enteral feeding volumes and oral feedings from the time that feeding volumes were increased (P = .89 and .08, respectively; Table 4). The incidence of feeding tolerance, as defined by reaching full feeding volumes within 10 days of initiating increasing feeding volumes, was also similar in the 2 groups (P = .68; Table 4).
These data show there is a higher risk for NEC among preterm infants who were fed advancing feeding volumes compared with those whose feeding volumes remained low during the first 10 days of feeding. This risk for NEC was so high that this study was closed early. Although infants whose feeding volumes were advanced achieved full enteral feeding volumes sooner and were less likely to have central venous lines placed,7,11,19–21 the risk-to-benefit ratio of increasing feeding volumes during the first 10 feeding days may not warrant this practice. These data provide an important gap in the information concerning feeding practices. Previous studies have shown that the use of small feedings seems to be as safe as using no feedings,3–6 and Rayyis et al11 showed that the incidence of NEC is similarly high among infants who have their feedings increased by differing increments. Nevertheless, no study to date had compared the safety of small feedings to increasingly larger feeding volumes, and one could speculate that infants in the study by Rayyis et al were given larger feeding volumes (although at 2 different rates). Thus, this study attempted to assess both the risk (incidence of NEC) and the benefit (maturation of gastrointestinal function) of using increasing feeding volumes. Obviously, larger feeding volumes increased the risk for NEC in the current study, but that increase in risk was not offset by an increase in benefit.
Small enteral feedings trigger maturation of gastrointestinal function in neonatal animals and preterm infants.3,6,10,22–24 Larger feedings in the current study triggered greater release of gastrin and motilin, but maturation of motor patterns and feeding outcomes were similar in the 2 feeding volume groups, as measured by the number of feeding days to reach full enteral feeding volume after feeding volumes were increased and the incidence of feeding tolerance, as defined by reaching full enteral feeding volumes within 10 days of attempts to advance volumes. Thus, neither maturation of small intestinal motor patterns nor feeding performance was further enhanced by using larger feeding volumes. Gastrin has trophic effects and plays an important role in the regulation of antral acid release and nutrient processing, but those functions were not assessed in the current study.
As expected, the prolonged use of small enteral feeding volumes resulted in a greater need for central venous line placement and prolonged the use of parenteral nutrition. Despite these outcomes, the incidence of late sepsis and cholestatic jaundice was not higher in the minimal volume group. The incidence of death was similar in both study groups, as was the composite incidence of NEC and death. We conclude from these findings that the use of advancing feeding volumes has an impact on NEC but not on death.
Although the prolonged use of small enteral feedings reduced the risk for NEC, it did so by delaying the establishment of full enteral feeding volumes by approximately 10 days and prolonging hospital stay. Using this strategy to prevent NEC requires that 10.6 infants unnecessarily be delayed in establishing full enteral feeding volumes for every case of NEC that is prevented. Preterm infants who develop NEC incur additional hospital charges of $74 000 to $186 000 above those incurred by gestational age-matched preterm infants who do not develop NEC.25 We estimate that using a feeding strategy that delays the time to establish full enteral feedings may still affect a cost savings for patients as well as NICUs. However, this study was not designed to assess this cost, as itemized charges for this NICU are not available. It is apparent that such an analysis warrants careful consideration, as the adoption of such a feeding regimen has the potential to increase costs for neonatal care.
This study did not attempt to identify the mechanism whereby the prolonged use of small feedings reduces the incidence of NEC. Enteral feeding poses an immunologic challenge to the immature gut, as many aspects of neonatal intestinal host defense are absent or diminished in capacity. We speculate that prolonged use of small feedings may stimulate—and possibly upregulate-1 or more aspects of host defense, whereas advancing volumes overwhelm these mechanisms. This speculation is consistent with the observation that NEC most commonly occurs when feeding volumes exceed 100 mL/kg/d.11,26 Feeding also triggers an obligatory postprandial hyperemic response in the splanchnic bed.27,28 An alternative explanation for our findings is that larger volumes may trigger local tissue needs for blood flow that exceed the neonatal vascular bed’s ability to meet this demand. This relative local tissue hypoxia, in turn, could trigger the release of local inflammatory mediators.
The findings reported here are limited to a single NICU, where feeding practices may vary from other NICUs. For this reason, there is a need to confirm these findings in a larger multicenter trial. Until then, however, it may be prudent for neonatologists to limit feeding volumes during the first 10 feeding days until the safety of advancing feeding volumes is assessed in a larger prospective trial.
We acknowledge the contributions of the following individuals: Drs Baruch Brody and O’Brian Smith, who composed the Data Safety Monitoring Board; Drs Thomas Jaksic, David Wesson, and Joseph Garcia-Prats, who reviewed radiographs and patient records; Dr George Klee at Mayo Clinic and Foundation, who performed radioimmunoassays; the faculty of the Newborn Section at Baylor College of Medicine and the neonatal nursing staff at Ben Taub General Hospital, who provided patient care; Elizabeth Baker-Wills, David Reed, Michelle Malanga, Sophia Lopez, and Mark Zeller, who provided technical assistance; and Drs Leonard Weisman and William C. Heird for review of the manuscript.
- Received July 25, 2002.
- Accepted September 10, 2002.
- Reprint requests to (C.L.B.) Mead Johnson Nutritionals, 2400 West Lloyd Expwy, Evansville, IN 47721-0001. E-mail:
Dr Berseth had no affiliation or financial agreement with any company when this study was performed and completed; she has been employed by Mead Johnson since September 5, 2002.
- ↵Tyson JE, Kennedy KA. Minimal enteral nutrition for promoting feeding tolerance and preventing morbidity in parenterally fed infants. Cochrane Database Syst Rev.2000;2 :CD000504
- ↵Schanler RJ. The low-birth-weight infant. In: Walker WA, Watkins JB, eds. Nutrition in Pediatrics: Basic Science and Clinical Application. 2nd ed. Hamilton, Ontario, Canada: BC Decker; 1996:392–412
- ↵Berseth CL, Nordyke C. Enteral nutrients promote postnatal maturation of intestinal motor activity in preterm infants. Am J Physiol.1992;103 :1523– 1528
- ↵Schanler RJ, Shulman RJ, Lau C, et al. Feeding strategies for premature infants: randomized trial of gastrointestinal priming and tube feeding method. Pediatrics.1999;103 :434– 439
- ↵Amarnath RP, Berseth CL, Malagelada J-R, Perrault J, Abell TL, Hoffman AD. Postnatal maturation of small intestinal motility in preterm and term infants. J Gastrointest Motil.1989;1 :138– 143
- ↵Dryburgh JR, Brown JC. Radioimmunoassay for motilin. Gastroenterology.1980;9 :657– 678
- ↵Berseth CL. Neonatal small intestinal motility motor responses to feeding in term and preterm infants. J Pediatr.1990;117:5 :777– 782
- ↵Caple JI, Armentrout DC, Huseby VD, et al. The effect of feeding volume on the clinical outcome in premature infants. Pediatr Res.1997;41 :231A
- Shulman DI, Konterek K. Gastrin, motilin, insulin, and insulin-like growth factor-1 concentrations in very low birth weight infants receiving enteral or parenteral nutrition. JPEN J Parenter Enteral Nutr.1993;17 :130– 135
- ↵Owens L, Burrin D, Klee G, Berseth CL. Minimal enteral feeding induces maturation of intestinal motor function but not mucosal growth in neonatal dogs. J Nutr.2002;132 :2717– 2722
- ↵Bisquera J, Cooper T, Berseth C. Impact of necrotizing enterocolitis on length of stay and hospital charges. Pediatrics.2002;109 :423– 428
- ↵Owens L, Berseth C. Is there a volume threshold for enteral feeding and necrotizing enterocolitis? Pediatr Res.1995;37 :315A
- ↵Yanowitz TD, Yao AC, Pettigrew KD, Werner JC, Oh W, Stonestreet BS. Postnatal hemodynamic changes in very-low-birthweight infants. J Physiol.1999;87 :370– 380
- Copyright © 2003 by the American Academy of Pediatrics