PEDIATRICS Vol. 105 No. 2 February 2000, pp. 379-384

The Incidence of Necrotizing Enterocolitis After Introducing Standardized Feeding Schedules for Infants Between 1250 and 2500 Grams and Less Than 35 Weeks of Gestation

Michael D. Kamitsuka, MD^*, Mary K. Horton, MEd^*, and Michelle A. Williams, ScD

From the Divisions of ^* Neonatal Medicine and  Perinatal Medicine, Swedish Medical Center, Seattle, Washington.

	ABSTRACT

Top Abstract Methods Results Discussion Conclusion References

Objective.  To evaluate the incidence of necrotizing enterocolitis (NEC) after implementing standardized feeding schedules.

Method.  This was a cohort study, which retrospectively reviewed the incidence of NEC for a 3-year period before implementing feeding schedules and prospectively evaluated the incidence of NEC for a 3-year period after implementing feeding schedules in infants with birth weights between 1250 and 2500 g and <35 weeks' gestation. The feeding schedules were comprised of 3 parts. First, no group was fed within the first 24 hours of life. Feeds were started between 24 to 72 hours of life based on birth weight. Second, the initial feed was full-strength breast milk or half-strength formula. Half-strength formula was changed to full-strength formula on the fourth day for all groups. Third, the daily feeding volume increase was no greater than 20 mL/kg for all groups.

Results.  In the 477 infants before the feeding schedules, there were 23 (4.8%) cases of definite NEC defined as pneumatosis on abdominal film or NEC confirmed at the time of surgery, and in the 466 infants after the feeding schedules, there were 5 (1.1%) cases of NEC. Before the feeding schedules, those who developed NEC started feeds sooner 1.5 ± 1.5 (± standard deviation [SD]) days versus 3.0 ± .7 (±SD) days, reached full feeds sooner 4.0 ± 1.8 (± SD) days versus 9.8 ± .5 (±SD) days and were more likely to have been started on formula than those who developed NEC after implementing the feeding schedules. The mean time for NEC to occur after the feeding schedules increased from 5.9 ± 4.1 (±SD) days to 19.4 ± 16.3 (± SD) days, although not statistically significant. The number of mothers who received prenatal steroids increased after the feeding schedules. The number of infants with NEC, however, significantly decreased whether their mothers were pretreated with steroids. The risk of NEC was reduced 84% after the introduction of feeding schedules as determined by multiple logistic regression analysis and adjusting for confounding by birth weight, white race, prenatal steroid exposure, day of life of first feed, day of life to reach full feeds, and breast milk.

Conclusion.  The incidence of NEC was significantly decreased after the implementation of standardized feeding schedules, which was independent of birth weight, prenatal steroid exposure, breast milk, day of life of first feed, and the number of days to reach full feeds.  Key words:  necrotizing enterocolitis, feeding schedules, prenatal steroids, breast milk.

Although a description of necrotizing enterocolitis (NEC) can be found as early as the 19th century¹ and the condition was named over 40 years ago by Schmid and Quaiser,² the understanding of the exact mechanism of injury of NEC remains obscure. The most accepted epidemiologic precursors for NEC are prematurity and gastrointestinal feeding.³ It is suspected that the resultant intestinal mucosal injury may be initiated by a number of different factors, including intestinal ischemia, infectious agents, and feeding practices.⁴ Various aspects of feeding practices have been implicated as possible causative factors, such as rate of daily enteral volume increase,^5,6 timing of first feed,⁶ breast milk versus formula,^7,8 and the osmolality of the first feed.⁹

In our Special Care Nursery, feeding practices have been uniform for infants <1250 g, because a continuous drip feeding schedule has been in place for infants <1250 g. Before January 1, 1995, however, feeding practices were varied for infants 1250 g. Feeds were full-strength formula or breast milk, were randomly started, and the volumes were indiscriminately increased. To reduce variability in our feeding practice, standardized feeding schedules were established for infants with a birth weight between 1250 to 2500 g and <35 weeks' gestation. Brown and Sweet¹⁰ and Spritzer et al¹¹ had reported that the incidence of NEC virtually vanished from their respective units once conservative feeding schedules were implemented. Therefore, the purpose of this study was to evaluate the incidence of NEC after implementing standardized feeding schedules in our unit.

	METHODS

Top Abstract Methods Results Discussion Conclusion References

Study Design

This was a cohort study from January 1, 1992 to December 31, 1997, which retrospectively reviewed the incidence of NEC before feeding schedules were implemented and prospectively evaluated the incidence of NEC after feeding schedules were implemented in a single unit. Infants were eligible for the study if they were admitted to the Special Care Nursery at Swedish Medical Center with birth weights between 1250 and 2500 g and gestational ages (GAs) <35 weeks. Three separate feeding schedules were devised for infants 1250 to 1500 g (group A), 1501 to 2000 g (group B), and 2001 to 2500 g (group C). These schedules were approved by the forms committee and were kept in the infant's chart after discharge.

Feeding Schedules

In developing the feeding schedules, 3 factors were taken into consideration: 1) when to begin feeds; 2) how rapidly to advance feeds; and 3) the concentration of the feeds. Each group was made non per os: group A for at least 72 hours, group B for at least 48 hours, and group C for at least 24 hours. At the physician's discretion, initiation of feeds may have been delayed further, based on the degree of asphyxia or clinical instability. No infant, however, was fed sooner than the designated times. The daily volume increase was no greater than 20 mL/kg. Once feeds were initiated, full volume (150 mL/kg/day) was reached in 10 days in group A, 9 days in group B, and 8 days in group C. First feed was either full-strength breast milk or formula diluted to half-strength using bottled water. The initial volume was 3 mL of breast milk or half-strength formula every 3 hours for infants in group A and B and 4 mL every 3 hours in group C. Half-strength formula was changed to full-strength formula on the fourth day of feeds for all groups. Similac Special Care 24 cal/oz or Similac 20 cal/oz (Ross Laboratories, Columbus, OH) formula was used from January 1, 1992 to December 31, 1994. For contractual reasons, the hospital changed to Enfamil Premature 24 cal/oz and Enfamil 20 cal/oz (Mead Johnson, Evansville, IN) on January 1, 1995.

Data Analyses

Mean differences in continuous variables were assessed using Student's t test statistics. Comparisons of categorical variables were made between the 2 study groups using ² or Fisher's exact tests. Fisher's exact tests were used if frequencies for categorical variables were below 5. ² tests were used for all other categorical variables.

The odds ratio (OR) was used as a measure of the association between the introduction of feeding schedules and NEC. Logistic regression procedures were used to calculate the maximum likelihood estimates for the coefficients. Their standard errors were used to calculate ORs and 95% confidence intervals (CI), adjusted for confounding factors. Confounding was assessed by entering variables into a logistic regression model 1 at a time and comparing the adjusted ORs.¹² Factors assessed as potential confounding factors included birth weight, white race, prenatal steroid exposure, the day of life feeds started, the number of days to reach full-volume feeds, consumption of only breast milk, consumption of only formula, and consumption of both breast milk and formula. The final logistic regression model included only covariates that altered the unadjusted OR by at least 10%.¹² All reported P values are 2-tailed.

	RESULTS

Top Abstract Methods Results Discussion Conclusion References

From January 1, 1992 to December 31, 1994, 503 infants were admitted to the Special Care Nursery between 1250 and 2500 g and <35 weeks of gestation. Twenty-six infants were excluded from the study because they were transferred to another hospital or they expired within the first 10 days of life of causes other than NEC. A total of 477 infants remained for analysis. From January 1, 1995 to December 31, 1997, 492 infants qualified as study candidates. Twenty-five infants were excluded from the study. A total of 467 infants were analyzed for this period. Both groups were similar except for birth weight (Table 1). The birth weight differed between groups because a greater proportion of infants heavier than 2001 g was in the study group before the feeding schedules (39.4% vs 30.4%).

Twenty-three (4.8%) cases of definite NEC were identified before implementing the feeding schedules and 5 (1.1%) cases were identified after implementing the feeding schedules for a 77.1% reduction in the incidence of NEC (P = .0006). Definite NEC was defined as pneumatosis on abdominal film or NEC confirmed at the time of surgery. In both groups, infants were not critically ill, based on the number requiring ventilatory support for >48 hours before developing NEC and were not depressed based on a 5-minute Apgar score <7. Factors that potentially could be predisposing for NEC were not statistically different between groups (Table 2). The distribution of NEC differed by GA between study periods. Before the feeding schedules, NEC occurred in 16 infants 32 weeks and in 3 infants <30 weeks. After the feeding schedules, NEC occurred in only 1 infant 32 weeks and in no infants <30 weeks. No more than 2 cases of NEC occurred during any month for the total study period.

When analyzing the total study population for the average age of starting feeds and the number of days to reach full feeds, those infants with definite NEC, intestinal atresia, or gastroschisis were excluded. The average age of starting feeds was 2.5 ± 2.3 (± standard deviation [SD]) days in infants (n = 452) before and 2.8 ± 2.4 (±SD) days in infants (n = 457) after the feeding schedules (P = .0744). The average number of days to reach full feeds was 7.2 ± 5.8 (±SD) days before and 9.2 ± 4.3 (±SD) days after the feeding schedules (P = .0001).

For those who developed NEC, the average age of starting feeds was 1.5 ± 1.5 (±SD) days before the feeding schedules and 3.0 ± 0.7 (±SD) days after the standardized feeding schedules (P = .005). After the feeding schedules, the mean time for NEC to occur after the first feed increased from 5.9 ± 4.1 (±SD) days to 19.4 ± 16.3 (±SD) days (P = .139), and the number of infants who failed to reach full feeds decreased from 9 (39%) to 1 (20%). The average number of days to reach full feeds was 4.0 ± 1.8 (±SD) days before the feeding schedules and 9.8 ± .5 (±SD) days after the feeding schedules (P < .001).

The distribution of formula as the first feed significantly differed between study periods. For the total study population, 73.3% had formula as the first feed before the feeding schedules, compared with 52.1% after the feeding schedules (P < .001). For those who developed NEC, 83% had formula as the first feed before the feeding schedules, compared with 20% after the feeding schedules (P = .007).

Also, the distribution of formula and breast milk intake significantly differed between study periods. For the total study population, 22.9% were fed only formula before the feeding schedules, compared with 14.8% after the feeding schedules (P < .001) and 20.5% were fed only breast milk before the feeding schedules, compared with 34.8% after the feeding schedules (P = .002). For those who developed NEC, 56.5% were fed only formula before the feeding schedules, compared with none after the feeding schedule and 8.7% were fed only breast milk before the feeding schedules, compared with 80% after the feeding schedules.

In those who developed NEC, oral medication was given to 2 (8.7%) infants before the feeding schedules and none after the feeding schedules. Before the feeding schedules, 6 (26.1%) infants required surgery and 1 (4.3%) infant died. Three mothers had received prenatal steroids. After the feeding schedules, 2 (40%) infants required surgery and no infant died. Both mothers had received prenatal steroids.

The effect of prenatal steroid exposure on the incidence of NEC was evaluated for both study groups (Table 3). After the feeding schedules, an increased number of mothers received prenatal steroids (P < .0001). The number of infants with NEC, however, significantly decreased regardless of whether their mothers were pretreated with steroids.

Multivariate analysis was performed using a multiple logistic regression model. The ORs were adjusted for confounding by birth weight, white race, prenatal steroid exposure, the day of life feeds started, the number of days to reach full-volume feeds, and consumption of only breast milk (Table 4). The risk of NEC significantly decreased by 84% after the feeding schedules, which was independent of the other confounding variables (adjusted OR: .16; 95% CI: .06-.47; P = .0008). Also, the risk of NEC significantly decreased by 7% each day it took the infant to reach full-volume feeds and by 60% in infants consuming only breast milk, although this finding was not significant. The risk of NEC, however, increased with white race (P = .011) and prenatal steroid exposure, although not quite reaching statistical significance (P = .0621) and was not significantly affected by birth weight or the day of life of first feed.

	DISCUSSION

Top Abstract Methods Results Discussion Conclusion References

The risk of NEC was reduced by 84% after the implementation of standardized feeding schedules. Those infants who developed NEC after the feeding schedules were implemented, started feeds significantly later (1.5 to 3.0 days), advanced to full feeds more slowly (4.0 to 9.8 days), and were less likely to have been started on formula. After implementing the feeding schedules, there were no deaths as a result of NEC and the mean time from first feed to diagnosis of NEC tripled (5.9 to 19.4 days).

In 1978, Brown and Sweet¹⁰ reported that a slowly progressive feeding regimen decreased their incidence of NEC from .8% to .4%, compared with an historic control. In 1988, Spritzer et al¹¹ reported that after they introduced a cautious feeding regimen, their incidence of NEC decreased from 9.6% to .5%. Their study, however, involved using a formula not available in the United States and they made no mention of using breast milk. We report a decreased incidence of NEC from 4.8% to 1.1% after implementing feeding schedules. Our feeding schedule differed from the previously reported feeding schedules in that the formulas we used were different and we used breast milk. We also evaluated the effect prenatal steroids had on the incidence of NEC, which was not done in either study.

Three factors were taken into consideration when we developed the feeding schedules: how rapidly to advance feeds, the concentration of feeds using breast milk and currently available premature and term formulas, and when to begin feeds. The daily increase in feeds was no greater than 20 mL/kg based on the conclusion by Anderson and Kleigman⁵ that rapid advancement of feeding volume predisposed premature infants to NEC. Recently, Rayyis et al¹³ prospectively evaluated daily feed advancement rates of 15 mL/kg and 35 mL/kg. They found that the incidence of NEC was not increased at the faster rate. Several differences between our study and that of Rayyis et al merit discussion. First, in their study, the majority of infants weighed <1250 g. Second, the type of feed was different. In their study, all infants were fed full-strength formula and they excluded all infants fed breast milk. Third, in our unit, we have a continuous drip feeding schedule for infants <1250 g. From 1992-1997, the incidence of NEC in infants <1250 g was 2.6% (10/371). Therefore, for infants <1250, our incidence of NEC is ~4 times lower than the 10% (19/188) reported by Rayyis et al. Fourth, the birth weight distribution was different between the 2 study populations with NEC. Hence, these differences in study design and population hinder direct comparisons of study results.

The concentration of the first feed was influenced by the work done by Crissinger et al¹⁴ in their piglet model of NEC. In a 1-day-old term piglet, the bowel was made ischemic by occluding the mesenteric artery for 1 hour, then reperfusing for 1 hour. Once premature infant formula was predigested and bile acid-solubilized, the formula was instilled into the small bowel lumen. A condition similar to NEC resulted. The investigators had previously demonstrated that the lipid component of the premature formula was responsible for causing mucosal injury.¹⁵ When lipids were removed from a premature infant formula and fed to 1-day-old piglets under similar ischemic conditions, no gross mucosal injury was noted. Term infant formula caused moderate villus injury, but less injury than premature infant formula.¹⁴ We are aware that the piglet model of NEC may not be applicable to humans because mucosal injury after premature formula feeds in human premature infants has not been studied. Nonetheless, if breast milk was not available, we reduced the lipid concentration in the formula by starting our infants on half-strength formula. Before the feeding schedules, the incidence of NEC was no different in infants being fed premature or term formula; therefore, both premature and term formulas were used. Ostertag et al¹⁶ concluded that the incidence of NEC was not increased when dilute feeds were advanced to full-strength feeds over a 7-day period; therefore, we arbitrarily selected the fourth day to change the concentration of the formula to full-strength. Another benefit of diluted formula may be decreased feed intolerance. Currao et al¹⁷ has shown that infants starting with half-strength formula reached full-volume feeds sooner than those starting with full-strength formula.

Two different formulas, Similac and Enfamil, were used during the study period. The substrate composition is very similar in both formulas. It is unlikely that changing from Similac to Enfamil accounted for the large observed differences in the incidence of NEC.

We found the risk for NEC was reduced by 60% in infants fed breast milk only. NEC has been reported to occur less frequently in infants fed breast milk than in those fed formula.⁷ Crissinger reported that sow milk caused less injury to piglet intestinal mucosa than premature or term formula.¹⁸ Possible protective factors in breast milk against NEC include leukocytes¹⁹ and immunoglobulins.²⁰

When to start feeds has been a controversial topic. The incidence of NEC may be decreased when feeds are delayed,⁶ but La Gamma et al²¹ reported that the incidence of NEC was not reduced when feeds were delayed for 2 weeks. Berseth²² noted that premature infants reached full feeds sooner and had fewer days of feeding intolerance when hyocaloric feeds were started within the first 3 to 5 days instead of at 10 to 14 days. In our physician group, some believed that the more mature infants should be fed sooner than the less mature infants. Indeed, Ostertag et al¹⁶ reported that starting dilute formula early, on the first day of life, did not increase the incidence of NEC, compared with a group fed late, on the seventh day of life. They did not study, however, the incidence of NEC in infants whose breast milk feeds were started early versus late. Therefore, infants in our study were arbitrarily stratified into 3 groups based on size. Feeds were started at 72 hours in the smaller, less mature infants and at 24 hours in the larger, more mature infants.

After the National Institutes of Health consensus statement on prenatal steroid use was published in February 1995,²³ prenatal steroid use significantly increased in our institution. Results from studies that examine the effect of prenatal steroid exposure on the incidence of NEC are varied. After prenatal steroid exposure, Halac et al²⁴ and Bauer et al²⁵ have reported a decreased incidence of NEC, although Uauy et al²⁶ was not able to demonstrate a decreased incidence. From our multivariate logistic regression analysis, prenatal steroid exposure was associated with a doubling in the risk of NEC.

Over a period of time, unrecognized changes in our clinical practice may have affected the incidence of NEC apart from the feeding schedules. Aside from the use of prenatal steroids, the number of infants being fed only formula, and the formula type, no other changes in our clinical practice could be identified. Multiple logistic regression analysis indicated that the most likely confounding variables were not strongly associated factors for NEC.

	CONCLUSION

Top Abstract Methods Results Discussion Conclusion References

In summary, the incidence of NEC was significantly decreased after the implementation of standardized feeding schedules. The reduction in NEC was independent of birth weight, prenatal steroid exposure, breast milk, day of life of first feed, and number of days to reach full feeds. Prospective studies will need to be performed to confirm our observations.

	ACKNOWLEDGMENTS

We would like to thank Dr Ivan D. Frantz III, Chief of Newborn Medicine, Tufts University School of Medicine, Boston, MA; Dr Karen D. Crissinger, Department of Pediatrics/Division of Gastroenterology, Vanderbilt University School of Medicine, Nashville, TN; and Dr Paula K. Raschko, Division of Neonatology, Swedish Medical Center, Seattle, WA, for their thoughtful review of this manuscript.

	FOOTNOTES

Received for publication Nov 20, 1998; accepted May 10, 1999.

This work was presented in part at the 21st Annual Conference of the American Academy of Pediatrics, District VIII, Section on Perinatal Medicine; May 24, 1996; Blaine, WA.

Reprint requests to (M.D.K.) Division of Neonatology, Swedish Medical Center, 747 Broadway, Seattle, WA 98122. E-mail: mkamitsuka{at}aol.com

	ABBREVIATIONS

NEC, necrotizing enterocolitis; GA, gestational age; OR, odds ratio; CI, confidence interval; SD, standard deviation.

	REFERENCES

Top Abstract Methods Results Discussion Conclusion References

1. Simpson JY Notices of cases of peritonitis in the foetus in utero. Edinburgh Med Surg J 1836; 15:390-416
2. Henleyn WL. History. In: Brown EG, Sweet AY, eds. Neonatal Necrotizing Enterocoilits. New York, NY: Grune & Stratton; 1980:1
3. Kleigman RM, Walker WA, Yolken RH Necrotizing enterocolitis: research agenda for a disease of unknown etiology and pathogenesis. Pediatr Res 1993; 34:701-708 [Medline]
4. Santulli TV, Schullinger JN, Heird WC, Acute necrotizing enterocolitis in infancy: a review of 64 cases. Pediatrics 1975; 55:376-387 [Abstract/Free Full Text]
5. Anderson DM, Kleigman RM The relationship of neonatal alimentation practices to the occurrence of endemic necrotizing enterocolitis. Am J Perinatol 1991; 8:62-67 [Medline]
6. McKeown RE, Marsch TD, Amarnath U, Role of delayed feeding and of feeding increments in necrotizing enterocolitis. J Pediatr 1992; 121:764-770 [CrossRef][Medline]
7. Lucas A, Cole TJ Breast milk and necrotizing enterocolitis. Lancet 1990; 336:1519-1523 [CrossRef][Medline]
8. Kleigman RM, Pittard WB, Fanaroff AA Necrotizing enterocolitis in neonates fed human milk. J Pediatr 1979; 95:450-453 [CrossRef][Medline]
9. Book LS, Herbert JJ, Atherton SO, Jung AL Necrotizing enterocolitis in low-birth weight infants fed an elemental formula. J Pediatr 1975; 87:602-605 [CrossRef][Medline]
10. Brown EG, Sweet AY Preventing necrotizing enterocolitis in neonates. JAMA 1978; 240:2452-2454 [Abstract]
11. Spritzer R, Koolen AM, Fetter WP, A prolonged decline in the incidence of necrotizing enterocolitis after the introduction of a cautious feeding regimen. Acta Paediatr Scand 1988; 77:909-911 [Medline]
12. Greenland S Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989; 79:340-349 [Abstract/Free Full Text]
13. Rayyis S, Ambalavanan N, Wright L, Carlo WA Randomized trial of "slow" versus "fast" feed advancements on the incidence of necrotizing enterocolitis in very low birth weight infants. J Pediatr 1999; 134:293-297 [CrossRef][Medline]
14. Crissinger KD, Burney DL, Velasquez OR, Gonzalez E An animal model of necrotizing enterocolitis induced by infant formula and ischemia in developing piglets. Gastroenterology 1994; 106:1215-1222 [Medline]
15. Crissinger KD, Tso P The role of lipids in ischemia/reperfusion-induced changes in mucosal permeability in developing piglets. Gastroenterology 1992; 102:1693-1699 [Medline]
16. Ostertag SG, La Gamma EF, Reisen CE, Ferrentino FL Early enteral feeding does not affect the incidence of necrotizing enterocolitis. Pediatrics 1986; 77:275-280 [Abstract/Free Full Text]
17. Currao, WJ, Cox C, Shapiro MD Diluted formula for beginning the feeding of premature infants. Am J Dis Child 1988; 142:730-731 [Abstract]
18. Crissinger KD, Burney DL Intestinal oxygenation and mucosal permeability with luminal mother's milk in developing piglets. Pediatr Res 1996; 40:269-275 [Medline]
19. Kliegman RM, Walsh MC Neonatal necrotizing enterocolitis: Pathogenesis, classification and spectrum of illness. Curr Probl Pediatr 1987; 17:213-288 [Medline]
20. Wolf HM, Eibl MM The relevance of immunoglobulin in the prevention of necrotizing enterocolitis. Adv Exp Med Biol 1991; 310:381-389 [Medline]
21. La Gamma EF, Ostertag SG, Birenbaum H Failure of delayed oral feedings to prevent necrotizing enterocolitis. Am J Dis Child 1985; 139:385-389 [Abstract]
22. Berseth CL Effect of early feeding on maturation of the preterm infant's small intestine. J Pediatr 1992; 120:947-953 [CrossRef][Medline]
23. NIH Consensus Development Panel Effect of corticosteroids for fetal maturation on prenatal outcomes. JAMA 1995; 273:413-417 [Abstract]
24. Halac E, Halac J, Begue EF, Prenatal and postnatal cortisone therapy to prevent neonatal necrotizing enterocolitis: a controlled trial. J Pediatr 1990; 117:132-138 [CrossRef][Medline]
25. Bauer CR, Morrison JC, Poole WK, A decreased incidence of necrotizing enterocolitis after prenatal glucocorticoid therapy. Pediatrics 1984; 73:682-688 [Abstract/Free Full Text]
26. Uauy RD, Fanaroff AA, Korones SB, Necrotizing enterocolitis in very low birth weight infants: Biodemographic and clinical correlates. J Pediatr 1991; 119:630-638 [CrossRef][Medline]