- Email: [email protected]
Randomized trial of “slow” versus “fast” feed advancements on the incidence of necrotizing enterocolitis in very low birth weight infants
R
Randomized trial of “slow” versus “fast” feed
advancements on the incidence of necrotizing enterocolitis in very low birth weight infants Suha F. Rayyis, MD, Namasivayam Ambalavanan, MD, Linda Wright, MD, and Waldemar A. Carlo, MD
Objective: To determine whether the rate of feed advancement affects the incidence of necrotizing enterocolitis (NEC). Study design: Prospective randomized controlled trial involving 185 formula-fed infants with birth weight 501 to 1500 g and gestational age ≤34 weeks. Infants were randomized into 2 groups: “slow” (n = 98), who received 15 cc/kg/d increments (a 10-day schedule to full feeds) and “fast” (n = 87), who received 35 cc/kg/d increments (a 5-day schedule to full feeds) of Similac Special Care 20 cal/oz. Feeds were increased only if well tolerated as defined by a protocol. Results: The incidence of NEC (Bell stage ≥II) was similar in both groups (slow 13% and fast 9%, P = .5). The incidence of perforation (Bell stage III) was also similar in both groups (slow 4% and fast 2%, P = .8). Feeds were started at a comparable postnatal age in both groups (median age: slow 5 days and fast 4 days, P = .9). Although the neonates in the fast group attained full enteral intake earlier (median days [25th and 75th percentiles]: slow 15 [12, 21] and fast 11 [8, 15], P < .001) and regained their birth weight earlier (slow 15 [11, 20] and fast 12 [8, 15], P < .05), their ages at discharge were not statistically different (slow 47 [31, 67] and fast 43 [29, 62], P = .3) Conclusions: A greater than twofold difference in the rate of feed advancement from 15 cc/kg/d to 35 cc/kg/d did not affect the incidence of NEC ≥ stage II. Factors other than feed advancement appear to be more important in the pathogenesis or progression of NEC. (J Pediatr 1999;134:293-7)
Necrotizing enterocolitis is a major cause of mortality and morbidity in preterm infants. The incidence of NEC in the National Institute of Child
Health and Development Neonatal Research Network has been 8% to 10% of very low birth weight infants, with center differences in incidence
From the Department of Pediatrics, University of Alabama at Birmingham, Birmingham, Alabama, and National Institute of Child Health and Human Development, Bethesda, Maryland.
Presented in part at the Pediatric Academic Societies Meeting, Washington, DC, May 2-6, 1997. Submitted for publication Aug 5, 1998; revision received Nov 4, 1998; accepted Dec 2, 1998. Reprint requests: Waldemar A. Carlo, MD, University of Alabama at Birmingham, 525 New Hillman Bldg, 619 South 19th St, Birmingham, AL 35233-7335. Copyright © 1999 by Mosby, Inc. 0022-3476/99/$8.00 + 0 9/21/96310
from 3.9% to 22.4%.1,2 The cause of NEC is inadequately understood, although many factors such as enteral feeds, hypoxia, ischemia, patent ductus arteriosus, and infection have been associated with an increased incidence of NEC.3-5 Several retrospective studies suggest that rapid advancement in feeds may be associated with an increased incidence of NEC.6-8 A case control study by the National Institute of Child Health and Development Neonatal Research Network of 249 NEC Necrotizing enterocolitis PDA Patent ductus arteriosus VLBW Very low birth weight
neonates with NEC revealed that the infants who had NEC received significantly more enteral nutrition in the 3 days before the diagnosis of NEC compared with a control group.9 In that study the incidence of NEC in individual centers ranged from 2% to 19%, and 2 centers accounted for 63% of the cases.9 During that time period the 16% incidence of NEC in our center was the second highest of the 8 clinical centers. At that time VLBW neonates frequently achieved full enteral intake in <5 days after initiation of feeds. The purpose of this study was to determine whether the rate of advancement of feeds influenced the incidence of NEC. We tested the hypotheses that in VLBW neonates: (1) slow feed advancement (15 cc/kg/d) decreases the overall incidence of NEC (Bell stage ≥ II) compared with fast feed advance293
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS MARCH 1999
Table I. Incidence of NEC (≥stage II) by weight subgroups
Birth weight (g) stratum 501-750 751-1000 1001-1250 1251-1500 Total
Slow NEC/Total (%)
Fast NEC/Total (%)
P value
2/13 (15) 7/30 (23) 3/22 (13) 1/33 (3) 13/98 (13)
0/9 (0) 3/24 (13) 4/25 (16) 1/29 (3) 8/87 (9)
NS NS NS NS NS
ment (35 cc/kg/d) and (2) slow feed advancement decreases the incidence of NEC with intestinal perforation (Bell stage III) compared with fast feed advancement.
METHODS Subjects From January 24, 1994, until December 30, 1996, 187 infants with birth weight 501 to 1500 g and gestational age ≤34 weeks admitted to the regional neonatal intensive care unit at the University of Alabama at Birmingham were prospectively enrolled in the trial. The exclusion criteria included (1) Apgar score <3 at 5 minutes, (2) hemodynamic instability requiring pressors to maintain blood pressure at the time of initiation of feeds, (3) presence of cyanotic congenital heart disease or other major malformation, (4) polycythemia (venous hematocrit >70%), (5) partial or double volume exchange transfusion before initiation of feeds, (6) human milk feeds (partial or total), and (7) multiple gestation of triplets or more. Small for gestational age infants, large for gestational age infants, infants undergoing mechanical ventilation, and infants who received indomethacin were not excluded. Trophic (minimal enteral) feeds were not used. Eligibility was assessed after the decision to initiate feeding was made by the clinical team. Informed parental consent was obtained. Randomization was performed according to weight stratification into 4 groups: 501 to 750 g, 751 to 1000 g, 1001 to 1250 g, and 1251 to 1500 g. Shuffled blocks of 2, 4, and 6 294
sealed envelopes for each weight group were used. The person obtaining parental consent was masked to the odds of a subsequent infant receiving either treatment. Twins were randomized to the same treatment mode according to randomization of the first eligible of the pair.
Protocol The protocol was approved by the institutional review board before the trial was begun. The patients were randomized into 2 groups: (1) “slow” feeding group: feeds were started at 20 cc/kg/d and advanced by increments of 15 cc/kg/d to a maximum of 160 cc/kg/d (a 10-day schedule to full feeds), (2) “fast” feeding group: feeds started at 35 cc/kg/d and advanced by increments of 35 cc/kg/d to a maximum of 160 cc/kg/d (a 5-day schedule to full feeds). Both groups started feeds with Similac Special Care 20 cal/oz formula and were switched to Similac Special Care 24 cal/oz formula when full feeds were reached. Oral medications were not given until full enteral feeds were attained. The patients were fed every 3 hours with an orogastric tube. Aspirates (residuals) from the orogastric tube were checked before every feed. If aspirates were ≤20% of total feed and the abdominal examination was normal, feeds were advanced according to the protocol. If aspirates were between 20% and 30% and the abdominal examination was normal, feeds were continued, but no increase in feeds was done on that day. Feeds were temporarily discontinued for 24 hours and
then continued at half the previous volume for aspirates between 30% and 40%. If the aspirates were ≥40% or the abdominal examination was abnormal regardless of the aspirates, a complete blood count and an abdominal radiograph were obtained. If both investigations were normal, feeds were temporarily discontinued for 24 hours and then resumed at half the previous volume and advanced at the previous day’s rate. An abnormal abdominal examination was defined as 1 or more of the following: (1) abdominal distention, (2) erythema of the abdominal wall, (3) abdominal tenderness, (4) decreased bowel sounds, (5) increased abdominal girth, and (6) gross or occult blood in the stools. NEC was defined according to the Bell stages.10 Bell stage II (proven NEC) includes radiographic findings of pneumatosis intestinalis or portal vein gas with clinical signs including poor feeding, increased residuals, or abdominal distention with bloody stools. Bell stage III includes radiographic observation of pneumoperitoneum with the previously described clinical signs associated with deterioration of vital signs and septic shock. Presence of a PDA, when suspected on clinical evaluation, was confirmed by echocardiography. Indomethacin was administered at standard doses (0.2 mg/kg per dose, 3 doses at 12hour intervals). There was no change in the feeding schedule in the presence of a PDA. The end point of the study was when the patient was discharged or had NEC (stage ≥ II) before discharge home. The results were analyzed by intention to treat.
Statistics The expected incidence of NEC (≥ stage II) in the fast feeding group was 16% based on the incidence at our center, and the anticipated incidence of NEC in the slow feeding group was estimated to be 5% based on the incidence at the centers with slower feed-
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS VOLUME 134, NUMBER 3 ing protocols.9 The sample size was estimated at a total of 185 with the Fisher-Irwin tables for sample size estimation with power of 80%, α of 0.05, and a small excess to compensate for small variations in the actual incidences from the expected incidences. Chi-squared analysis, t test, or Wilcoxon rank sum test was used for statistical analysis of the data, depending on whether the data were normally distributed. A P value < .05 was considered statistically significant.
RESULTS During the enrollment period 605 neonates with birth weight 501 to 1500 g were admitted to the neonatal intensive care unit. A total of 295 patients were excluded because of 1 or more of the following reasons: human milk feeds (n = 153), major malformations (n = 71), died before initiation of feeds (n = 54), fed before consent was obtained (n = 32), low Apgar scores (n = 26), congenital heart disease (n = 22), transferred to another hospital before initiation of feeds (n = 4), triplets or quintuplets (3 sets), unable to get consent because of maternal illness (n = 4), and transpyloric feeds (n = 3). Parents of 310 neonates were approached, and randomization was performed after informed consent was obtained on 185 infants (rate of 60%). There were 7 protocol violations including patients fed human milk at parent’s request (n = 3), patient transferred to another hospital before receiving full feeds (n = 2), and infants with multiple episodes of feeding intolerance taken off the study at the request of the attending physician (n = 2). Data on these infants were included in the analysis, because analysis was done on the basis of intention to treat. The data from 185 infants (slow 98, fast 87) were analyzed. Birth weight, gestational age, sex, and race were comparable in the 2 groups (Table I). Feedings were started at a comparable
Table II. Postnatal age in relation to nutrition and discharge
Age at starting feeds (d) Age at full feeds (d) Age when birth weight regained (d) Age at discharge (d)
Slow
Fast
5 (3-7) 15 (12-21) 15 (11-20) 47 (31-67)
4 (3-7) 11 (8-15)* 12 (8-15)* 43 (29-62)
Data are shown as median days (25th to 75th percentiles). *P < .05.
postnatal age in both groups (slow 5 days, fast 4 days, P = .9). No difference was seen in the incidence of NEC (Bell stage ≥ II) between both groups (slow 13%, fast 9%, P = .50) or for each of the weight subgroups (Table I). It should be noted that although subgroup analysis by weight is shown, the small sample size of each stratum makes type II error more likely. The incidence of intestinal perforation (Bell stage III) was also comparable in both groups (slow 4%, fast 2%, P = .8). No significant difference was seen in mortality resulting from NEC (in entire feeding group: slow 2%, fast 3%, P = .9 and in those who had NEC stage ≥ II: slow 2 of 13, fast 3 of 8, P = .32 by Fisher’s Exact test). The babies in the fast group had NEC at the age of 25 ± 14 days (mean ± SD), whereas the neonates in the slow group had NEC at 18 ± 9 days (P = .3). The average volume of feeds in the 3 days before NEC in the fast group was 122 ± 44 cc/kg/d (mean ± SD) and 96 ± 40 cc/kg/d in the slow group (P = .07). In the 3 days before the onset of NEC, no difference was noted in feeding intolerance (<15% of neonates in either group with more than 20% residuals). The incidence of NEC in the neonates enrolled in this trial (combined) was not significantly different from the incidence seen historically in the same unit (fast + slow 11% vs historical 16%, P = .4). Although not part of the trial, the incidence of NEC in the neonates excluded because of human milk feeding was 7% (fast + slow 11% vs human milk 7%, P = .4).
The age at which the diagnosis of NEC was made was similar in the 2 groups. Three infants in each group had NEC before reaching full feeds. No contribution by the epidemic form of NEC was noted, because there were never more than 2 cases of NEC (Bell stage ≥ II) per month (in both groups combined) during the study. Data on postnatal age were not normally distributed and are reported as median with the 25th and 75th percentile (Table II). As expected, a significant difference was seen between the 2 groups in postnatal age at full feeds (slow 15 days [12 to 21], fast 11 days [8 to 15]; P < .001) and the postnatal age to regain birth weight (slow 15 days [10 to 20], fast 12 days [8 to 15]; P < .05). When the number of days to achieve full feeds from initiation of feeds were compared, 46 (47%) achieved full feeds in 10 days in the “slow” group, and 49 (56%) achieved full feeds in 5 days in the fast group. However, the age at discharge was not statistically different (slow 47 days [31 to 67], fast 43 days [29 to 62], P = .3). The 2 groups were comparable in variables that may influence the incidence of NEC such as antenatal steroid exposure, postnatal indomethacin use (either as prophylaxis for intraventricular hemorrhage or as management of PDA), the presence of symptomatic PDA (all given indomethacin), and the placement of umbilical arterial catheter. The demographics of the VLBW neonates who were enrolled for this study were comparable to those of the total VLBW population admitted to 295
RAYYIS ET AL
the nursery at that time period (mean birth weight 1045 g, 28 weeks’ gestation, 49% male). The incidence of NEC in the VLBW neonates not enrolled in the study (9.5%) was also similar to the incidence of NEC in the study (11%). The median length of hospital stay was also similar in the babies who did not participate in the study (49 days) compared to those who participated in the study (45 days).
DISCUSSION The incidence of NEC (Bell stage ≥II) and NEC with intestinal perforation (Bell stage III) did not change with more than a twofold difference in the rate of feed advancement. Also, the timing of the development of NEC did not differ between both groups. Although the birth weight was regained earlier in the fast group, this did not translate into an earlier discharge. The fast group reached full feeds earlier, which indicates that this rate of feed advancement is equally tolerated compared to the slow feeding regimen. Although the study was a randomized controlled trial, it had some important limitations. It could not have been done in a masked manner. To decrease the influence of potential bias, feed intolerance was defined by strict criteria, and a radiologist independent of the study made the radiologic diagnosis of NEC. A different number of patients were noted to have been randomized to each group (slow 98, fast 87). Other than the effect of chance in the technique of block randomization, 1 factor that had an effect is the number of twins randomized to each group. Five sets of twins were randomized to the slow group, and 4 were randomized to the fast group. Another limitation is that although the sample size was calculated on a projected incidence of NEC at 16% in 1 group and 5% in the other group based on past studies, the actual incidences were 296
THE JOURNAL OF PEDIATRICS MARCH 1999 lower (slow 13%, fast 9%). However, the post-hoc power of the study indicating the groups were similar was 87%, and although a type II error is still possible, it is less likely (a sample size of 1850 is required to show that the difference in the incidence of NEC noted between the groups is statistically significant at a .05 and 80% power). The effect of feeding schedules on the incidence of NEC has been widely debated. In support of our observations 2 prospective randomized trials demonstrated no increase in the incidence of NEC when feed increments were increased. Book et al11 compared 2 feed increments, 10 cc/kg/d and 20 cc/kg/d. No significant difference was seen in the incidence of NEC between the 2 groups. However, the sample size was small (total n = 29), and a type II error was possible. A recent prospective randomized study by Caple et al12 showed no difference in the incidence of NEC between feed advancements of 20 cc/kg/d and 30 cc/kg/d. These results are consistent with our study, because we did not detect any increase in the incidence of NEC even with a wider range of feeding advancements of 15 cc/kg/d and 35 cc/kg/d. On the other hand, several retrospective case control studies suggest that aggressive increments in feeds may increase the incidence of NEC. Anderson and Kliegman7 showed that infants who had NEC compared with a matched control group had significantly larger feeding increments. Similar results were obtained in other retrospective case control studies by Covert et al6 and McKeown et al.8 When compared with a matched control group, the infants with NEC were fed earlier and received larger feeding volumes.8 These retrospective case control studies reveal results that may not be comparable with ours. There may be 2 reasons for this: (1) in some cases feed increments were greater than 35 cc/kg/d, and (2) the neonates with the NEC group were found to have started feeds earlier than the control group.
In our study both groups started feeds at a similar postnatal age. Four chart review studies revealed an association between the incidence of NEC and rapid feeding practices. Krouskop13 reported that more than 50% of infants who had NEC had received aggressive feeding increments. Book et al11 noted dramatic increases in the incidence of NEC when feeding protocols became faster. In a similar manner, Spritzer et al14 found dramatic decreases in the incidence of NEC in their neonatal care unit after introducing a slower feeding regimen. Finally, Goldman,15 in another chart review, found that patients who had NEC had larger feeding increments before the episode of NEC occurred. These studies have results that contradict our prospective randomized study. Chart review studies have the limitations common to other forms of retrospective studies. No control of other variables may influence the incidence of NEC. On the other hand, known and unknown confounders are likely to be equally distributed between groups in a randomized trial such as this study. This study has shown that feed increments of 35 cc/kg/d compared with 15 cc/kg/d were equally well tolerated and did not increase the incidence of NEC in VLBW neonates matched for weight and gestational age. Therefore factors other than the rate of feed advancement may be more important in the pathogenesis or progression of NEC. However, it does not follow that increments larger than 35 cc/kg/d are equally safe in the VLBW population. It is possible that larger increments, beyond a currently undefined value, may in fact alter the risk of NEC. Marked intercenter differences in the incidence of NEC2,15 indicate that further investigation of the determinants of NEC should include analysis of variations in clinical practice and sociodemographic variables in addition to the variables of prematurity, infection, hypoxia, ischemia, and feeding practices that are commonly implicated.
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS VOLUME 134, NUMBER 3
REFERENCES 1. Uauy RD, Fanaroff AA, Korones SB, Philips EA, Philips JB, Wright LL. Necrotizing enterocolitis in very low birth weight infants: biodemographic and clinical correlates. J Pediatr 1991; 119:630-8. 2. Hack M, Wright LL, Shankaran S, Tyson JE, Horbar JD, Bauer CR, et al. Very-low-birth-weight outcomes of the National Institute of Child Health and Human Development Neonatal Network, November 1989 to October 1990 Am J Obstet Gynecol 1995;172:457-64. 3. Stoll BJ. Epidemiology of necrotizing enterocolitis. Clin Perinatol 1994;21: 205-18. 4. La Gamma EF, Browne LE. Feeding practices for infants weighing less than 1500 g at birth and the pathogenesis of necrotizing enterocolitis. Clin Perinatol 1994;21:271-306. 5. Nowicki PT; Nankervis CA. The role
6.
7.
8.
9.
10.
of the circulation in the pathogenesis of necrotizing enterocolitis. Clin Perinatol 1994;21:219-34. Covert RF, Neu J, Elliot M, Rea JL, Gimolty PA. Factors associated with age of onset of necrotizing enterocolitis. Am J Perinatol 1989;614:455-60. Anderson D, Kliegman R. The relationship of neonatal alimentation practices to the occurrence of endemic NEC. Am J Perinatol 1991;8:62-7. McKeown RE, Marsh TD, Amaranth W, Garrison, CZ, Addy CL, Thompson SJ, et al. Role of delayed feeding and of feeding increments in necrotizing enterocolitis. J Pediatr 1992;121: 764-70. Wright LL, Uauy RD, Younes N, Fanaroff AA, Korones SB, Philips JB. Rapid advances in feeding increase the risk of necrotizing enterocolitis in very low birthweight in infants [abstract]. Pediatr Res 1993;33:313A. Bell MJ, Trenberg JL, Feigin RD, Keating JP, Marshall R, Barton L, et
11.
12.
13.
14.
15.
al. Neonatal necrotizing enterocolitis: therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1-7. Book LS, Herbst JJ, Jung AL. Comparison of fast- and slow-feeding rate schedules in the development of necrotizing enterocolitis. J Pediatr 1976;89: 463-6. Caple JI, Armentrout DC, Huseby VD, Halbrdier BM, Garcia J, Sparks JW. The effect of feeding volume on clinical outcome in premature infants [abstract]. Pediatr Res 1997:41:229A. Krouskop R. The relationship of feeding to NEC [abstract]. Pediatr Res 1974;8:109. Spritzer R, Koolen AMP, Bauts W, Fetter WPF, Lafeber HN, Sauer PJJ. A prolonged decline in the incidence of necrotizing enterocolitis after the introduction of a cautious feeding regimen. Acta Paediatr Scand 1988;77: 905-91. Goldman HI. Feeding and NEC. Am J Dis Child 1980;134:553-5.
AVAILABILITY OF JOURNAL BACK ISSUES As a service to our subscribers, copies of back issues of The Journal of Pediatrics for the preceding 5 years are maintained and are available for purchase from Mosby until inventory is depleted, at a cost of $15.00 per issue. The following quantity discounts are available: 25% off on quantities of 12 to 23, and one third off on quantities of 24 or more. Please write to Mosby, Inc., Subscription Services, 11830 Westline Industrial Drive, St. Louis, MO 63146-3318, or call (800)453-4351 or (314)453-4351 for information on availability of particular issues. If unavailable from the publisher, photocopies of complete issues may be purchased from UMI, 300 N. Zeeb Rd., Ann Arbor, MI 48106, (313)761-4700.
297
Randomized trial of “slow” versus “fast” feed
advancements on the incidence of necrotizing enterocolitis in very low birth weight infants Suha F. Rayyis, MD, Namasivayam Ambalavanan, MD, Linda Wright, MD, and Waldemar A. Carlo, MD
Objective: To determine whether the rate of feed advancement affects the incidence of necrotizing enterocolitis (NEC). Study design: Prospective randomized controlled trial involving 185 formula-fed infants with birth weight 501 to 1500 g and gestational age ≤34 weeks. Infants were randomized into 2 groups: “slow” (n = 98), who received 15 cc/kg/d increments (a 10-day schedule to full feeds) and “fast” (n = 87), who received 35 cc/kg/d increments (a 5-day schedule to full feeds) of Similac Special Care 20 cal/oz. Feeds were increased only if well tolerated as defined by a protocol. Results: The incidence of NEC (Bell stage ≥II) was similar in both groups (slow 13% and fast 9%, P = .5). The incidence of perforation (Bell stage III) was also similar in both groups (slow 4% and fast 2%, P = .8). Feeds were started at a comparable postnatal age in both groups (median age: slow 5 days and fast 4 days, P = .9). Although the neonates in the fast group attained full enteral intake earlier (median days [25th and 75th percentiles]: slow 15 [12, 21] and fast 11 [8, 15], P < .001) and regained their birth weight earlier (slow 15 [11, 20] and fast 12 [8, 15], P < .05), their ages at discharge were not statistically different (slow 47 [31, 67] and fast 43 [29, 62], P = .3) Conclusions: A greater than twofold difference in the rate of feed advancement from 15 cc/kg/d to 35 cc/kg/d did not affect the incidence of NEC ≥ stage II. Factors other than feed advancement appear to be more important in the pathogenesis or progression of NEC. (J Pediatr 1999;134:293-7)
Necrotizing enterocolitis is a major cause of mortality and morbidity in preterm infants. The incidence of NEC in the National Institute of Child
Health and Development Neonatal Research Network has been 8% to 10% of very low birth weight infants, with center differences in incidence
From the Department of Pediatrics, University of Alabama at Birmingham, Birmingham, Alabama, and National Institute of Child Health and Human Development, Bethesda, Maryland.
Presented in part at the Pediatric Academic Societies Meeting, Washington, DC, May 2-6, 1997. Submitted for publication Aug 5, 1998; revision received Nov 4, 1998; accepted Dec 2, 1998. Reprint requests: Waldemar A. Carlo, MD, University of Alabama at Birmingham, 525 New Hillman Bldg, 619 South 19th St, Birmingham, AL 35233-7335. Copyright © 1999 by Mosby, Inc. 0022-3476/99/$8.00 + 0 9/21/96310
from 3.9% to 22.4%.1,2 The cause of NEC is inadequately understood, although many factors such as enteral feeds, hypoxia, ischemia, patent ductus arteriosus, and infection have been associated with an increased incidence of NEC.3-5 Several retrospective studies suggest that rapid advancement in feeds may be associated with an increased incidence of NEC.6-8 A case control study by the National Institute of Child Health and Development Neonatal Research Network of 249 NEC Necrotizing enterocolitis PDA Patent ductus arteriosus VLBW Very low birth weight
neonates with NEC revealed that the infants who had NEC received significantly more enteral nutrition in the 3 days before the diagnosis of NEC compared with a control group.9 In that study the incidence of NEC in individual centers ranged from 2% to 19%, and 2 centers accounted for 63% of the cases.9 During that time period the 16% incidence of NEC in our center was the second highest of the 8 clinical centers. At that time VLBW neonates frequently achieved full enteral intake in <5 days after initiation of feeds. The purpose of this study was to determine whether the rate of advancement of feeds influenced the incidence of NEC. We tested the hypotheses that in VLBW neonates: (1) slow feed advancement (15 cc/kg/d) decreases the overall incidence of NEC (Bell stage ≥ II) compared with fast feed advance293
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS MARCH 1999
Table I. Incidence of NEC (≥stage II) by weight subgroups
Birth weight (g) stratum 501-750 751-1000 1001-1250 1251-1500 Total
Slow NEC/Total (%)
Fast NEC/Total (%)
P value
2/13 (15) 7/30 (23) 3/22 (13) 1/33 (3) 13/98 (13)
0/9 (0) 3/24 (13) 4/25 (16) 1/29 (3) 8/87 (9)
NS NS NS NS NS
ment (35 cc/kg/d) and (2) slow feed advancement decreases the incidence of NEC with intestinal perforation (Bell stage III) compared with fast feed advancement.
METHODS Subjects From January 24, 1994, until December 30, 1996, 187 infants with birth weight 501 to 1500 g and gestational age ≤34 weeks admitted to the regional neonatal intensive care unit at the University of Alabama at Birmingham were prospectively enrolled in the trial. The exclusion criteria included (1) Apgar score <3 at 5 minutes, (2) hemodynamic instability requiring pressors to maintain blood pressure at the time of initiation of feeds, (3) presence of cyanotic congenital heart disease or other major malformation, (4) polycythemia (venous hematocrit >70%), (5) partial or double volume exchange transfusion before initiation of feeds, (6) human milk feeds (partial or total), and (7) multiple gestation of triplets or more. Small for gestational age infants, large for gestational age infants, infants undergoing mechanical ventilation, and infants who received indomethacin were not excluded. Trophic (minimal enteral) feeds were not used. Eligibility was assessed after the decision to initiate feeding was made by the clinical team. Informed parental consent was obtained. Randomization was performed according to weight stratification into 4 groups: 501 to 750 g, 751 to 1000 g, 1001 to 1250 g, and 1251 to 1500 g. Shuffled blocks of 2, 4, and 6 294
sealed envelopes for each weight group were used. The person obtaining parental consent was masked to the odds of a subsequent infant receiving either treatment. Twins were randomized to the same treatment mode according to randomization of the first eligible of the pair.
Protocol The protocol was approved by the institutional review board before the trial was begun. The patients were randomized into 2 groups: (1) “slow” feeding group: feeds were started at 20 cc/kg/d and advanced by increments of 15 cc/kg/d to a maximum of 160 cc/kg/d (a 10-day schedule to full feeds), (2) “fast” feeding group: feeds started at 35 cc/kg/d and advanced by increments of 35 cc/kg/d to a maximum of 160 cc/kg/d (a 5-day schedule to full feeds). Both groups started feeds with Similac Special Care 20 cal/oz formula and were switched to Similac Special Care 24 cal/oz formula when full feeds were reached. Oral medications were not given until full enteral feeds were attained. The patients were fed every 3 hours with an orogastric tube. Aspirates (residuals) from the orogastric tube were checked before every feed. If aspirates were ≤20% of total feed and the abdominal examination was normal, feeds were advanced according to the protocol. If aspirates were between 20% and 30% and the abdominal examination was normal, feeds were continued, but no increase in feeds was done on that day. Feeds were temporarily discontinued for 24 hours and
then continued at half the previous volume for aspirates between 30% and 40%. If the aspirates were ≥40% or the abdominal examination was abnormal regardless of the aspirates, a complete blood count and an abdominal radiograph were obtained. If both investigations were normal, feeds were temporarily discontinued for 24 hours and then resumed at half the previous volume and advanced at the previous day’s rate. An abnormal abdominal examination was defined as 1 or more of the following: (1) abdominal distention, (2) erythema of the abdominal wall, (3) abdominal tenderness, (4) decreased bowel sounds, (5) increased abdominal girth, and (6) gross or occult blood in the stools. NEC was defined according to the Bell stages.10 Bell stage II (proven NEC) includes radiographic findings of pneumatosis intestinalis or portal vein gas with clinical signs including poor feeding, increased residuals, or abdominal distention with bloody stools. Bell stage III includes radiographic observation of pneumoperitoneum with the previously described clinical signs associated with deterioration of vital signs and septic shock. Presence of a PDA, when suspected on clinical evaluation, was confirmed by echocardiography. Indomethacin was administered at standard doses (0.2 mg/kg per dose, 3 doses at 12hour intervals). There was no change in the feeding schedule in the presence of a PDA. The end point of the study was when the patient was discharged or had NEC (stage ≥ II) before discharge home. The results were analyzed by intention to treat.
Statistics The expected incidence of NEC (≥ stage II) in the fast feeding group was 16% based on the incidence at our center, and the anticipated incidence of NEC in the slow feeding group was estimated to be 5% based on the incidence at the centers with slower feed-
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS VOLUME 134, NUMBER 3 ing protocols.9 The sample size was estimated at a total of 185 with the Fisher-Irwin tables for sample size estimation with power of 80%, α of 0.05, and a small excess to compensate for small variations in the actual incidences from the expected incidences. Chi-squared analysis, t test, or Wilcoxon rank sum test was used for statistical analysis of the data, depending on whether the data were normally distributed. A P value < .05 was considered statistically significant.
RESULTS During the enrollment period 605 neonates with birth weight 501 to 1500 g were admitted to the neonatal intensive care unit. A total of 295 patients were excluded because of 1 or more of the following reasons: human milk feeds (n = 153), major malformations (n = 71), died before initiation of feeds (n = 54), fed before consent was obtained (n = 32), low Apgar scores (n = 26), congenital heart disease (n = 22), transferred to another hospital before initiation of feeds (n = 4), triplets or quintuplets (3 sets), unable to get consent because of maternal illness (n = 4), and transpyloric feeds (n = 3). Parents of 310 neonates were approached, and randomization was performed after informed consent was obtained on 185 infants (rate of 60%). There were 7 protocol violations including patients fed human milk at parent’s request (n = 3), patient transferred to another hospital before receiving full feeds (n = 2), and infants with multiple episodes of feeding intolerance taken off the study at the request of the attending physician (n = 2). Data on these infants were included in the analysis, because analysis was done on the basis of intention to treat. The data from 185 infants (slow 98, fast 87) were analyzed. Birth weight, gestational age, sex, and race were comparable in the 2 groups (Table I). Feedings were started at a comparable
Table II. Postnatal age in relation to nutrition and discharge
Age at starting feeds (d) Age at full feeds (d) Age when birth weight regained (d) Age at discharge (d)
Slow
Fast
5 (3-7) 15 (12-21) 15 (11-20) 47 (31-67)
4 (3-7) 11 (8-15)* 12 (8-15)* 43 (29-62)
Data are shown as median days (25th to 75th percentiles). *P < .05.
postnatal age in both groups (slow 5 days, fast 4 days, P = .9). No difference was seen in the incidence of NEC (Bell stage ≥ II) between both groups (slow 13%, fast 9%, P = .50) or for each of the weight subgroups (Table I). It should be noted that although subgroup analysis by weight is shown, the small sample size of each stratum makes type II error more likely. The incidence of intestinal perforation (Bell stage III) was also comparable in both groups (slow 4%, fast 2%, P = .8). No significant difference was seen in mortality resulting from NEC (in entire feeding group: slow 2%, fast 3%, P = .9 and in those who had NEC stage ≥ II: slow 2 of 13, fast 3 of 8, P = .32 by Fisher’s Exact test). The babies in the fast group had NEC at the age of 25 ± 14 days (mean ± SD), whereas the neonates in the slow group had NEC at 18 ± 9 days (P = .3). The average volume of feeds in the 3 days before NEC in the fast group was 122 ± 44 cc/kg/d (mean ± SD) and 96 ± 40 cc/kg/d in the slow group (P = .07). In the 3 days before the onset of NEC, no difference was noted in feeding intolerance (<15% of neonates in either group with more than 20% residuals). The incidence of NEC in the neonates enrolled in this trial (combined) was not significantly different from the incidence seen historically in the same unit (fast + slow 11% vs historical 16%, P = .4). Although not part of the trial, the incidence of NEC in the neonates excluded because of human milk feeding was 7% (fast + slow 11% vs human milk 7%, P = .4).
The age at which the diagnosis of NEC was made was similar in the 2 groups. Three infants in each group had NEC before reaching full feeds. No contribution by the epidemic form of NEC was noted, because there were never more than 2 cases of NEC (Bell stage ≥ II) per month (in both groups combined) during the study. Data on postnatal age were not normally distributed and are reported as median with the 25th and 75th percentile (Table II). As expected, a significant difference was seen between the 2 groups in postnatal age at full feeds (slow 15 days [12 to 21], fast 11 days [8 to 15]; P < .001) and the postnatal age to regain birth weight (slow 15 days [10 to 20], fast 12 days [8 to 15]; P < .05). When the number of days to achieve full feeds from initiation of feeds were compared, 46 (47%) achieved full feeds in 10 days in the “slow” group, and 49 (56%) achieved full feeds in 5 days in the fast group. However, the age at discharge was not statistically different (slow 47 days [31 to 67], fast 43 days [29 to 62], P = .3). The 2 groups were comparable in variables that may influence the incidence of NEC such as antenatal steroid exposure, postnatal indomethacin use (either as prophylaxis for intraventricular hemorrhage or as management of PDA), the presence of symptomatic PDA (all given indomethacin), and the placement of umbilical arterial catheter. The demographics of the VLBW neonates who were enrolled for this study were comparable to those of the total VLBW population admitted to 295
RAYYIS ET AL
the nursery at that time period (mean birth weight 1045 g, 28 weeks’ gestation, 49% male). The incidence of NEC in the VLBW neonates not enrolled in the study (9.5%) was also similar to the incidence of NEC in the study (11%). The median length of hospital stay was also similar in the babies who did not participate in the study (49 days) compared to those who participated in the study (45 days).
DISCUSSION The incidence of NEC (Bell stage ≥II) and NEC with intestinal perforation (Bell stage III) did not change with more than a twofold difference in the rate of feed advancement. Also, the timing of the development of NEC did not differ between both groups. Although the birth weight was regained earlier in the fast group, this did not translate into an earlier discharge. The fast group reached full feeds earlier, which indicates that this rate of feed advancement is equally tolerated compared to the slow feeding regimen. Although the study was a randomized controlled trial, it had some important limitations. It could not have been done in a masked manner. To decrease the influence of potential bias, feed intolerance was defined by strict criteria, and a radiologist independent of the study made the radiologic diagnosis of NEC. A different number of patients were noted to have been randomized to each group (slow 98, fast 87). Other than the effect of chance in the technique of block randomization, 1 factor that had an effect is the number of twins randomized to each group. Five sets of twins were randomized to the slow group, and 4 were randomized to the fast group. Another limitation is that although the sample size was calculated on a projected incidence of NEC at 16% in 1 group and 5% in the other group based on past studies, the actual incidences were 296
THE JOURNAL OF PEDIATRICS MARCH 1999 lower (slow 13%, fast 9%). However, the post-hoc power of the study indicating the groups were similar was 87%, and although a type II error is still possible, it is less likely (a sample size of 1850 is required to show that the difference in the incidence of NEC noted between the groups is statistically significant at a .05 and 80% power). The effect of feeding schedules on the incidence of NEC has been widely debated. In support of our observations 2 prospective randomized trials demonstrated no increase in the incidence of NEC when feed increments were increased. Book et al11 compared 2 feed increments, 10 cc/kg/d and 20 cc/kg/d. No significant difference was seen in the incidence of NEC between the 2 groups. However, the sample size was small (total n = 29), and a type II error was possible. A recent prospective randomized study by Caple et al12 showed no difference in the incidence of NEC between feed advancements of 20 cc/kg/d and 30 cc/kg/d. These results are consistent with our study, because we did not detect any increase in the incidence of NEC even with a wider range of feeding advancements of 15 cc/kg/d and 35 cc/kg/d. On the other hand, several retrospective case control studies suggest that aggressive increments in feeds may increase the incidence of NEC. Anderson and Kliegman7 showed that infants who had NEC compared with a matched control group had significantly larger feeding increments. Similar results were obtained in other retrospective case control studies by Covert et al6 and McKeown et al.8 When compared with a matched control group, the infants with NEC were fed earlier and received larger feeding volumes.8 These retrospective case control studies reveal results that may not be comparable with ours. There may be 2 reasons for this: (1) in some cases feed increments were greater than 35 cc/kg/d, and (2) the neonates with the NEC group were found to have started feeds earlier than the control group.
In our study both groups started feeds at a similar postnatal age. Four chart review studies revealed an association between the incidence of NEC and rapid feeding practices. Krouskop13 reported that more than 50% of infants who had NEC had received aggressive feeding increments. Book et al11 noted dramatic increases in the incidence of NEC when feeding protocols became faster. In a similar manner, Spritzer et al14 found dramatic decreases in the incidence of NEC in their neonatal care unit after introducing a slower feeding regimen. Finally, Goldman,15 in another chart review, found that patients who had NEC had larger feeding increments before the episode of NEC occurred. These studies have results that contradict our prospective randomized study. Chart review studies have the limitations common to other forms of retrospective studies. No control of other variables may influence the incidence of NEC. On the other hand, known and unknown confounders are likely to be equally distributed between groups in a randomized trial such as this study. This study has shown that feed increments of 35 cc/kg/d compared with 15 cc/kg/d were equally well tolerated and did not increase the incidence of NEC in VLBW neonates matched for weight and gestational age. Therefore factors other than the rate of feed advancement may be more important in the pathogenesis or progression of NEC. However, it does not follow that increments larger than 35 cc/kg/d are equally safe in the VLBW population. It is possible that larger increments, beyond a currently undefined value, may in fact alter the risk of NEC. Marked intercenter differences in the incidence of NEC2,15 indicate that further investigation of the determinants of NEC should include analysis of variations in clinical practice and sociodemographic variables in addition to the variables of prematurity, infection, hypoxia, ischemia, and feeding practices that are commonly implicated.
RAYYIS ET AL
THE JOURNAL OF PEDIATRICS VOLUME 134, NUMBER 3
REFERENCES 1. Uauy RD, Fanaroff AA, Korones SB, Philips EA, Philips JB, Wright LL. Necrotizing enterocolitis in very low birth weight infants: biodemographic and clinical correlates. J Pediatr 1991; 119:630-8. 2. Hack M, Wright LL, Shankaran S, Tyson JE, Horbar JD, Bauer CR, et al. Very-low-birth-weight outcomes of the National Institute of Child Health and Human Development Neonatal Network, November 1989 to October 1990 Am J Obstet Gynecol 1995;172:457-64. 3. Stoll BJ. Epidemiology of necrotizing enterocolitis. Clin Perinatol 1994;21: 205-18. 4. La Gamma EF, Browne LE. Feeding practices for infants weighing less than 1500 g at birth and the pathogenesis of necrotizing enterocolitis. Clin Perinatol 1994;21:271-306. 5. Nowicki PT; Nankervis CA. The role
6.
7.
8.
9.
10.
of the circulation in the pathogenesis of necrotizing enterocolitis. Clin Perinatol 1994;21:219-34. Covert RF, Neu J, Elliot M, Rea JL, Gimolty PA. Factors associated with age of onset of necrotizing enterocolitis. Am J Perinatol 1989;614:455-60. Anderson D, Kliegman R. The relationship of neonatal alimentation practices to the occurrence of endemic NEC. Am J Perinatol 1991;8:62-7. McKeown RE, Marsh TD, Amaranth W, Garrison, CZ, Addy CL, Thompson SJ, et al. Role of delayed feeding and of feeding increments in necrotizing enterocolitis. J Pediatr 1992;121: 764-70. Wright LL, Uauy RD, Younes N, Fanaroff AA, Korones SB, Philips JB. Rapid advances in feeding increase the risk of necrotizing enterocolitis in very low birthweight in infants [abstract]. Pediatr Res 1993;33:313A. Bell MJ, Trenberg JL, Feigin RD, Keating JP, Marshall R, Barton L, et
11.
12.
13.
14.
15.
al. Neonatal necrotizing enterocolitis: therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1-7. Book LS, Herbst JJ, Jung AL. Comparison of fast- and slow-feeding rate schedules in the development of necrotizing enterocolitis. J Pediatr 1976;89: 463-6. Caple JI, Armentrout DC, Huseby VD, Halbrdier BM, Garcia J, Sparks JW. The effect of feeding volume on clinical outcome in premature infants [abstract]. Pediatr Res 1997:41:229A. Krouskop R. The relationship of feeding to NEC [abstract]. Pediatr Res 1974;8:109. Spritzer R, Koolen AMP, Bauts W, Fetter WPF, Lafeber HN, Sauer PJJ. A prolonged decline in the incidence of necrotizing enterocolitis after the introduction of a cautious feeding regimen. Acta Paediatr Scand 1988;77: 905-91. Goldman HI. Feeding and NEC. Am J Dis Child 1980;134:553-5.
AVAILABILITY OF JOURNAL BACK ISSUES As a service to our subscribers, copies of back issues of The Journal of Pediatrics for the preceding 5 years are maintained and are available for purchase from Mosby until inventory is depleted, at a cost of $15.00 per issue. The following quantity discounts are available: 25% off on quantities of 12 to 23, and one third off on quantities of 24 or more. Please write to Mosby, Inc., Subscription Services, 11830 Westline Industrial Drive, St. Louis, MO 63146-3318, or call (800)453-4351 or (314)453-4351 for information on availability of particular issues. If unavailable from the publisher, photocopies of complete issues may be purchased from UMI, 300 N. Zeeb Rd., Ann Arbor, MI 48106, (313)761-4700.
297