PATIENTS AND METHODS
Preterm infants with birth weight ≤ 1500 g were recruited if there was evidence of feeding intolerance defined by the following criteria: either failure to start minimal (non-nutritive) enteral feedings by 1 week of life or failure to attain at least 20 mL/kg/d enteral feeds 1 week after initiating feeds. Infants with overwhelming septicemia, major congenital malformations, or growth retardation plus antenatal Doppler studies documenting absent or reversed end-diastolic flow (higher risk of experiencing necrotizing enterocolitis) were excluded. Appropriateness for gestational age was determined by dating scans and confirmed by Dubowitz/Ballard scores. This study was approved by the hospital review board for human subject research, and informed written consent was obtained from the parents before randomization.
Eligible preterm infants were enrolled in the study from May to December 1999 and randomized to receive either oral erythromycin (5 mg/kg every 8 hours) or saline placebo. Erythromycin ethyl succinate (EES) (a white, relatively odorless powder) was reconstituted with water at the concentration of 50 mg/5 mL. Five % dextrose with 0.1% NaCl w/w and 5% Novolose (also a white powder) were used as the placebo. Both preparations were isotonic, with the osmolarity between 150 and 180 mosm/kg, and were packaged in bottles of the same size and color for blinding purposes. Allocation to the two groups was determined via sealed envelopes, with the assignments known only to the pharmacist, who was blinded to the subsequent care of the infants in the neonatal unit.
All infants were begun on parenteral nutrition (Vaminolact and 20% Intralipid, Fresenius Kabi, Stockholm, Sweden) on day 3 of life. Oral bolus milk feedings were usually started within the first 6 days of life at the discretion of the attending neonatologist. The increments in feeding volume were usually < 10 mL/kg/d for the first week of life and < 20 mL/kg/d subsequently based on the volume and nature of gastric residuals and the degree of abdominal distension as determined by department feeding protocol. Infants were fed mother's milk when possible, but preterm commercial milk formulas were also given as supplements when mother's milk was insufficient. All infants were examined at least three times a day and closely monitored for the occurrence of emesis, diarrhea, abdominal distension, and volume of gastric residuals. Oral feedings were discontinued if vomiting occurred more than twice in 24 hours; if the volume of gastric residuals exceeded 25% of the oral intake in the preceding 4 hours on two occasions within the same day; if there were clinical signs and symptoms indicative of necrotizing enterocolitis or other intra-abdominal pathology; or if repeated regurgitation and aspiration pneumonia were suspected. Continuous intragastric tube feeding was considered if at least 50% of bolus feedings were still not tolerated 2 weeks after starting enteral feedings. All infants were maintained in the prone position with the head elevated 30° postprandially for at least 1 hour. Use of the study drug was discontinued 1 week after full enteral feedings (at least 130 mL/kg/d) were attained and tolerated.
A Monocrystant Antimony pH probe was inserted transnasally for extended distal esophageal pH monitoring. The exact placement of the sensor from the naris was estimated by the formula 0.252 × length of infant + 5 cm and confirmed by chest radiograph (19). The percentage of monitored time that distal esophageal pH was <4 for at least 15 seconds (RI = reflux index), the total number of reflux episodes per day, the number of reflux episodes lasting >5 minutes per day, and the longest reflux episode (minutes) were recorded and computed by the digitrapper for each infant. Extended distal esophageal pH monitoring was done on each infant at the time of recruitment and 1 day after establishment of full enteral feedings.
The primary outcome variable was the time taken to attain full enteral feeding (at least 130 mL/kg/d). Additional information collected included gestational age; birth weight; gender; use of antenatal/postnatal steroids and duration of treatment; use of drugs with potential impact on gastrointestinal motility and/or gastroesophageal reflux, such as xanthine derivatives (either intravenous aminophylline or oral caffeine); age at which feedings were initiated; rate of feeding increments; type of feeding given; and mode of delivery (bolus or continuous drip). We considered the following occurrences as potential side effects of erythromycin and recorded them: cholestatic jaundice, cardiac dysrhythmias, infantile hypertrophic pyloric stenosis, necrotizing enterocolitis, and confirmed septicemia.
Our unit statistics revealed a consistently high prevalence of milk intolerance in preterm infants with very low birth weight during the 4 consecutive years before the study. The mean total time taken to attain full enteral feedings in our unit was 35 ± 5 days. Recruitment of 24 infants would therefore be sufficient to detect a 30% difference in the time taken to achieve full feedings (a reduction from 35 days to 25 days) between the two groups with a two-tailed test size of 5% (alpha = 0.05) and power of 80% (beta = 0.2).
Study data were analyzed on an intent-to-treat basis. The primary outcome variable was the time taken to attain full enteral feedings (at least 130 mL/kg/day). Continuous variables were analyzed by two-sample Wilcoxon rank-sum (Mann-Whitney) test, whereas categorical data (proportions) were analyzed by Pearson χ2 test. All statistical tests were performed by SPSS for Windows (Release 10.0; SPSS Inc, Chicago, Illinois), and the level of significance was set at 0.05.
Gestational age at birth was similar between the 2 groups (erythromycin, 27.1 ± 1.9 weeks versus placebo 27.5 ± 2.9 weeks) (Table 1). The erythromycin group had lower mean birth weight (806.3 ± 215.6 g) than did the placebo group (981.4 ± 285.4 g; P = 0.18). More erythromycin infants were small for gestational age (≤ 3rd centile) (4/13 = 31% vs 1/11 = 9%; P = 0.224), although the difference was not statistically significant. Infants in the erythromycin group were enrolled at a mean age of 19.7 days, and those in the placebo group 17.3 days (corrected gestation: erythromycin, 29.9 weeks; placebo, 30.0 weeks). There was no difference between the two groups with regard to the volume of feedings they were receiving at enrollment (median for both groups was 19.6 mL/kg/d), age or gestation at enrollment, age when first fed, or the requirements for antenatal, postnatal steroids, and xanthine derivatives. All but one infant in the erythromycin group satisfied both inclusion criteria, which were failure to start minimal (non-nutritive) enteral feeds by 1 week of life and failure to attain at least 20 mL/kg/d enteral feeds 1 week after initiating feeds.
After enrollment, the group receiving low-dose erythromycin attained full feeds earlier (24.9 ± 2.9 days) than did the placebo group (30.8 ± 4.1 days), although the difference did not reach statistical significance (P = 0.17) (Fig. 1). The daily feeding increment was similar in the two groups. Most of the infants received expressed breast milk (12 in the erythromycin group and 9 in the placebo group). The rest received a combination of breast milk and formula (erythromycin group: 1 premature formula; placebo group: 1 breast milk plus premature formula, and 1 premature plus term formula). Most of the infants (9/13 in the erythromycin group and 8/11 in the placebo group) received bolus feedings.
A higher proportion of neonates in the placebo group (7/11, 64%) experienced cholestatic jaundice than did those in the erythromycin group (4/13 = 31%; P = 0.113); the neonates in the erythromycin group received more days of parenteral nutrition (39.4 ± 13.8 days) than those in the placebo group (43.3 ± 18.3 days, P = 0.743) (Table 2). These differences were statistically insignificant. None of the infants in the erythromycin group had cardiac dysrhythmias, clinical symptoms of pyloric stenosis, or septicemia with multiresistant organisms. One infant in the placebo group developed necrotizing enterocolitis at 2 months of age, almost 1 month after having attained full enteral feedings.
The reflux indices at enrollment and after attaining full enteral feedings were not significantly different in the two groups (erythromycin, 7.3 ± 15.5 and 4.3 ± 7.1, versus placebo, 13.6 ± 17.3 and 0.3 ± 0.6, respectively). There was no difference between the 2 groups with regard to reduction of gastroesophageal reflux; that is, 3 of 13 infants in the erythromycin group improved compared with 4 of 11 infants in the placebo group (P = 0.565). Although the differences did not reach statistical significance, the low-dose erythromycin group regained birth weight faster, attained full feeds faster (from birth), needed fewer glycerin suppositories, and stayed fewer days in the hospital.
Feeding intolerance is a common problem for preterm infants that frequently results in dependence on prolonged parenteral nutrition, with the attendant complications of septicemia and cholestatic jaundice. This is the first randomized, controlled trial in preterm neonates with feeding intolerance using oral low-dose erythromycin instead of the standard antimicrobial doses. The low-dose erythromycin group attained full enteral feeds earlier without any major side effects although this difference did not reach statistical significance.
In preterm infants less than 32 weeks gestation MMCs do not occur, although plasma motilin concentrations are comparable with those of adults. In preterm infants older than 32 weeks gestation, normal MMCs occur with increasing frequency (12,20,21). This may explain the immature patterns of antroduodenal and small bowel motor activity and the consequent feeding intolerance of preterm infants. It is possible, however, that motilin receptors on smooth muscle become functional earlier in development than do neuronal motilin receptors, which are responsible for the generation of MMCs (10,11).
Erythromycin, a motilin receptor agonist, has been reported in three case series of preterm infants with gastrointestinal dysmotiltiy to decrease the time taken to attain full oral feedings when given at the standard antimicrobial doses (4–6). Earlier randomized trials comparing the effects of prophylactic erythromycin on establishment of feeding in preterm infants did not demonstrate a reduction in the time to establish full feeds (7,8). Significant differences might have resulted if only infants who had feeding intolerance were recruited, rather than all preterm infants, some of whom probably did not have feeding intolerance and therefore did not benefit from erythromycin.
A later trial by Ng et al. (9), evaluated the impact of erythromycin on preterm infants with moderately severe gastrointestinal dysmotility. Fifty-six preterm infants (birth weight <1500 g) who were consuming less than half their total daily fluid requirements enterally on day 14 of life were randomized to receive either oral erythromycin 12.5 mg/kg every 6 hours for 14 days or saline placebo. The erythromycin group attained half, three quarters, and full enteral feeds significantly sooner. Although severe arrhythmias or infantile hypertrophic pyloric stenosis did not occur in either group, the potential exists, especially with the standard antimicrobial doses used (15–17). In another double-blind randomized, crossover study (10), oral erythromycin (10 mg/kg every 8 hours) or placebo was given for 7 days to 20 preterm infants with feeding intolerance of 32 weeks median gestational age. Both antral contractility and whole gut transit time were significantly shorter during erythromycin treatment, and the authors reported no adverse effects such as arrhythmias.
Small, nonrandomized studies suggest that erythromycin at doses less than normally given for antimicrobial activity result in better prokinetic activity (2,11,12) without attendant complications. Oei and Lui (18) randomized preterm infants ≤ 32 weeks to receive either low-dose erythromycin or placebo for 10 days from the time of the first oral feeding. Although the erythromycin group had significantly fewer episodes of large gastric residuals and achieved full oral feedings earlier, it might not have been necessary to randomize some of these infants because they might not have had feeding intolerance. Dellagrammaticas and Iacovidou (22) reported encouraging results from their case series (no control subjects) on the effects of oral erythromycin (1.7 mg/kg given every 8 hours) on gastric aspirates of ventilated neonates younger than 32 weeks gestation with gastrointestinal dysmotility. Thus, we decided to give oral erythromycin at a dosage of 5 mg/kg every 8 hours (15 mg/kg/d), which was lower than the dosages previously reported (9,10), and we randomized only preterm infants with feeding intolerance.
More objective measures of gut motility, such as ultrasonography for gastric antral cross-sectional area (10,23,24), antroduodenal manometry (13), or carmine red dye (10,13) have been used to help define feed-intolerant study populations and to document the effects of erythromycin treatment. We did not use any of these methods to define feeding intolerance but recruited our preterm infants using subjective and noninvasive clinical criteria. Another limitation of our study was the unexpected heterogeneity in demographic characteristics of our two groups. The erythromycin group weighed less and also had more patients with intrauterine growth retardation. It is possible that the effects of erythromycin might have been more dramatic in groups that were less heterogeneous.
We conducted extended distal esophageal pH studies (19) on the infants in our study to see whether erythromycin would reduce gastroesophageal reflux, a potential contributory cause for feeding intolerance. Gastroesophageal reflux decreased in both groups over time and thus seems likely to be an effect of maturation rather than a result of erythromycin therapy.
More infants in the placebo group experienced cholestatic jaundice than in the erythromycin group. Cholestatic jaundice was more common in the placebo group possibly because the patients took a longer time to establish full feeds and thus received parenteral nutrition for a longer period. None of the infants experienced cardiac dysrhythmias, which have been reported in studies in which erythromycin (15,16) has been administered intravenously. Mahon et al. (17) reported an increased incidence of infantile hypertrophic pyloric stenosis with standard doses of erythromycin, but we did not encounter this in our patients. Concerns about prolonged courses of antibiotics causing changes in gut microflora and subsequent antibiotic resistance were not validated in our study. None of our patients on erythromycin developed septicemia with antibiotic resistant organisms or necrotizing enterocolitis (9,18).
In conclusion, our study shows that oral erythromycin at low doses did not reduce the time taken to attain full enteral feeds in preterm infants with very low birth weight and feed intolerance. Our results also indicate that oral erythromycin at the dose used here is safe and well-tolerated. We speculate that the heterogeneity in demographic characteristics between the two groups may have accounted for the less than dramatic effects of erythromycin on the primary outcome. These preliminary data justify repeated study in larger trials. Until such data are available, we caution that the use of erythromycin should remain experimental.
The authors thank Dr. V. Vijayan (Research and Administrative Unit) and Ms. E. Woo and W. C. Koi (Pharmacy Department), Kandang Kerbau Women's and Children's Hospital; and Dr. Adeline Seow, Department of Community, Occupational & Family Medicine, National University of Singapore, for helping in the design and subsequent conduct of this study.
1. Inatomi N, Satoh H, Maki Y, et al. An erythromycin derivative, EM-523, induces motilin-like gastrointestinal motility in dogs. J Pharmacol Exp Ther 1989; 251:707–12.
2. Otterson MF, Sarna SK. Gastrointestinal motor effects of erythromycin. Am J Physiol 1990; 259:G355–63.
3. Greenwood B, Dieckman D, Kirst HA, et al. Effects of LY267108, an erythromycin analogue derivative, on lower esophageal sphincter function in the cat. Gastroenterology 1994; 106:624–8.
4. Simkiss DE, Adams IP, Myrdal U, et al. Erythromycin in neonatal postoperative intestinal dysmotility. Arch Dis Child 1994; 71:F128–9.
5. Kubota M, Nakamura T, Motokura T, et al. Erythromycin improves gastrointestinal motility in extremely low birthweight infants. Acta Paediatr Jpn 1994; 36:198–201.
6. Ng PC, Fok TF, Lee CH, et al. Erythromycin treatment for gastrointestinal dysmotility in preterm infants. J Paediatr Child Health 1997; 33:148–50.
7. Stenson BJ, Middlemist L, Lyon AJ. Influence of erythromycin on establishment of feeding in preterm infants: observations from a randomized controlled trial. Arch Dis Child Fetal Neonatal Ed 1998; 79:F212–4.
8. Patole SK, Almonte R, Kadalraja R, et al. Can prophylactic oral erythromycin reduce time to full enteral feeds in preterm neonates? Int J Clin Pract 2000; 54( 8):504–8.
9. Ng PC, So KW, Fung KSC, et al. Randomised controlled trial of oral erythromycin for treatment of gastrointestinal dysmotility in preterm infants. Arch Dis Child Fetal Neonatal Ed 2001; 84:F177–82.
10. Costalos C, Gounaris A, Varhalama E, et al. Erythromycin as a prokinetic agent in preterm infants. J Pediatr Gastroenterol Nutr 2002; 34:23–5.
11. Tomomasa T, Miyazaki M, Koizumi T, et al. Erythromycin increases gastric antral motility in human premature infants. Biol Neonate 1993; 63:349–52.
12. Jadcherla SR, Klee G, Berseth CL. Regulation of migratory motor complexes by motilin and pancreatic polypeptide in human infants. Pediatr Res 1997; 42:365–9.
13. Berseth CL, ElHennawy A. Erythromycin fails to improve gastrointestinal function or feeding outcome in feeding intolerant preterm infants [Abstract]. Pediatr Res 2002; 51( 5):143A.
14. Cairns PA, Craig S, Tubman R, et al. Randomised controlled trial of low-dose erythromycin in preterm infants with feed intolerance [Abstract]. Pediatr Res 2002; 51( 5):379A.
15. Farrar HC, Walsh-Sukys MC, Kyllonen K, et al. Cardiac toxicity associated with intravenous erythromycin lactobionate: two case reports and a review of the literature. Pediatr Infect Dis J 1993; 12:688–91.
16. Gouyon BJ, Benoit A, Betremieux P, et al. Cardiac toxicity of intravenous lactobionate in preterm infants. Pediatr Infect Dis J 1994; 13:840–1.
17. Mahon BE, Rosenman MB, Kleiman MB. Maternal and infant use of erythromycin and other macrolide antibiotics as risk factors for infantile hypertrophic pyloric stenosis. J Pediatr 2001; 139( 3):380–4.
18. Oei J, Lui K. A placebo-controlled trial of low-dose erythromycin to promote feed tolerance in preterm infants. Acta Paediatr 2001; 90:904–8.
19. Ng SCY, Quak SH. Gastroesophageal reflux in preterm infants: norms for extended distal esophageal pH monitoring. J Pediatr Gastroenterol Nutr 1998; 27:411–4.
20. Berseth CL. Gastrointestinal motility in the neonate. Clin Perinatol 1996; 23( 2):179–90.
21. Jadcherla SR, Berseth CL. Efficacy of erythromycin on gastroduodenal contractile activity in developing neonates. J Pediatr Gastroenterol Nutr 2002; 34:16–22.
22. Dellagrammaticas HD, Iacovidou N, Megaloyanni E, et al. Effect of low-dose oral erythromycin on gastric aspirates in ventilated neonates less than 32 weeks of gestation. Preliminary results. Biol Neonate 2002; 81( 3):213–6.
23. Newell SJ, Chapman S, Booth IW. Ultrasonic assessment of gastric emptying in the preterm infant. Arch Dis Child 1993; 69:32–6.
24. Carlos MA, Babyn PS, Marcon MA, et al. Changes in gastric emptying in early postnatal life. J Pediatr 1997; 130:931–7.