Postnatal growth restriction has been recently recognised as a major and almost universal issue in preterm, especially extremely low birthweight neonates.^1–7 Optimising enteral nutrition in preterm neonates, however, is not easy because of the common occurrence of feed intolerance due to gastrointestinal hypomotility (“ileus of prematurity”) and the risk of necrotising enterocolitis, a potentially devastating illness. It is not uncommon to take several days or even weeks to establish full enteral feeds (FEFs) in high risk preterm neonates. Given this evidence and also the adverse effects associated with prolonged use of total parenteral nutrition,^8,9 establishing FEFs quickly in high risk preterm neonates has become a priority in neonatal intensive care.

The gastrointestinal motor effects of erythromycin are well documented in experimental and clinical studies.^10–16 After the early positive reports, the use of erythromycin as a prokinetic agent became fairly common in neonatal nurseries until recently.^17–20 An earlier systematic review that focused only on the use of low dose erythromycin (3–12 mg/kg/day) in preterm neonates ⩽36 weeks gestational age with feeding tolerance did not find any eligible studies at the time.²¹ The lack of a clear understanding of the basis of its prokinetic action, however, has recently resulted in a plethora of clinical trials using different doses, routes and modes of administration, and durations of treatment. Most of these trials used surrogate markers of gastrointestinal motility such as gastric residuals and time to FEFs to assess the efficacy of erythromycin as a prokinetic agent. Not surprisingly the results are conflicting. This systematic review aimed to study the efficacy and safety of erythromycin as a prokinetic agent in preterm neonates.

METHODS

Trials in preterm neonates with gestation ⩽37 weeks were considered eligible for inclusion. The primary outcome of interest was the time taken to reach FEFs of 150 ml/kg/day. The secondary outcomes of interest included the following: (a) erythromycin related adverse effects such as diarrhoea, cardiac arrhythmias, potentiation of theophylline toxicity, late onset infections, and hypertrophic pyloric stenosis; (b) duration of total parenteral nutrition; (c) duration of hospital stay; (d) weight at discharge from hospital; (e) incidence of necrotising enterocolitis of stage 2 or worse.^22,23 No restrictions were applied on the dose (low: 3–12 mg/kg/day; antimicrobial: ⩾12 mg/kg/6–8 hours) or route (oral or intravenous) or mode (prophylactic or rescue) of administration. The Cochrane Central Register of Controlled Trials (Central, The Cochrane Library, Issue 4, 2002), Medline, Embase, Cinahl databases, and proceedings of the Pediatric Academic Societies (published in Pediatric Research from 1980), European Society for Pediatric Research (ESPR) were searched in December 2003 and again in June 2004. Proceedings of the first and the second World Congress of Pediatric Gastroenterology, Hepatology, and Nutrition were also checked. The reference lists of identified randomised controlled trials, and personal files were searched. No language restriction was applied. The following key words were used: enteral, erythromycin feeding, neonates, and infants. Authors were contacted for additional specific data for meta-analysis or clarification of data. Data were independently extracted by the first two investigators and cross checked by all investigators to avoid any errors. Any inconsistencies were sorted out after discussions and agreement. Assessment of study quality was based on the guidelines of the Cochrane Neonatal Review Group. Low and antimicrobial doses of erythromycin have been known to have different effects on the gastrointestinal motility. The clinical implications of the prophylactic versus rescue approach are different and especially important, and the definition and significance of signs of “feed intolerance” are not clear. A subgroup analysis was therefore preplanned on the basis of the dose and mode of administration of erythromycin.

RESULTS

A total of seven studies involving 359 neonates were found eligible for inclusion in the analysis.^24–30Tables 1–6 show the characteristics and quality assessment of these studies. Two of the three studies involving the prophylactic approach (n  =  192) used the antimicrobial dose, whereas the third used the low dose of erythromycin. Three of the four studies involving the rescue approach (n  =  167) used the low dose, whereas the fourth used the antimicrobial dose of erythromycin. The doses, routes and modes of administration, and durations of erythromycin treatment as well as the feeding protocols and definition of feed intolerance varied in these trials (tables 5 and 6) Specific data required for meta-analysis (raw data/mean (SD)) were either inadequate or not available from certain authors.^25,26,29 In one case, the authors could not provide the data, as their manuscript was not yet published.^30,31 There was no reply from authors in one case.²⁷ Meta-analysis thus could not be performed.

DISCUSSION

The results of our systematic review indicate considerable variation in the use of erythromycin as a prokinetic in preterm neonates, making it difficult to reach any clear conclusions or recommendations. The trial designs, results, and authors’ conclusions reflect the poorly understood prokinetic actions of erythromycin under different conditions of use—for example, low versus antimicrobial dose—as well as the urgent need for preventing/minimising feed intolerance in preterm neonates. The issue of erythromycin related adverse effects also cannot be addressed adequately given the small sample sizes and insufficient data on long term follow up.

The dose (low versus antimicrobial) and route (intragastric versus intravenous) and mode (prophylactic versus rescue) of administration of erythromycin as well as the fasting versus fed state of the neonates has to be considered carefully before interpreting the results of such studies. The gestational and postnatal age³² at exposure to erythromycin, the type of feeds (breast milk versus formula),^33,34 method of feeding (intermittent boluses versus continuous infusion), the rate of bolus feed infusion,³⁵ and the definition and severity of feed intolerance³⁶ are also equally important.

The choice between prophylactic and rescue approach seems to be clear. Given that mild to moderate feed intolerance is almost universal in high risk preterm, especially extremely low birthweight neonates, prophylactic use of erythromycin will expose almost this entire population of neonates to a drug with adverse effects such as hypertrophic pyloric stenosis and cardiac arrest.^37–42 Researchers have already suggested that neonatologists should limit the use of erythromycin as a rescue rather than prophylactic treatment.^28,29

The oral route may be preferred for drug administration because all erythromycin related, life threatening, and fatal cardiac complications have been associated with the intravenous route. The safety dose limit of intravenous erythromycin in preterm neonates has also not been determined yet.⁴³ The choice between antimicrobial and low dose is, however, difficult. Exposure to antimicrobial doses for ⩾14 days in neonates up to 2 weeks old has been associated with a 10-fold increase in the risk of hypertrophic pyloric stenosis.³⁸ Oral administration of a low dose on the other hand may result in inadequate serum concentrations with no prokinetic effects. Studies comparing serum concentrations and gastrointestinal motor effects after different doses and routes of administration of erythromycin may thus be useful in preterm neonates with “significant” feed intolerance beyond 2 weeks of life. Careful attention to the postnatal age and type, mode, and method of feeding is needed in the design of such studies.

Erythromycin is a competitive motilin receptor agonist with high affinity for motilin receptors and mimics the effects of motilin on the proximal gastrointestinal tract.^44,45 It is also known to enhance the release of endogenous motilin, and stimulate cholinergic nerves of the gut at both preganglionic and postganglionic levels, leading to the release of calcium, which initiates contractions of the gut smooth muscle that are responsible for the forward propulsion of nutrients.^44–51 Understanding the two basic patterns (fasting versus fed) of small intestinal motor activity and the influence of gestational age on it is necessary before using erythromycin as a prokinetic agent in preterm neonates.^52–54 A cyclical pattern of antral and intestinal contractile activity, called migrating motor complex (MMC), progresses from the antrum to the ileum during fasting. It has been described as the housekeeper of the bowel, removing the indigestible solids and bacteria from the upper gastrointestinal tract. In adult humans, these MMCs consist of four phases. Phase I consists of no contractile activity, which is sequentially interrupted by periods of irregular contractions (phase II), followed by regular contractions at a rate of three per minute in the atrium or 12 per minute in the duodenum (phase III), and a brief period of irregular contractions before the return of quiescence (phase IV). The cycle repeats every 45–180 minutes. It is important to know that the MMC is interrupted by a meal, with an indistinct pattern of irregular contractions appearing until hours after the meal.^52–54 By 34 weeks these complexes are of variable length, with clear intervals and being increasingly propagated. The mature MMC at this stage has a periodicity of 20–40 minutes and is interrupted by feeding.^55,56 Ittmann et al⁵⁷ have shown that fasting antral motor activity per se is comparable in preterm and term neonates and that the degree of antroduodenal coordination improves simultaneously towards term. Except for the considerably shorter periodicity, the MMC has adult characteristics at term gestation.

Despite considerable research, the important issue of fasting versus fed status of the neonates in question remains neglected. As discussed above, MMCs are a property only of the fasting state, and are interrupted by a meal. If the prokinetic effects of erythromycin are indeed related primarily to induction of MMCs, they are expected to be unpredictable in neonates on intermittent bolus feeds given the difficulty in separating fasting versus fed states in relation to drug administration. The issue is even more complicated in the presence of continuous feeds where there is effectively no “fasting” state. The rate of infusion of intermittent bolus feeds may also be related to feed tolerance. Duodenal motor responses in preterm neonates fed by slow intragastric infusion over 120 minutes are more like those in adults, and their gastric contents are emptied faster and more completely than when they are fed with a rapid bolus.³⁵ This approach may provide clinical benefits by improving gastric hypomotility residuals. The problem of lower intestinal, including colonic, hypomotility, however, will not be solved.³⁶

The prokinetic effects of erythromycin are reported to be dose dependent.^12,13,51 At antimicrobially ineffective, intravenous low doses (1–3 mg/kg), premature MMCs are induced, whereas at higher (10 mg/kg) doses this effect is lost because of mechanisms that are poorly understood. The presence of two different types of motilin receptors may explain the difference in responses to a low or antimicrobial dose of erythromycin.^58–60 The “neural” receptor is stimulated by a low dose that triggers the MMCs.⁵⁹ The “muscle” receptor on the other hand is stimulated by higher doses of erythromycin triggering antral contractions and inhibiting MMCs.⁶⁰ Beneficial prokinetic effects, however, have been reported at therapeutic or high doses of intravenous and oral erythromycin. Researchers have proposed that effects of the drug are influenced by the nature of the underlying disorder.^59–64 These findings, however, may not necessarily apply to preterm neonates as these studies were conducted in either full term neonates and children or adults.

The effect of erythromycin on gastric antral motility/emptying may explain the clinical benefits in very preterm neonates in whom MMCs are expected to be either immature or absent. Tomomasa et al¹⁵ have studied this issue in healthy preterm neonates (gestation 23–30 weeks, birth weight 825–1408 g) who were 6–31 days old. They infused 0.75 mg/kg erythromycin intravenously for 15 minutes and compared gastric and duodenal contractions for 30 minutes between before and after the initiation of erythromycin infusion. The migrating complex was not present in these neonates and was not induced by erythromycin. However, erythromycin significantly increased non-propagating antral clusters of contractions in all six neonates. The antral motility index increased fourfold, indicating the presence of functioning motilin receptors in preterm neonates. Tomomasa et al⁶⁵ have also studied the effect of oral erythromycin (10 mg v 3 mg) in neonates (postnatal age 5–100 days, weight 2.5–4 kg) with mild to moderate gastric emptying delay while on ⩾40 ml of milk feeds. After aspirating and discarding gastric residuals, neonates were given either erythromycin (10 mg (n  =  8), 3 mg (n  =  6)) or an equal volume of distilled water on the first day during measurements. The order of the drugs was reversed on the second day of the study. Oral erythromycin at a dose of 10 mg but not at 3 mg significantly increased gastric emptying compared with control measurements. This effect was not explained by an increase in the number of antral contractions. The researchers proposed that either an increase in the tone of the proximal stomach or a decrease in the pyloric tone may have been responsible for such results. Findings such as these may be the basis of the improvement in feed intolerance noted by researchers using the “rescue-oral antimicrobial dose” approach.²⁷ Jadcherla et al,³² on the other hand, have reported that intragastric low dose erythromycin (0.75–3 mg/kg) failed to induce phase III MMCs in neonates <31 weeks gestation. It, however, induced them in a dose dependent manner in neonates with gestational age ⩾32 weeks (p<0.05). Erythromycin significantly increased the amplitude and frequency of antral contractions in term neonates and significantly increased the duodenal contraction amplitude in older preterm and term neonates, but these effects were absent from younger preterm neonates. They concluded that early use of erythromycin as a prokinetic agent may not be useful in very preterm neonates, partially useful in older preterm neonates, and useful in full term neonates.⁵³ ElHennawy et al²⁹ have also reported that low dose (1.5 mg/kg) intragastric erythromycin did not improve gastrointestinal motor function or feed tolerance in the short or the long term in preterm neonates. It is quite possible that such negative results are related to the low intragastric doses with resultant inadequate serum erythromycin concentrations for any prokinetic effect to occur. However, differences in neonatal characteristics including gestational/postnatal age and feeding type/methods may also be related.

In summary, current data indicate that the use of erythromycin should be reserved for only a very small subset of high risk preterm neonates with persistent/severe feed intolerance while limiting the duration of exposure and ensuring long term follow up. The definition and interpretation of the manifestations of “feed intolerance” are not clear.³⁶ However, the need for prokinetic agents in preterm neonates is probably influenced considerably by the degree of our tolerance to feed intolerance. The recent decline in the use of prokinetics such as erythromycin in preterm neonates indicates tolerance towards feed intolerance and/or an increased awareness of the drug related side effects.⁶⁶ Development of newer compounds with better safety profiles may mean that the chapter on macrolides as prokinetics in preterm neonates may not be closed yet.^45,67