Types of studies

Types of participants

Types of interventions

Types of outcome measures

The standard search strategy of the Cochrane Neonatal Group was used. Searches were made of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 1, 2009), MEDLINE (1966 - Feb 2009), EMBASE (1980 - Feb 2009), and CINAHL (1982 - Feb 2009) [via OVID] using the following text words and MeSH terms: [Infant, Newborn OR Infant, Premature OR Infant, Low Birth Weight OR infan* OR neonat*] AND "Infant-Nutrition"/ all subheadings OR Infant Formula OR milk OR formula OR trophic feeding OR minimal enteral nutrition OR gut priming ]. The search outputs were limited with the relevant search filters for clinical trials. No language restriction was applied.

References in previous reviews and studies were examined. The abstracts presented at the Society for Pediatric Research and European Society for Pediatric Research between 1990 and 2008 were searched. Trials reported only as abstracts were eligible if sufficient information was available from the report, or from contact with the authors, to fulfil the inclusion criteria. The meta-Register of clinical trials (http://www.controlled-trials.com/mrct/search.html) web site was searched for completed or ongoing trials.

1. The title and abstract of all studies identified by the above search strategy were screened and the full articles for all potentially relevant trials obtained. The full text of any potentially eligible reports was reassessed and those studies that did not meet all of the inclusion criteria were excluded. Any disagreements were discussed until consensus was achieved.
2. The criteria and standard methods of the Cochrane Neonatal Review Group were used to independently assess the methodological quality of any included trials in terms of allocation concealment, blinding of parents or caregivers and assessors to the intervention, and completeness of assessment in all randomised individuals. Additional information from the trial authors was requested to clarify methodology and results as necessary.
3. A data collection form was used to aid extraction of relevant information from each included study. Each review author extracted the data separately. Any disagreements were discussed until consensus was achieved. If data from the trial reports were insufficient, the trialists were contacted for further information.
4. Meta-analyses were performed using the fixed effects model. Relative risk and risk difference were calculated for dichotomous data and weighted mean difference for continuous data, with respective 95% confidence intervals (CI). The number needed to treat was determined for a statistically significant reduction in the risk difference. The treatment effects of individual trials and heterogeneity between trial results was examined by inspecting the forest plots. The impact of heterogeneity in any meta-analysis was assessed using the I² statistic. If statistical heterogeneity was noted, the possible causes (for example, differences in study quality, participants, intervention regimens, or outcome assessments) were explored using post-hoc subgroup analyses.

Sixteen articles identified by the above search strategy were reviewed. Seven articles were excluded (Berseth 1993; Berseth 1992; Berseth 2003; LaGamma 1985; Ostertag 1986; Slagle 1988; Weiler 2006; see table Characteristics of Exluded Studies).

Eight trials fulfilled the inclusion criteria (Becerra 1996; Dunn 1988; McClure 2000; Meetze 1992; Mosqueda 2008; Saenz de Pipaon 2003; Schanler 1999; Troche 1995; van Elburg 2004; see table, Characteristics of Included Studies).

Participants

The included studies were all undertaken since the late 1980s by investigators attached to neonatal units in Europe and North America. Most were small single-centre studies. 754 infants participated in total (range 29 - 190). Most participants were appropriate-for-gestational age VLBW infants or very preterm infants receiving standard intensive care interventions such as mechanical ventilation and parenteral nutrition. In van Elburg 2004, participants were infants of birth weight less than 2000 g who were small for gestational age (below 10th percentile for birth weight). We included this study because the vast majority (> 80%) of participating infants were of VLBW. Most of the other trials specifically excluded infants who were small for gestational age at birth and infants with congenital anomalies, gastrointestinal problems, or neurological problems.

Interventions

Early trophic feeding was generally started within the first three days after birth and continued for varying durations; either until infants were judged to be clinically stable (for example following endotracheal extubation or removal of umbilical catheters) or for pre-defined intervals, generally between seven and ten days after birth. Feeding volumes ranged from about 12 - 24 ml/kg/day. In most trials, infants received either expressed breast milk or formula milk (diluted or full-strength) or a mixture of breast milk and formula. In two trials, infants received only formula milk as trophic feeds (Dunn 1988; Meetze 1992).

The controls were not fed milk and received no enteral nutrition for at least one week after birth. Infants in both comparison groups received standard parenteral nutrition during the trial period. In most trials, infants received intermittent bolus gastric feeding. In Schanler 1999, participating infants were also allocated to either bolus or continuous feeding using a factorial design.

Outcomes

Most trials assessed feed tolerance and short-term growth as primary outcomes. Growth parameters were reported in a variety of ways, most commonly time to regain birth weight and weight gain during the neonatal period (either as median and range or as mean and standard deviation).

Most reports also gave information on adverse outcomes including feed intolerance (variously defined), incidence of necrotising enterocolitis, and mortality.

None of the trials reported long-term growth and neurodevelopmental outcomes for surviving infants.

Quality assessments are included in the table, Characteristics of Included Studies. Most of the trials had some methodological weaknesses. In five trials it was unclear whether randomisation was blinded. Caregivers were not blinded to treatment group in any trial. Few trials undertook blinded assessments for any of the outcomes, and several of the trials did not include results for all infants randomised.

EARLY TROPHIC FEEDING VS. ENTERAL FASTING (COMPARISON 1)

PRIMARY OUTCOMES

Feed intolerance: time to establish full enteral feeding (Outcome 1.1: 8 trials).

Meta-analysis of six trials that reported data as mean and standard deviation did not detect a statistically significant effect: weighted mean difference -0.97 [95% CI -2.47, 0.53] days. There was evidence of statistical heterogeneity in the meta-analysis: I² = 74%.

Two trials reported median and range data. Neither detected a statistically significant difference: 32 days vs. 32 days (Mosqueda 2008); 13 days vs. 13 days (van Elburg 2004).

Necrotising enterocolitis (Outcome 1.2: 9 trials). Meta-analysis did not detect a statistically significant effect: typical relative risk 1.07 (95% CI 0.67, 1.70); typical risk difference 0.01 (95% CI -0.03, 0.05). There was no evidence of heterogeneity: I² = 0%.

SECONDARY OUTCOMES

Mortality (Outcome 1.3: 5 trials): Meta-analysis did not detect a statistically significant effect: relative risk 0.77 (95% confidence interval 0.46, 1.30); risk difference -0.03 (95% CI -0.09, 0.03).

Growth (Outcome 1.4: 7 trials): None of the trials reported a statistically significant difference in the time to regain birth weight. Meta-analysis of five trials with data as mean and standard deviation: weighted mean difference -0.01 [95% CI -0.96, 0.95] days.

Two trials reported median and range data. Neither detected a statistically significant difference: 13 days vs. 12 days (Mosqueda 2008); 11 days vs. 10 days (van Elburg 2004).

McClure 2000 reported that the average rate of weight gain and head circumference gain during the six weeks after birth was borderline significantly higher in infants who hade received trophic feeds:

• Weight: reported mean difference 130 (95% CI 1, 250) grams/week.
• Head circumference: reported mean difference 0.7 (95% CI 0.1, 1.3) cm/week

Mosqueda 2008 reported no statistically significant difference in rates of weight gain during the trial period: mean difference -7.3 (95% CI -19.2, 4.6) grams/week.

Saenz de Pipaon 2003 reported that the weight above birth weight attained by day 21 was not statistically significantly different (188 g vs. 190 g).

Troche 1995 reported that infants in the trophic feeding group had a greater increase in weight to day 30 (223 [standard deviation 125] vs. 95 [standard deviation 161] grams.

Long-term growth parameters were not reported by any of the trials.

Neurodevelopment: None of the trials assessed neurodevelopmental outcomes.

Incidence of invasive infection (Outcome 1.5: 2 trials): McClure 2000 reported that, on average, infants in the trophic feeds group had statistically significantly fewer episodes of "culture-confirmed sepsis" (0.5 vs. 1.2 in control group). Mosqueda 2008 did not find a statistically significant difference in the incidence of invasive infection: relative risk 1.51 (95% CI 0.73, 3.16); risk difference 0.11 (95% CI -0.08, 0.30).

Duration of phototherapy (Outcome 1.6; 3 trials): Meta-analysis did not detect a statistically significant effect: weighted mean difference 0.35 [95% CI -0.29, 0.99] days.

Duration of hospital stay (Outcome 1.7: 4 trials): Meta-analysis of three trials that reported data as mean and standard deviation did not detect a statistically significant effect: weighted mean difference -3.8 [95% CI -12.2, 4.5] days.

One trial that reported median and range data did not find a statistically significant difference: 81 days vs. 79.5 days (Mosqueda 2008).

Subgroup analyses

1. Exclusively formula milk-fed infants: In two trials, infants received only formula milk as trophic feeds (Dunn 1988; Meetze 1992). In the other trials, infants received either breast milk or formula milk or a mixture. Subgroup data were not available.
2. Infants at least partially fed with breast milk: Subgroup data were not available.
3. ELBW or extremely preterm infants: One trial restricted participation to ELBW infants (Mosqueda 2008). In the other trials, It is likely that less than one-third of all participants were ELBW or extremely preterm but subgroup data were not available.
4. Infants with intrauterine growth restriction or infants with AREDFV: In those trials where birth weight less than the 10th percentile was not an exclusion criterion, subgroup data were not available. One trial restricted participation to infants who were small for gestational age (birth weight less than 10th percentile for reference population) (van Elburg 2004).

Substantial clinical uncertainty remains about the effect of trophic feeding on clinically-important outcomes in VLBW infants, and particularly in ELBW infants or infants who are growth-restricted. Despite plausible rationale, the available trial data do not provide evidence of an effect on feed tolerance, growth, or development. Reassuringly, there is also no evidence that trophic feeding has adverse effects, particularly on the risk of necrotising enterocolitis.

Further randomised controlled trials are needed to determine how the timing of introduction and rate of progression of enteral feeds affects important clinical outcomes in VLBW infants. Undertaking trials of feeding interventions in this population is problematic (Tyson 2007). It is difficult to design a pragmatic trial that will ensure that caregivers and investigators are unaware of the allocated feeding regimen. A priori agreements on objective definitions of feed intolerance and indications for interruption of enteral feeding and for investigation of necrotising enterocolitis may help minimise the impact of this source of bias.

Future trials should be simple and pragmatic to ensure high levels of acceptance and participation. Trials should aim to ensure the participation of ELBW and extremely preterm infants as well as infants with evidence of compromised intrauterine growth so that subgroup analyses can be planned for these populations at high risk of necrotising enterocolitis and should aim to assess more objective outcomes, principally mortality and long-term growth and development.

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table

Risk of bias table