1. Objectives

To evaluate the effect of avoidance of bottle feeds during the transition to breastfeeding in preterm infants on the establishment of successful breastfeeding, and to evaluate the safety of alternatives to bottle feeds

2. How studies were identified

The following databases were searched to July 2016:

• CENTRAL (The Cochrane Library 2016, Issue 3)
• MEDLINE
• CINAHL
• EMBASE
• ClinicalTrials.gov
• WHO International Trials Registry and Platform
• ISRCTN Registry

In addition, reference lists of published trials were handsearched

3. Criteria for including studies in the review

3.1 Study type

Randomized and quasi-randomized controlled trials

3.2 Study participants

Preterm infants (<37 weeks’ gestation) whose mothers intended to breastfeed and had not yet used any alternatives to breastfeeding that required the infant to suck (e.g., bottle)

3.3 Interventions

Complete avoidance of bottles during the establishment of breastfeeding was compared to a control of supplementing breast feeds with bottle feeds

(Instead of bottles, alternative feeding devices were used in the intervention group for supplementing breast feeds, including gavage tubes, cups, spoons, droppers, finger feeding, paladai or others)

3.4 Primary outcomes

• Full breastfeeding on discharge home, and at three and six months post-discharge, compared with no breastfeeding or partial breastfeeding
• Any breastfeeding on discharge home, and at three and six months post-discharge, compared with no breastfeeding

Secondary outcomes included time to reach full sucking feeds, average daily weight gain, length of hospital stay, duration of supplementary feed, volume of supplementary feed, cardio-respiratory stability, episodes of choking or gagging per feed, milk aspiration, episodes of infection per infant, and parental and health professional satisfaction with feeding method

4. Main results

4.1 Included studies

Seven trials enrolling 1152 infants were included in this review:

• Four trials enrolled infants from 32 to <34 or 35 weeks’ gestation, two trials enrolled infants from 32 to 37 weeks’ gestation, and one trial used a birth weight criterion of 1000 to 2500 grams; the mean gestational age overall was 32 weeks
• Five trials used cups as an alternative feeding device when supplementary feeds were required, and one trial used a tube feeding approach. One other trial used a valved teat which was similar in action to a breast in that milk only flowed when the infant created a vacuum by sucking; while the novel teat was attached to a bottle, the review authors deemed this to be acceptable as it was similar to breast feeding in action
• In all included studies, bottle feeds or alternative device feeds were given only when the mother was not available to breastfeed or if additional milk was required. A nasogastric tube for additional feeds remained in situ for both intervention and control groups in two studies. Dummy/pacifier use varied across the studies
• Sucking feeds were initiated according to hospital policy or at the discretion of healthcare personnel

4.2 Study settings

• Australia (2 trials), Brazil, Turkey, the United Kingdom of Great Britain and Northern Ireland (2 trials), and the United States of America
• Studies were all undertaken in neonatal units, and two studies in Australia and Turkey were multicentre trials

4.3 Study settings

How the data were analysed
Breastfeeding plus supplemental feeds given by device other than a bottle was compared with breastfeeding plus supplemental feeds given by bottle. Fixed effect meta-analysis was used to produce pooled relative risks (RR) with 95% confidence intervals (CI) for categorical data, and mean differences (MD) with 95% CI for continuous data. Where substantial heterogeneity was detected (I²>50%), random effects models were used. Potential sources of heterogeneity were examined in subgroup analyses by alternative feeding device type (cup, gavage tube, novel teat) and by country setting (low- and middle-income countries versus high-income countries)

Results
Breastfeeding with supplemental feeds by device other than bottle versus breastfeeding with supplemental feeds by bottle
Full breastfeeding at discharge
Full breastfeeding at discharge was increased by 47% among infants who were not bottle-fed compared with those who were (RR 1.47, 95% CI [1.19 to 1.80], p=0.0003; I²=52%; 6 studies/1074 infants). This finding remained statistically significant in studies using both cups and gavage tubes as alternative feeding devices (both p≤0.0014), but was not significant in the single trial using a novel teat (p=0.8).

Full breastfeeding at three and six months post-discharge
In pooled analysis of four trials including 986 infants, the likelihood of full breastfeeding at three months post-discharge was statistically significantly increased among infants given additional feeds via an alternative feeding device in comparison with a bottle (RR 1.56, 95% CI [1.37 to 1.78], p<0.00001; I²=37%). This finding remained statistically significant in subgroup analyses of either cups or gavage tubes (both p≤0.0054), but not in subgroup analysis including the trial using a novel teat (p=0.67). At six months post-discharge, the overall likelihood of full breastfeeding was increased by 64% among those infants who received additional feeds via gavage tube or cup (RR 1.64, 95% CI [1.14 to 2.36], p=0.0071; I²=52%), and the effect of both interventions remained statistically significant in subgroup analyses (both p≤0.0059).

Any breastfeeding at discharge
Overall, the likelihood of any breastfeeding at discharge was increased by 11% among infants who were not bottle-fed (RR 1.11, 95% CI [1.06 to 1.16], p=0.000039; I²=0%; 6 studies/1138 infants), although this became non-significant in subgroup analysis when restricted to those fed using a novel teat (RR 1.12, 95% CI [0.95 to 1.33], 1 trial/97 infants).

Any breastfeeding at three and six months post-discharge
In pooled analysis, the rate of any breastfeeding was statistically significantly increased with bottle avoidance at both three months (RR 1.31, 95% CI [1.01 to 1.71], p=0.041; I²=73%; 5 trials/1063 infants) and six months (RR 1.25, 95% CI [1.10 to 1.41], p=0.00051; I²=50%; 3 trials/886 infants) post-discharge. Results became non-significant at three months post-discharge when restricted to studies using cups or a novel teat as an alternative feeding device (both p≥0.20).

Additional outcomes
The number of days to reach full sucking feeds was not significantly different between infants fed with alternative devices and those fed with bottles (MD 2.6 days, 95% CI [-7.2 to 12.3], p=0.31; I²=73%; 3 trials/429 infants), and nor was the length of hospital stay (MD 2.3 days, 95% CI [-3.4 to7.9], p=0.43; I²=73%; 4 trials/1004 infants). Limited evidence from the two trials assessing cardio-respiratory stability suggested improved stability with avoidance of bottles, although the breastfeeding plus bottle group had significantly more episodes of apnoea and bradycardia that required stimulation in one trial (p=0.0001). Daily weight gain, duration of supplementary feeds, and number of episodes of infection were not different between treatment groups. None of the studies reported on volume of supplementary feeds, or episodes of choking or gagging, and no episodes of milk aspiration occurred in the three studies that reported this outcome. In one trial there was a high rate of non-compliance amongst the group assigned to supplemental feeds by cup (56%), which was commonly cited as being due to mother, staff, or infant dissatisfaction.

Additional analyses
Subgroup analyses including the two trials from upper-middle-income countries (Turkey and Brazil) did not produce any meaningfully different results.

5. Additional author observations*

Although blinding of participants was not possible, allocation concealment was adequate in six of the seven included studies. The study that employed gavage tube feeding was of poor quality, and therefore results from this trial should be interpreted with caution. Using GRADE criteria, the quality of the evidence for full breastfeeding outcomes was judged to be low or moderate, and for any breastfeeding, very low or moderate. Evidence for the outcomes length of hospital stay and episodes of infection was rated as very low and moderate, respectively. Downgrading in all cases was performed due to risk of attrition bias and moderate to high heterogeneity.

Overall, avoidance of bottle feeding increased the likelihood of breastfeeding on discharge and at three and six months post-discharge. Use of the novel teat showed no benefit on breastfeeding rates in subgroup analyses, and insufficient evidence was found to support the use of gavage tube feeds.

Further studies of high methodological quality are required to investigate the effects of supplementing breast feeds with cup and tube feeds on length of hospital stay; weight gain; rates of breastfeeding; and parental, staff, and infant satisfaction with feeding.