Article Text

Download PDFPDF

Infant-driven feeding in premature infants: a quality improvement project
  1. Abigail Wellington,
  2. Jeffrey M Perlman
  1. Department of Pediatrics, New York Presbyterian-Weill Cornell Medical College, New York, New York, USA
  1. Correspondence to Professor Jeffrey Perlman, Department of Pediatrics, Division of Newborn Medicine, New York Presbyterian Hospital, Weill Cornell Medical College, 525 East 68th St, Suite N-506, New York, NY 10065, USA; jmp2007{at}


Background Many neonatal units are adopting developmentally appropriate feeding practices such as cue-based or infant-driven feeding (IDF). There have been limited studies examining the clinical benefit of this approach.

Methods A quality improvement initiative was undertaken to introduce an IDF protocol for premature infants <34 weeks gestational age (GA). Data were abstracted to determine whether time to full feeds and time to discharge would be shortened when compared with traditional practitioner-driven feeding (PDF) approach. Baseline data on postmenstrual age (PMA) at first feed, full nipple feeds and at discharge prior to implementation were compared with data obtained after implementation of the IDF protocol. Infants were divided into three subgroups: <28, 28–316/7 and 32–336/7 weeks gestation. A questionnaire assessed provider's acceptance of the plan.

Results The PMA at full nipple feeds and at discharge was significantly lower in the IDF than PDF group. Infants <28 weeks GA in the IDF versus PDF group reached full nipple feeds 17 days sooner and were discharged 9 days earlier. Babies 28–316/7 weeks GA reached full nipple feeds 11 days sooner and were discharged 9 days earlier in the IDF versus PDF group. Babies 32–336/7 weeks GA reached full nipple feeds 3 days sooner and were discharged 3 days earlier in the IDF versus PDF group. Providers viewed the implementation of the plan favourably.

Conclusions The IDF approach was associated with significant reduction in time to full feeds and discharge, an effect that was most pronounced in infants >28 weeks GA. The downstream benefits included provider and parent satisfaction.

  • Neonatology
  • Nutrition
View Full Text

Statistics from

What is already known on this topic?

  • There is wide variation among neonatal units on when to introduce and how to advance nipple feeds in the premature infant.

  • In recent years, neonatal units have adopted a cue-based feeding strategy or infant-driven feeding (IDF) strategy to advance nipple feeds.

  • While considered to be a more ‘developmentally appropriate’ feeding method, there are little data to support whether IDF has measurable clinical benefits.

What this study adds?

  • A shift from the more traditional ‘practitioner-driven’ feeding plan to an ‘infant-driven’ plan requires a multidisciplinary approach.

  • In this study, IDF was associated with earlier attainment of full nipple feeds and earlier discharge in infants <34 weeks gestational age.

  • IDF may enhance both parent and provider satisfaction.

Recently, there has been a shift towards adopting developmentally supportive practices in neonatal intensive care units (NICUs). Examples include kangaroo care, lighting for circadian rhythms and cue-based or infant-driven feeding (IDF).1–6 While IDF practices are being adopted by many NICUs, there are limited studies examining clinical benefits to the infant when compared with the more traditional, practitioner-driven feeding (PDF) method.7 ,8

Preterm infants born at <35 weeks gestational age (GA) are often unable to safely and/or effectively nipple feed (NF) by bottle or breast. They may have an uncoordinated suck, swallow and breathing pattern, autonomic instability and less organised sleep–wake characteristics.9 These factors may predispose to desaturation or bradycardic events and/or to infants who are not awake and alert enough to NF. The optimal timing for introduction and progression of NF varies widely. Feeding practices are often based on custom rather than evidence. In a national survey of NICUs, >50% of respondents had no formal criteria regarding when to start NF. Physicians were cited as the most important factor in decision making and 75% of respondents placed most emphasis on GA and weight.10

The attainment of full NF is often one of the last milestones reached by a premature infant before discharge.11 Determining factors that reduce the time to full NF may ultimately allow for an earlier discharge. We hypothesised that using an IDF plan rather than the PDF approach would result in earlier attainment of full NF and earlier discharge. An IDF practice plan (PP) was introduced into the NICU in premature infants ≤34 weeks GA to determine whether (1) using infant feeding cues would shorten the time to full NF and lead to earlier discharge and (2) whether the use of a feeding assessment tool would identify infant feeding readiness, improve quality of feedings and enhance consistency of feeds among providers.


This was a quality improvement (QI) initiative involving premature infants ≤34 weeks GA admitted to the NICU at New York Presbyterian Hospital, Weill Cornell Medical College (WCMC), to determine whether an IDF method would enhance attainment of full NF. There were three phases to this initiative. The first was a baseline period (September 2010–December 2011), the second was a development and phasing in of the PP (July 2012–May 2013) and third was implementation of the PP (June 2013–May 2014). The Institutional Review Board of WCMC approved the QI and collection of data.

A retrospective chart review of all infants <34 weeks GA either inborn or transferred within the first week of life and discharged home on full NF was undertaken to establish baseline outcome data when using the PDF approach. The PDF method was based on physician or nurse practitioner orders during daily rounds. NF would start at one attempt per day until a full feed was taken, followed by an increase to two attempts, then three attempts per day and so on. This approach did not incorporate feeding cues from the baby. It was outcome oriented whereby a ‘good’ feeding was equated with a finished bottle.

Phase 2 included the development and introduction of the IDF method. With the IDF approach, the infant determines when and how much to NF with the remainder provided via gavage tube. Emphasis is placed on quality of feeding rather than quantity ingested. A previously published, but modified ‘Premature Infant Feeding Assessment’ flowsheet was introduced, which allows the bedside nurse to evaluate an infant prior to each feeding in order to determine nipple feeding readiness12 (figure 1). The flowsheet consists of two separate assessment scales. The first is a ‘Readiness Score’ (range 1–5) whereby a nurse evaluates an infant for signs of nipple readiness every 3 h as part of care. The nurse initiates charting Readiness Scores at 32 weeks postmenstrual age (PMA) but continues to gavage feed until an infant is at least 33 weeks PMA and is scoring 1–2 s on the Readiness Scale for at least half the day for 1–2 days. Following a NF, the nurse documents a ‘Quality’ Score (range 1–5). Infants feeding safely and effectively (1–3) are fed for a maximum of 20–25 min. The remaining volume, if necessary is gavage fed. If a baby initially nipples well then fatigues, or became uncoordinated, the feeding is stopped and the remainder is gavage feed. The amount of milk taken by mouth and/or by gavage as well as ‘Caregiver Techniques’ (eg, side lying, pacing, oral stimulation) are also documented.

Figure 1

Premature Infant Feeding Assessment flowsheet (adapted from Ludwig and Waitzman12). Note the three boxes to the assessment: Readiness Score (scale 1–5), Quality Score (scale 1–5) and Caregiver Techniques. The top timed box includes the three assessments as well the amount and duration of the different feeding techniques as well as events during feeds.

Several interdisciplinary meetings involving neonatal attendings, nurses, nurse practitioners and feeding therapists were held to facilitate the introduction of the feeding assessment flowsheet to the bedside. The bedside nurses were introduced to IDF and were given several months to evaluate and adjust prior to implementation. Small-group in-services involving all nurses were subsequently conducted as a ‘refresher’ reinforcement prior to implementation. Parents were also educated about the flowsheet and infant cues and were encouraged to feed their babies. Only the nurses filled out the flowsheet. This phasing-in period was of approximately 6 months duration. There was no overlap in time between comparison infants (PDF group) and the start of IDF education.

During the first 6 months following implementation, the primary investigator (AW) circulated the unit daily to assure that the flowsheets were completed correctly and to optimise implementation. At the conclusion, all nurses were given a questionnaire to determine their opinions on using the IDF approach (see table 1).

Table 1

Questionnaire given to the providers

For quality control, 60 random samples of flowsheets were assessed; 20 from the beginning, middle and end of the study period to determine both compliance in filling out the flowsheets and in appropriately feeding the baby based on the Readiness Scores.

Data retrieval and analysis

The following data were retrieved at baseline and following implementation: GA, birth weight (BW), race, sex, PMA at first NF, full NF and discharge. Additional clinical characteristics included delivery mode, antenatal steroid use, surfactant use, need for mechanical ventilation and patent ductus arteriosus (PDA) ligation. Exclusion criteria included: infants subsequently discharged home on oxygen, requiring gastrostomy tube, intraventricular haemorrhage >grade II,13 necrotising enterocolitis >stage 2,14 congenital heart disease or any condition that interfered with an infant's ability to feed, for example, cleft palate.

Subgroup analysis was undertaken for groups with GA <28 weeks, 28–316/7 weeks and 32–336/7 weeks. The primary outcomes were PMA at first NF, full NF and at discharge. Analysis was performed using SPSS. Two-tailed t tests were used for continuous data. Correction for potential confounders was performed using multivariable linear regression analysis. All values are presented as mean±SD.


General characteristics

Three hundred and nine charts were analysed that after exclusions included 153 babies in the PDF and 101 infants in the IDF group (see figure 2). Subgroup categories were as follows: <28 weeks GA: 10 infants in the PDF and 13 babies in the IDF group; 28–316/7 weeks GA: 60 and 31 infants in each group, respectively, and 32–336/7 weeks GA: 83 and 57 infants in each group, respectively. For infants <28 weeks GA, both GA and BW were significantly lower in the PDF than the IDF cohort. For infants 28–316/7 weeks GA, BW did not differ between cohorts, but GA was 4 days less in the PDF versus IDF cohort. For infants ≥32 weeks GA, the BW did not differ between cohorts, but the GA was 1 day higher in the PDF versus IDF cohort. There were no differences between cohorts for race, delivery mode, antenatal steroid use, surfactant administration or mechanical ventilation. PDA ligation in the 28-week GA subgroup to 316/7-week GA subgroup was different (p=0.045) (table 2).

Table 2

Baseline characteristics in the subgroups <28 weeks, 28–32 weeks and >32 weeks for the physician-driven feeding (PDF) and infant-driven feeding (IDF) methods

Figure 2

Schema depicting the subjects included and excluded during the practitioner-driven feeding (PDF) method as well as during the infant-driven feeding (IDF) method. Note the breakdown into groups <28 weeks, 28–31 6/7 weeks and 32–33 6/7 weeks gestational age.

PMA at first NF and full NF

There was no difference in PMA at first NF between any subgroups. However, there was a significant difference in PMA at full NF across all subgroups (table 3). For infants <28 weeks GA, the IDF versus PDF group reached full NF 17 days sooner (374/7 vs 40 weeks; p=0.03). For infants 28–316/7 weeks GA, the IDF versus PDF group reached full NF 11 days sooner (354/7 vs 371/7 weeks; p<0.001, respectively) and for infants ≥32 weeks GA, the IDF versus PDF group reached full NF 3 days sooner (354/7 vs 351/7 weeks; p=0.04).

Table 3

Postmenstrual age (PMA) at initial nipple, PMA at full nipple feeds and PMA at discharge in the subgroups <28 weeks, 28–316/7 weeks and >32 weeks for the physician-driven feeding (PDF) and infant-driven feeding (IDF) cohorts

PMA at discharge

For infants <28 weeks GA, there was no difference in PMA at discharge between the IDF versus PDF group (413/7 vs 394/7 weeks; p=0.10). For infants 28–316/7 weeks GA, the IDF versus PDF group were discharged 9 days earlier (366/7 vs 381/7 weeks; p<0.001). For infants ≥32 weeks GA, the IDF versus PDF group were discharged 3 days earlier (36 weeks vs 363/7 weeks; p=0.048) (see table 3).

Controlling for confounders

Multivariable linear regression analysis was undertaken to assess the significance of feeding plan alone when controlling for BW, GA and sex between cohorts. For GA <28 weeks, the IDF plan did not remain significant for PMA at full NF or at discharge. For GA 28–316/7 weeks, the IDF plan remained significantly associated with earlier attainment of full NF and earlier discharge (p<0.01 for both outcomes). For GA ≥32 weeks, the IDF plan remained a significant influence on both PMA at full feeds and at discharge (p≤0.01 for both outcomes).


Thirty-eight questionnaires were filled out by nurses (n=33), attendings (n=4) and a feeding therapist (n=1). The responses were favourable: 37/38 (97%) were positive about the IDF approach; 35/38 (92%) indicated that the flowsheet helped in understanding an infant's feeding ability or problem; 30/33 (91%) of nurses indicated using the flowsheet was not a heavy time burden and 35/38 (92%) favoured continuing the IDF plan rather than reverting to the PDF method.

Quality control and quality assurance

In the early period following implementation, 75% of forms were filled out completely and correct actions were taken 80% of the time. In the middle period, 75% were filled out completely and correct actions were taken 95% of the time. At the end, 85% of forms were filled out completely and correct actions were taken 95% of the time.


This study demonstrated that implementation of an IDF PP was associated with an earlier attainment of full NF and earlier discharge in premature infants <34 weeks GA. This effect was most pronounced in infants born at ≥28 weeks. The initial NF was attempted at the same PMA for both methods across all GA subgroups. However, infants in the IDF group reached full NF at a significantly earlier PMA.

The ability to achieve full NF sooner in the IDF group likely reflects the fact that the PDF method may have limited skill progression by not using infant cues to guide the initiation and progression of feeds. It may also be due to the increased opportunities to participate in positive feeding experiences, as infants in the IDF group were not limited in the frequency and/or amount of daily cue-based attempts. Conversely, there were a few infants in the IDF cohort who initiated NF at a later PMA than would have been predicted (>35–36 weeks). Interestingly, when these infants eventually showed readiness, they quickly progressed to full feeds and were discharged soon thereafter (data not shown). In the PDF approach, such infants would have been expected to begin NF at a predetermined PMA, and ‘pushed’ to do so, even if not showing readiness.

A literature review indicates great interest in establishing feeding protocols based on an infant's readiness or behavioural state, although few studies have evaluated the clinical value of using such an approach.4 ,5 ,15 McCain et al conducted a randomised study comparing a PDF with an experimental method (non-nutritive suck for 10 min prior to feeding and infant readiness assessment) and found that the experimental group reached full NF 5 days earlier than the control group.7 Kirk et al8 conducted a small before and after study comparing 23 control infants (PDF approach) with 28 experimental infants (feeds advanced based on readiness signs and adequate weight gain) and found that the experimental group reached full NF 6 days sooner than the PDF group. Our findings of a decrease in time to achieve full NF are consistent with both studies. However, both were limited by smaller sample sizes, were without subgroup GA categories and did not examine the impact on time to discharge. Several studies have shown achievement of full NF to be an important modulator of discharge.9 ,16 ,17 In our study, infants born between 28 and 32 weeks in the IDF cohort went home 9 days sooner, although they tended to be less sick (asymptomatic PDA, less surfactant administration), and infants born at ≥32 weeks were discharged 3 days sooner than those in the PDF group. The practical implications of decreased length of stay include minimising parent/infant separation, reduced exposure to iatrogenic infection and savings in healthcare dollars.

While adoption of the IDF method was ultimately successful, there were many challenges to implementing such a cultural change in the NICU. There was initial resistance among both nurses and physicians. Many nurses considered a flowsheet unnecessary to assess an infant's readiness and felt that it would be time consuming, while physicians questioned whether premature infants could reliably provide feeding cues. Ultimately, nurses reported that they spent less time feeding when using the IDF approach. The consistency of the flowsheets indicates that premature infants are able to exhibit signs of nipple readiness to guide the caregiver. The flowsheet was essential to standardising objective criteria for readiness, given a frequent change of bedside nursing assignment. Moreover, the flowsheet helped parents understand the complexities involved in an infant's ability to NF, which in turn enhanced understanding and satisfaction.

A major limitation of this study is that it was an observational, rather than a randomised trial. This design was chosen because it did not seem prudent to train nurses to feed infants based on cues, only to revert back to the PDF method upon randomisation. While controlling for confounding factors, there is always the possibility that some uncontrolled differences between groups, such as changes in practice over time, may have influenced feeding ability. Another limitation is the different sample sizes between cohorts and across subgroups. Infants born at <28 weeks were limited in numbers because of exclusion criteria. Moreover, this subgroup is likely to have significant feeding issues related to comorbidities.17 Although there were differences in GA in the larger infant groupings, the IDF remained a significant influence on time to achieve full NF and discharge when adjusting for confounders. Last, the introduction of a standardised feeding protocol in of itself could have contributed to earlier feeds regardless of whether the feeding plan was infant driven.


An IDF approach was associated with an earlier attainment of full NF and as a consequence, earlier discharge. This was achieved using a standardised approach, which provided a more objective, consistent and infant-guided assessment of readiness for NF. The cultural change in feeding practice required time, education and an interdisciplinary approach to achieve success. The benefits included increased provider and parent satisfaction as well as potential for reduced hospital costs.


View Abstract


  • Contributors AW: conceptualised and designed the study, oversaw the training and implementation of the feeding protocol, drafted the initial manuscript and approved the final manuscript as submitted. JMP: contributed to study design, aided in data analysis, reviewed and revised the manuscript and approved the final manuscript as submitted.

  • Competing interests None declared.

  • Ethical approval This study has been approved by the Institutional Review Board of WCMC.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.