Improving Time to Independent Oral Feeding to Expedite Hospital Discharge in Preterm Infants

The 120-bed, level IV NICU at The University of Alabama at Birmingham admits >1400 infants per year, of which up to 200 infants are born at <29 weeks’ GA. Before the initiative, the institutional review board deemed the initiative exempt, and a multidisciplinary team of neonatologists, speech therapists, nurse managers, nurses, and a parent representative was formed. Our global aim was to optimize oral feeding development in convalescent preterm infants, with key drivers depicted in Fig 1. Six months of baseline data (January 2019 to June 2019) were used for successive PDSA cycle comparisons. Infants were included if inborn between 25 0/7 and 32 6/7 weeks’ GA. Infants born at <25 weeks’ GA were excluded given a higher prevalence of bronchopulmonary dysplasia,⁷  for which respiratory support limits oral feeding before 33 weeks’ PMA. Infants who were transferred to another hospital, died before 48 weeks’ PMA, developed spontaneous intestinal perforation or necrotizing enterocolitis (NEC) stage ≥2,⁸  or had major congenital anomalies or genetic syndromes were excluded. Infants with spontaneous intestinal perforation or NEC were excluded because these infants were transferred to the surgical unit.

Oral feeding initiation before the initiative began with speech therapy assessment performed by postnatal day 5, with ongoing weekly assessments for oral feeding readiness. The frequency of oral feeding attempts depended on ongoing provider assessment, physiologic milestone achievement, and progress related to feeding coordination and endurance. To better characterize oral feeding progression before the initiative, the frequency of oral feeding attempts by PMA was analyzed across 20 patients between 29 and 32 weeks’ PMA during the first PDSA cycle before introduction of cue-based feeding. In this cohort, the median percentages of daily feeds in which oral feeding was attempted by postmenstrual week were 32 weeks (6%), 33 weeks (50%), 34 weeks (88%), 35 weeks (90%), and 36 weeks (100%).

Starting in June 2019, guidelines were established for respiratory support and status precluding oral feeding. Limited evidence is available to guide safe respiratory support criteria for oral feeding^9,10  with substantial practice variation¹¹ ; our consensus was to allow infants stable on ≤4 L high-flow nasal cannula with normal respiratory effort to orally feed. Given that infants on higher respiratory support (eg, continuous positive airway pressure) were not permitted to orally feed, respiratory support weaning criteria derived from randomized trials^12,13  included a positive-end expiratory pressure ≤5, a fraction of inspired oxygen <30%, and infrequent apneic events requiring stimulation.

The initial intervention targeted earlier oral feeding initiation in preterm infants and consisted of weekly huddles between members of the multidisciplinary team to (1) identify infants at or approaching 32 0/7 weeks’ PMA, (2) identify infants on respiratory support precluding oral feeding and meeting weaning criteria, and (3) facilitate speech therapy assessment of eligible infants guided by established respiratory support criteria and initiation of oral feeding. Infants were still included in analyses if on respiratory support precluding oral feeding at 33 weeks’ PMA.

Cue-based feeding, wherein an infant is assessed for signs of feeding readiness before an oral feeding attempt, was introduced during the second PDSA. In choosing a cue-based method, a commonly cited method in the literature included an established program entitled Infant Driven Feeding^2,14,15  that utilizes a readiness score to determine whether to orally feed an infant and a quality score to determine length of a feeding attempt. However, given the limited validation for this approach, a bedside, cue-based flowsheet was established (Supplemental Fig 5) documenting the following assessments: (1) infants’ respiratory support and effort, (2) infants’ responsiveness to care assessment, (3) interventions implemented during the feed (eg, side-lying position), and (4) the reason for which a feeding was discontinued. For infants of breastfeeding mothers, nurses performed an oral feeding readiness assessment before oral feeding initiation. An infant was considered responsive to care if any of the following were noted: rooting, flexion in arms and legs, awake status, or sucking on pacifier or hands. Each of these components was assessed for every feed via bedside logs, and review of previous interventions enabled nurses to better identify ongoing feeding techniques and personalize oral feeding strategies.¹⁶ 

After targeting oral feeding initiation and cue-based feeding, we identified a population of infants who had not achieved independent oral feeding by 36 weeks’ PMA and were not orally fed with every feed. After multidisciplinary discussions, it was decided to transition selected infants once 36 weeks’ PMA from cue-based feeding to practitioner-driven feeding, wherein infants attempted oral feedings with every feed. Infants were progressed to this feeding strategy if orally fed for at least 3 weeks with >50% of daily feedings attempted by mouth. In addition to this feeding strategy, a concurrent standardization of nasogastric tube withdrawal after reaching 120 mL/kg/day by mouth occurred during this PDSA. These determinations were integrated into ongoing weekly huddles between the multidisciplinary team.

Process measures for this initiative included (1) oral feeding initiation by 33 weeks’ PMA and (2) bedside oral feeding–log assessments. Day of oral feeding initiation was defined as the first day when an infant was fed either by nipple or by breast. Bedside oral feeding–log assessments were deemed compliant if >50% of documented feedings had cue-based feeding assessments recorded for the first 2 weeks after oral feeding initiation. Compliance rates were shared with nursing staff throughout the initiative. The PMA of independent oral feeding achievement was defined as the day on which the feeding tube was removed and not subsequently replaced. The PMA at discharge from the hospital was defined as the PMA associated with the date of discharge or 48 weeks’ PMA for infants still hospitalized at this postnatal age.

Our initiative’s aim was to decrease the PMA at discharge from the hospital by 1 week within 12 months and sustain this reduction for an additional 6 months. Other outcome measures included PMA at independent oral feeding achievement and extreme length of stay defined as continued hospitalization at 48 weeks’ PMA. The PMA of other physiologic milestones, including thermoregulation, home respiratory support, caffeine discontinuation, and apnea resolution, were also monitored. All-cause rehospitalization within 2 weeks after discharge was used as a balancing measure, as used in other initiatives to reduce length of stay.¹⁷ 

All process, outcome, and balancing measures were analyzed by using statistical process control charts. Given reported differences in the postnatal age of independent oral feeding by GA at birth,²  and the inclusion of infants born between 25 weeks and 28 6/7 weeks’ GA more likely to have clinical characteristics contributory to late oral feeding achievement,⁷  these measures were initially analyzed by two strata: (1) 25 0/7 to 28 6/7 and (2) 29 0/7 to 32 6/7 weeks’ GA. To combine these strata, consecutive infants born at 29 0/7 to 32 6/7 weeks’ GA were first plotted by date of birth in discrete denominators (n = 10). Infants born at the lower GA strata were then added to relevant date ranges based on the date of birth. The median GA and included number of infants were then plotted per monitoring period.

Specific process control charts per measure included XmR charts for the PMA at independent oral feedings and discharge, p-charts for initiation of oral feeds by 33 weeks’ PMA and percentage of infants receiving cue-based feeding by monitoring period, and t-charts for days between infant births with extreme length of stay and readmission. Standard control chart rules were used to identify special-cause variation during the action period, which corresponded to 8 consecutive points above or below the mean.¹⁸  Pareto charts quantifying each physiologic milestone’s contribution to the terminal barrier to discharge were constructed for each PDSA cycle. A milestone was classified as the terminal barrier to discharge if the milestone was the last achieved during hospitalization on the basis of the relevant day of physiologic maturity or clinical stability. More specifically, the respective date for the following milestones were used: apnea countdown completion, 48 hours after removal of a gastric feeding tube, 48 hours in an open crib, and home respiratory support achievement. Regarding practices of characterizing and managing apnea and bradycardic events, unit guidelines are to count a bradycardic event if requiring stimulation or if an event lasts ≥10 seconds and does not occur in context to a feed. Infants must be event free for 5 days before discharge.

Demographic and clinical characteristics were analyzed by PDSA cycle. Nominal data were analyzed by using χ² analysis, and continuous data were analyzed by using either one-way analysis of variance or Kruskal-Wallis test, with Dunn multiple comparison test after tests for normality.

A total of 614 infants were born between 25 and 32 weeks’ GA during the baseline, PDSA cycles, and sustainment period. Of these infants, 62 were excluded: 23 infants developed NEC or intestinal perforation or were transferred for other indication, 20 infants had major congenital anomalies, 13 infants died before 48 weeks’ PMA, and 6 infants had genetic conditions. The median GA of included infants was 30.1 (interquartile range [IQR] 28–31.9), with a median birth weight of 1310 (IQR 1014–1580). A total of 186 infants were born at <29 weeks’ GA. Baseline and clinical characteristics did not differ between the cycles except that more white infants were born during the second PDSA. There were no differences in morbidities by study phase (Table 1).

During the baseline period, the rate of oral feeding initiation before 33 weeks’ PMA was 47%, which increased to 80% after the first PDSA targeting oral feeding initiation through weekly huddles (Fig 2). Cue-based feeding assessments were 90% throughout the initiative (Supplemental Fig 6). The PMA at discharge decreased from 38.8 weeks to 37.7 weeks’ PMA in June 2019, with 8 consecutive points below the mean, corresponding with our process measure of increased oral feeding initiation before 33 weeks’ PMA (Fig 3). The frequency of extreme length of stay decreased after the third PDSA implementing practitioner-driven feeds at 36 weeks’ PMA (Supplemental Fig 7A). The PMA at independent oral feeding decreased from 37.4 to 36.5 weeks’ PMA after the first PDSA (Fig 4). During the baseline period, oral feeding was the most frequent terminal barrier to discharge in 56.3% of hospitalizations and decreased to 33.7% during PDSA 3. Apnea resolution became the predominant terminal barrier to discharge during PDSA 2 (Supplemental Fig 8). All-cause readmissions did not increase during the initiative (Supplemental Fig 7B).

In this multidisciplinary quality improvement initiative on oral feeding in preterm infants born between 25 and 32 weeks’ GA, the PMA at discharge decreased by 1 week from 38.8 to 37.7 weeks’ PMA. In addition, the PMA at which infants achieved independent oral feeding decreased from 37.4 to 36.5 weeks’ PMA. This decrease in the postnatal age of oral feeding achievement and discharge were associated with the first PDSA cycle during which weekly huddles performed between speech therapy and the care team increased the frequency of oral feeding initiation by 33 weeks’ PMA. Furthermore, although oral feeding was the predominant terminal barrier to discharge during the baseline monitoring period, apnea resolution represented the most frequent terminal barrier in subsequent PDSA cycles.

There is limited evidence evaluating the impact of oral feeding initiation timing and the PMA at independent oral feeding and discharge. In a previous observational study in moderately preterm infants, researchers reported that for every week earlier that oral feeding was initiated, there was an associated decrease in the PMA of independent oral feeding achievement by 5 days.⁶  In a randomized trial of infants born at <30 weeks’ GA (N = 29) in which researchers compared oral feeding via a structured protocol to oral feeding at the provider’s discretion, researchers reported that early introduction of oral feeding reduced the time to independent oral feeding.¹⁹  In contrast, randomized trials of oral feeding initiation at 30 weeks’ PMA compared with 33 weeks’ PMA in infants born at <29 weeks’ GA²⁰  or 32 weeks compared with 34 weeks’ PMA in infants with an average GA of 30 weeks²¹  did not report a different PMA at discharge. Initiation of oral feeding before 33 weeks’ PMA in the present initiative, a consistent time of initiation in a review of 39 NICU oral feeding guidelines,⁵  was associated with a lower PMA at achievement of oral feeding independence. The later timing of oral feeding initiation in other Infant Driven Feeding initiatives (between 34 and 36 weeks’ PMA^2,3 ) may explain why cue-based feeding in the present initiative was not associated with a further decrease in the time to independent oral feeding or discharge. Establishing guidelines for both respiratory support on which infants can orally feed and timely weaning of infants from respiratory support precluding oral feeding may have contributed to this earlier postnatal age of oral feeding initiation.

In previous initiatives that analyzed the impact of Infant Driven Feeding on the PMA at independent oral feedings and discharge, the majority of infants have been moderate to late preterm. In addition, these initiatives have used interrupted time series, as compared to statistical process control charts, as used in the present initiative. A quality improvement initiative introducing Infant Driven Feeding in preterm infants born at <34 weeks’ GA reported a decrease in the PMA at independent oral feeding and discharge in infants born at 28 to 32 weeks’ and >32 weeks’ GA.²  A retrospective study comparing the PMA at independent oral feedings between infants receiving Infant Driven Feeding compared with practitioner-driven feeding reported a lower PMA at independent oral feeding but not discharge in moderately preterm infants receiving Infant Driven Feeding.²²  Although the present initiative observed no further reduction in the PMA at independent oral feeding or discharge, the implementation of practitioner-driven feeding in infants not achieving independent oral feeding by 36 weeks’ PMA was associated with a reduction in extreme length of stay. Therefore, continuing cue-based feeding for the entirety of the hospitalization may be associated with more extended hospitalizations.

In the present initiative, independent oral feeding achievement occurred at 36.5 week’s PMA, a postnatal age similar to that previously reported in preterm infants with morbidities such as BPD.¹  In reporting the relative impact of other physiologic milestones on discharge, oral feeding represented only the most common barrier to discharge at the beginning of the initiative, suggesting that the terminal barrier to discharge may be modifiable. As apnea now represents our most common terminal barrier to discharge, efforts to standardize caffeine discontinuation may lead to further reductions in length of stay.²³  Such hospital-specific and international differences in clinical practice may impact the large differences in length of stay of up to 3 weeks in extremely preterm infants.²⁴ 

The cost per hospitalization in infants born between 25 and 32 weeks’ GA ranges from ∼$400 000 to ∼$100 000, respectively.²⁵  For the 3 PDSA cycles comprised of 302 infants, the average PMA at discharge decreased by 1 week. Using an average hospital-day cost of $1500, this decrease in length of stay may have saved $10 500 per hospitalization and, in total, >$3 million without an increase in the frequency of readmission.

The strengths of this quality improvement initiative include process measures for the implementation of cue-based feeding, the application of interventions with high levels of reliability, analyses using statistical process control charts, and demonstration of sustainability. Although this initiative excluded infants born at <25 weeks’ GA and infants with prespecified morbidities, including NEC, it did include 185 infants born between 25 and 28 6/7 weeks’ GA. One limitation of the study is that the frequency of oral feeding attempts by PDSA was not systematically monitored, which could impact the duration of hospitalization given the influence of missed oral feeding opportunities on hospitalization.²⁶  The ability to assess the balancing measure of readmission may have also been limited because infants may have been readmitted at alternative hospitals in the community.

In this multidisciplinary quality improvement initiative, an increase in the frequency of oral feeding initiation by 33 weeks’ PMA was associated with a decrease in the PMA at discharge from 38.8 to 37.7 weeks’ PMA. The introduction of cue-based feeding during the second PDSA was not associated with a further reduction in the PMA at which independent oral feeding or discharge occurred, suggesting that more evidence may be needed before routine recommendation of this practice. In addition, the transition from cue-based feeding to practitioner-driven feeding at 36 weeks’ PMA was associated with a decrease in extreme length of stay, suggesting this transition may be warranted in infants unable to achieve independent oral feeding by this postnatal age.

We thank the nursing staff at The University of Alabama at Birminghim for their dedication in implementing these oral feeding practices throughout and since this initiative. We also thank Susan LaMantia for her mentorship throughout the initiative, as well as the Intermediate Improvement Science Series at Cincinnati Children’s for continuous input and support instrumental in this initiative’s design and execution, as well as tools used for data analyses.

GA

gestational age

IQR

interquartile range

PDSA

plan-do-study-act

PMA

postmenstrual age