A Quality Improvement Project to Reduce Time to Full Enteral Feeds for Very Low Birth Weight Neonates Free

The study took place in a 35-bed NICU within a level IV regional perinatal center with an average of 3400 annual deliveries and 700 admissions each year. Neonates were included in this study if they were <1500 g and <32 weeks at birth. We excluded neonates with a known gastrointestinal illness requiring early surgery before delivery. Patient care is provided by a vast multidisciplinary team, including neonatologists, neonatal fellows, pediatric residents, advanced practitioners, registered nurses, pharmacists, respiratory therapists, registered dieticians, lactation specialists, and social workers. For multiple aspects in the care of the very low birth weight neonate, the NICU has set standards via protocols and these protocols also cover nutritional support. These protocols align with the potentially better practice recommendations of the Vermont Oxford Network “Got Milk” focus group. This QI initiative was conducted over a 12-month period between May 2020 and April 2021.

A QI team consisting of important stakeholders, including a physician champion, a NICU dietician, a representative of the nursing leadership, and advanced NICU practitioners was created. Understanding the process involved in decision making for neonatal feeding in the NICU was crucial and this was elucidated with contributions from the major stakeholders, including an understanding of protocol deviation.

Meetings were held every 2 months between stakeholders to reinforce project goals and ensure that activities related to the processes were being appropriately monitored and implemented. A resident physician lead was included in the stakeholder meetings.

The main intervention was the creation of an R Shiny web application, which can be found at https://abiola.shinyapps.io/montefeedingcalc/.²¹  It was modeled directly from the existing nutrition and feeding policy and created in conjunction with the NICU dietician. This application could be accessed from both desktops and mobile devices. After inputting a participant's data, including weight, total fluid goals, and date of feed initiation, the application then calculated the feeding volume of a newborn from feeding initiation until full enteral feeds. A user could also access tabs that held the official feeding policy, guidance on how to manage feeding intolerance, and references for trace elements in total parenteral nutrition and caloric density of commonly used preterm formulas. Education was provided on the usage of the electronic R Shiny application at multiple teaching events to nurses, advanced practitioners, fellows, and residents.

The second intervention included creating a printed copy of a participant’s feeding schedule. For each study participant, the NICU dietician created a feeding schedule in the R Shiny application. This feeding schedule was then printed and posted at the bedside for providers to reference. This intervention was done to assess the feasibility of the application’s use in a busy unit. It also reduced the barrier of accessing the application online if access to a computer was not available. Nursing questions, feedback, and comfort with the feeding schedule were assessed directly during rounds. Provider education on the use of the tool was also implemented at available opportunities.

The third intervention included sending e-mails to neonatal fellows. The fellow lead distributed emails to the fellows on service describing the R Shiny application’s use and reminded fellows to use it when creating feeding goals for patients. Reminders were sent at the beginning of each service month block.

The fourth intervention included making links to the R Shiny application more accessible. Desktop computers in the unit were labeled by the fellow lead with links to the R Shiny application. E-mails were also sent to the advanced practitioner teams with links to the application. This intervention was done so that every provider could easily find and use the application when the NICU dietician was away or if patients were not covered by fellows.

The fifth intervention included encouraging resident involvement and incorporating the application to the EMR. The resident physician lead, in conjunction with the unit Epic representative, placed a hyperlink to the R Shiny application in an Epic Smartphrase. This Smartphrase was used for the resident handoff communication tool that was used daily on rounds. As the resident team rounded with Workstation on Wheels, this provided ease of access to the application in nutrition and feeding decision making. The resident physician lead sent monthly e-mail reminders to residents rotating through the unit. E-mails were sent at the beginning of each rotation block explaining the application utility and where to access it. Periodic check-ins were performed with the residents midrotation to answer questions about the application and to assess if the application was being used.

SQUIRE 2.0 reporting guidelines were used in the writeup of this project.²² 

Our primary outcome was mean time to full enteral feeds. Time to full enteral feeds was defined as time to tolerate enteral fluids of 160 ml/kg/day and recorded in days.^23,24  The secondary outcome was the median duration of central line use. The mean duration of central line utilization was calculated in days. This measure was chosen as the feeding protocol provided guidance for when to discontinue central line usage. The rate of central line-associated bloodstream infections (CLABSI) was also observed. In the baseline phase of the study, the rate of CLABSI was zero.

The process measure was the proportion of participants that were being fed in accordance with the standardized feeding protocol. Quantitatively, we measured the number of participants that had initiation, advancement, and fortification of feeding volume per the standardized feeding protocol and divided that by the total number of participants. Qualitatively, walk-throughs of the unit identified if each study participant had a printed standardized feeding regimen at their bedside. Having a physical printed copy of the feeding regimen was believed to improve compliance with the protocol.

The balancing measures included average weight gain on DOL 28, the occurrence of NEC, and occurrence of postnatal growth failure. The occurrence of NEC was defined to include all stages of necrotizing enterocolitis and was recorded as a dichotomous outcome as determined by reviewing notes of the attending on service. Z scores at 36 weeks were measured as a marker of postnatal growth failure. A decrease in Z score of more than −1.28 between birth and discharge was defined as postnatal growth failure. The balancing measures were chosen to evaluate for unintended adverse outcomes of the intervention. The development of NEC during enteral feed advancement created apprehension among providers to adopt the feeding protocol. Thus, we wanted to evaluate rates of NEC as a balancing measure to the primary outcome. Growth served as a balancing measure for the secondary outcome of central line usage duration because early removal of the central line may result in reduced fluid intake and thus growth.

Clinical characteristics were also collected, including sex, gestational age in weeks, birth weight in grams, mode of delivery, and whether the neonate was small for gestational age (defined as less than the 10th percentile on Fenton curves).

Data on all participants were collected prospectively, deidentified, password-protected, and stored in Microsoft Excel.

Differences between baseline and postintervention groups were analyzed by using the independent samples t test to compare the means of continuous variables and the χ² test for categorical variables. Medians were compared by using Mann-Whitney U tests. Run charts and statistical control charts were created in Excel. Special cause variations occurred when there was a change in the center line and its control limits. Because the processes were stable in the baseline period, any changes that occurred during the implementation phase were attributed to the activities of project implementation.

The study protocol was reviewed by the institutional review board and determined to be a QI study.

Baseline data collection showed a median adherence to initiation of enteral feeds of 23%. Median adherence to following advancement volumes per the protocol was 27%. Baseline fortification of feeds to 24 kilocalories (kcal) at the target volumes occurred in 82% of neonates. The baseline time to enteral feeds was 24.7 days. In addition, the median central line use duration was 19 days.

The activities in this QI project were introduced in 4 plan-do-study-act cycles that followed our primary drivers (Fig 1). Actions found to be impactful were carried into the following cycles.

Table 1 reveals the clinical characteristics of the neonates in the baseline and intervention phases of the study. There were no significant differences in gestational age, birth weight, and small for gestational age status during the course of this quality initiative.

Our main outcome measure was the average time to full enteral feeds, which was measured biweekly in all neonates who met our inclusion criteria. During the project, there was a reduction in the days to full enteral feeds of 160 ml/kg per day from a baseline of 24.7 days to 17.7 days. We had established normal variation during the baseline period of December 2019 to April 2020. Special cause variation was noted through the course of the project from May 2020 to April 2021 (Fig 2). We also showed a significant reduction in mean central line duration over the course of the project from a baseline of 19 days to 14.5 days, P = .037. (Table 2). In addition, we had no CLABSI in the 12 months during which these QI interventions took place.

These changes were observed despite no change in significant clinical features of the neonates in the baseline and intervention phases. Our balancing measures were not adversely affected. Specifically, mean growth velocity to day of life 28 in the baseline compared with intervention phases were 16.6 (± 5.1) compared with 16.8 (± 5.7) g/kg per day (P = .88). The Z scores did not fall below the level for postnatal growth failure of −1.28. The Z scores were −1.03 (± 0.99) and −1.15 (± 0.83) for baseline and intervention phases, respectively (P = .53). During the baseline and intervention phases, we had no cases of NEC in our population (Table 2).

Table 2 also reveals that during the course of the project, the proportion of initial feeding volumes and advancement volumes that were based on the NICU standardized feeding protocol increased from 23% and 27%, respectively, to 60% and 63%, respectively (P = .0002).

However, compliance with fortification decreased from 82% in the baseline group to 78% during the intervention phase (P = .80).

The aim of the authors of this project was to reduce the time to full enteral feeds using our major intervention of an electronic nutritional calculator. During the course of our QI project, we achieved a reduction of days to full enteral feeds by 6 days. The major strength of the project included a simple and user-friendly tool to aid in clinical decision making. In addition, the collaboration between unit leadership and practitioners helped to increase the utilization of the tool and helped to standardize care in our NICU.

As in many hospital units, implementing change and encouraging adherence is universally challenging. This conundrum was addressed by having conversations with the neonatology leadership who supported the project to standardize the care of newborns. In addition, demonstration of the ease of use of the calculator tool as well as its potential to reduce the manual process of calculating feeding goals daily increased buy-in and helped to ensure project progress.

The intervention was meant to be systematic and simple so that it could be sustainable. The positive feedback received during the project and the ease of use of the calculator lends to its sustainability. Bedside nurses were given confidence and ownership of feeding plans. In our unit, the nutritional calculator link remains incorporated in the EMR, and practitioners continue to use it to guide their clinical practice. Future considerations could include building an order set for automatic feeding initiation and advancement.

There exists wide variability in neonatology practice with different units possessing different feeding protocols. Our nutritional calculator tool was designed specifically around our unit protocols, limiting the generalizability of its use. However, the concept of designing a calculator tool is one that can be applied to other units and practices.

The determination of balancing measures of 36-week Z scores and 28-day growth velocity was done to see whether early central line removal would lead to worse overall growth. In the first few patients enrolled, we saw an initial downward trend in weight; however, this was not sustained. It is common practice in our unit to supplement calories from 24 kcal to 26 kcal when poor weight gain is noted. During our project, we did not specifically measure whether neonates were getting fortification beyond 24 kcal, and this may have contributed to the weight that was relatively sustained in study participants. This could be a variable to monitor in future similar projects.

During the study period, no patients <32 weeks and <1500 g developed NEC. Our unit uses donor breast milk in most very low birth weight neonates rather than cow’s milk-based formula; this may have contributed to the low rates of adverse events seen.

There are multiple ongoing initiatives directed at improving the quality of neonatal care in our unit. Therefore, the direct attributability of certain outcomes to our interventions is limited. Specifically, we have QI projects that directly involve the use of care bundles to prevent central line infections. Therefore, the absence of central line infections in the baseline and intervention phases may be the direct result of the central line care bundle and not only a consequence of earlier line removal that resulted from this QI project. Additionally, most neonates included as participants in this project reach full feeds long before they are discharged; as such, time to full enteral feeds does not directly correlate with length of stay.

Through key interventions, we were able to reduce the time to full enteral feeds in neonates without any increase in adverse events of NEC or poor weight gain. Our major process of improving adherence is one that would continue to require timely audits to ensure that the benefits of the QI project are sustained over time.