A Quality Improvement Initiative to Reduce Necrotizing Enterocolitis in Very Preterm Infants

This multisite project was conducted within the Calgary Health Region (CHR), Canada. The CHR comprises 2 level III NICUs with 53 beds and 3 level II NICUs with 75 beds. An average of 320 very preterm (≤32^6/7 weeks’ gestation) infants are admitted annually. This project was approved by the research ethics board at the University of Calgary.

All inborn infants born at ≤32^6/7 weeks’ gestation admitted to the CHR were included. Infants with significant congenital anomalies were excluded. The initiative period was from January 1, 2013 to December 31, 2021. The preintervention, intervention, and sustainment periods were from January 1, 2013 to June 30, 2016, July 1, 2016 to December 31, 2018, and January 1, 2019 to December 31, 2021, respectively. We also followed the NEC rate in the postinitiative year of 2022.

After we reviewed the incidence of NEC before 2016, a multidisciplinary team of neonatologists, neonatal-perinatal fellows, neonatal nurses, a QI nurse, clinical nurse educators, dietitians, pharmacists, and lactation consultants was formed. The initiative was called the NEC-Free NICU. After a literature review, a key driver diagram was created to determine practices that may contribute to NEC (Fig 1).

The working group developed guidelines for the CHR on the basis of the best available evidence-based practice. A level of evidence was assigned to each component by using the grading quality of evidence and strength of recommendation. The dissemination of guidelines across the CHR began in June 2016, including interactive educational sessions for all team members. The group met monthly during the implementation phase to review NEC cases and identify improvement areas. The dates of the interventions are summarized in Table 1.

Multi-strain probiotics containing lactobacillus and bifidobacterium are associated with a decreased risk of NEC in preterm infants with a birth weight of >1000 g.¹¹  We introduced FloraBABY (Renew Life Canada, Brampton, Ontario, Canada) for infants with a birth weight of >1000 g in July 2016. FloraBABY contains bifidobacterium and lactobacillus species and is approved by Health Canada. Given the limited evidence to support probiotics in extremely low birth weight (ELBW; <1000 g) infants, we did not introduce it as a standard intervention in this population. However, some ELBW infants received probiotics during this QI initiative as a part of a randomized trial conducted in our NICU to examine probiotics in ELBW infants.¹⁵  Buccal administration of colostrum (oral immune therapy) was already adopted in our NICUs because of the association with reduced NEC.¹⁶ 

We introduced antibiotic stewardship in September 2017 (Supplemental Table 4). Our antibiotic stewardship emphasized (1) improved evaluation for infants at risk for early-onset sepsis, (2) limiting empirical antimicrobial therapy to only infants with sepsis risk factors, (3) antimicrobial therapy review after 24 hours and when culture results become available (“Antibiotic Time-Out” in which, after discussion, the pharmacist entered a stop time of 48 hours on the electronic medical record [EMR]), and (4) treatment guided by organism susceptibility profile with infection disease consultation. Given the increased risk of NEC when histamine-2 blockers or proton pump inhibitors are used, we also limited their use in preterm infants.¹⁷ 

We recommended the avoidance of severe anemia and standardizing practices related to blood transfusion in July 2016. Previous practice varied widely and included restrictive blood transfusion despite severe anemia, holding feeds (nil per os [NPO]), and holding bovine-source products during transfusion. We adopted standardized thresholds to avoid severe anemia (Supplemental Table 5).¹⁸  We standardized practice to continue feeding but avoid bovine-based products around the time of transfusion. To date, evidence does not support withholding feeds during a blood transfusion.¹⁹  One study suggested infants who developed transfusion-associated NEC are more likely to have been fed formula during transfusion.²⁰ 

This intervention targeted improving breast milk production in the first week after delivery and sustaining milk volume during the hospital stay. The previous initiative to provide exclusive MOM to very preterm infants had become the standard of care in January 2013.¹³  However, the rate of exclusive MOM feeding declined after introducing DHM in April 2012. In November 2016, we reinforced strategies to promote exclusive MOM. The following best practices were included in the education sessions and materials: (1) early pumping and expression, (2) the importance of hand expression, (3) pumping routine, every 3 hours during daytime and at least once overnight, (4) the value of colostrum (every drop counts), and (5) the important protective effects of breast milk. The NICU parents advisory council team provided directions on strategies to empower parents to advocate for following standardized feeding and treatment practices. Additionally, we introduced antenatal lactation consultation for mothers expecting preterm infants in March 2017.

Standardized feeding tables were adopted in 2011; however, variable practices existed for feeding advancement and time of fortification. We emphasized following standardized feeding advancement and the introduction of bovine-milk-derived human milk fortifiers (HMFs) in November 2016. The weight-specific feeding advancement table was kept at the bedside until full enteral feeds were tolerated. We followed compliance by comparing the actual time of reaching full enteral feeding (as defined by 140 mL/kg/day sustained for 3 days) with the intended time to reach full feeding based on the infant’s weight-specific feeding advancement table (Supplemental Table 6). Variable compliance with feeding tables continued despite multiple educational sessions. A root cause analysis identified feeding intolerance as the driver for diverging from our feeding regimens. Therefore, we implemented a feeding intolerance algorithm (Supplemental Fig 5) to standardize our practice in January 2018.

We introduced conservative feeding guidelines for preterm infants with a high risk of impaired intestinal blood flow or relative intestinal hypoxia in May 2017. This intervention was aimed to avoid a 2-hit injury by restricting exposure to bovine-based products in infants undergoing medical treatment of hemodynamically significant patent ductus arteriosus (hsPDA).²¹  The 2-hit injury occurs when the intestinal oxygenation is compromised by impaired blood flow because of hsPDA and its treatment (first injury),²²  followed by an increase in the demand for oxygen due to feeding bovine-based products (second injury).²¹  Unlike indomethacin, ibuprofen has less effect on mesenteric blood flow and splanchnic tissue oxygenation.^22–24  Oral ibuprofen is associated with a reduced risk of NEC when compared with intravenous ibuprofen and indomethacin.²⁵  Withholding feeds during pharmacological hsPDA treatment remains controversial.²⁶  Given the lack of robust evidence, we developed conservative practices for infants with hsPDA, which include avoiding advancing feed volume during treatment in infants with impaired mesenteric blood flow, encouraging ibuprofen (over indomethacin), utilizing oral ibuprofen when feed volume is sufficient (≥60 mL/kg/day), and holding bovine-based products (ie, HMF or formula) during pharmacological treatment.

Infants were identified by using the EMR. Preterm infants born at ≤32^6/7 weeks’ gestation and admitted to CHR were included. The primary outcome was NEC, defined according to the Canadian Neonatal Network manual, which is based on Bell’s criteria of NEC stage II and III that is diagnosed on radiograph, intestinal ultrasound, or pathology.²⁷  Abdominal radiographs were reviewed independently by 2 to 3 neonatologists, and if initial interpretation differed, agreement by consensus was pursued. If requested, intestinal ultrasounds were reported by an attending radiologist. NEC was diagnosed if intestinal pneumatosis or portal vein gas was present on an abdominal radiograph or intestinal ultrasound or if there was a surgical or autopsy diagnosis of NEC. To make a diagnosis of NEC, 2 to 3 neonatologists reviewed all available data (clinical features, blood tests, radiograph, ultrasound, and pathology, when available). A diagnosis of spontaneous intestinal perforation (SIP) was made in infants with pneumoperitoneum without pneumatosis or portal vein gas on abdominal radiograph or ultrasound if clinical and laboratory findings of NEC were absent or when intestinal pathology (in infants undergone laparotomy) revealed isolated perforation with no evidence of NEC.²⁷  The cases of SIP were not included in the NEC numerator.

We tracked several process measures for each preventive domain. Balancing measures included growth anthropometrics at 36 weeks’ postmenstrual age (PMA), late-onset sepsis, SIP, length of hospital stay, length of parenteral nutrition (PN) support, and rate of blood transfusion (proportion of infants transfused in the preceding 6 months). To support sustainability, each case of NEC was reviewed by using a standardized review form (Supplemental Information). Each form was completed by the QI nurse and the attending clinical team.

Lastly, we assessed the effects of our initiative on various racial groups, aiming to identify any potential disparities, develop targeted strategies to address the specific health care needs of different racial groups, and enhance cultural competency within our NICUs. We reported the outcomes on the basis of prenatal self-reported race and ethnicity of the mother as specified in the EMR. Infants were classified into white, Asian (South or Southeast), Black (African or African American), and others (First Nations of Canada, Latino, Arab, or unknown). Because of their small sizes, it was not feasible to report the other subgroups separately.

We used statistical process control charts created in Excel with QI Macros add-in (KnowWare, Denver, CO) for data interpretation. The outcome measures, process measures, and balancing measures were assessed with p-, t-, and XmR-charts, as appropriate. Control limits were set at 3 SDs from the mean. The centerline was shifted with sustained special cause variation (eg, ≥8 values greater or less than the baseline centerline).²⁸  Descriptive statistics were used to describe the patients’ characteristics. Categorical variables, including race and ethnicity, were evaluated by using χ² and Fisher’s exact tests. Continuous variables were analyzed by using a 1-way analysis of variance or Kruskal-Wallis, as appropriate. P values <.05 were considered statistically significant. Stata version 16.0 software (Stata Corp, College Station, TX) was used for analyses.

During the study period, 2850 infants were born at ≤32^6/7 weeks’ gestation. Sixty-three infants were excluded because of congenital anomalies, with 2787 infants included in the study. A total of 1105, 763, and 919 infants were included in the preintervention, intervention, and sustainment periods, respectively. Table 2 compares characteristics between the 3 time periods. There was a significant increase in maternal hypertension and cesarian delivery rate, with a decrease in multiple births and SNAP 2 scores. Exposure to ibuprofen increased during the intervention period and appeared related to an increased, although statistically insignificant, rate of patent ductus arteriosus (PDA) treatment rather than decreased indomethacin use. The level III NICUs participated in a cluster randomized controlled trial (RCT) to treat PDA, with random assignment to indomethacin for the first treatment course. The rate of bronchopulmonary dysplasia (BPD) increased in the intervention and sustainment periods.

We found a reduction in NEC from the second half of 2016, with a decrease from 5.6% to 1.9% in the intervention and sustainment periods (Fig 2A). This reduction remained consistent at 1.8% when considering the postinitiative year of 2022. The time between NEC cases increased from mostly <50 days in 2013 to 2014 to often >100 days after 2017 (Fig 2B). There was a modest decrease in the rate of surgical NEC from 1.6% to 0.9% (Supplemental Fig 6). The downtrend of NEC was apparent in all gestational age categories (Supplemental Fig 7). Although the downward decline of NEC was evident in white, Black, and Asian infants in the intervention period, only white and Black sustained the low-rate pattern (Supplemental Fig 8).

Process measures are summarized in Fig 3. Breast milk feeding rates increased at hospital discharge (Fig 3A). However, the pattern of improvement differed among racial groups, with Black and white infants showing greater uptrends compared with Asian infants (Supplemental Fig 9). The difference between the actual and intended time to reach full enteral feed significantly decreased by 1.5 days after the introduction of the standardized management for feeding intolerance (Fig 3B). Table 3 summarizes other nutritional outcomes of the project participants. There was a significant decrease in the median age of ordering MOM feeds (the age when the physician or nurse practitioner’s order to commence feeds is entered in the EMR) and reaching full enteral feeds. The composition of the first enteral feed was available in our EMR in the sustainment period only. The median age at starting the first MOM feed was 23 (interquartile range [IQR] 14–34) and 14 (IQR 8–23) hours in preterm infants ≤28^6/7 and 29°^/7–32^6/7 weeks’ gestation, respectively.

The lowest hemoglobin level was higher in the intervention and sustainment periods (Fig 3C). Supplemental Fig 10 compares the lowest hemoglobin level preNEC in infants with NEC and during the NICU stay in infants without. The lowest hemoglobin was lower in infants with NEC. The duration of initial empirical antibiotics declined from 63.1 hours in the preintervention period to 55.5 hours in the second half of 2019 (Fig 3D). Probiotics were given to all infants >1000 g at birth. Of note, 31 preterm infants with birth weight <1000 g received probiotics because of enrollment in the probiotic RCT.¹⁵  One of the 31 infants developed NEC.

We observed alterations in balancing measures. The median duration of PN support decreased from 8.6 to 6.7 days after introducing feeding intolerance guidelines (Table 3, Fig 4A). Weight z scores at 36 weeks’ PMA significantly improved (Table 3). The rate of blood transfusion increased in the intervention and sustainment periods (Fig 4B). The rate of SIP did not change significantly (Supplemental Fig 11).

Through the implementation of evidence-based intestinal protective strategies for very preterm infants, we reduced the rate of NEC by approximately two-thirds and sustained improvement over 3 years. We observed a reduced number of NPO days, an increased rate of blood transfusion, and improved weight z scores at 36 weeks’ PMA.

Our results reinforce the importance of promoting MOM.²⁹  It is possible that this approach was the driving factor for the prevention of classic NEC.³⁰  Our previous initiative to promote MOM had increased breast milk feeding at discharge to 91%.¹³  This decreased after introducing DHM in early 2012 to 83%. Although the availability of DHM has been associated with increased breast milk feeding,³¹  the effect varies between NICUs, perhaps due to different baseline breast milk feeding rates. Although DHM remains the preferred choice when MOM is unavailable, studies have revealed questionable benefits, particularly for surgical NEC.³²  Our baseline rate of MOM feeding is high compared with other NICUs in Canada and the United States, which likely protected more infants from NEC.^33,34 

Our results indicate increased ibuprofen and acetaminophen use to treat PDA, with a stable rate of indomethacin use. The recurrent shortage of indomethacin and our center’s involvement in a cluster RCT to treat PDA have limited our certainty about compliance and the benefits of ibuprofen as the first choice.

Our preintervention practice regarding feeding during PDA treatment was widely variable, with a tendency to stop or decrease enteral feeding during treatment. We introduced the concept of “avoidance of two-hit injury” and standardized our practice to withhold bovine-based products (ie, formula or HMF) in infants with impaired mesenteric blood flow. Withholding only bovine-based products, rather than withholding all feed, may have contributed to improved weight z scores.

There was an increase in blood transfusion. The potential association between blood transfusion and NEC has led to fear of transfusion and adopting permissive anemia awaiting improved hematopoiesis. Standardizing blood transfusion thresholds resulted in the prevention of severe anemia. This explains the increased blood transfusion rate. Avoidance of bovine-based products during and 24 hours after blood transfusion, instead of withholding feed, probably contributed to improved weight z scores.

We observed an increased rate of BPD. BPD is a multifactorial disease with a variety of prenatal and postnatal factors.³⁵  Our results were reassuring with no association with factors that contribute to BPD (eg, growth). The number of blood transfusions³⁶  and maternal pregnancy-induced hypertension have been associated with BPD.^37,38  Although the increase in blood transfusion was modest, a significant increase in pregnancy-induced hypertension over time might explain the increase in BPD.

The variation in clinical practices between multiple NICUs was initially seen as a hindrance to our initiative. This variation was recognized as a significant factor contributing to inconsistent care delivery across CHR NICUs. By gaining support from each NICU’s leadership, the QI team successfully implemented system-wide changes and promoted resource sharing. In addition, the presence of multiple NICUs facilitated learning from one another’s experiences.

The standardization of feeding protocols across several NICUs was challenging because 2 units had their own protocols. Given that the rate of feeding advancement within these protocols was within the range of our project, we changed fortification practices only. Providing adequate education for staff in level II NICUs was difficult. Clinical nurse educators, dietitians, and pharmacists in each NICU educated bedside staff and provided updates on our initiative. Sustainability was a major challenge. A standardized review form helped maintain sustainability by facilitating group discussion, educating staff, and identifying potential contributing factors. All elements of our project have become standard care practices at the end of the project.

Our project revealed a significant decrease in the rate of NEC in all racial groups during the intervention period, with a return to the preintervention rate in Asian infants in the sustainment period. The magnitude of the increase in breast milk feeding at discharge in Asian infants was lower than that in white and Black infants. Several reports indicated the racial disparities of NEC, with a higher incidence in Black infants.^39,40  The rate of breast milk feeding at discharge is shown to account for 22% and 44% of the total risk of NEC in non-white and Black infants, respectively, compared with white infants.^33,40  The NICU team needs to pursue culturally competent interventions that serve the needs of all racial groups.

Efforts aimed to reduce NEC have had varying success and have often included interventions with limited evidence or certainty.^41,42  Mavis et al reported a small QI initiative that reduced NEC from 19.5% to 6% in infants born at <30°^/7 weeks’ gestation, focusing primarily on using an exclusive human milk diet (human milk-derived fortifiers), prioritizing MOM, preventing intestinal dysbiosis, and improving compliance with feeding and transfusion guidelines.⁴²  Although the key drivers in this initiative were similar to those in our project, we did not use an exclusive human milk diet, given the limited evidence from RCTs.⁴³  Our previous QI initiative to promote MOM feeding before DHM availability resulted in a significant decrease in NEC from 8.9% to 4.7% in infants born at <32°^/7 weeks’ gestation.¹³  Although it is difficult to identify the main driver in reducing the relatively high rate of NEC in the initiative by Mavis et al, our previous initiative proposed the promotion of MOM as the key to preventing NEC. Nesterenko et al reported a decrease in NEC from 5.3% to 3.5% in very low birth weight infants using a QI project focused on standardized feeding practices, blood transfusion thresholds, and spreading the medications’ osmolar load.⁴¹  Although high osmolarity originating from nutrients is associated with NEC, there is insufficient data to support medications posing a similar risk.^44,45 

Our project has several limitations. We did not follow compliance for many process measures, including the number of antenatal consultations, DHM use, adherence to blood transfusion recommendations, or PDA guidelines because of workload limitations within the QI team. Overlapping interventions made it difficult to identify the extent of each intervention’s contribution to improvement. Given the nature of the initiative, all improvements are associations rather than causations.

The implementation of multidisciplinary practice guidelines to improve MOM volumes, promote adherence to feeding regimens, standardize the treatment of hsPDA, standardize protocols for blood transfusion, avoid intestinal injury during PDA treatment and blood transfusion, and promote a healthy intestinal microbiome was associated with a sustained reduction in NEC in very preterm infants.

We acknowledge all members of the Neonatal Gastroenterology and Nutrition committee, Neonatal Hemodynamics group, and Infection Control and Prevention group whose tireless work and dedication were fundamental to success. We also acknowledge HOIF, MNCY, and AHS for their grant support of this work.

BPD

bronchopulmonary dysplasia

CHR

Calgary Health Region

DHM

donor human milk

ELBW

extremely low birth weight

EMR

electronic medical record

HMF

human milk fortifier

hsPDA

hemodynamically significant patent ductus arteriosus

IQR

interquartile range

MOM

mother’s own milk

NEC

necrotizing enterocolitis

NPO

nil per os

PDA

patent ductus arteriosus

PMA

postmenstrual age

PN

parenteral nutrition

QI

quality improvement

RCT

randomized controlled trial

SIP

spontaneous intestinal perforation