A Quality Improvement Bundle to Improve Outcomes in Extremely Preterm Infants in the First Week

The University of Alabama at Birmingham’s (UAB) regional NICU is an academic quaternary referral center with ∼1400 admission annually, including 130 inborn extremely preterm infants from 22 + 0/7 to 27 + 6/7 weeks’ gestation. The average daily census is 90 infants, and 85% of infants are inborn. Our setting has ∼210 staff (nurses and respiratory therapists), 90 pediatric residents, 9 neonatal fellows, 22 neonatal faculty members, 5 maternal-fetal medicine fellows, and 22 maternal-fetal medicine faculty members.

A comprehensive literature review was conducted to develop evidence-based guidelines defining our care of extremely preterm infants.¹³  A multidisciplinary core committee was created including a nurse educator, a nurse manager, a respiratory therapy manager, a neonatal fellow, and an attending neonatologist serving as the program director. Primary and secondary key drivers were identified (Fig 1). Perinatal care, delivery room care, admission processes, staff handovers, and patient rounds were mapped. This quality improvement initiative was approved by the UAB Institutional Review Board.

We included all actively treated inborn extremely preterm infants from 22 + 0/7 to 27 + 6/7 weeks’ gestation weighing ≥400 g without congenital anomalies from April 2014 to September 2020. Intervention phase 1 lasted 12 months from April 2015 to March 2016. The multidisciplinary team met monthly to review outcome and process measure data and implement practice changes using rapid plan-do-study-act cycles (Supplemental Table 3). In April 2015, fellows were trained in new standardized patient care guidelines incorporating evidence-based potentially better practices. In May 2015, a new standardized electronic handover document was implemented among residents and fellows.¹⁴  A twice daily fellow-driven rounding checklist, which included a patient flowchart for ventilator settings, blood gas values, and fluid and electrolyte balance, was introduced.¹⁵  Nurse-led checklists were developed to standardize admission and reduce the time to line placement.

Joint meetings were held with obstetric and maternal-fetal medicine physicians, and delayed cord clamping was agreed in June 2015.¹⁶  Antenatal corticosteroid administration was recommended for all pregnancies ≥22 + 0/7 (previous threshold of 22 + 5/7) if resuscitation was planned.^17–19  Standardized delivery room care practices were implemented for respiratory support,^20,21  monitoring,²²  and thermoregulation.²³  In July 2015, avoidance of bolus of normal saline²⁴  and bicarbonate^25–27  in the first days after birth was recommended. In September 2015, we expanded indomethacin prophylaxis²⁸  for all extremely preterm infants without contraindications regardless of weight (previous threshold <750 g) using a single dose.²⁹ 

In October 2015, we implemented midline head position, with minimal head movement during the first 72 hours.^30,31  Transcutaneous carbon dioxide (Co₂) monitors are routinely used in our unit, and in November 2015, we began to place them immediately after admission before umbilical line placement.³²  In November 2015, we also introduced quarter normal saline flushes in place of normal saline flushes³³  and changed to an arterial line fluid containing half normal sodium acetate instead of normal saline.^25–27  In February 2016, we introduced Golden Week Program teaching modules for all staff, which were delivered over 2 months followed by monthly follow-up classes until October 2019. These classes included education on minimal handling during diaper changes (avoiding lifting body or legs) and avoidance of flushing or rapid withdrawal to or from central lines.

In phase 2, starting in April 2016, specific standardized electronic order sets were launched for admission, pharmacy, laboratory, nutrition, nursing, and respiratory therapy orders. In May 2016, cardiac electrode leads were added in the delivery room for a faster detection of the heart rate.³⁴  In June 2016, we started weighing infants every 12 hours in the first 72 hours to assist fluid³⁵  and electrolyte management, coinciding with serum sodium measurements every 12 hours and diaper checks every 6 hours. In September 2016, we adjusted staff education to emphasize the importance of trends in clinical and laboratory values. In October 2018, we began a Golden Week specific service, and in January 2019, a dedicated team of Golden Week nurse practitioners began to care for these infants. In August 2019, we switched from a pressure control ventilator strategy to a volume control strategy.³⁶ 

Clinical data were collected prospectively until hospital discharge by trained research personnel using standardized definitions from the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network.¹  The primary outcome was death or severe ICH in the first 7 days after birth. Delivery room deaths and deaths within 12 hours after birth that received any active postnatal care were included. Severe ICH was defined as ventricular enlargement with concurrent or previous blood in the ventricles or blood and/or echodensity in the parenchyma.²  Our global aim, survival without major morbidity, was defined as survival to discharge without severe ICH, necrotizing enterocolitis greater than stage 2 of modified Bell’s criteria,³⁷  physiologic bronchopulmonary dysplasia,³⁸  or retinopathy of prematurity greater than stage 3 or treated with vascular-endothelial growth factor or laser.

The balancing measures included the rate of spontaneous intestinal perforation and the rate of acute kidney injury in the first week. Acute kidney injury was defined as serum creatinine ≥1.5 mmol/L or a ≥0.3-mmol/L rise in the first week after birth.³⁹  An electronic medical record audit was used to assess compliance with delayed cord clamping (30–60 seconds), electronic order sets, indomethacin prophylaxis, normal saline bolus, sodium bicarbonate administration, inotrope use, and blood transfusions. Data on weight changes, daily total fluid intake and urine output, laboratory results, admission temperature, and mean arterial blood pressure were collected from the electronic medical record by the UAB clinical data warehouse. Anemia was defined as a hematocrit <36%. Pco₂ ≥40 and ≤60 mm Hg, pH ≥7.20 and ≤7.39, and bicarbonate ≥15 mmol/L were considered within limits. Hypoglycemia was defined as glucose <40 mg/dL. Admission temperature <36.0°C and >37.9°C was considered abnormal. Adequate time to line placement was indicated by a first blood gas result within 90 minutes of birth. Low mean arterial pressure was defined as less than gestational age in weeks.

Data for the primary outcome, process measures, and balancing measures analyzed based on groups of 20 consecutive eligible infants were plotted on p-charts by using QI Macros (KnowWare International Inc., Denver, Colorado; version 2018) to determine if and when special cause variation occurred in relation to plan-do-study-act cycles.⁴⁰  We compared preintervention, phase 1, and phase 2 results using Fisher’s exact test or χ² test for categorical data and Student’s t test for continuous data. Relative risk and 95% confidence intervals were calculated comparing preintervention with phase 1 and phase 2 and comparing phase 1 with phase 2. No adjustments were made for baseline characteristics, which were compared among preintervention, phase 1, and phase 2 by using χ² test for categorical data and analysis of variance for continuous data. A 2-sided P value < .05 was considered statistically significant.

There were 820 consecutive inborn extremely preterm infants included. The mean (SD) gestational age was 25 + 3/7 (11/7) weeks and birth weight was 744 g (211 g). There were no significant baseline differences in gestational age, birth weight, sex, receipt of antenatal corticosteroids, multiple birth, and mode of delivery among the preintervention, phase 1, and phase 2 groups (Table 1). There was a significant difference in the racial and ethnic distribution which was primarily driven by a lower rate of Black infants and a higher rate of White infants in phase 1.

The mean rate of severe ICH or death in the first week after birth in the year before the intervention was 40 of 146 (27.4%). From April 2015 until June 2018, the rate of death or ICH was 21.4%. A special cause variation was noted in June 2018, after which 8 consecutive points were below the centerline. The mean rate of severe ICH or death in the first week after birth decreased to 15.0% (Fig 2). This change corresponded with improvements in targeting Co₂ and pH and decreased use of inotropes. Goal Co₂ targeting (40–60 mm Hg) increased from 42.9% to 69.4%. Overall the median Pco₂ level was ∼3 mm Hg lower on days 3 and 4 postintervention (Supplemental Fig 4). The rate of successful pH targeting in the first week initially decreased from 33.6% to 26.1% before increasing to 47.2%. Overall, the median pH was 0.02 to 0.06 higher from days 2 to 5 after birth postintervention (Supplemental Fig 4). Inotrope use decreased from 23.6% to 16.6% after the introduction of our bundle. Inotrope use decreased further to 8.9% in June 2018 (Supplemental Fig 5). Decreases in the use of inotropes were not associated with changes in the rate of low initial mean arterial pressure (Supplemental Fig 5) or normal saline bolus use (Supplemental Fig 6), which both decreased after commencing our bundle.

The preintervention rate of severe ICH was 16.4%. After the introduction of our bundle, the rate of severe ICH was 14.3%. In June 2017, a special cause variation was observed and the rate of severe ICH decreased to 10.0% (Fig 3). The change in the rate of ICH corresponded with increased compliance with delayed cord clamping, expanded use of indomethacin prophylaxis, and decreased use of sodium bicarbonate in the first 72 hours. Delayed cord clamping went from 14.3% to 57.9% (Supplemental Fig 7). The number of infants with a hematocrit <36% in the first 72 hours decreased over time from 22.5% to 14.3%. The median hematocrit was 2% higher on day 1 and 3% higher on days 2 and 3 after birth postintervention. Indomethacin prophylaxis increased from 57.6% to 88.8% (Supplemental Fig 5). Sodium bicarbonate use initially increased from a mean rate of 17.8% to 20.9% before decreasing to a rate of 9.3% in June 2017 (Supplemental Fig 6). Although infants were exposed to less sodium bicarbonate, the mean serum bicarbonate was 1 mmol/L higher after the introduction of our bundle (Supplemental Fig 8) and the rate of infants with a bicarbonate <15 mmol/L went from 16.3% to 8.0% (Supplemental Fig 6).

The preintervention rate of death in the first week after birth was 13.7%. After the introduction of our bundle, the mean rate of death in the first week after birth was 9.7% without any special cause variation observed (Supplemental Fig 9). The rate of survival without major morbidity did not differ among preintervention, phase 1, and phase 2 (Table 2). Preintervention, 81 of 146 (55.5%) infants had at least 1 creatinine measured in the first week after birth. Postintervention, 541 of 614 (88.1%) infants had at least 1 creatinine measured. The rate of acute kidney injury went from a baseline rate of 31.2% of all infants preintervention to 22.1% of all infants postintervention (Supplemental Fig 5). The rate of spontaneous intestinal perforation did not differ among preintervention, phase 1, and phase 2 (Table 2).

Compliance with the use of electronic admission, pharmacy, and respiratory order sets was >90% (Supplemental Fig 10). Compliance with laboratory order sets increased from an initial 30% to 90% within 6 months. The mean number of infants exposed to ≥1 blood transfusions in the first week also decreased from 67.9% to 52.9% (Supplemental Fig 7). Sodium concentrations on days 2 and 3 were lower postintervention (Supplemental Fig 8). Infants had a lower cumulative total fluid intake of ∼40 mL/kg over the first 72 hours postintervention. Cumulative urine output was lower by ∼1 mL/kg per hour in the first 72 hours after birth. The number of infants with a blood gas result within 90 minutes went from 26.4% to 42.0% after introducing our bundle. There was a further increase in the number of infants with blood gas results within 90 minutes after birth to 54.6% from July 2016 (Supplemental Fig 6).

This comprehensive evidence-based quality improvement initiative was associated with improved outcomes in the first week after birth. A special cause variation in the rate of death or severe ICH was temporally related to improved targeting of Co₂ and pH levels and lower use of inotropes. A special cause variation in the rate of severe ICH was temporally associated with improved adherence to delayed cord clamping, expanding indomethacin prophylaxis, and avoidance of sodium bicarbonate administration.

In this study, we introduced multiple potentially better practices. It is possible that multiple practice changes cumulatively lowered the risk of ICH. Bundles have been used to improve delivery room care and perinatal survival.⁴¹  Care bundles have been shown to reduce the rate of central line–associated bloodstream infections.⁴²  Bundles targeting major morbidities among preterm infants had mixed results possibly because of the small number of intervention changes and the choice of potentially better practices selected, some of which later turned out to be ineffective in randomized controlled trials.^43,44  Recently, a care bundle including avoidance of flushing or rapid withdrawal of blood, midline head positioning, and minimal handling during diaper changes was associated with a lower rate of ICH, cystic periventricular leukomalacia, or death in very preterm infants.⁴⁵  Our program also included these interventions.

Compliance with specific interventions was associated with clinical improvements in the current study. Delayed cord clamping and indomethacin prophylaxis are known to decrease the rate of ICH.^16,28  Order sets were used to sustain improvement in the higher use of indomethacin. Although large fluctuations in Co₂ levels with corresponding changes in pH have been associated with ICH,³²  randomized controlled trials of pH-controlled permissive hypercapnia have not observed an increased risk of brain injury.^20,46,47  Changes in Co₂ and pH may have been sustained by early use of transcutaneous Co₂ monitors and acetate containing line fluids. In the absence of evidence from randomized controlled trials of inotropes⁴⁸  we recommended against treatment of hypotension if there were signs of good perfusion.⁴⁹  We avoided sodium bicarbonate on the basis of studies indicating a lack of efficacy and potential for harm.^25–27  We used monthly staff education to reinforce recommendations limiting inotropes and sodium bicarbonate. It is also possible that the introduction of the Golden Week specific service may have improved our standardization and adherence to guidelines.

Some interventions were introduced in the current study to deal with specific issues noted during implementation including hypernatremia and acidosis. The effect of quarter normal saline flushes and half sodium acetate in line fluids requires further study. However, these interventions may have helped reduce total fluid intake³⁵  and avoid sodium bicarbonate administration.^25–27  Interventions such as vitamin A therapy⁵⁰  and automatic antibiotic stop dates⁵¹  were included in the order sets as part of our efforts to improve our global aim of survival without major morbidity. However, despite improvements in outcomes in the first week after birth, the rate of survival without major morbidity did not differ likely because of a trend toward higher rates of bronchopulmonary dysplasia over time in our center. These higher rates of bronchopulmonary dysplasia may have been partly related to a trend toward higher survival to 36 weeks’ postmenstrual age.

It is unlikely that the results of this quality improvement initiative are due solely to the Hawthorne effect, although special cause variation was observed during phase 1 with increased scrutiny of processes and outcomes. It is possible that the educational efforts improved bundle compliance and the associated outcomes. We noted that there was a trend toward higher rates of our primary outcome within a few months of discontinuing monthly education. Whereas the upward trend in the rate of severe ICH was also temporally related to the introduction of volume control ventilation, randomized clinical trials have found decreased rates of ICH with volume targeting.³⁶  In the current study, the electronic medical record–embedded order sets were associated with improved compliance with specific evidence-based potentially better practices. Furthermore, electronic medical record–based data collection of compliance data replaced manual data collection without noticeable adverse effects.

This study was conducted at a single academic US medical center and the results may not be generalizable to other settings, although most units in the United States now use electronic order sets that could be harnessed to implement potentially better practices. It is possible that some of the potentially better practices we selected may prove ineffective in randomized controlled trials, but by introducing multiple practices, we may have mitigated the problem of implementing ineffective practices. With this study, we observed improvement in death or ICH in the first week after birth. It is not known whether this will translate into lower rates of death or neurodevelopmental impairment at follow-up, although severity of ICH predicts later outcomes.⁴ 

A comprehensive bundle of evidence-based potentially better practices in the care of extremely preterm infants implemented by an interdisciplinary team using specific electronic order sets was associated with a lower rate of death or severe ICH in the first week after birth.

We thank our research coordinators for their assistance with outcome data collection and the assistance of all the staff at the UAB regional NICU for their contribution. In particular, we thank Shirley Cosby, RN; Monica Collins, RN; Patricia Smith, RN; Susan Roberts, RT; Bradley Evans, RT; Melissa Ohnich, BSN, RN; Jennifer Heatherly, BSN, RN; Amanda Hoover, RN; Kimberley Ross, PharmD; Ashley Pruitt, RN; Andrea Rice, RN; Madison Drinkard, RN; Leah Cupp, RN; Jill Everette; Sandra Milstead; Cecilia Oree; and Cassandra Hudson for their assistance and contributions to the success of the Golden Week Program.

Co₂

carbon dioxide

ICH

intracranial hemorrhage

UAB

University of Alabama at Birmingham