Reduction of Severe Intraventricular Hemorrhage in Preterm Infants: A Quality Improvement Project

This study was conducted at the UCSF Benioff Children’s Hospital in San Francisco, a tertiary care teaching hospital with a 58-bed level IV ICN. An average of 40 ELBW infants are admitted per year, with ∼75% inborn through our high-risk obstetrics service. Infants are cared for by a multidisciplinary team of neonatologists, neonatology fellows, neonatal nurse practitioners (NNPs), residents, nurses, dieticians, respiratory therapists (RTs), occupational and physical therapists, pharmacists, and social workers.

This project was submitted to the UCSF Institutional Review Board, who considered it not human subject research and exempt from its review.

All inborn infants born at <28 weeks’ gestation or <1000 g and admitted to the ICN were included in this study. Infants with significant congenital anomalies were excluded. The study period was from January 1, 2012, to June 30, 2020. January 1, 2012, to March 31, 2016, served as the baseline preintervention period, with the start date chosen for when our current electronic medical record (Epic Systems, Verona, WI) was created. April 1, 2016, to December 31, 2018, served as the intervention period. January 1, 2019, to June 30, 2020, served as the sustainment period because improvement was established and there were no new interventions during that time.

After review of local outcomes for ELBW infants before 2015, a multidisciplinary team of neonatal nurses, neonatologists, NNPs, RTs, dieticians, pharmacists, social workers, parents, and other community members was formed with a shared goal to reduce the rate of sIVH. The initiative came to be called the Premie Project (P2), and the QI team named the “P2 committee.”

The P2 committee met monthly starting in January 2015. After an extensive literature review, a key driver diagram was created to determine the practices that may contribute to sIVH in premature infants (Fig 1). Five primary drivers were identified: (1) impaired cerebral autoregulation, (2) disturbance in cerebral blood flow, (3) underdeveloped germinal matrix, (4) platelet and/or coagulation disturbances, and (5) unknown factors. We identified secondary drivers and developed a series of potential interventions.

On review of the key driver diagram, it became clear that there were gaps in our current practice. Before this, a QI framework was rarely employed in our care of ELBW infants. It was determined that standardized practice with a multipronged approach was needed to reduce the incidence of sIVH.

The P2 committee developed unit-specific guidelines for the care of ELBW infants. In addition to specific practice changes described below, this included recommendations for potentially better neurodevelopmental care (Table 1).¹²  Guidelines were broken up by chapter, each containing domains for evidence-based practices prioritized by our key driver diagram.

The P2 committee began wide dissemination of the guidelines in March 2016, which included mandatory interactive educational sessions for nurses, providers, RTs, and other team members. Hard copies of the document were kept in binders throughout the unit, with an electronic copy available on our institutional intranet. The P2 committee met monthly during the implementation period to review the rate of IVH, identify improvement areas, and update guidelines.

The second intervention targeted the resuscitation and first hour after delivery of an ELBW infant, termed the “golden hour.” These guidelines were adapted from the California Perinatal Quality Care Collaborative (CPQCC) Delivery Room Toolkit.²⁰  Several practice interventions associated with reduction in IVH were recommended, including DCC for >30 seconds,^9,21,22  prevention of hypothermia²³  and hypoglycemia,²⁴  and promoting success with noninvasive ventilation (NIV).²⁵  An updated local golden hour checklist was developed, with reminder signs displayed in the resuscitation room.

The third intervention included specific recommendations to encourage the use of NIV, shown to be safe and potentially beneficial compared with universal invasive ventilation.^26,27  Given the association of intubation with sIVH among ELBW infants, especially in the first 72 hours,^6,28  a standardized criterion for intubation was created. This guideline, adapted from similar protocols in trials studying NIV in preterm infants,^27,29  recommended intubation only after failure with NIV, defined as significant hypoxemia (fraction of inspired oxygen: >0.45), hypercarbia (carbon dioxide: >70 mmol/L), acidosis (pH: <7.1), or prolonged apnea. This was a higher threshold for intubation than some providers were practicing, although, before this, there was large variation in practice. Providers and RTs participated in interactive educational sessions biannually starting in March 2016 dedicated to ventilation strategies for ELBW infants.

In July 2014, our obstetrics team began giving a second “rescue” course of betamethasone to eligible women at risk for preterm labor if the first dose of the initial course was >14 days prior.³⁰  Adherence to this protocol was marginal, with baseline data showing 57% of eligible mothers receiving this intervention. Additionally, new data emerged revealing an increased incidence of sIVH in ELBW infants born to mothers who received a single course of betamethasone >10 days before delivery, with potential reversal of this effect if a rescue course was given.^31,32 

The fourth intervention rolled out in December 2016, with a joint effort between the P2 committee and obstetrics groups to improve the rate of rescue betamethasone administration for women at risk for labor at <28 weeks’ gestation. Simultaneously, a practice change was disseminated to include patients receiving the first dose of the initial betamethasone course >7 to 10 days prior. Interventions included education at joint practice conferences, updated protocols, and inclusion of rescue betamethasone eligibility in the obstetrics sign-out and workroom checklists.

The fifth intervention began in July 2017 after review of current evidence and local data revealing that prophylactic indomethacin may reduce the risk of sIVH in select extremely preterm infants.^{10,11,33–35}  The neonatology group developed a clinical consensus protocol for indomethacin prophylaxis in the highest risk patients for IVH. Specifically, the protocol recommended indomethacin administration for any infant born <25 weeks’ gestation as well as high-risk infants born between 25 + 0/7 and 27 + 6/7 weeks’ gestation who remained intubated or newly intubated beyond 24 hours and were <72 hours old.

The final intervention was implemented in June 2018, with the opening of a small baby unit (SBU), called “The Grove.”^36,37  Every infant <28 weeks’ gestation or <1000 g was admitted to the SBU, located in a dedicated zone of the ICN chosen for its dark rooms. To achieve care consistency, NNPs specially trained in clinical management of ELBW infants provided the role of the front-line provider. To limit educational consequences for trainees, residents had the opportunity to take a “Grove Elective” to gain experience in the care of ELBW infants.

Patients were identified by using a quarterly report generated from the electronic medical record. Any inborn neonate delivered at <28 weeks’ gestation or <1000 g and admitted to the ICN was included in this report, which provided baseline and outcome data. Measures not included were gathered via retrospective chart review.

The primary outcome was sIVH, defined as any grade III or IV IVH diagnosed on head ultrasound per the Papile definition.²  Per our standard practice, screening head ultrasounds were performed at 1 week and 1 month for ELBW infants. Occasionally, head ultrasounds were performed earlier or more frequently in the setting of a clinical change or severe illness. Pediatric radiologists interpreted the sonograms, and there were no major changes in the framework for interpretation throughout the study period. Mortality was included as a secondary outcome, in addition to sIVH or mortality to account for infants who died before a head ultrasound.

We tracked several process measures based on adherence to practice recommendations. First was the rate of intubation within 24 hours of birth, intended to track success with optimization of NIV and creation of a standard criterion for intubation. The second was the overall rate of DCC for a minimum of 30 seconds. Third was the rate of eligible mothers who received any rescue betamethasone, which included the eligibility change in December 2016. The fourth process measure was the rate of eligible infants who received indomethacin prophylaxis on the basis of our protocol.

Our balancing measures included global outcomes for ELBW infants. Specifically, these were cystic periventricular leukomalacia (PVL), chronic lung disease (CLD), necrotizing enterocolitis (NEC) (Bell’s stage IIA or higher),³⁸  focal intestinal perforation (FIP), any retinopathy of prematurity (ROP), and corrected gestational age (CGA) at discharge. Outcomes were defined by the 2020 CPQCC data manual of definitions,³⁹  see the Table 2 footnotes for details.

Data were interpreted by using statistical process control charts created in Excel with QI Macros add-in (KnowWare, Denver, CO). The outcome measures, process measures, and balancing measures were assessed with p-charts. Control limits were set at 3 standard deviations from the mean, and the centerline was shifted when sustained special cause variation (eg, ≥8 values above or below the baseline centerline) was established.⁴⁰  Characteristics between groups were analyzed with 2-tailed t tests or χ² tests where appropriate. P values of <.05 were considered statistically significant.

During the project period, there were 280 inborn infants admitted to the ICN born <28 weeks’ gestation or <1000 g. Twelve patients with congenital anomalies were excluded. Of the 268 included patients, 127 were in the preintervention period, and 141 were in the postintervention period. There was no difference in the majority of baseline characteristics (Table 3), with the exception of significantly more male infants born in the postintervention period.

There was significant improvement in the primary outcome of the rate of sIVH. Special cause variation was achieved in quarter 4 of 2017, with ≥8 points below the centerline, resulting in a decrease in sIVH from 14% to 1.2% postintervention (Fig 2). The secondary outcome also showed significant improvement, with mortality decreasing from 20% to 10% since quarter 4 of 2017 (Fig 3). Finally, there was significant improvement in the combined outcome of sIVH or death, with a decrease from 23% to 14% since quarter 4 of 2017. Five infants died before a head ultrasound was performed (1 preintervention; 4 postintervention), and these were included in the secondary outcome analysis.

All process measures showed improvement during the implementation period. The rate of intubation within 24 hours significantly decreased postintervention, from 64% to 36% since quarter 1 of 2018 (Fig 4). Additionally, postintervention, there has been 100% compliance in eligible women receiving rescue betamethasone and eligible infants receiving prophylactic indomethacin. Although the rates of DCC showed significant improvement during the implementation period from 35% to 54%, this improvement did not continue during the sustainment period (Fig 5).

Finally, there were no differences in the balancing measures of PVL, CLD, NEC, or any ROP (Table 2). There was a significant reduction in FIP from 8% to 1% (P = .02), whereas CGA at discharge has increased postintervention, from 39.2 weeks to 40.4 weeks (P = .02).

Through the implementation of evidence-based neuroprotective strategies for the care of extremely preterm infants, including a golden hour bundle, promotion of NIV, consistent use of rescue antenatal betamethasone and indomethacin prophylaxis, and creation of a dedicated SBU, we significantly reduced the rate of sIVH by 91%, from 14% to 1.2%, with sustained improvement over 2 and a half years. We exceeded our original target of 7%, in addition to reducing the mortality rate by 50% during the same time period. This was accomplished without increasing markers of morbidity in this at-risk patient population.

sIVH is a strong risk factor for both death and adverse neurodevelopmental outcomes in preterm infants, and a review of baseline data revealed a higher rate of sIVH in our ICN, compared with that in national data. Although there has been mixed success with efforts aimed to reduce sIVH in preterm infants,^16–18  some centers have shown significant improvement after QI work focused on delivery room management and bedside neuroprotective strategies.^12–14,41  We set out to emulate these units, often in community hospital settings, that have shown improved outcomes for ELBW infants after rigorous QI implementations. Because there was little previous QI work in our ICN, we chose a multidisciplinary approach to our problem with all staff and providers working together to review evidence, create guidelines, and learn from each other about how to employ a QI framework specific to our unit. We expanded on golden hour and IVH bundles with additional layered potentially better practices. It is possible this approach was associated with our success, with a dramatic reduction in sIVH to 1.2%, well below the national average.

In this QI initiative, the timing of sIVH reduction is correlated with success in reducing intubation within 24 hours. A 2018 analysis of CPQCC data found that a decrease in delivery room intubation in preterm infants was one of the primary factors contributing to a reduction in sIVH over time.⁶  Other studies have shown the number of intubation attempts also contribute to the development of sIVH.²⁸  In our center, developing a criterion for intubation allowed for consistency in practice when previously there was none. Additionally, the involvement of the RT team was integral to the success in this process measure because they were the primary educators and champions for the use of NIV.

Significant improvement in our outcome and process measures was not attained until 18 months after the launch of P2, highlighting the challenge with an implementation of this magnitude. Multiple mandatory educational sessions in the ICN over the first year were necessary to disseminate new protocols and promote buy-in from staff. Once improvement was established, sustainability became the biggest challenge. In our center, the creation of the SBU, called “The Grove,” solidified practice changes from P2. Although SBUs alone have not yet been shown to improve the rate of sIVH, care of ELBW infants by a consistent team in a developmentally appropriate physical space has been shown improve outcomes.^37,38  In our particular ICN with a diverse patient population and many rotating trainees, the SBU created a space with an intentional focus on the unique QI processes that may improve outcomes for preterm infants, including sIVH.

Sustainability requires ongoing work, exemplified by the rate of DCC in which our initial improvement has waned. Although DCC has been shown to have many benefits in preterm infants, including reduction in any IVH, it has not been rigorously shown to reduce sIVH.⁴²  Therefore, it is possible that this intervention was not an impactful driver to our primary outcome. DCC is truly a multidisciplinary practice that relies on combined efforts between obstetric and pediatric teams who may have different care priorities.⁴³  Since review of these data, we have launched a joint QI effort with both disciplines to improve our DCC rate.⁴⁴ 

The study of balancing outcomes is essential when introducing interventions for ELBW infants. Morbidities in preterm infants are multifactorial and interrelated, with the possibility that improvement in one outcome can negatively influence another.^45,46  Since implementation of P2, there has been no increase in other morbidities such as PVL, ROP, NEC, and CLD. Interestingly, the rate of FIP has decreased after the introduction of risk-based indomethacin prophylaxis, despite described side effects with this intervention.^47,48  Our analysis did show that CGA at discharge has increased, which is consistent with national trends.⁴⁹  Length of stay is a mixed metric that may reflect changes in criteria for discharge or transfer, increased survival in this patient population, or practice changes associated with P2.⁵⁰  Regardless, the change in this balancing outcome warrants further investigation.

This study has several limitations. First, we did not track adherence to several of our process measures, including the golden hour checklist and recommendations for bedside neuroprotective care. This, in addition to overlapping interventions, makes it difficult to determine the degree each factor contributed to the observed improvement. Second, all improvements in outcome are association, not causation, as is the nature for QI initiatives. Finally, we do not know if reducing sIVH will improve patient outcomes. IVH is only one risk factor for poor neurodevelopment,⁵¹  and long-term follow-up is needed to determine the associated impact of our interventions.

Implementation of neuroprotective bedside potentially better practice guidelines, in addition to the consistent use of rescue betamethasone, indomethacin prophylaxis, and early NIV, was associated with a sustained reduction in the incidence of sIVH and mortality in ELBW infants. Monitoring of long-term neurodevelopmental outcomes is warranted.

We acknowledge all members of the P2 committee, whose tireless work and dedication were integral to success. We also acknowledge Dr Ronald Clyman for his participation and work to improve outcomes for our smallest infants.

CGA

corrected gestational age

CLD

chronic lung disease

CPQCC

California Perinatal Quality Care Collaborative

DCC

delayed cord clamping

ELBW

extremely low birth weight

FIP

focal intestinal perforation

ICN

Intensive Care Nursery

IVH

intraventricular hemorrhage

NEC

necrotizing enterocolitis

NNP

neonatal nurse practitioner

PVL

periventricular leukomalacia

P2

Premie Project

QI

quality improvement

ROP

retinopathy of prematurity

RT

respiratory therapist

SBU

small baby unit

sIVH

severe intraventricular hemorrhage

SPCC

statistical process control chart

UCSF

University of California, San Francisco