Reducing Severe Intraventricular Hemorrhage in Preterm Infants With Improved Care Bundle Adherence

Our NICU is a 47-bed level IV unit that serves a large referral area for high-risk maternal fetal medicine, with ∼500 admissions per year. Approximately 80 to 120 infants are admitted to our NICU with birth weights ≤1250 g or ≤30 gestational weeks annually, with ∼60% of those infants born extremely preterm (<28 weeks’ gestation). During the time of this study, we offered resuscitation starting at 23 0/7 weeks’ gestation. Approximately 95% of the admissions are inborn. NICU providers included attending neonatologists, advanced practice providers (APPs), and resident and fellow physicians. At our center, cranial ultrasound (CUS) is routinely recommended for all newborns who are either ≤30 weeks’ gestation or ≤1250 g at birth at 7 to 10 days after birth, although an earlier CUS can be obtained if a clinician deems it appropriate. If an IVH is detected on a CUS, repeat CUS testing is performed every 1 to 2 weeks pending stabilization and/or regression. IVH was classified using Volpe staging.¹⁰ 

We organized a focus group to determine factors related to high rates of severe IVH. An interprofessional team composed of NICU physicians, APPs, nurses, respiratory therapists, and occupational therapists formulated this focus group, which reviewed barriers to implementing the IVHPB and potential interventions on a monthly basis. Longitudinal data tracking for outcomes (IVH rates) and interventions provided real-time feedback and helped identify areas in need of ongoing improvement. Primary and secondary drivers were identified by key stakeholders; they are depicted in the key driver diagram (Fig 1). Plan-Do-Study-Act (PDSA) cycles were completed sequentially on the basis of the preceding PDSA cycle.

Our primary process measure was adherence to the IVHPB. Our primary outcome measure was incidence of severe IVH. The diagnosis of severe IVH included unilateral or bilateral grade III or IV IVH.¹⁰  We also had PDSA-cycle specific process measures that included the following: Completion rate of Research Electronic Data Capture (REDCap) survey, completion rates of in-service training, admission time, goal thermal regulation rates, number of times IVHPB was interrupted, and completion rates of audit.

Lack of adherence to the IVHPB was considered a possible reason for the lack of beneficial effect of the IVHPB. We surveyed all neonatal nursing and clinical staff with an institutional review board (IRB)-approved REDCap questionnaire to assess their knowledge, attitudes, and perceived barriers to implementation of the IVHPB. This questionnaire served as our initial adherence measure for the IVHPB. Only 24% of responders reported the IVHPB was frequently or almost always adhered to as written, despite the majority of NICU staff reporting they believed the IVHPB could contribute to a lower incidence of IVH with robust adherence. Overall, 62% reported that the IVHPB was either very important or somewhat important in decreasing incidence of IVH. The survey also identified multiple opportunities to promote adherence to the IVHPB, including improved multidisciplinary communication regarding the coordination of hands-on care. For PDSA cycle 1, we collected completion rate of the survey to assess impact.

On the basis of the survey of NICU caregivers, 89% of responders observed that in-service training of bedside staff was important to address barriers to implementing the IVHPB. Our initial intervention was to improve this knowledge. On the basis of postsession feedback, these trainings improved knowledge and provided practical tips to properly implement the IVHPB. For PDSA cycle 2, we collected completion rate of in-service training to assess impact.

This PDSA cycle was focused protocol change to implement start of IVHPB in the delivery room (DR) and to clearly delineate and execute effective bundle components. During the training sessions, we identified a lack of containment, excess head rotation with stabilizing the airway, and a high stimulation environment in the DR as potential areas for improvement. We realized that coordination among clinical teams (respiratory therapy, nursing, and providers) was essential in providing role clarity with specific tasks. We also ensured that all necessary equipment was available in the DR so there would be no need to change equipment or positioning of the infant after admission to the NICU. To improve adherence of the NICU staff to the IVHPB, we modified the clinical practice guidelines to clearly delineate strategies lessening stimulation and unnecessary interventions during the high-risk, initial 72-hour time period. These changes included deferring daily weights until after 72 hours, placement of cerebral near-infrared spectroscopy probes in the DR, infusion of feeds using automatic pumps, identification of team members to perform 2-person cares, and routine use of oral sucrose or breast milk to help calm the infant during procedures. For PDSA cycle 3, we subjectively reviewed thermal regulation, admission time, and understanding of protocol changes in focus groups to assess impact.

To improve awareness and compliance to the updated IVHPB, the phrase and mnemonic “making care times COUNT” was created by nursing staff to help with recall of critical components of the IVHPB. Specifically, “COUNT” refers to the following: Containment during 2-person cares, only disrupt sleep when medically necessary, use birth weight for 72 hours (no daily weighing), neuroprotective positioning starts in DR, and thoroughly assess infant from head to toe. Posters of the mnemonic COUNT were placed throughout the NICU to promote increased recall and adherence to the bundle.

Multidisciplinary care coordination and 2-person cares were identified as largest barrier to sustained adherence; therefore, during daily multidisciplinary huddles, IVHPB participants were identified by room number to help coordinate interdisciplinary care times and enlist extra personnel to complete 2-person cares to promote containment and minimize stimulation.

For the last 6 months of the initiative, charge nurses collected adherence data from bedside nurses every 12 hours during the 72-hour IVHPB for each neonate in the audit. For PDSA cycle 6, we collected completion rate of IVHPB participants to those who had an audit completed.

Three time periods are described: Preimplementation of the IVHPB before May 2016, preintervention period from May 2016 to June 2018, and intervention period from July 2018 to May 2020. Analysis for this QI initiative occurred between the preintervention period and intervention period. We collected baseline incidence of severe IVH among infants with birth weights ≤1250 g or ≤30 gestational weeks admitted to our NICU over 2 years (preintervention). Infants were excluded if they: (1) died within 24 hours after birth because of limited exposure to the IVHPB, or (2) died before obtaining head imaging. IRB-approved data were collected retrospectively from electronic medical records. Baseline patient characteristics were obtained, including sex, singleton, median gestational age, median birth weight, antenatal corticosteroid exposure, pressor use in first week after birth as a proxy to severity of illness, invasive ventilation, mortality, and severe IVH incidence.

We used focus group feedback throughout the interventions and objective metrics to assess completion of interventions such as in-service training rate. An audit assessing adherence to the IVHPB was performed during the last 6 months of the initiative. Specifically, the audit evaluated the following specifics: Number of occurrences the IVHPB was interrupted when not medically necessary, compliance of care times by the multidisciplinary team, and 2-person containment at all cares. The audit incorporated the adherence questionnaire from the initial REDCap survey to compare adherence using a fixed metric. “Medically necessary interruptions” were interruptions that the bedside nurse warranted as appropriate because of a clinical need of the neonate. A median adherence score (1–3; 1 = lowest adherence, 3 = highest adherence) was assigned to each neonate in the IVHPB during the audit period. Adherence scores were computed using feedback from the bedside nurse on an audit form, which was a combination of questions (same questions as on the preintervention questionnaire), plus the number of interruptions not medically necessary during the 12-hour shift. Adherence scores were assigned every 12 hours, resulting in 6 scores per participant over the 72-hour period; the median of these 6 scores was used (Supplemental Table 2).

The proper implementation and strict adherence of IVHPBs during the first 72 hours of life could result in poor bonding between mother and infant (lack of holding and minimal stimulation policy), which could ultimately reduce the breastfeeding rates. Another concern was that the IVHPB could demotivate providers from regular assessments and timely weaning off positive pressure ventilatory support because of minimal stimulation practice guidelines. We therefore documented the incidence of any breastfeeding at discharge, length of mechanical ventilation, and rates of bronchopulmonary dysplasia (BPD) in these infants as balancing measures.

Statistical analyses were preformed using GraphPad 2020 (San Diego, CA) and RStudio Version 1.2.5033.²⁴  Baseline patient characteristics in pre- and postintervention groups were compared using Fisher’s exact test and the Mann-Whitney U test. Data were plotted on a process control chart using Microsoft Excel 2013 (Redmond, WA). Between-group differences were calculated using a Mann-Whitney U test. The incidence of severe IVH was tracked monthly and plotted on a control chart after fixing the mean from baseline data 2016 to 2018. We used 8 consecutive points below the initial baseline as our indicator of special cause variation for our rate of IVH (Fig 2). The number of participants in IVHPB between severe IVH cases was tracked and plotted on a g-chart (Fig 3). We used 2 out of 3 points >2 SDs above mean as our indicator of special cause variation for our number of IVHPB infants without severe IVH in between severe IVH cases. Logistic regression adjusting for weight and gestational age was performed to determine the association between adherence score and incidence of IVH. A P value <.05 was considered statistically significant.

The University of Washington Medical Center IRB reviewed this project, and it was determined to be nonhuman subjects’ research because it was a QI initiative.

We received 70 responses to the survey, including 87% (26 of 30) of NICU providers (NICU attending physicians, fellows, and APPs) and 71% (44 of 62) of core NICU nurses.

Ninety-five percent of NICU nurses (n = 110 of 116) and 88% of respiratory therapists (n = 22 of 25) underwent in-service training, led by APPs with specialized training in positioning and containment of preterm newborns, and hands-on teaching utilizing a neonatal manikin.

Initiating IVHPB to start in the DR resulted in a shortened admission time, improved thermal regulation, and fewer interruptions to the IVHPB subjectively when reviewed in our multidisciplinary focus groups. Protocol changes outside of the DR were embedded into an already extensive clinical care guideline; therefore, feedback during our monthly QI meeting concluded this change had minimal impact because of poor visibility resulting in our next PDSA cycle.

The audit was completed at random on 78% (41 of 52) infants that qualified for the IVHPB during the last 6 months of the QI initiative.

The preintervention cohort included 240 neonates exposed to the IVHPB between May 2016 and June 2018. The postintervention cohort included 185 infants monitored between July 2018 and May 2020. Baseline patient characteristics were similar in pre- and postintervention groups (Table 1). Antenatal corticosteroid exposure (preintervention group 91.7%, postintervention group 93.0%; P = .94) and vasopressor use within the first week of life (preintervention group 21.6%, postintervention group 19.0%; P = .63) did not differ between the 2 groups (Table 1).

Regarding the primary outcome measure, the rate of severe IVH dropped from 9.8% to 2.4% on the basis of special cause variation during the intervention period (Fig 2). After the start of the QI initiative, the average number of patients between severe IVH events increased from 6.0 to 16.5, representing a greater than twofold improvement (Fig 3). Regarding the primary process measure, adherence to the IVHPB at start of the project, as measured by the retrospective survey, was 24%. Adherence to the IVHPB over final 6 months of implementation, as measured by an audit inclusive of same survey questions, was 88%. Higher adherence to the IVHPB was associated with a significant decrease in severe IVH (Fig 4). A lower adherence score was associated with an infant having a diagnosis of IVH (2.0 vs 3.0, P = .01) (Supplemental Table 3). However, the number of medically necessary interruptions was not associated with developing IVH (Supplemental Table 4).

No concerning changes in balancing measures were noted between baseline and intervention periods. Use of mother’s own milk at the time of discharge was 12.4% during the baseline period and 16.7% during the intervention period. The use of mechanical ventilation after initial resuscitation was 47% during the baseline period and 42% during the intervention period. Similarly, rates of BPD were 23% during baseline period and 24% during the intervention period.

By undergoing 6 PDSA cycles, we were able to significantly improve adherence to our center’s IVHPB for neonates born ≤30 weeks’ gestation or ≤1250 g. This improvement in adherence was in turn associated with a decrease in severe IVH from 10.4% to 3.8%. With QI interventions including education of staff, prompt initiation of the IVHPB, revision of guidelines, and process standardization, IVH rates decreased, without unintended consequences such as decreasing breast milk use at discharge, increased rates of mechanical ventilation, or increased incidence of BPD.

There is conflicting literature on the efficacy of IVHPBs.^18,20–22  For example, a recent study by Persad et al in 2021 showed no improvement in rates of severe IVH in patients exposed to an IVHPB; however, compared with our pre-IVHPB group, their IVHPB-exposed group had significantly lower birth weights and gestational ages, both known to be independently associated with higher IVH rates.²⁰  Another study by Murthy et al in 2020 reported improved outcomes with decreased death or severe brain injury in infants exposed to an IVHPB that included specific respiratory and hemodynamic components not present in many IVHPBs.²¹  Thus, differences in the efficacy of IVHPBs could be accounted for by variations in patient populations, by the heterogeneity of components included in IVHPBs, and through variation in adherence to these IVHPBs. Our findings describe the latter as a novel mechanism to reconcile the discordant outcomes on the variable efficacy of IVHPBs.^20,21 

This concept of the importance of following clinical guidelines to decipher efficiency is generalizable.²⁵  After instituting a series of interventions to increase adherence to the IVHPB, we saw incremental positive effects with an ultimate decrease in the rate of severe IVH. Thus, we learned that adherence to clinical guidelines is an independent factor, which has been shown to have a critical impact on improving health outcomes.^26–28  Similarly, lack of compliance to guidelines has demonstrated significant variability and waste in the US health care system,^29,30  or more simply stated, we may be disregarding optimal care solutions because of suboptimal implementation. By standardizing care and reducing variability of clinical practice, we achieved a positive impact on clinical outcomes. We also found that auditing compliance of guidelines was useful, and the feedback cycles allowed for continuous improvement ideas that could be implemented to further improve outcomes.

Study strengths included a multidisciplinary approach involving multiple PDSA cycles. We involved all important stakeholders in this effort and, by doing so, received a diversity of input that was very valuable, particularly from frontline staff. We set a specific, measurable, achievable, and realistic aim. We developed a wide range of factors we could address and ensured that each were considered. We had the support of the QI team for methodology and in helping us display data to best effect. We have shared our work with our leadership teams and continue to incorporate many of the change ideas into practice. We anticipate that sharing our experience with other teams in similar settings, we can improve outcomes more widely.

However, our study was not without limitations. This was a single-center QI initiative, which may limit the findings generalizability to other NICUs. Moreover, we only evaluated adherence metrics at the beginning and end of this study. The lack of evaluation of IVHPB adherence rates throughout the entire duration of the study could have introduced bias. However, our findings of a significant reduction in IVH suggest otherwise. Additionally, there may have been a severity of illness bias in that infants with greater severity of illness have an increased likelihood of medical intervention, decreasing bundle adherence, and are also more likely to suffer from IVH. Decreased bundle adherence may be associated with severe IVH in those infants despite not being causative. Furthermore, pressor use in the first week after birth was used as a proxy for severity of illness; however, this metric captures only 1 data point per patient and thus has limitations. Efforts were made to keep adherence metrics standardized, objective, and excusing for “medically necessary” interruptions during the audit. Stakeholders from different disciplines were included to mitigate bias and identify any process gaps; however, we failed to include the parental perception, which is a shortcoming of this study. During the study period, our CUS protocol did not include mastoid fontanel views, and therefore we were unable to comment on cerebellar hemorrhage. In addition, because our CUS protocol doesn’t include frequent and early CUS examinations, milder and early forms of white matter injury may not have been noted.

Overall, improving health care provider adherence to an evidence-based IVHPB was associated with a decreased rate of severe IVH from 9.8% to 2.4%, a 76% reduction in the incidence of severe IVH in ≤1250 g birth weight or ≤30 gestational week-old infants. Efforts focused on evaluating and promoting adherence to evidence-based guidelines, and prevention bundles are necessary to optimize their effectiveness on preventing adverse health outcomes.

We thank the nursing staff and leadership at the University of Washington Medical Center in Seattle, Washington, who provided care to the neonates throughout this QI initiative.

APP

advanced practice provider

BPD

bronchopulmonary dysplasia

CUS

cranial ultrasound

DR

delivery room

IRB

institutional review board

IVH

intraventricular hemorrhage

IVHPB

intraventricular hemorrhage prevention bundle

PDSA

Plan-Do-Study-Act

QI

quality improvement

REDCap

Research Electronic Data Capture