Resuscitative events in a level 4 NICU: Prevalence,characteristics,and outcomes of compressive versus non-compressive events

Abbreviations

BPD

Bronchopulmonary dysplasia

1 Introduction

Resuscitation events in the neonatal intensive care unit (NICU) are relatively rare occurrences. Depending on the population studied, events requiring cardiopulmonary resuscitation range from 10% to 21% in the extremely low birth weight population [1, 2] to a lower incidence of 1–6% in historical NICU studies [3–5] and 2% in modern-day quaternary-level NICU populations [6]. Most studies have found especially poor outcomes for extremely low birth weight (ELBW) infants [3, 7]. Some of the risk factors associated with poor outcome in these patients include lower gestational age [4], lower birth weight [3, 6–8], renal failure [5], sepsis [5], longer duration of cardiopulmonary resuscitation [9, 10], and use of vasopressors [2, 8]. Looking at a more diverse NICU population, a 2017 study from the level 4 NICU at Children’s Hospital of Philadelphia showed similar results with lower survival to discharge in those infants receiving vasopressors or antibiotics prior to the code event and in those with longer duration of chest compressions [6]. However, no published data exist on attributes associated with resuscitation events requiring positive pressure ventilation but no chest compressions (aka “staff assist” or “rapid response”). Literature on the characteristics and outcomes of resuscitation events in a quaternary level NICU also remains limited.

The objective of this study is to describe characteristics and outcomes of neonates who have compressive vs. non-compressive resuscitative events.

2 Methods

3 Results

During the study period, a total of 5,135 babies were admitted to the Children’s Mercy NICU, and a total of 641 resuscitation events occurred in 317 unique patients. Of the resuscitation events, 481 were rapid responses and 160 events required chest compressions. Of the events requiring chest compressions, 89 were long code events and 71 were short code events.

In the entire sample, the median gestational age at birth was 28.4 weeks and median birth weight was 1030 g (Table 1). Gestational age at birth and birth weight were not significantly different across groups. Median age at time of resuscitation event was 57 days. Those neonates with a long code event were younger, with a median age at time of code event of 25 days, followed by 54 days in the rapid response group and 99 days in the short code group (p < 0.0001).

RSS is defined as the product of fraction of inspired oxygen and the mean airway pressure and is an estimate of severity of neonatal lung disease. The median RSS prior to the event was 5.2 in the rapid response group, 6.0 in the short code group, and 9.0 in the long code group (p = 0.0007) (Table 2). Post-hoc tests found the RSS was significantly higher in the long code group compared to the rapid response group. The short code group was not significantly different from the other two groups.

In the 24 hours before the event, there was no statistical difference between groups’ BUN, creatinine, or urine output (Table 2). In the 24 hours following the event, urine output was significantly higher in the rapid response group (median 3.60 ml/kg/hr) as compared to the long code (median 2.70 ml/kg/hr) and short code (median 3.50 ml/kg/hr) groups (p = 0.0002). Creatinine in the 24 hours following the event was significantly higher in the long code group compared to the rapid response group (p = 0.0203). There was no difference between groups in BUN.

The incidence of positive blood, urine, cerebrospinal fluid, or tracheal aspirate cultures in the 48 hours prior to the event was not statistically different between groups (Table 3). In the 48 hours following the event, there was an increased incidence of overall positive cultures in the short code group (28.2%) and the long code group (28.1%) compared to the rapid response group (15.6%, p = 0.0002). However, there was no difference when the individual source of positive culture was analyzed.

Medication administration in the 12 hours prior to the event was evaluated between the three groups (Table 3). Patients in the long code group were significantly more likely to have received any vasopressor than patients in the short code and rapid response group (28.1% long code vs. 8.5% short code and 7.3% rapid response, p < 0.0001). Specifically, the long code group was significantly more likely to have received dopamine compared to the other two groups (p < 0.0001) and was significantly more likely to have received dobutamine (p = 0.0003) and epinephrine (p < 0.0001) compared to the rapid response group. Epinephrine administration did not include any epinephrine that was administered during the code event itself. The long code group was more likely to have received a fluid bolus (31.5% long code vs. 11.3% short code and 7.5% rapid response, p < 0.0001). The short code group was more likely to be on diuretics compared to the long code group and rapid response group (p = 0.0180), but the long code group was more likely to require IV electrolytes (p < 0.0001) compared to the short code group and rapid response group. The short code group was more likely to be on prednisone (p = 0.0255), inhaled steroids (p = 0.0188), or inhaled bronchodilators (p = 0.0007) compared to the rapid response group. Finally, the long code group was more likely to be on hydrocortisone compared to the short code group and rapid response group (27% vs. 11.3% vs. 13.5%, p = 0.0032).

Regarding the primary outcome, survival to discharge, those in the long code group were significantly less likely to survive to discharge compared to the short code and rapid response groups (p < 0.0001) (Table 4). The secondary outcome, survival in the 24-hour period after the code event, was statically significantly different between the three groups with the fewest survivors in the long code group followed by the short code group (p < 0.0001) (Table 4). Conversely, the long code group had a significantly shorter length of stay compared to the short code group and rapid response group (p = 0.0092) (Table 4). Regarding cause of death, the largest proportion of those with a long code had “Other” listed as their cause of death, followed by multi-organ system failure and then respiratory failure (Table 5). Those with short codes most frequently died of multi-organ system failure, followed by respiratory failure and other causes. For the rapid response group, the largest proportion died due to respiratory failure, followed by multi-organ system failure and then congenital anomaly/syndrome.

This retrospective cohort study described types, characteristics, and outcomes of resuscitative events within a level 4 NICU from 2012 to 2017. During the study period, there were 641 resuscitative events in 317 unique patients. The largest group by far was the rapid response group. However, those in the long code group had the worst outcomes, with decreased survival in the 24-hour period following the code event, decreased survival to discharge, and therefore a shorter length of stay. When analyzing factors leading up to the event, we found that the long code group was more likely to have a higher RSS, and more likely to require medications such as vasopressors, fluid bolus, IV electrolytes, and hydrocortisone. Following the code event, the long code group was more likely to have a higher creatinine.

There was no statistical difference among the groups in gestational age at birth or birth weight. The long code group were the youngest neonates, and the short code group were the oldest. In addition, there were differences noted between the short code group and rapid response group. Not only were those in the short code group older at the time of the event, they were also more likely to be on chronic medications such as diuretics, prednisone, inhaled steroids, and inhaled bronchodilators.

Consistent with prior studies, our study found a lower survival to discharge in those requiring a longer duration of chest compressions and those receiving vasopressors prior to the event [6]. Contrary to prior studies [3, 6–8], we found no statistical difference among the groups in terms of gestational age or birth weight. However, these prior studies were isolated to ELBW babies, whereas our study reflects all-comers in the NICU with a resuscitative event.

From our data, we hypothesize that the long code group is the youngest group because patients are most critically ill at the time of admission, reflected by the maximal medications and respiratory support and worse survival outcomes. Interestingly, the rapid response group was younger than the short code group; this difference may be explained in several ways. There may not be a clinical difference in the rapid response group and the short code group, in that the short code group may not have always required chest compressions since they were given for a brief amount of time. The short code group may be babies who have a slower recovery, and therefore the care teams thought it was necessary to start chest compressions; however, if given more time, these babies may have recovered with only increased respiratory support. Conversely, the short code group could represent a distinct cohort –babies with more established bronchopulmonary dysplasia (BPD), who require more time to recover from any inciting event. When assessing the diagnoses at hospital discharge, we found that those in the short code group and rapid response group did have a higher rate of BPD diagnosis, but it is not clear from our data set if these babies had that diagnosis at the time of the resuscitation event.

RSS was statistically higher in the long code group compared to the rapid response group but not compared to the short code group. Following our above hypotheses, the younger long code group may be more critically ill, requiring more oxygen and mean airway pressure. The older short code group may be infants with BPD requiring higher mean airway pressures depending on the severity of their particular lung pathology. A higher mean airway pressure would increase the RSS. So, although the RSS is not significantly different between these two groups, the reason for the elevated RSSs may be different.

Several medications were significantly more commonly used in the short code group compared to the rapid response groups: diuretics, prednisone, inhaled steroids, and inhaled bronchodilators. These medications are commonly used in the medical management of BPD. This finding corroborates with the older age and BPD diagnoses in the short code group; the younger rapid response group may ultimately have BPD but not yet be on chronic medications at the time of their event. The long code group was more likely to require IV electrolytes, which is seen more often when immediate electrolyte correction is required. These medications are common when a neonate is critically ill and not receiving enteral nutrition consistent with our previous hypotheses.

This study has several limitations. First, our study is retrospective and therefore limited by the quality of charting and data extraction. The documentation improved as the study period progressed, but at times it was difficult to determine the duration of chest compressions. This study attempted to differentiate short code events from long code events based on duration of chest compressions, but this delineation may be hampered by the accuracy of documentation. Time periods in which the data were assessed were arbitrary. For instance, medications were assessed in the 12 hours leading up to the event and RSS in the 6 hours leading up to the event. These time periods were chosen because they were thought to be sufficient to see a trend and potentially act on in clinical practice. However, it is possible that with different time thresholds, the data analysis may have yielded different results. Finally, the patient population is from a single institution’s level 4 NICU, so findings may not be generalizable to NICUs of differing levels of care or with a different patient population. At Children’s Mercy, the ELBW population comprises a smaller proportion of the census and a higher proportion of infants with genetic anomalies and other complex medical conditions.

A strength of this study is the comprehensive Code Blue Patient Safety Work Product database that the Children’s Mercy NICU maintains. This database keeps records of each resuscitative event that occurs within the NICU. The database is updated in real time and contains comprehensive data. The committee holds monthly meetings to discuss trends in events, determine follow-up needs, and assess reasons for the codes and whether interventions are necessary to change trends.

This study is important because it describes the characteristics and differences of those who have various types of resuscitative events in the NICU, with those who require chest compressions for longer than 1 minute having the worst clinical outcomes. Future research should include assessing similar data in a multi-institutional collaboration. A case-control study would be important to determine underlying risk factors for a code event. The characteristics that were statistically significant in this study could be a starting point to determine if they are present in the NICU population that does not have a resuscitation event during hospitalization.