Acute Neuro-Functional Morbidity Upon Discharge From the Pediatric Intensive Care Unit After Critical Bronchiolitis Free

With the approval of our institutional review board, medical records of all patients <24 months of age admitted to our pediatric ICU from 2014 to 2016 with a primary diagnosis of bronchiolitis were reviewed. During the study period, all children with bronchiolitis at our institution requiring high-flow nasal cannula (HFNC), noninvasive ventilation, or invasive mechanical ventilation were cared for exclusively in the ICU. Children with preexisting tracheostomies were excluded, given their associated clinical characteristics (eg, severe neuro-functional comorbidities) and our local PICU admission criteria that may be unique and negatively impact the generalizability of our findings. Variables were collected from either our internal “Virtual PICU” quality improvement database (Virtual Pediatrics Systems, LLC; Los Angeles, CA) or our hospital’s electronic medical record (EMR). Virtual PICU data are entered by trained data collectors at our site, including primary diagnosis. Other collected variables included demographics, weight, past medical history, invasive mechanical ventilation (IMV), HFNC settings, ICU length of stay (LOS), and the lowest oxygen saturation (SpO₂) during ICU admission.

Race was categorized as African American, White, or other. Children at “high-risk” of ANFM were defined a priori by the presence of any of (1) a previous acute neurologic illness requiring hospitalization (seizures, hypoxic ischemic encephalopathy, intraventricular hemorrhage, or traumatic brain injury), (2) a chronic neurologic condition (eg, spastic quadriplegia, epilepsy), (3) history of feeding intolerance or oromotor dysfunction, or (4) a previous diagnosis associated with a substantial incidence of neurologic sequela (<29 weeks’ gestation, congenital heart disease, or chromosomal abnormalities).^14–16  Children not meeting any of these high-risk criteria were termed “low-risk” children.

The primary outcome was ANFM, a constellation of symptoms within the PICS umbrella. On the basis of previous work in the field,¹²  we defined ANFM at PICU discharge as documentation from an attending physician on the day of discharge from the PICU to the general ward or any of the subsequent 2 calendar days with any 1 of the following: (1) swallow difficulty (eg, language like “choked with oral feeds” or “uncoordinated swallowing”), (2) the use of nasogastric (NG) tube feeds, (3) hypotonia, or (4) lethargy. The presence of ANFM at hospital discharge was also collected from EMR documentation, including clinician notes and diagnostic studies, from the day of hospital discharge.

The overall rate of ANFM is reported using descriptive analyses. Categorical variables were compared between children with and without ANFM by using χ² tests. Continuous variables were compared with Wilcoxon rank tests. Analyses were repeated for the low-risk and high-risk subgroups of children. Data are shown as n (%) and median (IQR). Variables with a P value <.10 in univariate analyses were included in a multivariate logistic regression model with ANFM as the outcome variable. All statistical analysis was performed by using SigmaPlot v12.5 (San Jose, CA).

There were 417 children who met our inclusion criteria with a median age of 6 (2–11) months and median ICU LOS of 2.2 (1.0–4.3) days (Table 1). More than half (59%) of the children were male and 43% were African American. IMV was required in 14% of the subjects, and most of the noninvasively supported children received HFNC (236 [57%]). Among those children, the median peak flow rate was 1.1 (0.8–1.4) liters per minute per kilogram, and the peak fraction of inspired oxygen while on that maximum flow rate was 0.40 (0.3375–0.45).

In the whole cohort, one-sixth of subjects (70 [17%]) had documentation that met our criteria for ANFM near PICU discharge. The majority of these patients had swallow difficulties (96%), followed by NG requirement (46%), hypotonia (13%), and lethargy (1%) (Fig 1).

ANFM was associated with younger age, lower weight, high-risk status, use of IMV, and longer ICU LOS (Table 1). In the multivariate logistic regression model, male sex was associated with decreased risk of ANFM, whereas high-risk subgroup, invasive mechanical ventilation, and longer ICU LOS were associated with increased risk of ANFM (Table 4). Among the 70 children with ANFM near the time of PICU discharge, 32 (46%) had ANFM on the day of hospital discharge.

In the low-risk subgroup (n = 362), ANFM was identified in 44 (12%) children. In this subgroup, ANFM was associated with younger age, lower weight, use of IMV, and longer ICU LOS (Table 2). The majority of patients with ANFM in the low-risk subgroup had swallow difficulties (98%), followed by NG placement (52%), hypotonia (9%), and lethargy (2%) (Fig 1). Of these 44 children in the low-risk subgroup, 13 (30%) had ANFM at the time of discharge from the hospital.

In the high-risk subgroup (n = 55), ANFM was observed in 26 (47%) children. In this subgroup, the only studied variable associated with ANFM was longer ICU LOS (Table 3), and swallow difficulty was the most common sign of ANFM (92%). Most (73%) of these 26 children in the high-risk subgroup had signs of persistent morbidity at the time of hospital discharge.

In this retrospective observational study, we found that 1 out of every 6 children with a primary diagnosis of critical bronchiolitis had signs of ANFM at PICU discharge and almost half of them persisted to hospital discharge. Factors associated with ANFM at PICU discharge included preexisting comorbidities, need for endotracheal intubation, and longer PICU LOS. However, even after excluding the children with high-risk comorbidities, we found that 1 out of every 8 children had evidence of ANFM, including some with lower severity of illness who did not require invasive ventilation. As such, our data suggest that a substantial number of children, even those who may be deemed previously healthy or have moderate acute illness severity, may be at risk for developing PICU. Given that bronchiolitis is one of the most common indications for PICU admission, additional investigation is urgently needed to measure ANFM using even more precise measures in a prospective, multicenter cohort.^11,17 

There are few previous studies of PICS among survivors of critical bronchiolitis to which we can compare our findings. In one study of 236 previously healthy children cared for 20 years ago, ∼10% of survivors of critical bronchiolitis had signs of neurologic deficits at the time of PICU discharge, defined as lethargy, feeding difficulties, abnormalities of tone, or strabismus.¹²  This is comparable to our finding that 12% of children in our low-risk subgroup had signs of ANFM, although somewhat lower than the ∼18% rate of ANFM we previously observed in an administrative database.¹³  Many previous studies of heterogeneous PICU cohorts likely included patients with bronchiolitis, but the rate of PICS for the subgroup of children with bronchiolitis is not often reported.¹⁸  In one recent study from 31 US PICUs, ∼13% of bronchiolitis/asthma patients had a decline in functional status that had not returned to baseline 6 months after PICU discharge, a rate of morbidity that is similar to ours and, frankly, even more concerning given the chronicity and robust methodology.¹⁹  Although there may be value in comparing this study to ours, the outcomes are not directly comparable given that symptoms of ANFM could improve between discharge and 6 months, or even worsen if the child suffered an additional insult. In a much older study from Australia, 35 of 432 subjects had bronchiolitis and > 25% of those patients had poor to fair quality of life after PICU discharge.²⁰  It is also hard to compare our findings to previous studies that reveal that overall outcomes (eg, mean quality of life scores) are worse than control populations but do not reveal an incidence of unfavorable outcome.²¹  Clearly more work is needed to better understand long-term outcomes of critical bronchiolitis.

Our study-specific definition of high-risk status was created to identify children who may have poor long-term outcomes based on preexisting comorbidities (eg, prematurity, congenital heart disease) and children who may have had preexisting developmental delay.^14–16  Therefore, it is not surprising that ANFM was 4 times more common in our high-risk subjects. We observed a univariate trend toward female patients having an increased rate of ANFM (21% vs 14%) that reached statistical significance in the multivariate logistic regression model. This warrants additional assessment, especially given the recent finding that female PICU patients are less likely to receive mobilization therapy during critical illness.²²  Children with bronchiolitis may be an ideal population to study such risk factors because it is such a common illness and patients often have no pertinent comorbid conditions.

This is the largest study to date of outcomes among survivors of critical bronchiolitis. However, as a single-center retrospective study, there is a need for additional studies to understand the generalizability of our findings to different settings and larger samples. There are several limitations to this study. First, our ability to capture feeding difficulties, hypotonia, and lethargy was limited to data collected in the electronic health record for clinical purposes. It is possible that other children had these elements of ANFM but were not identified, necessitating prospective studies using precise definitions and thorough data collection. It is also possible that reliance on physician documentation of somewhat subjective findings may have overestimated some outcomes, and the clinical significance of a single mention of uncoordinated swallowing or mild hypotonia needs to be prospectively evaluated. Second, we did not have detailed information pertaining to pre-PICU neurofunctional status. It is likely that many subjects, particularly the high-risk patients, had abnormal development before PICU admission.

In addition, our description of high-risk subgroup and ANFM are somewhat novel, although they are based on previous literature of risk factors for severe bronchiolitis and unfavorable outcomes.^12,17  Third, we did not assess long-term outcomes and it is likely that at least some of the observed morbidities improved with time, although half of the post-PICU morbidities persisted at least to hospital discharge. Fourth, although our institution did not have mandatory protocols at the time of the study that could have falsely increased the rate of ANFM (eg, mandatory swallowing evaluations after mechanical ventilation, mandatory usage of NG tubes while in the PICU), local practice patterns by individual clinicians may have influenced our rate of ANFM. Prospective, multicenter studies with consistent utilization of diagnostic tests and clear criteria of impairments of different severities are needed to better establish generalizability, understand if clinical-practice- or patient-level factors are predominantly driving our results, and identify the best way to assess clinically important outcomes. Fifth, clinicians may use NG tubes on the basis of the degree of respiratory distress to avoid aspiration and not be representative of ANFM, although we only included tubes used at the time of transfer out of the PICU, at which time all children had improved their respiratory status to the point of coming off of HFNC or positive pressure ventilation, thereby increasing the likelihood that usage was related to oromotor dysfunction and not respiratory status. Sixth, some of the risk factors for ANFM we assessed are not modifiable, and additional studies are needed to identify ways to prevent and treat ANFM after critical bronchiolitis. Seventh, we selected variables to include in the multivariate model using results of univariate analyses, and not directed acyclic graphs.

EMR documentation of acute neuro-functional morbidity after discharge from the PICU is common with critical bronchiolitis, present in 1 out of 6 children in our study and persisting in half of these children to the time of hospital discharge. The risk factors for developing post-PICU ANFM included longer ICU length of stay, invasive mechanical ventilation, and previous neurologic symptoms. These physical impairments were often present even in the low-risk subgroup and among those who did not require invasive mechanical ventilation, suggesting a risk of postintensive care syndrome despite moderate acuity. Prospective studies using rigorous methods are urgently needed to better describe the long-term outcomes of critical bronchiolitis and identify modifiable risk factors to plan trials aimed at reducing morbidity from this common disease.