Neurologic and Functional Morbidity in Critically Ill Children With Bronchiolitis Steven L. Shein, MD1; Katherine N. Slain, DO1; Jason A. Clayton, MD, PhD1; Bryan McKee, MD1; Alexandre T. Rotta MD1; Deanne Wilson-Costello, MD2 Objectives: Neurologic and functional morbidity occurs in ~30% of PICU survivors, and young children may be at particular risk. Bronchiolitis is a common indication for PICU admission among children less than 2 years old. Two single-center studies suggest that greater than 10–25% of critical bronchiolitis survivors have neurologic and functional morbidity but those estimates are 20 years old. We aimed to estimate the burden of neurologic and functional morbidity among more recent bronchiolitis patients using two large, multicenter databases. Design: Analysis of the Pediatric Health Information System and the Virtual Pediatric databases. Setting: Forty-eight U.S. children’s hospitals (Pediatric Health Information System) and 40 international (mostly United States) children’s hospitals (Virtual Pediatric Systems). Patients: Previously healthy PICU patients less than 2 years old admitted with bronchiolitis between 2009 and 2015 who survived and did not require extracorporeal membrane oxygenation or cardiopulmonary resuscitation. Interventions: None. Neurologic and functional morbidity was defined as a Pediatric Overall Performance Category greater than 1 at PICU discharge (Virtual Pediatric Systems subjects), or a subsequent hospital encounter involving developmental delay, Division of Pediatric Critical Care Medicine, Department of Pediatrics, Rainbow Babies and Children’s Hospital, Cleveland, OH. 2 Division of Neonatology, Department of Pediatrics, Rainbow Babies and Children’s Hospital, Cleveland, OH. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ pccmjournal). Dr. Shein disclosed that, in a secondary analysis, the authors report associations with our primary outcome and prescription of common analgesics (fentanyl and morphine), sedatives (dexmedetomidine, lorazepam, and midazolam), neuromuscular blocking agents, and diuretics, where many of these drugs are used off-label in the PICU. Dr. Rotta received funding from BD/Carefusion, Vapotherm, and Elsevier. The remaining authors have disclosed that they do not have any potential conflicts of interest. Address requests for reprints to: Steven L. Shein, MD, Division of Pediatric Critical Care Medicine, 11100 Euclid Avenue, Rainbow Babies and Children’s Hospital, 3rd Floor, Pediatric Critical Care Medicine, Cleveland, OH 44106. E-mail: [email protected]
feeding tubes, MRI of the brain, neurologist evaluation, or rehabilitation services (Pediatric Health Information System subjects). Measurements and Main Results: Among 3,751 Virtual Pediatric Systems subjects and 9,516 Pediatric Health Information System subjects, ~20% of patients received mechanical ventilation. Evidence of neurologic and functional morbidity was present at PICU discharge in 707 Virtual Pediatric Systems subjects (18.6%) and more chronically in 1,104 Pediatric Health Information System subjects (11.6%). In both cohorts, neurologic and functional morbidity was more common in subjects receiving mechanical ventilation (27.5% vs 16.5% in Virtual Pediatric Systems; 14.5% vs 11.1% in Pediatric Health Information System; both p < 0.001). In multivariate models also including demographics, use of mechanical ventilation was the only variable that was associated with increased neurologic and functional morbidity in both cohorts. Conclusions: In two large, multicenter databases, neurologic and functional morbidity was common among previously healthy children admitted to the PICU with bronchiolitis. Prospective studies are needed to measure neurologic and functional outcomes using more precise metrics. Identification of modifiable risk factors may subsequently lead to improved outcomes from this common PICU condition. (Pediatr Crit Care Med 2017; XX:00–00) Key Words: bronchiolitis; pediatrics; postintensive care syndrome
1
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: 10.1097/PCC.0000000000001337
Pediatric Critical Care Medicine
B
ronchiolitis is the most common indication for hospitalization among young children in the United States, and related hospital costs exceed $1 billion annually (1–5). In a recent study of 43 U.S. children’s hospitals that contribute to the Pediatric Health Information System (PHIS) database, more than 20% of inpatients with bronchiolitis received care in a PICU (6). Invasive mechanical ventilation (MV) is frequently required and was used in ~25–40% of PICU patients with bronchiolitis in recent multicenter studies (6, 7). Though the risk of mortality is low in bronchiolitis, any critically ill patient who requires MV is at risk for neurologic and functional morbidity (NFM) after their critical illness resolves, sometimes referred to as “postintensive care syndrome (PICS)” (8, 9). In one study of 1,032 PICU patients, chronic NFM was observed in 29.2% of all survivors, including 32.7% of survivors of respiratory illness and 25.1% of children with a low (< 1%) www.pccmjournal.org
1
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
risk of mortality (10). Young children may be at particular risk for NFM due to age-related increased susceptibility for neurotoxicity from triggers such as hypoxia and commonly used sedatives, but they have been excluded from many studies of PICS (9, 11–13). In addition, respiratory syncytial virus (RSV), the most common cause of bronchiolitis, is associated with impaired learning in mice (14). Despite these risks and the high prevalence of bronchiolitis in the PICU, the neurofunctional outcomes of survivors of critical bronchiolitis have been poorly described in the literature. The little data available include one study in which ~10% of previously healthy critical bronchiolitis patients had acute neurologic or functional deficits at time of PICU discharge, and another study in which greater than 25% of critical bronchiolitis patients had a “fair” or worse quality of life measured months after PICU discharge (15, 16). In order to better understand the burden of post-PICU morbidity among survivors of critical bronchiolitis, we used two large, quality-controlled, multicenter databases to identify signs and symptoms of NFM in previously healthy children who were admitted to a PICU with bronchiolitis. Our primary goal was to estimate the burden of both acute and more long-term NFM in this population. We also performed limited exploratory analyses to identify risk factors associated with the presence of NFM.
METHODS This study was reviewed and approved by the Institutional Review Board of University Hospitals of Cleveland and the review boards for the Children’s Hospital Association (Overland Park, KS) and the Virtual Pediatric System LLC (Los Angeles, CA). Data Sources Data were obtained from two separate databases. The PHIS is an administrative database that contains inpatient, emergency department, ambulatory surgery, and observation encounterlevel data from over 45 not-for-profit, tertiary care pediatric hospitals in the United States. These hospitals are affiliated with the Children’s Hospital Association. Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and participating hospitals. Portions of the data submission and data quality processes for the PHIS database are managed by Truven Health Analytics (Ann Arbor, MI). For the purposes of external benchmarking, participating hospitals provide discharge/encounter data including demographics, diagnoses, and procedures. Nearly all of these hospitals also submit resource utilization data (e.g., pharmaceuticals, imaging, and laboratory) into the PHIS database. Data are deidentified at the time of data submission and are subjected to a number of reliability and validity checks before being included in the database. For this study, data from 48 hospitals were included. The PHIS database contains several “Flags” that use Clinical Transaction Classification and International Classification of Diseases, 9th Edition (ICD-9) codes to identify patients meeting criteria of interest. The Virtual PICU Systems (VPS) "Virtual PICU" database (Virtual Pediatric Systems, LLC, Los Angeles, CA) is a prospectively constructed, quality-controlled database with 2
www.pccmjournal.org
greater than 100 contributing ICUs. Most contributing centers are in the United States, in addition to centers in Canada (n = 3) and Saudi Arabia (n = 3). Trained data coordinators at participating institutions perform standardized collection of a mandatory dataset, which includes demographics, diagnoses, Pediatric Index of Mortality (PIM)-2 score components, and outcomes. Each center may elect to collect nonmandatory variables as well, including the Pediatric Overall Performance Category (POPC) score at admission and/or discharge. Every 3 months, each center decides a priori to either collect or not collect each nonmandatory variable for all subjects admitted to their PICU. For this study, 40 centers collected the necessary variables to contribute data to our analysis. Subjects Inclusion criteria for this study were admission between 2009 and 2015 to an ICU at a participating center at age less than 2 years old with a primary diagnosis code for bronchiolitis. Specifically, all included subjects from the VPS database had a primary diagnosis with a VPS diagnosis code of 466 (“bronchitis/bronchiolitis”), and all included subjects from the PHIS database had both a primary diagnosis with an ICD-9 code of 466.11 (bronchiolitis due to RSV) or 466.19 (bronchiolitis due to other organism) and an All Patient Refined DiagnosisRelated Group code of 138 (“bronchiolitis and RSV pneumonia”) (17). PICU admission was identified using the ICU Flag for PHIS subjects; only PICU subjects are included in the VPS database. For each identified subject, only the first admission meeting inclusion criteria was included in the analysis. Exclusion criteria were the presence of a significant comorbidity, use of extracorporeal membrane oxygenation (ECMO), cardiac arrest, or death prior to discharge. The presence of a significant comorbidity was identified using specific diagnosis codes that were “present at admission” for subjects in the VPS database (for list of diagnoses, see Appendix, Supplemental Digital Content 1, http://links.lww.com/PCC/A533) and the presence of any “complex chronic condition” for subjects in the PHIS database (18). Children treated with ECMO were identified using the procedure code for ECMO (code number 135) for subjects in the VPS database and the ECMO Flag for subjects in the PHIS database. Children with cardiac arrest were identified using the procedure code for cardiopulmonary resuscitation (133) for VPS subjects and specific ICD-9 codes (427.5 [cardiac arrest], 779.85 [cardiac arrest of newborn], 997.1 [cardiac complications; includes intraoperative cardiac arrest], and V12.53 [sudden cardiac arrest]) for PHIS subjects. Children with preexisting NFM were also excluded. In the VPS database, NFM was defined as a POPC score greater than 1. POPC scores range from 1 to 6. A POPC score of 1 indicates “good overall performance” of activities of daily life and “normal” cognitive function; higher scores indicate dysfunction (Appendix, Supplemental Digital Content 1, http://links.lww. com/PCC/A533) (19). Only children who had a documented POPC score of 1 at PICU admission were included in the analysis. Children with missing POPC data at PICU discharge were also excluded. In the PHIS database, NFM was defined as XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
a hospital encounter (inpatient or outpatient) that contained a code for developmental delay, feeding tubes, MRI of the brain, neurologist evaluation, or rehabilitation services (e.g., physical therapy, occupational therapy, and speech therapy) (see Appendix for specific definitions, Supplemental Digital Content 1, http://links.lww.com/PCC/A533). Children with an encounter that met those criteria before their bronchiolitis admission were excluded. Outcomes and Analysis The primary outcome was the presence of NFM, defined as a POPC score of greater than 1 at PICU discharge (VPS) or a subsequent hospital encounter after the bronchiolitis admission that met the above criteria (PHIS). Variables collected from both data sources included demographics and use of invasive MV (identified using the MV Flag for PHIS subjects, or a procedure code for endotracheal intubation [2] or MV [15] for VPS subjects). PICU length of stay (LOS) was extracted for VPS subjects, and hospital LOS was extracted for PHIS subjects. For each database, variables were compared between subjects with NFM and those without NFM using chi-square for categorical variables and Wilcoxon rank-sum for continuous variables. Variables that were associated with NFM in either database (p < 0.05) were included in multivariable logistic regression models with NFM as the dependent variable. In a secondary analysis, we aimed to identify risk factors associated with NFM. For VPS subjects, we extracted available laboratory results that were collected as part of the PIM2 score. Though PIM2 scores only incorporate data from the first 60 minutes of PICU care, some subjects had more than one value available, so both minimum and maximum values were analyzed using Wilcoxon rank-sum test. For PHIS subjects, prescription of common analgesics (fentanyl and morphine), sedatives (dexmedetomidine, lorazepam, and midazolam), neuromuscular blocking agents, and diuretics was collected using pharmacy billing codes. Please note that many of the evaluated drugs are used off-label. Associations between drug prescription and NFM were analyzed using chi-square among the subset of subjects who received MV. All data are presented TABLE 1.
as n (%) or median (interquartile range). All analyses were performed using SigmaPlot version 12.5 (Systat Software, San Jose, CA), and p value of less than 0.05 was considered significant.
RESULTS Data were analyzed for 3,751 subjects from the VPS database and 9,516 subjects from the PHIS database. Demographics for each cohort are shown in Table 1. In both cohorts, median age was 3 months, ~60% were male subjects, nearly half of subjects were Caucasian, and ~20% of subjects received MV. Evidence of NFM was present at PICU discharge in 707 VPS subjects (18.6%) (three of whom had POPC increase by ≥ 2 points) and at a subsequent hospital encounter in 1104 PHIS subjects (11.6%). In the PHIS database, rehabilitation was the most common specific encounter type (613 [6.4%]), followed by feeding tubes (407 [4.3%]), developmental delay (200 [2.1%]), brain MRI (155 [1.6%]), and neurologist evaluation (118 [1.2%]). Results of the bivariate analyses are shown in Tables 2 and 3. In both cohorts, NFM was more common in subjects receiving MV (27.5% vs 16.5% in VPS; 14.5% vs 11.1% in PHIS; both p < 0.001). Age, gender, and race/ethnicity were all associated with NFM in at least one of the cohorts, but no single demographic variable was associated with increased NFM in both cohorts. NFM was associated with longer hospital LOS in the PHIS cohort but not with PICU LOS in the VPS subjects. Results of the multivariate models are shown in Table 4. Again, MV was the only variable that was associated with increased NFM in both cohorts. As seen in Table 5, all evaluated medications with the exception of lorazepam were associated with NFM. Prescription of diuretics, fentanyl, midazolam, morphine, and neuromuscular blocking agents was all associated with increased rates of NFM, whereas prescription of dexmedetomidine was associated with a decreased rate of NFM. Among the VPS subjects, NFM was associated with lower minimum pH, higher Paco2 (both minimum Paco2 and maximum Paco2), and higher maximum sodium levels. There was a trend between the presence of severe acidosis (pH < 7.15) and NFM (35/83 [40.2%] vs 154/522 [29.5%]; p = 0.06).
Demographics for Both Cohorts
Variables
n Age (mo), median (IQR)
Pediatric Health Information System Dataset
9,516 3.3 (1.5–8.7)
Virtual Pediatric Systems Dataset
3,751 2.7 (1.3–7.1)
Male, n (%)
5,673 (59.6)
2,264 (60.4)
African-American, n (%)
1,659 (17.4)
539 (14.4)
Caucasian, n (%)
4,304 (45.2)
1,691 (45.1)
Hispanic, n (%)
2,163 (22.7)
899 (24.0)
Other, n (%)
1,390 (14.6)
622 (16.6)
Mechanically ventilated, n (%)
1,461 (15.4)
804 (21.4)
Length of stay (d), median (IQR)
Hospital: 4 (3–7)
ICU: 2.0 (0.9–4.3)
IQR = interquartile range.
Pediatric Critical Care Medicine
www.pccmjournal.org
3
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
TABLE 2. Bivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in the Virtual Pediatric Systems Database Variables
n (%) Age (mo), median (IQR)
Morbidity
No Morbidity
707 (18.8)
3,044 (81.2)
2.4 (1.2–6.5)
2.7 (1.4–7.3)
0.040 0.483
Gender, n (%) Male
418 (18.5)
1,846 (81.5)
Female
289 (19.4)
1,198 (80.6) < 0.001
Race/ethnicity, n (%) African-American
104 (19.3)
435 (80.7)
Caucasian
235 (13.9)
1,456 (86.1)
Hispanic
190 (21.1)
709 (78.9)
Other
178 (28.6)
444 (71.4) < 0.001
Respiratory support, n (%) Mechanically ventilated
221 (27.5)
583 (72.5)
Not mechanically ventilated
486 (16.5)
2,461 (83.5)
PICU length of stay, median (IQR)
p
2.0 (0.8–4.8)
2.0 (1.0–4.2)
0.617
IQR = interquartile range. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test.
TABLE 3. Bivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in the Pediatric Health Information System Database Variables
n (%) Age (mo), median (IQR)
Morbidity
No Morbidity
1,104 (11.6)
8,412 (88.4)
3.25 (1.7–7.4)
3.3 (1.5–9.0)
0.474 0.002
Gender, n (%) Male
705 (12.4)
4,968 (87.6)
Female
399 (10.4)
3,444 (89.6) < 0.001
Race/ethnicity, n (%) African-American
264 (15.9)
1,395 (84.1)
Caucasian
484 (11.2)
3,820 (88.8)
Hispanic
208 (9.6)
1,955 (90.4)
Other
148 (10.6)
1,242 (89.4) < 0.001
Respiratory support, n (%) Mechanically ventilated
212 (14.5)
1,249 (85.5)
Not mechanically ventilated
892 (11.1)
7,163 (88.9)
6 (4–8)
4 (3–6)
Hospital length of stay, median (IQR)
p
< 0.001
IQR = interquartile range. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test.
DISCUSSION In this study, we used two large, multicenter, quality-controlled databases to estimate the rates of NFM in previously healthy children who developed critical illness due to bronchiolitis. 4
www.pccmjournal.org
Even after excluding children at particularly high risk of neurologic injury due to cardiac arrest or use of ECMO, we found 18.6% of subjects had evidence of acute NFM and that 11.6% had evidence of more chronic morbidity (20, 21). Given the XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
TABLE 4. Multivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in Both Databases Pediatric Health Information System Variables
Coefficient
Age
–0.02
p
Virtual Pediatric Systems p
OR (95% CI)
Coefficient
OR (95% CI)
< 0.001
0.98 (0.97–0.99)
–0.0157
0.061
0.98 (0.97–1.00)
0.329
0.92 (0.78–1.09)
Male
0.202
0.002
1.22 (1.07–1.40)
–0.0846
African-American
0.475
< 0.001
1.61 (1.30–2.00)
–0.503
< 0.001
0.61 (0.46–0.80)
Caucasian
0.0421
0.673
1.04 (0.86–1.27)
–0.939
< 0.001
0.39 (0.31–0.49)
–0.105
0.355
0.90 (0.72–1.13)
–0.417
< 0.001
0.66 (0.52–0.84)
0.282
< 0.001
1.33 (1.13–1.56)
0.651
< 0.001
1.92 (1.59–2.32)
Hispanic Mechanically ventilated
OR = odds ratio. Results of multivariate logistic regression analyses of both cohorts.
large burden of bronchiolitis patients in the PICU, these estimated rates support that a substantial number of children are at risk for developing NFM. Overall PICU mortality is improving, and mortality from critical bronchiolitis is quite low (6, 9). This has led to an increasing focus on diagnosing and preventing PICS and chronic morbidity among survivors of critical illness (22, 23). Bone et al (24) also used changes in POPC score among VPS database subjects to identify newly acquired disability. They observed an overall morbidity rate of ~10%, though ~18% of subjects with an admission POPC score of 1 had a discharge POPC score greater than 1. There are many possible negative sequelae of NFM among survivors of critical bronchiolitis, even if the NFM is only short term. There are possible side effects to feeding tubes, including patient discomfort, parental anxiety, tube dislodgement, and incorrect placement leading to aspiration (25, 26). Young children are at increased risk for adverse effects from conscious sedation, which is generally required to obtain a brain MRI (27). Rehabilitation service appointments have direct financial costs and may have indirect costs if they impair a parent’s ability to work. Hospitalization for bronchiolitis negatively impacts the family’s emotional and physical well being, and this is likely to be worsened if the child has NFM (9, 28). If NFM persists beyond the short term, the child’s long-term growth and development may be negatively impacted. Though not the primary focus of our study, we attempted to identify risk factors associated with NFM. More severe respiratory acidosis at presentation is likely a marker of more severe disease, but it is important to assess in future studies if permissive hypercapnea, which is recommended for children with severe acute respiratory failure, is associated with unfavorable outcomes (29). Midazolam, which was associated with NFM in our study, is one of the several sedatives that the U.S. Food and Drug Administration recently warned against using in young children due to concerns of neurotoxicitiy during the age of peak synaptogenesis (30, 31). Prescription of opiates and benzodiazepines for 2 or more days has been associated with delusional memories of survivors of PICU care (32). We also found an association between NFM and diuretics, and a conservative Pediatric Critical Care Medicine
“dry lung” fluid strategy is associated with unfavorable chronic neurologic outcomes in adults with the acute respiratory distress syndrome (33). Alternatively, the association between dexmedetomidine and reduced NFM we observed supports prospective data that this drug may be neuroprotective in critical illness (34). Other risk factors that may influence neurologic and functional outcomes after critical illness include hypoxia, hypotension, inflammation, glucose dysregulation, and nutrition provided (23). Given our methods, these associations cannot be interpreted as causation. Prospective studies are needed to identify if any of these variables are modifiable risk factors for postbronchiolitis NFM. There are several limitations to our study, and our data are best viewed as estimates of the rate of NFM that can be used to generate hypotheses and design prospective studies of outcomes of children with critical bronchiolitis. First, our ability to identify NFM was dependent on variables available in the databases. Though POPC is a commonly used PICU outcome scale, and while feeding difficulties, neurology consultation, and rehabilitation services have been used to identify neurologic morbidity among bronchiolitis patients previously, we did not have access to more precise measures of NFM such as the Bayley Scales of Infant and Toddler Development and the AmielTilson Neurological Assessment (15, 35–38). POPC scores may be particularly difficult to categorize in infants and small children, though the original description of POPC reported validity in the subgroup of patients less than 1 year old (19) and a follow-up study showed that POPC scores in children less than or equal to 42 months old were significantly associated with Vineland Adaptive Behavioral Scores (39). Prospective studies using more ideal measures are needed. Similarly, we do not have specific results of neuroimaging or neurology consultation, and some children likely had negative examination findings. That is why we only included the utilization of these resources during subsequent hospital encounters in the definition for NFM and not utilization during the bronchiolitis admission. By only assessing for the use of these resources during subsequent hospital encounters, we identified children who a practitioner was sufficiently concerned about the presence of NFM even after the child was well enough to be discharged from the hospital www.pccmjournal.org
5
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
TABLE 5. Clinical Risk Factors Associated With Neurofunctional Morbidity Among Subjects in Both Databases Variables
Morbidity
No Morbidity
p
Pediatric Health Information System dataset, n (%) Dexmedetomidine
80 (11.6)
612 (88.4)
132 (17.2)
637 (82.8)
80 (23.7)
258 (76.3)
132 (11.8)
991 (88.2)
95 (22.4)
330 (77.7)
No fentanyl
117 (11.3)
919 (88.7)
Lorazepam
81 (17.1)
394 (82.9)
No lorazepam
131 (13.3)
855 (86.7)
Midazolam
106 (18.1)
481 (91.9)
No midazolam
106 (12.1)
768 (87.9)
51 (23.5)
166 (76.5)
161 (12.9)
1083 (86.7)
91 (22.9)
306 (77.1)
121 (11.4)
943 (88.6)
Pediatric Index of Mortality 2 (probability) (%)
0.75 (0.60–0.95)
0.78 (0.59–1.09)
0.175
Minimum pH
7.25 (7.185–7.32)
7.29 (7.21–7.35)
< 0.001
No dexmedetomidine Diuretic No diuretic Fentanyl
Morphine No morphine NMB No NMB
0.003 < 0.001 < 0.001 0.066 0.002 < 0.001 < 0.001
Virtual Pediatric Systems dataset, median (IQR)
Minimum Pao2 (mm Hg)
99 (70–111)
87 (66–116)
0.517
Maximum Paco2 (mm Hg)
59 (50–73)
56 (47.25–69.85)
0.016
49 (41.4–56.5)
46 (39.3–54.4)
0.017
Maximum WBC count (in cells x 10 /L)
9.0 (6.2–12.6)
8.5 (6.4–11.8)
0.615
Minimum WBC count (in cells x 10 /L)
8.3 (6.0–12.3)
8.3 (6.2–11.7)
0.605
Maximum blood sodium level (in mEq/L)
138 (134–141)
136 (133–139)
< 0.001
Minimum blood sodium level (in mEq/L)
135 (133–138)
135 (133–138)
0.865
Minimal Paco2 (mm Hg) 9
9
Maximum bicarbonate (mEq/L)
25 (23–28)
26 (23–28.6)
0.532
Minimum bicarbonate(mEq/L)
25 (23–27)
25 (22.2–27.55)
0.604
Maximum blood urea nitrogen (in mg/dL)
7 (4–9)
7 (5–9)
0.703
Maximum creatinine (mg/dL)
0.25 (0.2–0.3)
0.26 (0.2–0.3)
0.294
Maximum glucose (mg/dL)
108 (91–145)
106 (87.75–132.25)
0.113
Minimum glucose (mg/dL)
92 (79–112)
95 (77–117)
0.562
IQR = interquartile range, NMB = neuromuscular blocking agent. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test. Displayed p values are not adjusted for multiple comparisons; to adjust for multiple comparisons, the cut-off p value for significance would be 0.007 for the Pediatric Health Information System dataset and 0.003 for the Virtual Pediatric Systems dataset.
to bring the child back to the hospital for further evaluation. Utilization of these resources is much higher during the bronchiolitis admission, though likely a less specific marker for morbidity (40). Similarly, by only evaluating use of a feeding tube during a subsequent hospital encounter, we have likely identified children with NFM and not patients requiring a temporary 6
www.pccmjournal.org
feeding tube during their acute illness because of concerns about respiratory muscle fatigue, but data describing indication for feeding tubes were not available. Second, though both databases have strict quality control measures in place, any study using large datasets is at risk for inaccuracies. However, both PHIS and VPS employ methodology to reduce erroneous data XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
and have been used for research purposes (6, 24). Regardless, this could have led to false inclusion of subjects not meeting our inclusion/exclusion criteria, misclassification of the presence of NFM, and inaccuracies in the presence of risk factors for NFM. In database studies, these limitations are balanced by a robust sample size, and our study includes 50-fold more subjects than the largest previously published study of neurologic outcomes in children with critical bronchiolitis (15). Third, subjects from the PHIS database who had subsequent encounters at nonPHIS hospitals that would have otherwise met criteria for NFM would not have been captured in our study. Thus, the true rate of NFM among PHIS subjects may be higher than our estimate. Fourth, some centers participate in both VPS and PHIS so there is likely some overlap between cohorts. In conclusion, in two large, multicenter databases, neurologic or functional morbidity may be common among previously healthy children admitted to the PICU with bronchiolitis who did not suffer cardiac arrest or require treatment with ECMO. Risk factors may include disease severity and medication usage. Prospective studies are needed to measure neurologic and functional outcomes using more precise metrics. Identification of modifiable risk factors may subsequently lead to improved outcomes from this common PICU condition.
REFERENCES
1. Hasegawa K, Tsugawa Y, Brown DF, et al: Trends in bronchiolitis hospitalizations in the United States, 2000-2009. Pediatrics 2013; 132:28–36 2. Pelletier AJ, Mansbach JM, Camargo CA Jr: Direct medical costs of bronchiolitis hospitalizations in the United States. Pediatrics 2006; 118:2418–2423 3. Stang P, Brandenburg N, Carter B: The economic burden of respiratory syncytial virus-associated bronchiolitis hospitalizations. Arch Pediatr Adolesc Med 2001; 155:95–96 4. Stockman LJ, Curns AT, Anderson LJ, et al: Respiratory syncytial virus-associated hospitalizations among infants and young children in the United States, 1997–2006. Pediatr Infect Dis J 2012; 31:5–9 5. Yorita KL, Holman RC, Sejvar JJ, et al: Infectious disease hospitalizations among infants in the United States. Pediatrics 2008; 121:244–252 6. Gupta P, Beam BW, Rettiganti M: Temporal trends of respiratory syncytial virus-associated hospital and ICU admissions across the United States. Pediatr Crit Care Med 2016; 17:e343–e351 7. Pierce HC, Mansbach JM, Fisher ES, et al: Variability of intensive care management for children with bronchiolitis. Hosp Pediatr 2015; 5:175–184 8. Valentine SL, Randolph AG: Long-term outcomes after mechanical ventilation in children. In: Pediatric and Neonatal Mechanical Ventilation: From Basics to Clinical Practice. Rimensberger PC (Ed). Berlin, Heidelberg, Springer Berlin Heidelberg, 2015, pp 1489–1499 9. Herrup EA, Wieczorek B, Kudchadkar SR: Characteristics of postintensive care syndrome in survivors of pediatric critical illness: A systematic review. World J Crit Care Med 2017; 6:124–134 10. Taylor A, Butt W, Ciardulli M: The functional outcome and quality of life of children after admission to an intensive care unit. Intensive Care Med 2003; 29:795–800 11. Loepke AW: Developmental neurotoxicity of sedatives and anesthetics: a concern for neonatal and pediatric critical care medicine? Pediatr Crit Care Med 2010; 11:217–226 12. Ikonomidou C, Mosinger JL, Salles KS, et al: Sensitivity of the developing rat brain to hypobaric/ischemic damage parallels sensitivity to N-methyl-aspartate neurotoxicity. J Neurosci 1989; 9:2809–2818 13. Yager JY, Shuaib A, Thornhill J: The effect of age on susceptibility to brain damage in a model of global hemispheric hypoxia-ischemia. Brain Res Dev Brain Res 1996; 93:143–154
Pediatric Critical Care Medicine
14. Espinoza JA, Bohmwald K, Céspedes PF, et al: Impaired learning resulting from respiratory syncytial virus infection. Proc Natl Acad Sci U S A 2013; 110:9112–9117 15. Kho N, Kerrigan JF, Tong T, et al: Respiratory syncytial virus infection and neurologic abnormalities: Retrospective cohort study. J Child Neurol 2004; 19:859–864 16. Morrison AL, Gillis J, O’Connell AJ, et al: Quality of life of survivors of pediatric intensive care. Pediatr Crit Care Med 2002; 3:1–5 17. Parikh K, Hall M, Teach SJ: Bronchiolitis management before and after the AAP guidelines. Pediatrics 2014; 133:e1–e7 18. Feudtner C, Feinstein JA, Zhong W, et al: Pediatric complex chronic conditions classification system version 2: Updated for ICD-10 and complex medical technology dependence and transplantation. BMC Pediatr 2014; 14:199 19. Fiser DH: Assessing the outcome of pediatric intensive care. J Pediatr 1992; 121:68–74 20. Moler FW, Silverstein FS, Holubkov R, et al; THAPCA Trial Investigators: Therapeutic hypothermia after in-hospital cardiac arrest in children. N Engl J Med 2017; 376:318–329 21. Joffe AR, Lequier L, Robertson CM: Pediatric outcomes after extracorporeal membrane oxygenation for cardiac disease and for cardiac arrest: A review. ASAIO J 2012; 58:297–310 22. Pollack MM, Holubkov R, Funai T, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Collaborative Pediatric Critical Care Research Network: Pediatric intensive care outcomes: Development of new morbidities during pediatric critical care. Pediatr Crit Care Med 2014; 15:821–827 23. Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: Report from a stakeholders’ conference. Crit Care Med 2012; 40:502–509 24. Bone MF, Feinglass JM, Goodman DM: Risk factors for acquiring functional and cognitive disabilities during admission to a PICU. Pediatr Crit Care Med 2014; 15:640–648 25. Wood EG, Bunchman TE, Khurana R, et al: Complications of nasogastric and gastrostomy tube feedings in children with end stage renal disease. Adv Perit Dial 1990; 6:262–264 26. Meert KL, Daphtary KM, Metheny NA: Gastric vs small-bowel feeding in critically ill children receiving mechanical ventilation: A randomized controlled trial. Chest 2004; 126:872–878 27. Cravero JP, Beach ML, Blike GT, et al; Pediatric Sedation Research Consortium: The incidence and nature of adverse events during pediatric sedation/anesthesia with propofol for procedures outside the operating room: A report from the Pediatric Sedation Research Consortium. Anesth Analg 2009; 108:795–804 28. Lapillonne A, Regnault A, Gournay V, et al: Impact on parents of bronchiolitis hospitalization of full-term, preterm and congenital heart disease infants. BMC Pediatr 2012; 12:171 29. Pediatric Acute Lung Injury Consensus Conference Group: Pediatric acute respiratory distress syndrome: Consensus recommendations from the Pediatric Acute Lung Injury Consensus Conference. Pediatr Crit Care Med 2015; 16:428–439 30. U.S. Food and Drug Administration: FDA Drug Safety Communication: FDA review results in new warnings about using general anesthetics and sedation drugs in young children and pregnant women. 2016. Available at: http://www.fda.gov/Drugs/DrugSafety/ucm532356.htm. Accessed February 24, 2017 31. Lin EP, Soriano SG, Loepke AW: Anesthetic neurotoxicity. Anesthesiol Clin 2014; 32:133–155 32. Colville G, Kerry S, Pierce C: Children’s factual and delusional memories of intensive care. Am J Respir Crit Care Med 2008; 177:976–982 33. Mikkelsen ME, Christie JD, Lanken PN, et al: The adult respiratory distress syndrome cognitive outcomes study: Long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012; 185:1307–1315 34. Su X, Meng ZT, Wu XH, et al: Dexmedetomidine for prevention of delirium in elderly patients after non-cardiac surgery: A randomised, double-blind, placebo-controlled trial. Lancet 2016; 388:1893–1902 35. Bohmwald K, Espinoza JA, González PA, et al: Central nervous system alterations caused by infection with the human respiratory syncytial virus. Rev Med Virol 2014; 24:407–419 www.pccmjournal.org
7
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al 36. Silverstein FS, Slomine BS, Christensen J, et al; Therapeutic Hypothermia to Improve Survival After Cardiac Arrest Trial Group: Functional outcome trajectories after out-of-hospital pediatric cardiac arrest. Crit Care Med 2016; 44:e1165–e1174 37. Farris RW, Weiss NS, Zimmerman JJ: Functional outcomes in pediatric severe sepsis: Further analysis of the researching severe sepsis and organ dysfunction in children: A global perspective trial. Pediatr Crit Care Med 2013; 14:835–842 38. Vohr BR, Wright LL, Dusick AM, et al: Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National
8
www.pccmjournal.org
Institute of Child Health and Human Development Neonatal Research Network, 1993–1994. Pediatrics 2000; 105:1216–1226 39. Fiser DH, Long N, Roberson PK, et al: Relationship of pediatric overall performance category and pediatric cerebral performance category scores at pediatric intensive care unit discharge with outcome measures collected at hospital discharge and 1- and 6-month follow-up assessments. Crit Care Med 2000; 28:2616–2620 40. Shein S, McKee B, Slain K, et al: 1052: Changing patterns in neurologic medications among mechanically ventilated bronchiolitis patients. Critical Care Medicine 2016; 44:339
XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
feeding tubes, MRI of the brain, neurologist evaluation, or rehabilitation services (Pediatric Health Information System subjects). Measurements and Main Results: Among 3,751 Virtual Pediatric Systems subjects and 9,516 Pediatric Health Information System subjects, ~20% of patients received mechanical ventilation. Evidence of neurologic and functional morbidity was present at PICU discharge in 707 Virtual Pediatric Systems subjects (18.6%) and more chronically in 1,104 Pediatric Health Information System subjects (11.6%). In both cohorts, neurologic and functional morbidity was more common in subjects receiving mechanical ventilation (27.5% vs 16.5% in Virtual Pediatric Systems; 14.5% vs 11.1% in Pediatric Health Information System; both p < 0.001). In multivariate models also including demographics, use of mechanical ventilation was the only variable that was associated with increased neurologic and functional morbidity in both cohorts. Conclusions: In two large, multicenter databases, neurologic and functional morbidity was common among previously healthy children admitted to the PICU with bronchiolitis. Prospective studies are needed to measure neurologic and functional outcomes using more precise metrics. Identification of modifiable risk factors may subsequently lead to improved outcomes from this common PICU condition. (Pediatr Crit Care Med 2017; XX:00–00) Key Words: bronchiolitis; pediatrics; postintensive care syndrome
1
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: 10.1097/PCC.0000000000001337
Pediatric Critical Care Medicine
B
ronchiolitis is the most common indication for hospitalization among young children in the United States, and related hospital costs exceed $1 billion annually (1–5). In a recent study of 43 U.S. children’s hospitals that contribute to the Pediatric Health Information System (PHIS) database, more than 20% of inpatients with bronchiolitis received care in a PICU (6). Invasive mechanical ventilation (MV) is frequently required and was used in ~25–40% of PICU patients with bronchiolitis in recent multicenter studies (6, 7). Though the risk of mortality is low in bronchiolitis, any critically ill patient who requires MV is at risk for neurologic and functional morbidity (NFM) after their critical illness resolves, sometimes referred to as “postintensive care syndrome (PICS)” (8, 9). In one study of 1,032 PICU patients, chronic NFM was observed in 29.2% of all survivors, including 32.7% of survivors of respiratory illness and 25.1% of children with a low (< 1%) www.pccmjournal.org
1
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
risk of mortality (10). Young children may be at particular risk for NFM due to age-related increased susceptibility for neurotoxicity from triggers such as hypoxia and commonly used sedatives, but they have been excluded from many studies of PICS (9, 11–13). In addition, respiratory syncytial virus (RSV), the most common cause of bronchiolitis, is associated with impaired learning in mice (14). Despite these risks and the high prevalence of bronchiolitis in the PICU, the neurofunctional outcomes of survivors of critical bronchiolitis have been poorly described in the literature. The little data available include one study in which ~10% of previously healthy critical bronchiolitis patients had acute neurologic or functional deficits at time of PICU discharge, and another study in which greater than 25% of critical bronchiolitis patients had a “fair” or worse quality of life measured months after PICU discharge (15, 16). In order to better understand the burden of post-PICU morbidity among survivors of critical bronchiolitis, we used two large, quality-controlled, multicenter databases to identify signs and symptoms of NFM in previously healthy children who were admitted to a PICU with bronchiolitis. Our primary goal was to estimate the burden of both acute and more long-term NFM in this population. We also performed limited exploratory analyses to identify risk factors associated with the presence of NFM.
METHODS This study was reviewed and approved by the Institutional Review Board of University Hospitals of Cleveland and the review boards for the Children’s Hospital Association (Overland Park, KS) and the Virtual Pediatric System LLC (Los Angeles, CA). Data Sources Data were obtained from two separate databases. The PHIS is an administrative database that contains inpatient, emergency department, ambulatory surgery, and observation encounterlevel data from over 45 not-for-profit, tertiary care pediatric hospitals in the United States. These hospitals are affiliated with the Children’s Hospital Association. Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and participating hospitals. Portions of the data submission and data quality processes for the PHIS database are managed by Truven Health Analytics (Ann Arbor, MI). For the purposes of external benchmarking, participating hospitals provide discharge/encounter data including demographics, diagnoses, and procedures. Nearly all of these hospitals also submit resource utilization data (e.g., pharmaceuticals, imaging, and laboratory) into the PHIS database. Data are deidentified at the time of data submission and are subjected to a number of reliability and validity checks before being included in the database. For this study, data from 48 hospitals were included. The PHIS database contains several “Flags” that use Clinical Transaction Classification and International Classification of Diseases, 9th Edition (ICD-9) codes to identify patients meeting criteria of interest. The Virtual PICU Systems (VPS) "Virtual PICU" database (Virtual Pediatric Systems, LLC, Los Angeles, CA) is a prospectively constructed, quality-controlled database with 2
www.pccmjournal.org
greater than 100 contributing ICUs. Most contributing centers are in the United States, in addition to centers in Canada (n = 3) and Saudi Arabia (n = 3). Trained data coordinators at participating institutions perform standardized collection of a mandatory dataset, which includes demographics, diagnoses, Pediatric Index of Mortality (PIM)-2 score components, and outcomes. Each center may elect to collect nonmandatory variables as well, including the Pediatric Overall Performance Category (POPC) score at admission and/or discharge. Every 3 months, each center decides a priori to either collect or not collect each nonmandatory variable for all subjects admitted to their PICU. For this study, 40 centers collected the necessary variables to contribute data to our analysis. Subjects Inclusion criteria for this study were admission between 2009 and 2015 to an ICU at a participating center at age less than 2 years old with a primary diagnosis code for bronchiolitis. Specifically, all included subjects from the VPS database had a primary diagnosis with a VPS diagnosis code of 466 (“bronchitis/bronchiolitis”), and all included subjects from the PHIS database had both a primary diagnosis with an ICD-9 code of 466.11 (bronchiolitis due to RSV) or 466.19 (bronchiolitis due to other organism) and an All Patient Refined DiagnosisRelated Group code of 138 (“bronchiolitis and RSV pneumonia”) (17). PICU admission was identified using the ICU Flag for PHIS subjects; only PICU subjects are included in the VPS database. For each identified subject, only the first admission meeting inclusion criteria was included in the analysis. Exclusion criteria were the presence of a significant comorbidity, use of extracorporeal membrane oxygenation (ECMO), cardiac arrest, or death prior to discharge. The presence of a significant comorbidity was identified using specific diagnosis codes that were “present at admission” for subjects in the VPS database (for list of diagnoses, see Appendix, Supplemental Digital Content 1, http://links.lww.com/PCC/A533) and the presence of any “complex chronic condition” for subjects in the PHIS database (18). Children treated with ECMO were identified using the procedure code for ECMO (code number 135) for subjects in the VPS database and the ECMO Flag for subjects in the PHIS database. Children with cardiac arrest were identified using the procedure code for cardiopulmonary resuscitation (133) for VPS subjects and specific ICD-9 codes (427.5 [cardiac arrest], 779.85 [cardiac arrest of newborn], 997.1 [cardiac complications; includes intraoperative cardiac arrest], and V12.53 [sudden cardiac arrest]) for PHIS subjects. Children with preexisting NFM were also excluded. In the VPS database, NFM was defined as a POPC score greater than 1. POPC scores range from 1 to 6. A POPC score of 1 indicates “good overall performance” of activities of daily life and “normal” cognitive function; higher scores indicate dysfunction (Appendix, Supplemental Digital Content 1, http://links.lww. com/PCC/A533) (19). Only children who had a documented POPC score of 1 at PICU admission were included in the analysis. Children with missing POPC data at PICU discharge were also excluded. In the PHIS database, NFM was defined as XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
a hospital encounter (inpatient or outpatient) that contained a code for developmental delay, feeding tubes, MRI of the brain, neurologist evaluation, or rehabilitation services (e.g., physical therapy, occupational therapy, and speech therapy) (see Appendix for specific definitions, Supplemental Digital Content 1, http://links.lww.com/PCC/A533). Children with an encounter that met those criteria before their bronchiolitis admission were excluded. Outcomes and Analysis The primary outcome was the presence of NFM, defined as a POPC score of greater than 1 at PICU discharge (VPS) or a subsequent hospital encounter after the bronchiolitis admission that met the above criteria (PHIS). Variables collected from both data sources included demographics and use of invasive MV (identified using the MV Flag for PHIS subjects, or a procedure code for endotracheal intubation [2] or MV [15] for VPS subjects). PICU length of stay (LOS) was extracted for VPS subjects, and hospital LOS was extracted for PHIS subjects. For each database, variables were compared between subjects with NFM and those without NFM using chi-square for categorical variables and Wilcoxon rank-sum for continuous variables. Variables that were associated with NFM in either database (p < 0.05) were included in multivariable logistic regression models with NFM as the dependent variable. In a secondary analysis, we aimed to identify risk factors associated with NFM. For VPS subjects, we extracted available laboratory results that were collected as part of the PIM2 score. Though PIM2 scores only incorporate data from the first 60 minutes of PICU care, some subjects had more than one value available, so both minimum and maximum values were analyzed using Wilcoxon rank-sum test. For PHIS subjects, prescription of common analgesics (fentanyl and morphine), sedatives (dexmedetomidine, lorazepam, and midazolam), neuromuscular blocking agents, and diuretics was collected using pharmacy billing codes. Please note that many of the evaluated drugs are used off-label. Associations between drug prescription and NFM were analyzed using chi-square among the subset of subjects who received MV. All data are presented TABLE 1.
as n (%) or median (interquartile range). All analyses were performed using SigmaPlot version 12.5 (Systat Software, San Jose, CA), and p value of less than 0.05 was considered significant.
RESULTS Data were analyzed for 3,751 subjects from the VPS database and 9,516 subjects from the PHIS database. Demographics for each cohort are shown in Table 1. In both cohorts, median age was 3 months, ~60% were male subjects, nearly half of subjects were Caucasian, and ~20% of subjects received MV. Evidence of NFM was present at PICU discharge in 707 VPS subjects (18.6%) (three of whom had POPC increase by ≥ 2 points) and at a subsequent hospital encounter in 1104 PHIS subjects (11.6%). In the PHIS database, rehabilitation was the most common specific encounter type (613 [6.4%]), followed by feeding tubes (407 [4.3%]), developmental delay (200 [2.1%]), brain MRI (155 [1.6%]), and neurologist evaluation (118 [1.2%]). Results of the bivariate analyses are shown in Tables 2 and 3. In both cohorts, NFM was more common in subjects receiving MV (27.5% vs 16.5% in VPS; 14.5% vs 11.1% in PHIS; both p < 0.001). Age, gender, and race/ethnicity were all associated with NFM in at least one of the cohorts, but no single demographic variable was associated with increased NFM in both cohorts. NFM was associated with longer hospital LOS in the PHIS cohort but not with PICU LOS in the VPS subjects. Results of the multivariate models are shown in Table 4. Again, MV was the only variable that was associated with increased NFM in both cohorts. As seen in Table 5, all evaluated medications with the exception of lorazepam were associated with NFM. Prescription of diuretics, fentanyl, midazolam, morphine, and neuromuscular blocking agents was all associated with increased rates of NFM, whereas prescription of dexmedetomidine was associated with a decreased rate of NFM. Among the VPS subjects, NFM was associated with lower minimum pH, higher Paco2 (both minimum Paco2 and maximum Paco2), and higher maximum sodium levels. There was a trend between the presence of severe acidosis (pH < 7.15) and NFM (35/83 [40.2%] vs 154/522 [29.5%]; p = 0.06).
Demographics for Both Cohorts
Variables
n Age (mo), median (IQR)
Pediatric Health Information System Dataset
9,516 3.3 (1.5–8.7)
Virtual Pediatric Systems Dataset
3,751 2.7 (1.3–7.1)
Male, n (%)
5,673 (59.6)
2,264 (60.4)
African-American, n (%)
1,659 (17.4)
539 (14.4)
Caucasian, n (%)
4,304 (45.2)
1,691 (45.1)
Hispanic, n (%)
2,163 (22.7)
899 (24.0)
Other, n (%)
1,390 (14.6)
622 (16.6)
Mechanically ventilated, n (%)
1,461 (15.4)
804 (21.4)
Length of stay (d), median (IQR)
Hospital: 4 (3–7)
ICU: 2.0 (0.9–4.3)
IQR = interquartile range.
Pediatric Critical Care Medicine
www.pccmjournal.org
3
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
TABLE 2. Bivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in the Virtual Pediatric Systems Database Variables
n (%) Age (mo), median (IQR)
Morbidity
No Morbidity
707 (18.8)
3,044 (81.2)
2.4 (1.2–6.5)
2.7 (1.4–7.3)
0.040 0.483
Gender, n (%) Male
418 (18.5)
1,846 (81.5)
Female
289 (19.4)
1,198 (80.6) < 0.001
Race/ethnicity, n (%) African-American
104 (19.3)
435 (80.7)
Caucasian
235 (13.9)
1,456 (86.1)
Hispanic
190 (21.1)
709 (78.9)
Other
178 (28.6)
444 (71.4) < 0.001
Respiratory support, n (%) Mechanically ventilated
221 (27.5)
583 (72.5)
Not mechanically ventilated
486 (16.5)
2,461 (83.5)
PICU length of stay, median (IQR)
p
2.0 (0.8–4.8)
2.0 (1.0–4.2)
0.617
IQR = interquartile range. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test.
TABLE 3. Bivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in the Pediatric Health Information System Database Variables
n (%) Age (mo), median (IQR)
Morbidity
No Morbidity
1,104 (11.6)
8,412 (88.4)
3.25 (1.7–7.4)
3.3 (1.5–9.0)
0.474 0.002
Gender, n (%) Male
705 (12.4)
4,968 (87.6)
Female
399 (10.4)
3,444 (89.6) < 0.001
Race/ethnicity, n (%) African-American
264 (15.9)
1,395 (84.1)
Caucasian
484 (11.2)
3,820 (88.8)
Hispanic
208 (9.6)
1,955 (90.4)
Other
148 (10.6)
1,242 (89.4) < 0.001
Respiratory support, n (%) Mechanically ventilated
212 (14.5)
1,249 (85.5)
Not mechanically ventilated
892 (11.1)
7,163 (88.9)
6 (4–8)
4 (3–6)
Hospital length of stay, median (IQR)
p
< 0.001
IQR = interquartile range. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test.
DISCUSSION In this study, we used two large, multicenter, quality-controlled databases to estimate the rates of NFM in previously healthy children who developed critical illness due to bronchiolitis. 4
www.pccmjournal.org
Even after excluding children at particularly high risk of neurologic injury due to cardiac arrest or use of ECMO, we found 18.6% of subjects had evidence of acute NFM and that 11.6% had evidence of more chronic morbidity (20, 21). Given the XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
TABLE 4. Multivariate Analyses of Factors Associated With Neurofunctional Morbidity Among Subjects in Both Databases Pediatric Health Information System Variables
Coefficient
Age
–0.02
p
Virtual Pediatric Systems p
OR (95% CI)
Coefficient
OR (95% CI)
< 0.001
0.98 (0.97–0.99)
–0.0157
0.061
0.98 (0.97–1.00)
0.329
0.92 (0.78–1.09)
Male
0.202
0.002
1.22 (1.07–1.40)
–0.0846
African-American
0.475
< 0.001
1.61 (1.30–2.00)
–0.503
< 0.001
0.61 (0.46–0.80)
Caucasian
0.0421
0.673
1.04 (0.86–1.27)
–0.939
< 0.001
0.39 (0.31–0.49)
–0.105
0.355
0.90 (0.72–1.13)
–0.417
< 0.001
0.66 (0.52–0.84)
0.282
< 0.001
1.33 (1.13–1.56)
0.651
< 0.001
1.92 (1.59–2.32)
Hispanic Mechanically ventilated
OR = odds ratio. Results of multivariate logistic regression analyses of both cohorts.
large burden of bronchiolitis patients in the PICU, these estimated rates support that a substantial number of children are at risk for developing NFM. Overall PICU mortality is improving, and mortality from critical bronchiolitis is quite low (6, 9). This has led to an increasing focus on diagnosing and preventing PICS and chronic morbidity among survivors of critical illness (22, 23). Bone et al (24) also used changes in POPC score among VPS database subjects to identify newly acquired disability. They observed an overall morbidity rate of ~10%, though ~18% of subjects with an admission POPC score of 1 had a discharge POPC score greater than 1. There are many possible negative sequelae of NFM among survivors of critical bronchiolitis, even if the NFM is only short term. There are possible side effects to feeding tubes, including patient discomfort, parental anxiety, tube dislodgement, and incorrect placement leading to aspiration (25, 26). Young children are at increased risk for adverse effects from conscious sedation, which is generally required to obtain a brain MRI (27). Rehabilitation service appointments have direct financial costs and may have indirect costs if they impair a parent’s ability to work. Hospitalization for bronchiolitis negatively impacts the family’s emotional and physical well being, and this is likely to be worsened if the child has NFM (9, 28). If NFM persists beyond the short term, the child’s long-term growth and development may be negatively impacted. Though not the primary focus of our study, we attempted to identify risk factors associated with NFM. More severe respiratory acidosis at presentation is likely a marker of more severe disease, but it is important to assess in future studies if permissive hypercapnea, which is recommended for children with severe acute respiratory failure, is associated with unfavorable outcomes (29). Midazolam, which was associated with NFM in our study, is one of the several sedatives that the U.S. Food and Drug Administration recently warned against using in young children due to concerns of neurotoxicitiy during the age of peak synaptogenesis (30, 31). Prescription of opiates and benzodiazepines for 2 or more days has been associated with delusional memories of survivors of PICU care (32). We also found an association between NFM and diuretics, and a conservative Pediatric Critical Care Medicine
“dry lung” fluid strategy is associated with unfavorable chronic neurologic outcomes in adults with the acute respiratory distress syndrome (33). Alternatively, the association between dexmedetomidine and reduced NFM we observed supports prospective data that this drug may be neuroprotective in critical illness (34). Other risk factors that may influence neurologic and functional outcomes after critical illness include hypoxia, hypotension, inflammation, glucose dysregulation, and nutrition provided (23). Given our methods, these associations cannot be interpreted as causation. Prospective studies are needed to identify if any of these variables are modifiable risk factors for postbronchiolitis NFM. There are several limitations to our study, and our data are best viewed as estimates of the rate of NFM that can be used to generate hypotheses and design prospective studies of outcomes of children with critical bronchiolitis. First, our ability to identify NFM was dependent on variables available in the databases. Though POPC is a commonly used PICU outcome scale, and while feeding difficulties, neurology consultation, and rehabilitation services have been used to identify neurologic morbidity among bronchiolitis patients previously, we did not have access to more precise measures of NFM such as the Bayley Scales of Infant and Toddler Development and the AmielTilson Neurological Assessment (15, 35–38). POPC scores may be particularly difficult to categorize in infants and small children, though the original description of POPC reported validity in the subgroup of patients less than 1 year old (19) and a follow-up study showed that POPC scores in children less than or equal to 42 months old were significantly associated with Vineland Adaptive Behavioral Scores (39). Prospective studies using more ideal measures are needed. Similarly, we do not have specific results of neuroimaging or neurology consultation, and some children likely had negative examination findings. That is why we only included the utilization of these resources during subsequent hospital encounters in the definition for NFM and not utilization during the bronchiolitis admission. By only assessing for the use of these resources during subsequent hospital encounters, we identified children who a practitioner was sufficiently concerned about the presence of NFM even after the child was well enough to be discharged from the hospital www.pccmjournal.org
5
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al
TABLE 5. Clinical Risk Factors Associated With Neurofunctional Morbidity Among Subjects in Both Databases Variables
Morbidity
No Morbidity
p
Pediatric Health Information System dataset, n (%) Dexmedetomidine
80 (11.6)
612 (88.4)
132 (17.2)
637 (82.8)
80 (23.7)
258 (76.3)
132 (11.8)
991 (88.2)
95 (22.4)
330 (77.7)
No fentanyl
117 (11.3)
919 (88.7)
Lorazepam
81 (17.1)
394 (82.9)
No lorazepam
131 (13.3)
855 (86.7)
Midazolam
106 (18.1)
481 (91.9)
No midazolam
106 (12.1)
768 (87.9)
51 (23.5)
166 (76.5)
161 (12.9)
1083 (86.7)
91 (22.9)
306 (77.1)
121 (11.4)
943 (88.6)
Pediatric Index of Mortality 2 (probability) (%)
0.75 (0.60–0.95)
0.78 (0.59–1.09)
0.175
Minimum pH
7.25 (7.185–7.32)
7.29 (7.21–7.35)
< 0.001
No dexmedetomidine Diuretic No diuretic Fentanyl
Morphine No morphine NMB No NMB
0.003 < 0.001 < 0.001 0.066 0.002 < 0.001 < 0.001
Virtual Pediatric Systems dataset, median (IQR)
Minimum Pao2 (mm Hg)
99 (70–111)
87 (66–116)
0.517
Maximum Paco2 (mm Hg)
59 (50–73)
56 (47.25–69.85)
0.016
49 (41.4–56.5)
46 (39.3–54.4)
0.017
Maximum WBC count (in cells x 10 /L)
9.0 (6.2–12.6)
8.5 (6.4–11.8)
0.615
Minimum WBC count (in cells x 10 /L)
8.3 (6.0–12.3)
8.3 (6.2–11.7)
0.605
Maximum blood sodium level (in mEq/L)
138 (134–141)
136 (133–139)
< 0.001
Minimum blood sodium level (in mEq/L)
135 (133–138)
135 (133–138)
0.865
Minimal Paco2 (mm Hg) 9
9
Maximum bicarbonate (mEq/L)
25 (23–28)
26 (23–28.6)
0.532
Minimum bicarbonate(mEq/L)
25 (23–27)
25 (22.2–27.55)
0.604
Maximum blood urea nitrogen (in mg/dL)
7 (4–9)
7 (5–9)
0.703
Maximum creatinine (mg/dL)
0.25 (0.2–0.3)
0.26 (0.2–0.3)
0.294
Maximum glucose (mg/dL)
108 (91–145)
106 (87.75–132.25)
0.113
Minimum glucose (mg/dL)
92 (79–112)
95 (77–117)
0.562
IQR = interquartile range, NMB = neuromuscular blocking agent. Categorical variables compared with χ2 test. Continuous variables compared with Wilcoxon rank-sum test. Displayed p values are not adjusted for multiple comparisons; to adjust for multiple comparisons, the cut-off p value for significance would be 0.007 for the Pediatric Health Information System dataset and 0.003 for the Virtual Pediatric Systems dataset.
to bring the child back to the hospital for further evaluation. Utilization of these resources is much higher during the bronchiolitis admission, though likely a less specific marker for morbidity (40). Similarly, by only evaluating use of a feeding tube during a subsequent hospital encounter, we have likely identified children with NFM and not patients requiring a temporary 6
www.pccmjournal.org
feeding tube during their acute illness because of concerns about respiratory muscle fatigue, but data describing indication for feeding tubes were not available. Second, though both databases have strict quality control measures in place, any study using large datasets is at risk for inaccuracies. However, both PHIS and VPS employ methodology to reduce erroneous data XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Feature Article
and have been used for research purposes (6, 24). Regardless, this could have led to false inclusion of subjects not meeting our inclusion/exclusion criteria, misclassification of the presence of NFM, and inaccuracies in the presence of risk factors for NFM. In database studies, these limitations are balanced by a robust sample size, and our study includes 50-fold more subjects than the largest previously published study of neurologic outcomes in children with critical bronchiolitis (15). Third, subjects from the PHIS database who had subsequent encounters at nonPHIS hospitals that would have otherwise met criteria for NFM would not have been captured in our study. Thus, the true rate of NFM among PHIS subjects may be higher than our estimate. Fourth, some centers participate in both VPS and PHIS so there is likely some overlap between cohorts. In conclusion, in two large, multicenter databases, neurologic or functional morbidity may be common among previously healthy children admitted to the PICU with bronchiolitis who did not suffer cardiac arrest or require treatment with ECMO. Risk factors may include disease severity and medication usage. Prospective studies are needed to measure neurologic and functional outcomes using more precise metrics. Identification of modifiable risk factors may subsequently lead to improved outcomes from this common PICU condition.
REFERENCES
1. Hasegawa K, Tsugawa Y, Brown DF, et al: Trends in bronchiolitis hospitalizations in the United States, 2000-2009. Pediatrics 2013; 132:28–36 2. Pelletier AJ, Mansbach JM, Camargo CA Jr: Direct medical costs of bronchiolitis hospitalizations in the United States. Pediatrics 2006; 118:2418–2423 3. Stang P, Brandenburg N, Carter B: The economic burden of respiratory syncytial virus-associated bronchiolitis hospitalizations. Arch Pediatr Adolesc Med 2001; 155:95–96 4. Stockman LJ, Curns AT, Anderson LJ, et al: Respiratory syncytial virus-associated hospitalizations among infants and young children in the United States, 1997–2006. Pediatr Infect Dis J 2012; 31:5–9 5. Yorita KL, Holman RC, Sejvar JJ, et al: Infectious disease hospitalizations among infants in the United States. Pediatrics 2008; 121:244–252 6. Gupta P, Beam BW, Rettiganti M: Temporal trends of respiratory syncytial virus-associated hospital and ICU admissions across the United States. Pediatr Crit Care Med 2016; 17:e343–e351 7. Pierce HC, Mansbach JM, Fisher ES, et al: Variability of intensive care management for children with bronchiolitis. Hosp Pediatr 2015; 5:175–184 8. Valentine SL, Randolph AG: Long-term outcomes after mechanical ventilation in children. In: Pediatric and Neonatal Mechanical Ventilation: From Basics to Clinical Practice. Rimensberger PC (Ed). Berlin, Heidelberg, Springer Berlin Heidelberg, 2015, pp 1489–1499 9. Herrup EA, Wieczorek B, Kudchadkar SR: Characteristics of postintensive care syndrome in survivors of pediatric critical illness: A systematic review. World J Crit Care Med 2017; 6:124–134 10. Taylor A, Butt W, Ciardulli M: The functional outcome and quality of life of children after admission to an intensive care unit. Intensive Care Med 2003; 29:795–800 11. Loepke AW: Developmental neurotoxicity of sedatives and anesthetics: a concern for neonatal and pediatric critical care medicine? Pediatr Crit Care Med 2010; 11:217–226 12. Ikonomidou C, Mosinger JL, Salles KS, et al: Sensitivity of the developing rat brain to hypobaric/ischemic damage parallels sensitivity to N-methyl-aspartate neurotoxicity. J Neurosci 1989; 9:2809–2818 13. Yager JY, Shuaib A, Thornhill J: The effect of age on susceptibility to brain damage in a model of global hemispheric hypoxia-ischemia. Brain Res Dev Brain Res 1996; 93:143–154
Pediatric Critical Care Medicine
14. Espinoza JA, Bohmwald K, Céspedes PF, et al: Impaired learning resulting from respiratory syncytial virus infection. Proc Natl Acad Sci U S A 2013; 110:9112–9117 15. Kho N, Kerrigan JF, Tong T, et al: Respiratory syncytial virus infection and neurologic abnormalities: Retrospective cohort study. J Child Neurol 2004; 19:859–864 16. Morrison AL, Gillis J, O’Connell AJ, et al: Quality of life of survivors of pediatric intensive care. Pediatr Crit Care Med 2002; 3:1–5 17. Parikh K, Hall M, Teach SJ: Bronchiolitis management before and after the AAP guidelines. Pediatrics 2014; 133:e1–e7 18. Feudtner C, Feinstein JA, Zhong W, et al: Pediatric complex chronic conditions classification system version 2: Updated for ICD-10 and complex medical technology dependence and transplantation. BMC Pediatr 2014; 14:199 19. Fiser DH: Assessing the outcome of pediatric intensive care. J Pediatr 1992; 121:68–74 20. Moler FW, Silverstein FS, Holubkov R, et al; THAPCA Trial Investigators: Therapeutic hypothermia after in-hospital cardiac arrest in children. N Engl J Med 2017; 376:318–329 21. Joffe AR, Lequier L, Robertson CM: Pediatric outcomes after extracorporeal membrane oxygenation for cardiac disease and for cardiac arrest: A review. ASAIO J 2012; 58:297–310 22. Pollack MM, Holubkov R, Funai T, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Collaborative Pediatric Critical Care Research Network: Pediatric intensive care outcomes: Development of new morbidities during pediatric critical care. Pediatr Crit Care Med 2014; 15:821–827 23. Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: Report from a stakeholders’ conference. Crit Care Med 2012; 40:502–509 24. Bone MF, Feinglass JM, Goodman DM: Risk factors for acquiring functional and cognitive disabilities during admission to a PICU. Pediatr Crit Care Med 2014; 15:640–648 25. Wood EG, Bunchman TE, Khurana R, et al: Complications of nasogastric and gastrostomy tube feedings in children with end stage renal disease. Adv Perit Dial 1990; 6:262–264 26. Meert KL, Daphtary KM, Metheny NA: Gastric vs small-bowel feeding in critically ill children receiving mechanical ventilation: A randomized controlled trial. Chest 2004; 126:872–878 27. Cravero JP, Beach ML, Blike GT, et al; Pediatric Sedation Research Consortium: The incidence and nature of adverse events during pediatric sedation/anesthesia with propofol for procedures outside the operating room: A report from the Pediatric Sedation Research Consortium. Anesth Analg 2009; 108:795–804 28. Lapillonne A, Regnault A, Gournay V, et al: Impact on parents of bronchiolitis hospitalization of full-term, preterm and congenital heart disease infants. BMC Pediatr 2012; 12:171 29. Pediatric Acute Lung Injury Consensus Conference Group: Pediatric acute respiratory distress syndrome: Consensus recommendations from the Pediatric Acute Lung Injury Consensus Conference. Pediatr Crit Care Med 2015; 16:428–439 30. U.S. Food and Drug Administration: FDA Drug Safety Communication: FDA review results in new warnings about using general anesthetics and sedation drugs in young children and pregnant women. 2016. Available at: http://www.fda.gov/Drugs/DrugSafety/ucm532356.htm. Accessed February 24, 2017 31. Lin EP, Soriano SG, Loepke AW: Anesthetic neurotoxicity. Anesthesiol Clin 2014; 32:133–155 32. Colville G, Kerry S, Pierce C: Children’s factual and delusional memories of intensive care. Am J Respir Crit Care Med 2008; 177:976–982 33. Mikkelsen ME, Christie JD, Lanken PN, et al: The adult respiratory distress syndrome cognitive outcomes study: Long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012; 185:1307–1315 34. Su X, Meng ZT, Wu XH, et al: Dexmedetomidine for prevention of delirium in elderly patients after non-cardiac surgery: A randomised, double-blind, placebo-controlled trial. Lancet 2016; 388:1893–1902 35. Bohmwald K, Espinoza JA, González PA, et al: Central nervous system alterations caused by infection with the human respiratory syncytial virus. Rev Med Virol 2014; 24:407–419 www.pccmjournal.org
7
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
Shein et al 36. Silverstein FS, Slomine BS, Christensen J, et al; Therapeutic Hypothermia to Improve Survival After Cardiac Arrest Trial Group: Functional outcome trajectories after out-of-hospital pediatric cardiac arrest. Crit Care Med 2016; 44:e1165–e1174 37. Farris RW, Weiss NS, Zimmerman JJ: Functional outcomes in pediatric severe sepsis: Further analysis of the researching severe sepsis and organ dysfunction in children: A global perspective trial. Pediatr Crit Care Med 2013; 14:835–842 38. Vohr BR, Wright LL, Dusick AM, et al: Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National
8
www.pccmjournal.org
Institute of Child Health and Human Development Neonatal Research Network, 1993–1994. Pediatrics 2000; 105:1216–1226 39. Fiser DH, Long N, Roberson PK, et al: Relationship of pediatric overall performance category and pediatric cerebral performance category scores at pediatric intensive care unit discharge with outcome measures collected at hospital discharge and 1- and 6-month follow-up assessments. Crit Care Med 2000; 28:2616–2620 40. Shein S, McKee B, Slain K, et al: 1052: Changing patterns in neurologic medications among mechanically ventilated bronchiolitis patients. Critical Care Medicine 2016; 44:339
XXX 2017 • Volume XX • Number XXX
Copyright © 2017 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies. Unauthorized reproduction of this article is prohibited
