Proadrenomedullin Predicts Severe Disease in Children With Suspected Community-acquired Pneumonia

Abstract

Background

Proadrenomedullin (proADM), a vasodilatory peptide with antimicrobial and anti-inflammatory properties, predicts severe outcomes in adults with community-acquired pneumonia (CAP) to a greater degree than C-reactive protein and procalcitonin. We evaluated the ability of proADM to predict disease severity across a range of clinical outcomes in children with suspected CAP.

Methods

We performed a prospective cohort study of children 3 months to 18 years with CAP in the emergency department. Disease severity was defined as mild (discharged home), mild–moderate (hospitalized but not moderate–severe or severe), moderate–severe (eg, hospitalized with supplemental oxygen, broadening of antibiotics, complicated pneumonia), and severe (eg, vasoactive infusions, chest drainage, severe sepsis). Outcomes were examined using proportional odds logistic regression within the cohort with suspected CAP and in a subset with radiographic CAP.

Results

Among 369 children, median proADM increased with disease severity (mild: median [IQR], 0.53 [0.43–0.73]; mild–moderate: 0.56 [0.45–0.71]; moderate–severe: 0.61 [0.47–0.77]; severe: 0.70 [0.55–1.04] nmol/L) (P = .002). ProADM was significantly associated with increased odds of developing severe outcomes (suspected CAP: OR, 1.68; 95% CI, 1.2–2.36; radiographic CAP: OR, 2.11; 95% CI, 1.36–3.38) adjusted for age, fever duration, antibiotic use, and pathogen. ProADM had an AUC of 0.64 (95% CI, .56–.72) in those with suspected CAP and an AUC of 0.77 (95% CI, .68–.87) in radiographic CAP.

Conclusions

ProADM was associated with severe disease and discriminated moderately well children who developed severe disease from those who did not, particularly in radiographic CAP.

Proadrenomedullin (proADM) predicts severe outcomes in adults with community-acquired pneumonia (CAP); pediatric data are limited. This prospective study of 369 children with suspected CAP found that proADM was associated with disease severity, particularly in radiographic CAP.

Community-acquired pneumonia (CAP) in children results in costly and frequent emergency department (ED) visits and hospitalizations [1]. Although accurate assessment of disease severity is critical to making management decisions in children with CAP, prognostic tools remain elusive. Clinical decisions are made by subjective impressions that do not accurately predict clinical outcomes.

Biomarkers offer objective information about the host response, potentially overcoming the limitations of gestalt or unreliable signs and symptoms. Conventional biomarkers, including white blood cell count (WBC), C-reactive protein (CRP), and procalcitonin (PCT), have had conflicting results for predicting severe outcomes in pediatric CAP, with the greatest potential to assist with ruling out only the most severe outcomes [2]. None of these markers were shown to be strong discriminators of overall disease severity in children with CAP.

More recently, midregional proadrenomedullin (proADM), a vasodilatory peptide with antimicrobial and anti-inflammatory properties, has shown promise in predicting illness severity in sepsis and lower respiratory tract infection (LRTI) in adults. Adrenomedullin (ADM) is upregulated by proinflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-1β [3, 4], as well as hypoxia [5]. Adrenomedullin is challenging to measure due to its short half-life and existence of a binding protein; however, the midregional proADM fragment is stable and representative of ADM levels [6]. Several meta-analyses have found proADM to be a robust predictor of short-term complications and mortality in adults with CAP, stronger than CRP and PCT in most cases [7–9]. The addition of proADM to existing severity scores improves their ability to predict severe outcomes [10–12].

To date, proADM has been inadequately studied in children. Three studies performed outside of the United States found that proADM was predictive of pediatric CAP complications, while one found no difference in proADM between severe and nonsevere disease [13–16]. Although proADM is a promising prognostic marker, these studies are limited by their small sample sizes, or narrow or inconsistent severity definitions.

Using a prospective cohort of children with suspected CAP, we sought to evaluate the association and predictive ability of proADM with disease severity across a range of clinical outcomes. If proADM predicts severity, it contributes to shifting clinical decision making from use of subjective features towards objective measures to improve risk stratification. It would focus disposition and resource-use decisions on those at most risk while avoiding unnecessary hospitalizations and therapies in those who do not require them.

METHODS

Study Design, Setting, and Participants

Catalyzing Ambulatory Research in Pneumonia Etiology and Diagnostic Innovations in Emergency Medicine (CARPE DIEM) was a prospective cohort study that enrolled children age 3 months to 18 years with signs and symptoms of LRTI who received a chest radiograph (CXR) for suspicion of CAP and presented to the ED from July 2013 to December 2017. The study was approved by the Cincinnati Children’s Hospital Medical Center Institutional Review Board. Written informed consent was obtained from all legal guardians and verbal assent was obtained from children aged 11 years or older. Details of this study have been previously published [2, 17].

Children enrolled in CARPE DIEM who had focal CXR findings indicating suspected CAP and had blood drawn were included in this analysis. We excluded children hospitalized 14 days or less before the index ED visit and those with immunocompromising or chronic medical conditions that predispose to severe or recurrent pneumonia (eg, immunodeficiency, chronic corticosteroid use, cystic fibrosis, chronic lung disease, malignancy, sickle cell disease, congenital heart disease, tracheostomy-dependent patients, neuromuscular disorders affecting the lungs). Children with a history of aspiration or aspiration pneumonia were excluded. Patients previously enrolled within 30 days of their study ED visit were excluded to ensure a distinct infection episode.

Study Procedures

After informed consent, the patient and/or parent provided demographic and historical information. Clinicians reported clinical signs and disposition decisions. Blood, urine, and/or nasopharyngeal swabs were collected at enrollment. Samples for proADM analysis were drawn into either a Li-heparin or ethylenediaminetetraacetic acid (EDTA) tube and placed into a 4°C refrigerator immediately after collection. Plasma was separated from serum and frozen at −80°C within 48 hours of collection. Clinical data were extracted from the electronic medical record. Patients received a follow-up phone call 7–10 days after they were discharged to assess disease course.

Given the variability in CXR interpretation, 2 radiologists masked to clinical information independently reviewed all study radiographs after the ED visit. Radiographic results were classified as “normal,” “definite/probable atelectasis,” “atelectasis vs pneumonia,” or “definite/probable pneumonia.” Radiologic CAP was defined as both radiologists classifying CXR findings as “atelectasis vs pneumonia” or “definite/probable pneumonia” [18]. In cases of disagreement, the attending radiologist’s read from the ED visit was used as a tiebreaker. If there was persistent disagreement, the radiologists came to a consensus on a final interpretation during an in-person meeting.

Outcome Measurements

Our primary outcome was disease severity, as assessed after the ED visit [2]. Mild disease was defined as discharge from the ED without return for hospitalization within 7 days. Mild–moderate disease was defined as those hospitalized on initial visit or upon revisit within 7 days but not meeting moderate–severe or severe criteria. Moderate–severe disease was classified as hospitalization with at least 1 of the following: at least 1 intravenous (IV) fluid bolus, continuous IV fluids for more than 12 hours, supplemental oxygen, broadening of antibiotics from aminopenicillin to any other antibiotic class, complicated pneumonia (moderate–large pleural effusion, metastatic infection, lung abscess, or lung necrosis), or presumed sepsis (systemic inflammatory response syndrome with receipt of antibiotics and ≥40 mL/kg of IV bolus fluid). Severe disease required at least 1 of the following: treatment in the intensive care unit (ICU), positive-pressure ventilation (continuous positive-pressure ventilation, bilevel positive-pressure ventilation, or intubation with mechanical ventilation), vasoactive infusions, chest drainage, extracorporeal membrane oxygenation (ECMO), severe sepsis/septic shock (using validated International Classification of Diseases, 9th revision, codes), or death [2, 19]. In an effort to distinguish the most severe outcomes while also encompassing the full spectrum of clinical outcomes, we examined disease severity in 2 ways: (1) a 4-tiered outcome classified as mild, mild–moderate, moderate–severe, and severe and (2) a group of binary comparisons (mild with the other categories; mild and mild–moderate with moderate–severe and severe; and severe with the other categories).

Biomarker Measurements

Details of the WBC, CRP, and PCT assays have been described previously [2]. Plasma extracted from blood stored in EDTA or Li-heparin tubes was frozen at −80°C and shipped on dry ice prior to thawing and proADM assay. ProADM values were assayed using TRACE (Time-Resolved Amplified Cryptate Emission) technology using the B.R.A.H.M.S Kryptor Compact (Brahms, Hennigsdorf, Germany), with an analytical detection limit of 0.08 nmol/L [20]. No proADM measurements in our cohort fell below the limit of detection (LOD). For CRP and PCT, biomarker measurements below the LOD were replaced with estimates equal to the LOD divided by the square root of 2 [21].

Statistical Analysis

The median biomarker concentrations with 25th and 75th percentiles were reported. Differences across outcome categories were tested using a Wilcoxon rank-sum test. Statistical tests were not conducted if any outcome category had a prevalence of less than 5% [22].

Biomarker concentrations were transformed using a log₂ transformation for modeling, as they were right-skewed. Odds ratios (ORs) were thus interpreted as the multiplicative change in odds for a doubling of each biomarker concentration. Logistic regression models were used to calculate unadjusted and adjusted ORs for binary comparisons in children with CAP and radiographic CAP. Proportional odds logistic regression models were used by considering disease severity an ordered factor. Multivariate models were adjusted for age (as a natural cubic spline with 3 degrees of freedom), receipt of antibiotics prior to arrival, duration of fever, and viral pathogen detection, as these variables were hypothesized a priori to affect biomarker concentrations and disease severity outcomes.

We characterized the discriminatory ability of proADM by creating receiver operator curves (ROCs) and calculating the area under the curve (AUC). Using an empirical cutpoint estimation method for maximizing classification accuracy, we calculated the optimal threshold for classification and calculated the resulting test characteristics with 95% confidence intervals (CIs) [23].

Statistical computing was conducted in R (version 3.5.3; R Foundation for Statistical Computing), SAS (version 9.4; SAS Institute), and Stata (version 16; StataCorp).

RESULTS

There were 1142 children enrolled in CARPE DIEM, of which proADM was available in 369 for the current analysis. Their mean age was 6.1 years (SD, 4.7 years) and 177 (48%) were female (Table 1, Supplementary Figure 1). Of the 369 children, 94 (25.5%) had mild disease, 99 (26.8%) had mild–moderate disease, 135 (36.6%) had moderate–severe disease, and 41 (11.1%) had severe disease.

Table 1.

 Cohort Characteristics

	Overall Cohort (N = 369)
Age, mean (SD), years	6.1 (4.7)
Male sex, n (%)	192 (52)
Race (n = 368), n (%)
White	249 (67.7)
Black	99 (26.9)
Other	20 (5.4)
Home antibiotics, n (%)	145 (39.3)
Home antibiotic class,a n (%)
Aminopenicillin	79 (55.2)
Cephalosporin	26 (18.2)
Macrolide	24 (16.8)
Other	37 (25.9)
Fever, n (%)	324 (87.8)
Duration of fever, mean (SD), days	4.1 (5.1)
Chest radiograph results,b n (%)
No atelectasis or pneumonia	13 (3.5)
Definite/probable atelectasis	189 (51.2)
Atelectasis vs pneumonia	28 (7.6)
Definite/probable pneumonia	139 (37.7)
Virus detected (n = 338 for viral testing performed),a n (%)	183 (54.1)
Rhinovirus/enterovirus	79 (23.4)
Respiratory syncytial virus	57 (16.9)
Influenza	22 (6.5)
Parainfluenza	10 (3)
Human metapneumovirus	17 (5)
Adenovirus	1 (0.3)
Bocavirus	6 (1.9)
Coronavirus	9 (2.7)
Mycoplasma pneumoniae (n = 326 for Mycoplasma testing performed), n (%)	24 (7.4)
Disease severity, n (%)
Mild	94 (25.5)
Mild–moderate	99 (26.8)
Moderate–severe	135 (36.6)
Severe	41 (11.1)

^a Patients may have received >1 antibiotic class or had >1 virus detected.

^bAs determined by 2 radiologists independently assessing chest radiographs.

The overall median proADM concentration was 0.587 nmol/L (interquartile range [IQR], 0.458–0.764; range, 0.2–18.3). Median proADM concentrations increased with increasing disease severity, with median concentrations of 0.53 nmol/L (IQR, 0.43–0.73) in mild, 0.56 nmol/L (IQR, 0.45–0.71) in mild–moderate, 0.61 nmol/L (IQR, 0.47–0.77) in moderate–severe, and 0.70 nmol/L (IQR, 0.55–1.04) in severe disease (P = .002) (Figure 1).

Figure 1.

 Proadrenomedullin and disease severity in pediatric suspected (A) and radiographic CAP (B). The median proADM concentration is represented by the middle line in each box. The lower and upper borders of the box represent the 25th and 75th percentiles, respectively. The whiskers represent (1.5 ×IQR). A, n = 369; B, n = 167. “Outside values” defined as >75th percentile + 1.5 × IQR: Mild, 1.0384, 1.2922, 1.458, 2.732; Mild–Moderate: 1.242, 1.541; Moderate–Severe: 1.218, 1.3557, 1.435, 1.735, 1.799, 4.475; Severe: 13.704. Abbreviations: CAP, community-acquired pneumonia; IQR, interquartile range; proADM, proadrenomedullin.

In examining the individual components of the severity outcome, proADM was statistically elevated in patients who were hospitalized, required the ICU, had hospital length of stay of more than 24 hours, receipt of IV fluid boluses or continuous IV fluids for more than 12 hours, and receipt of supplemental oxygen (Supplementary Table 1). For outcomes with a prevalence of less than 5%, statistical comparisons were not pursued; however, proADM was elevated in those with complicated pneumonia, sepsis, vasoactive infusions, chest drainage procedures, positive-pressure ventilation, ECMO, and death.

In adjusted analyses, proADM was statistically associated with severe outcomes (Table 2). When proADM was included in a single adjusted model with CRP and PCT, only proADM was associated with severe disease in children with CAP (OR, 2.02; 95% CI, 1.06–4.03), while CRP (OR, .98; 95% CI, .74–1.31) and PCT (OR, 1.05; 95% CI, .83–1.31) were not. Similar results were found in those with radiographic CAP (proADM: OR, 3.88; 95% CI, 1.57–11.53; PCT: OR, .85; 95% CI, .61–1.18; CRP: OR, 1.04; 95% CI, .64–1.76).

Table 2.

 Association of Proadrenomedullin With Disease Severity in Children With Community-acquired Pneumonia

	Unadjusted OR (95% CI)	Adjusted ORa (95% CI)
Suspected CAP
Mild vs (mild–moderate + moderate–severe + severe)	1.55 (1.04–2.38)	1.49 (.95–2.43)
Mild + (mild–moderate vs moderate–severe + severe)	1.74 (1.23–2.51)	1.53 (1.06–2.28)
Mild + (mild–moderate + moderate–severe vs severe)	2.22 (1.4–3.64)	2.15 (1.27–3.78)
Proportional odds logistic regression	1.82 (1.34–2.49)	1.68 (1.2–2.36)
Radiographic CAPb
Mild vs (mild–moderate + moderate–severe + severe)	1.77 (1.06–3.18)	1.83 (1.03–3.48)
Mild + (mild–moderate vs moderate–severe + severe)	1.93 (1.21–3.25)	1.9 (1.15–3.36)
Mild + (mild–moderate + moderate–severe vs severe)	2.95 (1.6–6.07)	2.56 (1.35–5.42)
Proportional odds logistic regression	2.16 (1.44–3.36)	2.11 (1.36–3.38)

Abbreviations: ED, emergency department; CAP, community-acquired pneumonia; CI, confidence interval; OR, odds ratio.

^aAdjusted for age, duration of fever, antibiotics prior to ED arrival, and viral pathogen detected.

^bAs determined by agreement of 2 radiologists independently reviewing chest radiographs after study ED visit.

A proADM threshold of 0.66 nmol/L was found to maximize sensitivity and specificity. At this threshold, proADM had an AUC of 0.64 (95% CI, .56–.72) in those with suspected CAP and an AUC of 0.77 (95% CI, .68–.87) in those with radiographic CAP (Table 3).

Table 3.

 Performance Characteristics of Proadrenomedullin to Predict Severe Versus Nonsevere Outcomes in Children With Community-acquired Pneumonia

	Threshold	AUC (95% CI)	Sensitivity, % (95% CI)	Specificity, % (95% CI)	PPV, % (95% CI)	NPV, % (95% CI)	LR+ (95% CI)	LR− (95% CI)
Suspected CAP (n = 369)	0.66	0.64 (0.56, 0.72)	61 (44.5, 75.8)	67.1 (61.7, 72.1)	18.8 (12.5, 26.5)	93.2 (89.2, 96.1)	1.85 (1.39, 2.47)	0.58 (0.39, 0.86)
Radiographic CAP (n = 167)	0.66	0.77 (0.68, 0.87)	83.3 (58.6, 96.4)	71.1 (63.2, 78.3)	25.9 (15.3, 39)	97.2 (92.2, 99.4)	2.89 (2.1, 4)	0.23 (0.08, 0.66)

Abbreviations: AUC, area under the curve; CAP, community-acquired pneumonia; CI, confidence interval; LR+ = likelihood ratio positive, LR- = likelihood ratio negative; NPV, negative-predictive value; PPV, positive-predictive value.

DISCUSSION

In this prospective cohort study, proADM measured at the time of ED visit was associated with the development of severe outcomes in children with CAP, with the greatest effect seen in those with radiographic CAP. When included in a model with CRP and PCT, only proADM was associated with disease severity. The high negative predictive value and low negative likelihood ratio for severe disease in radiographic CAP suggests that proADM may be most useful to rule out severe disease; further study is required to further understand the clinical utility of proADM in pediatric CAP.

In adults with LRTI, proADM is a stronger discriminator of severe outcomes compared with other conventional biomarkers, including WBC, CRP, and PCT [7, 8, 24]. The proADM assay is performed on the B·R·A·H·M·S KRYPTOR instrument, with an assay time of 29 minutes, a sample volume requirement of 26 μL of EDTA plasma, and is stable for 72 hours at room temperature [25]. The assay is CE marked in Europe, but not yet Food and Drug Administration cleared. One meta-analysis of 12 studies of adults with CAP found a pooled OR of 6.8 (95% CI, 4.7–10.1) for short-term mortality and 5.0 (95% CI, 3.9–6.5) for short-term complications with elevated proADM. The AUC for short-term mortality ranged from .72 to .89 [8]. Another meta-analysis of 12 studies of adults with CAP in the ED found moderate discriminatory ability of proADM for mortality (AUC, 0.76; 95% CI, .72–.8) and short-term complications (AUC, 0.74; 95% CI, .7–.78) [7]. The addition of proADM to existing clinical severity scores improved score performance [8, 26].

Four published studies examine proADM in children with CAP. One study of 433 previously healthy children hospitalized for radiographic CAP in Italy found that CRP, PCT, and proADM all had an AUC of 0.65 or less for severe disease [13]. This study included children with both alveolar and nonalveolar densities, and used the British Thoracic Society’s severity criteria. These criteria have limited sensitivity and only fair ability to discriminate diagnoses and interventions that require hospitalization [27]. In another study of 66 hospitalized children, proADM and CRP were associated with severity, defined by a respiratory clinical score, with proADM having a greater discriminatory ability over CRP [16]. Two small studies, one in the ED (n = 88) and another in hospitalized children (n = 50), found that proADM concentrations were higher in complicated versus uncomplicated CAP [14, 15]. We attempted to overcome these limitations of small samples and narrow severity definitions by enrolling a sufficient sample size and evaluating relevant outcomes across a range of disease severity. Our results expand on past results by examining clinical outcomes across the entire disease severity spectrum and by evaluating the ability of proADM to predict severity in a cohort of children with suspected CAP, in addition to a subset with radiologist-confirmed CAP.

When compared with CRP and PCT, proADM is more strongly associated with severe disease [2]. One possible explanation may be etiology. The association between proADM and severity appears to be less dependent on etiology compared with PCT. In adults, plasma levels of proADM were similar among patients infected with viruses, typical bacteria, and mixed infections, with no association between proADM and etiology [24]. This is in contrast with PCT, which is elevated in typical bacterial infections, but not in viral or atypical bacterial infections [28]. The lack of PCT increase in viral infections is thought to be due to virus-stimulated macrophage synthesis of interferon-α, which then inhibits TNF synthesis [29] and PCT expression. Most cases of CAP in young children are presumed to be viral [30]; therefore, the lack of association of PCT with disease severity in children may be due to the high prevalence of viral infections where expression of PCT is blunted. This hypothesis is supported by this and our prior work, where PCT was substantially elevated in severe outcomes that are known complications of bacterial CAP (eg, empyema), but was not elevated in patients with other outcomes, such as supplemental oxygen requirement or hypoxia, which can be seen regardless of etiology [2]. Due to limitations of current etiologic testing in CAP, only 2 children in CARPE DIEM had typical bacteria detected by culture. Therefore, we cannot draw conclusions about differences in proADM in children with definitive viral versus bacterial disease. Adrenomedullin (and thus proADM) has been found to be upregulated by hypoxia [5]. Thus, another hypothesis for our findings is that we observed hypoxia-induced expression of proADM, as hypoxia is associated with disease severity in CAP regardless of etiology.

Our results suggest that proADM has potential to be used during an ED visit to improve clinical decision making about disposition and treatment intensity in children with radiographic CAP, in combination with clinical judgment. With high negative-predictive value (NPV) and low negative likelihood ratio, perhaps its greatest utility is to rule out severe disease in those with low levels; however, as this is one of the few studies to examine proADM in children with CAP, validation of these findings is necessary before implementation into clinical practice. Although proADM discriminates severe disease moderately well and to a similar degree as in adults, it does not perform well enough to be used in isolation, similar to other biomarkers. Extrapolating from the adult CAP literature, it is likely that the combination of clinical features and 1 or more biomarkers will provide improved performance over either alone. In 877 adults with CAP, proADM had an AUC of 0.73 and the Risk of Early Admission to the ICU score had an AUC of 0.76 for requiring mechanical ventilation or vasoactive infusions within 3 days of ED presentation. When combined, the AUC was 0.81 with a 20% net reclassification of patients with the addition of proADM [10]. Another study of 491 adults with CAP compared the addition of biomarkers with 3 commonly used clinical scores (Severe Community Acquired Pneumonia, Pneumonia Severity Index, Confusion-Urea-Respiratory Rate-Blood Pressure-age 65 [CURB-65] score) to predict pneumonia-associated complications. ProADM demonstrated the greatest improvement in discrimination for all 3 scores compared with CRP and PCT [11]. As risk stratification scores are developed in children with CAP, evaluation of how biomarkers, such as proADM, improve these severity scores is essential.

Our study has several limitations. First, heterogeneity may exist within composite outcomes and the impact of individual outcomes on the whole may vary by outcome. Although sample size was small for some of the individual outcomes, we analyzed each outcome individually to assess its role in the results. In addition, our results were consistent across several sensitivity analyses, including proportional odds regression and different groupings of the severity outcomes. Second, limited sample size for some of the individual outcomes preclude definitive conclusions on the role of proADM in predicting those outcomes. However, in many cases, the point estimates show a substantial difference. Third, selection bias may have occurred in those patients who consented for blood. Children who provided blood were older and more likely to be admitted than those who did not (Supplementary Table 2), suggesting that they were more likely to have venipuncture for clinical care. They were more likely consent to obtain blood alongside clinical care, as opposed to consenting to a separate research venipuncture. This may have enriched our study for increased severity, theoretically increasing the rate of true positives for proADM. However, the opposite occurred in which our NPV was high, suggesting that selection bias had minimal effect on our outcomes. Fourth, the generalizability of these results warrants evaluation. Despite these limitations, the strengths of this study include its relatively large sample size and breadth of relevant clinical outcomes examined compared with prior studies evaluating proADM.

In conclusion, proADM measurement at the time of presentation to the ED was associated with the development of severe disease in children with suspected CAP and discriminated moderately well those who developed severe disease from those who did not, particularly those with radiographic CAP. Our study also suggests that proADM is more predictive of disease severity than the more conventional biomarkers CRP and PCT. Given these promising results and the relative novelty of proADM, further research is necessary to validate these findings and evaluate how proADM should be integrated with clinical findings to maximize predictive ability for children with CAP.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors acknowledge Judd Jacobs and Jessi Lipscomb for their role in data management for the CARPE DIEM study. They are grateful to the entire research team and patient services staff in the Divisions of Emergency Medicine and Hospital Medicine at Cincinnati Children’s Hospital Medical Center for their assistance with study procedures. Finally, they are especially grateful to the patients and families who enrolled in CARPE DIEM.

Disclaimer. The funders did not have any role in study design, data collection, statistical analysis, or manuscript preparation.

Financial support. This study was supported by the National Institutes of Health (NIH)/National Institute of Allergy and Infectious Diseases (grant numbers K23AI121325 to T. A. F. and K01AI125413 to L. A.), the Gerber Foundation (to T. A. F.), NIH/National Center for Research Resources (NCRR) and Cincinnati Center for Clinical and Translational Science and Training (grant number 5KL2TR000078 to T. A. F.), and the Cincinnati Children’s Hospital Medical Center Division of Emergency Medicine Small Grant program.

Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.