76. Predicting Severe Outcomes in Children Hospitalized with Community-acquired Pneumonia
Session: Oral Abstract Session: Pediatric Bacterial Infections and Colonization
Thursday, October 8, 2015: 9:15 AM
Room: 32--ABC

Background: Severity assessments to assist with clinical decision making for children hospitalized with pneumonia are lacking.  We sought to develop a childhood pneumonia clinical prediction rule (CPR) that accurately estimates risk for severe outcomes.

Methods: Data from children enrolled in the Etiology of Pneumonia in the Community Study, a prospective study of community-acquired pneumonia hospitalizations in 3 US children's hospitals were used to derive a CPR.  The outcome variable was a 3-level ordinal severity scale: severe (invasive mechanical ventilation or shock requiring vasoactive medications or drainage of a parapneumonic effusion); moderate (intensive care admission without severe features); or mild (children without moderate or severe features).  Expert consensus a priori selected 10 severity predictors measured at admission: age, number of comorbidities, PaO2:FiO2 (PF) ratio, heart and respiratory rate percentiles, hypotension, altered mental status, chest indrawing, infiltrate pattern, and parapneumonic effusion. Ordinal logistic regression was used for CPR development. Predictive accuracy of the CPR was measured using discrimination (c-index).  Partial chi-square (% of total chi-square) indicated the relative importance of each predictor. A separate model included microbiologic etiology. 

Results: Outcome frequencies for the 2240 included children were severe (10%), moderate (13%), and mild (77%). CPR predictive accuracy was moderately high (c-index 0.77, 95% CI 0.74-0.79). Age, PF ratio, altered mental status, and parapneumonic effusion were the most important predictors (combined partial chi-square >70%).  The addition of pathogen testing results (2109 children) improved discrimination (c-index 0.80, CI 0.78-0.82); this predictor had the highest partial chi-square (32%).  An illustrative case using the CPR (without etiology) is shown (Figure).

Conclusion: Using widely available predictors, we developed a CPR that accurately estimates risk for severe outcomes among children hospitalized with pneumonia.  Pathogen testing results improved rule performance. Validation is ongoing.  If validated, this CPR (with or without etiology) could be used to assess disease severity and inform clinical decision-making.



Derek J. Williams, MD, MPH1, Yuwei Zhu, MD, MS2, Wesley H. Self, MD, MPH3, Carlos G. Grijalva, MD, MPH4, Frank Harrell, PhD5, Carrie Reed, PhD6, Chris Stockmann, MSc7, Sandra R. Arnold, MD8, Krow Ampofo, MD, FIDSA, FPIDS7, Evan Anderson, MD9, Anna M. Bramley, MPH10, Lyn Finelli, DrPH, MS6, Richard Wunderink, MD11, Jonathan a. Mccullers, MD, FIDSA12, Andrew Pavia, MD, FIDSA, FSHEA, FPIDS7, Seema Jain, MD, MPH6 and Kathryn Edwards, MD, FIDSA13, (1)Pediatrics, Vanderbilt University School of Medicine, Nashville, TN, (2)Biostatistics, Vanderbilt University School of Medicine, Nashville, TN, (3)Emergency Medicine, Vanderbilt University School of Medicine, Nashville, TN, (4)Department of Health Policy, Vanderbilt University Medical Center, Nashville, TN, (5)Department of Biostatistics, Vanderbilt University School of Medicine, Nashville, TN, (6)Centers for Disease Control and Prevention, Atlanta, GA, (7)Department of Pediatrics, Division of Pediatric Infectious Diseases, University of Utah School of Medicine, Salt Lake City, UT, (8)Pediatrics, University of Tennessee Health Sciences Center, Memphis, TN, (9)Children's Memorial Hospital and Northwestern Memorial Hospital, Chicago, IL, (10)Centers for Disease Control and Prevention (CDC), Atlanta, GA, (11)Northwestern University Feinberg School of Medicine, Chicago, IL, (12)St. Jude Children's Research Hospital, Memphis, TN, (13)Pediatrics, Vanderbilt University Medical Center, Nashville, TN


D. J. Williams, None

Y. Zhu, None

W. H. Self, None

C. G. Grijalva, None

F. Harrell, None

C. Reed, None

C. Stockmann, None

S. R. Arnold, None

K. Ampofo, None

E. Anderson, Abbvie: Scientific Advisor , Consulting fee

A. M. Bramley, None

L. Finelli, None

R. Wunderink, None

J. A. Mccullers, None

A. Pavia, None

S. Jain, None

K. Edwards, None

Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 7th with the exception of research findings presented at the IDWeek press conferences.