1033. Chest Radiograph Impressions to Identify Pneumonia among Hospitalized Patients with Influenza: Experience from the Influenza Emerging Infections Program
Session: Poster Abstract Session: Diagnostic Procedures in Clinical Practice
Saturday, October 22, 2011
Room: Poster Hall B1
Background:  Pneumonia is a common influenza-related complication and an important predictor of disease severity among hospitalized patients.  Chest radiographs are essential for confirming a pneumonia diagnosis.  We assessed which radiographic key terms abstracted from chest radiograph reports are best predictors of pneumonia among hospitalized patients with influenza reported by the Emerging Infections Program (EIP) Network.

Methods:  The EIP Network conducts surveillance for laboratory-confirmed influenza hospitalizations in 10 states and represents 7% of the adult U.S. population.  Starting in the 2008-2009 influenza season, EIP surveillance staff reviewed chest radiograph reports available at admission and abstracted radiographic key terms from the final impression. The key terms were “bronchopneumonia,” “consolidation,” “airspace density,” “infiltrate,” or “pleural effusion.”  We assessed which single or combination of radiographic key terms best predict pneumonia diagnosis using sensitivity and specificity analyses.  Pneumonia was defined as presence of an ICD-9 CM code 480-487.0 and a discharge summary diagnosis of pneumonia. 

Results:  From October 2008 to December 2009, we identified 4105 adults with chest radiograph impressions available for review, 1652 (40%) of whom had pneumonia. Sensitivity, specificity, and positive predictive value (PPV), respectively, were 22%, 99% and 94% for “bronchopneumonia,” 10%, 99% and 89% for “consolidation,” 31%, 96% and 84% for “infiltrate,” and 24%, 94%, and 73% for “airspace density.”  “Pleural effusion” had a high specificity (98%) but low sensitivity (7%) and PPV (67%) and was almost always reported in combination with another key term; it accounted for only 2% of pneumonia cases as a single term.  Using a combination of all terms but excluding “pleural effusion” increased sensitivity (65%) and had a reasonable specificity of 89% and a PPV of 80%. 

Conclusion: “Bronchopneumonia,” “consolidation,” “infiltrate,” and “airspace density” are the best radiographic predictors of pneumonia. These terms can identify cases with radiographically confirmed pneumonia with high specificity and therefore can be used to monitor trends in disease severity among patients hospitalized with influenza.

Subject Category: J. Clinical practice issues

Anna M. Bramley, MPH1, Fatimah S. Dawood, MD2, Saumil Doshi, MD1, Arthur Reingold, MD3, Ken Gershman, MD4, James Meek, MPH5, Kathryn E. Arnold, MD6, Monica Farley, MD7, Patricia Ryan, MS8, Ruth Lynfield, MD9, Emily B. Hancock, MS10, Shelley Zansky, PhD11, Nancy Bennett, MD, MS12, Ann Thomas, MD, MPH13, William Schaffner, MD14, Laurie Kamimoto, MD, MPH1, Sandra Chaves, MD, MSc1, Lyn Finelli, DrPH, MS1 and Seema Jain, MD1, (1)Centers for Disease Control and Prevention, Atlanta, GA, (2) , Centers for Disease Control and Prevention, Atlanta, GA, (3)California Emerging Infections Program, Berkeley, CA, (4)Colorado Department of Public Health and Environment, Denver, CO, (5)Yale University, Connecticut Emerging Infections Program, New Haven, CT, (6)Georgia Division of Public Health and Georgia Emerging Infections Program, Atlanta, GA, (7)Emory University School of Medicine and Atlanta VAMC, Atlanta, GA, (8)MD Dept Health Mental Hygiene, Baltimore, MD, (9)Minnesota Department of Health, St. Paul, MN, (10)New Mexico Department of Health, Santa Fe, NM, (11)Emerging Infections Program, New York State Department of Health, Albany, NY, (12)Center for Community Health and Department of Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY, (13)Oregon Public Health Division, Portland, OR, (14)Department of Preventive Medicine, Vanderbilt University School of Medicine, Nashville, TN


A. M. Bramley, None

F. S. Dawood, None

S. Doshi, None

A. Reingold, None

K. Gershman, None

J. Meek, None

K. E. Arnold, None

M. Farley, None

P. Ryan, None

R. Lynfield, None

E. B. Hancock, None

S. Zansky, None

N. Bennett, None

A. Thomas, None

W. Schaffner, None

L. Kamimoto, None

S. Chaves, None

L. Finelli, None

S. Jain, None

Findings in the abstracts are embargoed until 12:01 a.m. EST Thursday, Oct. 20 with the exception of research findings presented at IDSA press conferences.