1583. Community Acquired Pneumonia (CAP) Requiring Hospitalization in HIV Infected (HIV+) and Un-infected (HIV-) Patients: Evaluation of Patients Identified by ICD-9 Codes
Session: Poster Abstract Session: Community Acquired Pneumonia
Saturday, October 10, 2015
Room: Poster Hall
  • Rodriguez-Barradas-ID week2015-Poster#1583.pdf (236.3 kB)
  • Background: CAP remains one of the most frequent pulmonary complications in HIV+ patients. We assessed the reliability of ICD-9 codes for identification of CAP compared to the gold standard of chart review, and determined the frequency of risk factors for health-care associated pneumonia (HCAP) and of microbiologically confirmed diagnosis of bacterial CAP by HIV status, using data from the Veterans Aging Cohort Study (VACS) of HIV+ patients and demographically similar HIV- controls.

    Methods: We identified patients with ≥1 inpatient ICD-9 codes consistent with CAP after VACS enrollment. Chart reviews used standardized data abstraction forms. Events were categorized as CAP present or not on admission and microbiologic diagnosis as definitive, presumed, or suspected.  Positive predictive value (PPV) was calculated as the number of chart review confirmed pneumonia events divided by the number identified by ICD-9 codes.

    Results: Between 2003-08, of 434 HIV+ and 115 HIV- patients with a CAP ICD-9 code, 74% and 66%, respectively, had pneumonia confirmed by chart review, 18%  and 25% had no pneumonia, 2%  and 4% had hospital-acquired pneumonia, and 6% and 4% were uncertain (p=0.1); 9% of HIV+ and 18% of HIV- confirmed CAP cases had HCAP risk factors (P=0.018). PPV of ICD-9 codes ranged from 0-100%; it was 74% for the most frequent code (pneumonia due to organism unspecified). Indeed, the majority of CAP patients had suspected bacterial pneumonia (BP) without microbiologic confirmation; only in 13% and 5% of HIV+ and HIV- was a definitive BP diagnosis established based on identification of a pathogen, and in 8% and 9%, respectively, the diagnosis was presumed based on compatible gram stain and culture results from expectorated or induced sputum, or from endotracheal aspirate (P=0.3).

    Conclusion: Overall, 72% of patients with a CAP ICD-9 code had clinically confirmed pneumonia by chart review, a definitive microbiologic diagnosis of BP was infrequently established. HIV- patients were more likely to have HCAP.  In this cohort, ICD-9 codes for CAP perform similarly in HIV+ and HIV- patients, and identified patients with clinically confirmed CAP requiring hospitalization with good reliability. Future work will assess modified algorithms to improve PPV of ICD-9 codes to identify patients with CAP.

    Maria C. Rodriguez-Barradas, MD, FIDSA1,2, Kathleen Akgun, MD3, Sheldon Brown, MD4, Adeel Butt, MD, MS5, Michael J. Fine, MD, MSc6, Matthew Bidwell Goetz, MD7, Christopher Graber, MD, MPH, FIDSA8, Laurence Huang, MD9, Kathleen Mcginnis, MS5, David Rimland, MD, FIDSA10, Amy C. Justice, MD, PhD11 and Kristina Crothers, MD12, (1)Infectious Diseases, Michael E. DeBakey VA Medical Center, Houston, TX, (2)Medicine, Baylor College of Medicine, Houston, TX, (3)Yale University, New Haven, CT, (4)Mt Sinai School of Medicine, Bronx, NY, (5)VA Pittsburgh Healthcare System, Pittsburgh, PA, (6)Center for Health Equity Research and Promotion (CHERP), VA Pittsburgh Healthcare System, Pittsburgh, PA, (7)Infectious Diseases, VA Greater Los Angeles Healthcare System, Los Angeles, CA, (8)VA Greater Los Angeles Healthcare System, Los Angeles, CA, (9)UCSF, San Francisco, CA, (10)Atlanta VA Medical Center, Decatur, GA, (11)Yale University and VA Connecticut Healthcare System, West Haven, CT, (12)University of Washington, Seattle, WA


    M. C. Rodriguez-Barradas, None

    K. Akgun, None

    S. Brown, None

    A. Butt, Gilead: Grant Investigator , Research grant
    AbbVie: Grant Investigator , Research grant

    M. J. Fine, None

    M. Bidwell Goetz, None

    C. Graber, None

    L. Huang, None

    K. Mcginnis, None

    D. Rimland, None

    A. C. Justice, None

    K. Crothers, 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.