1367. Use of death certificate data in healthcare-associated infection (HAI) surveillance
Session: Poster Abstract Session: HAI: Epidemiologic Methods
Friday, October 28, 2016
Room: Poster Hall
  • Lyman IDWeek--HAI Death Certificate.pdf (79.3 kB)
  • Background: Despite known limitations, death certificate (DC) data are commonly used to monitor frequency and trends in causes of death (CODs). The use of DC data in nationwide healthcare associated infection (HAI) surveillance has not been assessed recently in the US. We evaluated if DC data could be used to identify mortality attributable to HAIs.

    Methods: Among patients who underwent chart review for presence of an HAI during a 2011 hospital HAI point prevalence survey (PS) and had a death registry search performed, we identified those who died on or within 90 days of PS hospitalization discharge date. Death date, death location (hospital, long-term care, hospice, home), and up to 4 CODs which were assigned International Classification of Diseases (ICD-10) codes were obtained from DC and merged with HAI status and HAI type from the PS. We evaluated the frequency of CODs for HAI terminology (e.g., infection type or organism) and using Fisher’s Exact Test, compared the 15 leading CODs, as reported by the U.S. National Vital Statistics System, in patients with and without HAIs.

    Results: A total of 638 patients died on or within 90 days of PS discharge date, of whom 92 (14%) had a HAI (30% gastrointestinal, 22% pneumonia, 16% urinary tract). There was no difference in number of CODs listed for patients with and without an HAI (p=0.2). CODs for HAI terminology were infrequent (6 [1%] Clostridium difficile infection; 3 [<1%] Staphylococcus aureus infection; 2 [<1%] procedure/device related infection). The most common COD categories were heart disease, malignant neoplasm, and influenza/pneumonia. Significant differences in COD were identified for septicemia, which was more common in patients with HAIs than those without (23% vs 11%, p=0.004), and chronic lower respiratory disease, which was less common in patients with HAIs than those without (3% vs 12%, p=0.01).

    Conclusion: While 14% of patients in this cohort had a known HAI, only 1% had CODs with HAI terminology. CODs did not differ among patients with and without HAIs, with only septicemia more common in patients with HAIs. Because COD rarely contained HAI terminology and did not differentiate patients with HAIs from those without, our findings suggest DC data are of limited use in identifying mortality attributable to HAIs.

    Meghan Lyman, MD, Epidemic Intelligence Service, Center for Disease Control and Prevention, Atlanta, GA; National Center for Emerging and Zoonotic Infectious Diseases Division for Healthcare Quality Promotion, Center for Disease Control and Prevention, Atlanta, GA, Lisa Laplace, MPH, Centers for Disease Control and Prevention, Atlanta, GA, Nicolai Buhr, MPH, Office of Infectious Disease Epidemiology and Outbreak Response, Maryland Department of Health and Mental Hygiene, Baltimore, MD, Cathleen Concannon, MPH, Center for Community Health, University of Rochester Medical Center, Rochester, NY, Katherine Ellingson, PhD, Positive Deviance MRSA Prevention Partnership, Atlanta, GA, Helen Johnston, MPH, Colorado Department of Public Health and Environment, Denver, CO, Marion Kainer, MBBS, MPH, FSHEA, Tennessee Department of Health, Nashville, TN, Brittany Martin, MPH, California Emerging Infections Program, Oakland, CA, Meghan Maloney, MPH, Connecticut Department of Public Health, Hartford, CT, Lewis Perry, DrPH, MPH, RN, Georgia Emerging Infections Program, Atlanta, GA, Linn Warnke, RN, MPH, Minnesota Department of Health, St. Paul, MN and Nicola D. Thompson, PhD, MSc, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, Atlanta, GA


    M. Lyman, None

    L. Laplace, None

    N. Buhr, None

    C. Concannon, None

    K. Ellingson, None

    H. Johnston, None

    M. Kainer, None

    B. Martin, None

    M. Maloney, None

    L. Perry, None

    L. Warnke, None

    N. D. Thompson, None

    Findings in the abstracts are embargoed until 12:01 a.m. CDT, Wednesday Oct. 26th with the exception of research findings presented at the IDWeek press conferences.