5025.0: Wednesday, October 24, 2001 - Board 10

Abstract #28904

Concordance between ICD-9 and ICD-10 coded death causes for Baltimore City, 1999

Joy P. Nanda1, Nkossi Dambita2, Jon Mark Hirshon, MD, MPH3, and James Teisl2. (1) Population and Family Health Sciences, Johns Hopkins Medical Institutions, 624 North Broadway, Baltimore, MD 21205, 410-614-3485, jnanda@jhsph.edu, (2) Office of Grants, Research, Surveillance and Evaluation, Baltimore City Health Department, 210 Guilford Ave 2nd Floor, Baltimore, MD 21202, (3) Division of Emergency Medicine, University of Maryland Medical Systems, 1062 River Bay Rd., Annapolis, MD 21401

Classification of death causes using ICD codes allows global comparison of death statistics. It also enables local agencies to respond to unexpected death rates. Implementation of new ICD codes however, results in discrepancies in death causes between two versions. From both program and policy perspective, assessing biases introduced by coding system change is therefore, critical. This study determines concordance rates between ICD-9 and ICD-10 coding of death in Baltimore City.

All deaths occurring in Baltimore City in 1999 were coded by the same nosologist using both the ICD-9 and ICD-10 rules. Coding by State DHMH for 1999 deaths using ICD-10 were then merged to the BCHD database.

The City reported 10,527 deaths for 1999. Agreement among the three sources for 15 leading causes ranged from 26% for pneumonia to 99.6% for homicide, when comparing individual codes. Concordance was higher between the ICD-10 coding of BCHD and DHMH, than between the ICD-9 coding and ICD-10 coding systems. Adjustment for late changes in NCHS rules for pneumonia increased the concordance to 50%, but not as desired. When the individual codes were summarized by organ system or complex causes, overall concordance also improved. In general, agreement between ICD-9 and ICD-10 was low, when the underlying cause was broad or non-specific. Classification by ICD-10 appeared to be more specific for death cause categories.

Changes in coding rules and system may influence community-programs and policy makers who rely on trends in mortality based on ICD coding systems, for determining funding as well as outcome measurement.

Learning Objectives: At the end of the presentation the participant will be able to 1. recognize the implications of changes in death coding 2. evaluate the relative merits of these changes

Keywords: ICD, Mortality

Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: Johns Hopkins University School of Public Health Baltimore City Health Department University of Maryland Medical Systems
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

The 129th Annual Meeting of APHA