Date of Award

January 2015

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Judith H. Lichtman


Background/ Purpose: Epidemiology and health services research often use International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes to translate clinical information contained in administrative databases. Since errors in ICD-9-CM codes can affect the interpretation of results from these studies, we sought to expand upon existing research by determining if there are systematic variations in concordance between stroke patient clinical diagnoses and ICD-9-CM codes by hospital characteristics and degrees of stroke severity.

Methods: We used patient records with a discharge date in 2013 from the Paul Coverdell National Acute Stroke Program (PCNASP). Our primary analysis quantified the concordance between the attending physician’s clinical diagnosis and the primary ICD-9-CM billing code. Hospital characteristics data were used to examine concordance by presence/absence of a stroke unit and stroke team, hospital bed size categories, and urban/rural status of the hospital’s location. Furthermore, concordance by stroke severity (NIHSS) categories was compared for ischemic stroke and TIA patients.

Results: The overall sensitivity was 93.8% for all stroke and TIA diagnosis groups. Concordance was relatively high for each diagnosis category except “stroke not otherwise specified”. Carotid endarterectomy was a common reason for discordances between the clinical diagnosis and ICD-9-CM code. Concordance was highest for larger metropolitan hospitals with stroke units and teams, and more severe strokes.

Conclusions: Systematic variations in the coding accuracy of stroke patients’ diagnoses by hospital and patient characteristics have implications for hospital reimbursements and stroke case identification in epidemiologic studies and quality metrics.


This is an Open Access Thesis.

Open Access

This Article is Open Access