Date on Master's Thesis/Doctoral Dissertation

5-2008

Document Type

Doctoral Dissertation

Degree Name

Ph. D.

Department (Legacy)

School of Public Health and Information Sciences

Committee Chair

Muldoon, Susan B.

Author's Keywords

Syndromic surveillance; Bioterrorism; Emergency medical services; Kappa; Interrater reliability; Validity

Subject

Emergency medicine--Diagnosis; Emergency medical services

Abstract

A retrospective study involving the secondary analysis of public health surveillance records was undertaken to characterize the reliability and validity of an EMS dispatch data-based scheme for assigning emergency patients to surveillance syndromes in relation to two other schemes, one based on hospital ED clinicians' manual categorization according to patients' chief complaint and clinical presentation, and one based on ICD-9 coded hospital ED diagnoses. Comparisons of a sample of individual emergency patients' syndrome assignments according to the EMS versus each of the two hospital categorization schemes were made by matching EMS run records to their corresponding emergency department patient encounter records. This new, linked dataset was analyzed to assess the level of agreement beyond chance between the three possible pairs of syndrome categorization schemes in assigning patients to a respiratory or non-respiratory syndrome and to a gastrointestinal or non-gastrointestinal syndrome. Cohen's kappa statistics were used to measure chance-adjusted agreement between categorization schemes (raters). Z-tests and a chi-square-like test based on the variance of the kappa statistic were used to test the equivalence of kappa coefficients across syndromes, population subgroups and pairs of syndrome assignment schemes. The sensitivity, specificity, predictive value positive and predictive value negative of EMS dispatch and chief complaint-based categorization schemes were also calculated, using the ICD-9-coded ED diagnosis-based categorization scheme as the criterion standard. Comparisons of all performance characteristic (i.e. sensitivity, specificity, predictive value positive and predictive value negative) values were made across categorization schemes and surveillance syndromes to determine whether they were significantly different. The use of EMS dispatch codes for assigning emergency patients to surveillance syndromes was found to have limited but statistically significant reliability in relation to more commonly used syndrome grouping methods based on chief complaints or ICD-9 coded ED diagnoses. The reliability of EMS-based syndrome assignment varied significantly by syndrome, age group and comparison rater. When ICD-9 coded ED diagnosis-based grouping is taken as the criterion standard of syndrome definition, the validity of EMS-based syndrome assignment was limited but comparable to chief complaint-based assignment. The validity of EMS-based syndrome assignment varied significantly by syndrome.

Download

Share

COinS