Measuring quality of care in syncope: case definition affects reported electrocardiogram use but does not bias reporting.

Abstract

The objective was to calculate agreement between syncope as a reason for visiting (RFV) an emergency department (ED) and as a discharge diagnosis (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9]), to determine whether syncope case definition biases reported electrocardiogram (ECG) usage, a national quality measure. The authors analyzed the ED portion of the National Hospital Ambulatory Medical Care Survey (NHAMCS), 1993-2004, for patients age >or=18 years. A visit was defined as being for syncope if it received one of three RFV or ICD-9 codes. Agreement between RFV and ICD-9 codes was calculated, and the percentages of syncope patients (RFV vs. ICD-9) who had an ECG were compared using chi-square and multivariate logistic regression. Raw agreement between syncope as an RFV and as an ICD-9 diagnosis code was 30.1% (95% confidence interval [CI] = 32.6% to 35.5%), representing only moderate agreement beyond chance (kappa = 0.50). ECG utilization was lower among visits defined by RFV (64.1%; 95% CI = 62.0% to 66.3%) than for ICD-9 diagnosis (73.6%; 95% CI = 71.4% to 75.8%). There was no meaningful variation in adjusted ECG use by patient, visit, or hospital characteristics between case definitions. Adjusted ECG use was lower under both case definitions among female patients and discharged patients and increased with age (p < 0.05). Despite only moderate agreement, syncope case definition should not bias reported ECG rate by patient, visit, or hospital characteristics. Among ED patients with syncope, ECG is performed less frequently in women, a potentially important disparity.