Inter-rater Reliability of the Diamond & Forrester Score in Emergency Department Chest Pain Observation Unit Patients.

BACKGROUND

Cardiology consensus guidelines recommend use of the Diamond & Forrester (D&F) score in augmenting the decision to pursue stress testing. We have recently shown that it may have value in safely reducing stress utilization in an emergency department chest pain unit (CPU). However, full application necessitates demonstration of a good inter-rater reliability of the D&F score in the CPU setting. We hypothesized that D&F pretest probability would have good inter-rater reliability in CPU patients.

METHODS

This was a chart review of randomly selected patients from a previously collected prospective observational trial of admitted CPU patients in a large-volume academic urban emergency department. Inclusion criteria were: age>18 years, American Heart Association low/intermediate risk, nondynamic electrocardiograms, and normal initial troponin I. Exclusion criteria were: age>75 years with coronary artery disease. A D&F score for likelihood of coronary artery disease was calculated on each patient by 2 trained chart abstractors using a standardized data abstraction instrument. Abstractors were trained to specifically categorize presenting symptoms as fitting 1 of 3 types of chest pain symptoms: nonanginal, atypical, or anginal based on previously published prespecified criteria. Approximately 20% of charts in a CPU registry were abstracted by 2 chart abstractors who were blind to each other's categorization, the patient outcomes, and the study hypothesis. The primary outcome was the kappa statistic for agreement between the 2 raters.

RESULTS

The charts of 705 random patients were reviewed. The mean age was 55.1±11.8 years, 52% were female. Forty four percentage of patients received stress testing, and 2.4% of patients had acute coronary syndrome. The mean D&F score was 39±24. There was good inter-rater agreement of chest pain characteristics (κ=0.77, 95% confidence interval, 0.72-0.81; P<0.01).

CONCLUSION

This study supports the use of the D&F score as a reliable indicator of pretest probability in CPU patients by demonstrating that there is good inter-rater reliability. Prospective validation is necessary at the point of patient assessment, in conjunction with application of the D&F score to augment stress utilization decision making.