The ability of a computer program based on the Marquette Matrix-12 short measurement matrix to replicate coding by the Minnesota ECG coding laboratory

The study was undertaken to determine whether a computer program that uses "short measurement matrix" data from the Marquette Matrix-12 system can rep licate Minnesota electrocardiogram (ECG) coding laboratory interpretations. An agreement was found between coding of median complex ECGs at the Minnes ota ECG coding laboratory and coding based on Marquette Matrix-12. short me asurement matrix. The comparison was based on 763 ECGs plus chest pain hist ory and serum enzyme values for a stratified random sample of 141 patients hospitalized in 1990 or 1991 for an event coded as HICDA 410.x (acute myoca rdial infarction), 411 (other acute and subacute forms of ischemic heart di sease), 413 (angina pectoris), or 796.9 (other ill defined and unknown caus es of morbidity and mortality). The population was reconstructed from the s tratified random sample to enable population-based inferences. Exact agreem ent between Matrix-12 and Minnesota coding laboratory interpretation on 4 E CG patterns (evolving diagnostic, diagnostic, equivocal, or other ECG patte rn) was 74.5% (Kappa = 0.63 +/- 0.05) for the stratified random sample and 78.8% (Kappa = 0.66 +/- 0.05) for the reconstructed population. For coding myocardial infarction based on the EGG, serum enzyme levels, and ischemic c hest pain, agreement was 91.5% (Kappa = 0.85 +/- 0.04) for the stratified r andom sample and 90% (Kappa = 0.83 +/- 0.04) for the reconstructed populati on. Although ECG interpretation by a computer program based on the short me asurement matrix of the Matrix 12 system results in better agreement than p rior attempts to replicate the Minnesota coding laboratory, interpretation remains unacceptably discordant.