Cardiac troponin T concentrations and patient-specific risk of myocardial infarction using the novel PALfx parameter
Artikel i vetenskaplig tidskrift, 2019
Myocardial infarction (MI) is more likely if the heart damage biomarker cardiac troponin T (cTnT) is elevated in a blood sample from a patient with chest pain. There is no conventional method to estimate the risk of MI at a specific cTnT concentration. The purpose of this study was to evaluate the performance of a novel method that converts cTnT concentrations to patient-specific risks of MI. Methods: Admission cTnT measurements in 15,425 ED patients from three hospitals with a primary complaint of chest pain, with or without a clinical diagnosis of MI, were Box-Cox-transformed to normality density functions to calculate the percentage with MI among patients with a given cTnT concentration, the parametric predictive value among lookalikes (PALfx). The ability of the PALfx to generate stable risk estimates of MI was examined by bootstrapping and expressed as the coefficient of variation (CV). Results: Four age and sex-specific subgroups above or below 60 years of age with distinct cTnT distributions were identified among patients without MI. The cTnT distributions across subgroups with MI were similar, allowing us to use all admissions with MI to calculate the PALfx in the four subgroups. For instance, at a baseline cTnT concentration of 7 ng/L, a female patient < 60 years would have a 0.5% risk of MI whereas a male patient > 60 years would have a 1.9% risk of MI. To assess the stability of the PALfx method we bootstrapped smaller and smaller subsets of the 15,422 ED visits. We found that 1950 patients without MI and 50 patients with MI were sufficient to limit the variation of the PALfx with a CV of 0.8–5.4%, close to the CV using the entire dataset. The MI risk estimates were similar when data from the three hospitals were used separately to derive the PALfx equations. Conclusions: The PALfx can be used to estimate the risk of MI at patient-specific cTnT concentrations with acceptable margins of error. The patient-specific risk of disease using the PALfx could complement decision limits.