Derivation and Validation of Simple Routine Parameter Score to Estimate Left Ventricular Ejection Fraction

摘要

Background. Left ventricular ejection fraction (LVEF) is an important parameter related to the diagnosis and prognosis of patients with heart failure. The calculation of LVEF usually performed using echocardiography. Unfortunately, such device is not widely available. Moreover, an expert is required to operate the device. Therefore, an alternative diagnostic modality to assess LVEF is mandatory, given no access to echocardiography.Aims. The aim of the study was derived a scoring system which predict LVEF through simple routine parameter and determine its internal validityMethod. This study was a cross sectional analytical study held at RSUP dr. Sardjito Yogyakarta from January 2016 to April 2016. The scoring system was derived from multivariate linear regression with logit LVEF transformation. The internal validity was assessed accordingly.Result. There were 42 subjects recruited, with mean age was 59 years, consisted of 67% male and 33% female. LVEF was not normally distributed, with median 49.5%. Multivariate linear regression with logit LVEF transformation was conducted to the initial 10 independent variables, resulting in 6-variable model. The score derived from the model was able to predict LVEF with good validity (r = 0.78; R2 = 0.61). The performance of the score to discriminate systolic (LVEF < 40%) from diastolic (LVEF > 40%) dysfunction was considered good (AUC= 0.82, sensitivity 76.47%, specificity 88%, positive predictive value 89.02%, and negative predictive value 74.61%), and in good agreement with echocardiography κ = 0.65). Although the Lin’s concordance coefficient indicated poor agreement (ρc= 0.77), the Bland-Altman plot showed no residual systematic bias.Conclusion. The clinical score to estimate LVEF had a good predictive and diagnostic value. It was shown to have a good internal validity, and therefore can be used locally to estimate LVEF as an alternative to echocardiography.