Possibilities of Predicting Peak Oxygen Consumption in Patients With Chronic Heart Failure According to the 6‑Minute Walk Test

Abstract

Aim. To determine the correlation between the results of the 6-minute walk test (6MWT) and peak oxygen consumption (VO2peak) for populations of patients with chronic heart failure with pronounced clinical and demographic differences; to study a possibility of indirect measurement of VO2peak based on the results of 6MWT using the formulas available from the literature.Material and methods. Two databases were analyzed: 50 patients included in the AEROFIT study (group A), and 31 patients from the Almazov National Medical Research Center (group B). The inclusion criteria were the availability of data from the cardiopulmonary stress test and the 6MWT. The possibility of predicting VO2peak was calculated based on the results of 6MWT using the formulas reported in the literature (L. P. Cahalin et al., 1996; R. M. Ross et al., 2010; R. A. Adedoyin et al., 2010). The predictive accuracy of the models was assessed using the coefficient of determination (R2). The relationship between functional and clinical-demographic indicators was assessed using the Pearson or Spearman correlation analysis.Results. The study groups differed significantly in all parameters, except for the proportion of men and the mean VO2peak. Group B patients were 20 years younger than group A patients, had a lower left ventricular ejection fraction (24.06±7.75 and 41.52±10.48 %, respectively; p<0.001), and covered a 130 m shorter distance in the 6MWT. Despite the absence of a significant difference in VO2peak between groups A and B (13.6 and 13.1 ml / kg / min, respectively; p=0.6581), 61 % of group B patients and 20% of group A belonged to Weber functional class IV. In group A, the 6MWT distance correlated closely with VO2peak (R=0.78; p<0.01) and weakly with age (R=0.4) and body mass index (R=0.3). In group B, the 6MWT distance correlated only with VO2peak (R=0.77; p<0.01). For group A, the R.M. Ross et al. model demonstrated high accuracy in determining the mean VO2peak value with a 0.06% prediction error normalized to measured VO2peak. For group B, none of the models showed satisfactory predictive accuracy. The Ross and Cahalin models showed the best coefficients of determination for groups A and B: Group A, Ross et al. (R2=0.58) and Cahalin et al. (R2=0.59); Group B, Ross et al. (R2=0.59) and Cahalin et al. (R2=0.6).Conclusion. In two groups of patients with a statistically insignificant difference in the mean values of VO2peak, the mean values of 6MWT distance were significantly different, although these indicators correlated closely. The VO2peak prediction models showed satisfactory accuracy for estimation of mean VO2, but poor accuracy for estimation of individual values. A better predictive accuracy is determined by similar clinical and demographic characteristics between the training and testing populations, and likely also by models based on larger, more diversified populations.