Peak oxygen uptake (VO2peak) is an important measurement of maximal functional capacity in patients with chronic heart failure (CHF). It is also a predictor of survival and an indicator for heart transplantation in these patients.1–3 Although cardiopulmonary exercise testing (CPET) is considered the ideal process to evaluate oxygen consumption in subjects with CHF,4 the high cost of the technique may limit access to it.

The six-minute walk test (6MWT) has been used as an alternative to the CPET to evaluate functional capacity in these individuals due to its feasibility and lower cost.5,6 Prediction equations have been suggested to estimate VO2peak in subjects with CHF7–9 based on the six-minute walking distance (6MWD). These equations have demonstrated good correlation with anthropometric characteristics,7,8 cardiac and lung function at rest,7 6MWD,7–9 and VO2peak, but the determination coefficient varies widely. In addition, the standard error of the estimate (SEE) reaches values close to 1 MET, which may represent a large percentage of the maximal functional capacity of patients with CHF. These results have led us to hypothesize that the VO2peak estimated from the 6MWT does not indicate the real VO2peak of these patients. Furthermore, no studies evaluated the agreement between measured and estimated VO2peak values in a sample of individuals with CHF. The evaluation of agreement between two measurements allows for quantifying the differences between these measures and establishing limits of agreement and bias.10

Therefore, the present study aims to evaluate the agreement between the measured VO2peak and the VO2peak estimated by four prediction equations based on the 6MWD in subjects with CHF.

The inclusion criteria were established to obtain a homogeneous sample of CHF patients that included patients clinically diagnosed with CHF (left ventricular ejection fraction <50%). Those who were stable for at least two months in optimal pharmacological treatment, aged between 25 and 59 years, with a body mass index <30kg/m2 and not practicing regular exercise were included in the study. Older adults, adolescents, or obese patients and those practicing regular exercise (because it affects disease severity and control) were not included in order to avoid other mechanisms potentially being involved in the pathophysiology of the disease. Patients were excluded if: they were classified as class IV according to the New York Heart Association (NYHA IV), with limiting disorders to physical exercise, recent history of pulmonary disease, unstable angina, uncontrolled arrhythmias or peripheral arterial disease, or if they did not reach a respiratory exchange ratio (RER) greater than or equal to 1.0 during CPET.11Table 1 shows the main characteristics of the samples in the different studies.

This study was approved by the ethics committee of the institution (protocol number 050/09, Comitê de Ética em Pesquisa, Universidade Federal de Minas Gerais (UFMG), Belo Horizonte, MG, Brazil) in which the research was conducted in accordance with the Declaration of the World Medical Association. Volunteers signed a written informed consent form before participation.

Table 1 shows the comparisons among the samples of the studies that developed equations 1–4. The respiratory exchange ratio (RER) was similar to equations 1 and 2. The cardiac function (LVEF) and the maximal functional capacity of the patients were higher than equations 1–3. The study of equation 4 did not report these values.

The physical and clinical characteristics of the 36 subjects with CHF assessed in this study are shown in Table 2. This sample consisted primarily of male subjects (29 men). About 17% were in NYHA functional class I (n=6), 47% in NYHA class II (n=17), and 36% in NYHA class III (n=13). The VO2peak achieved in the CPET and the 6MWD were statistically different among the NYHA functional classes (p<0.0001 and p=0.002, respectively).

Fig. 1 shows significant differences between three of the four prediction equations (equations 2–4) and the VO2peak values measured in the CPET (p<0.0001). Although all equations presented moderate levels of correlation with the measured VO2peak (r>0.5; p<0.001), equations 2–4 underestimated the results by 15–51% (Table 3). Equation 1 presented the highest correlation with the measured VO2peak (r=0.70; p<0.0001).

Figure 1.

Results of peak oxygen uptake (VO2peak) measured by cardiopulmonary exercise testing (CPET) and estimated by four equations. *p<0.0001 compared with CPET. (Repeated-measures ANOVA followed by Bonferroni's post hoc test.)

(0.06MB).

The Bland–Altman analysis did not reveal agreement among the VO2peak values in the CPET and those estimated by equations 2–4 because the difference between the measurements obtained by both methods showed a bias that was statistically different from zero (p=0.007) and very large limits of agreement (Fig. 2). Furthermore, the linear relationship between the mean and the difference of the measurements indicates improved NYHA class and consequently improved maximal functional capacity. There is also an increase in the VO2peak prediction error.

Figure 2.

Bland–Altman analysis of peak oxygen uptake (VO2peak) measured with CPET and estimated with equations 1–4. CPET, cardiopulmonary exercise testing; VO2peak, peak oxygen uptake; ULA, upper limit of agreement; LLA, lower limit of agreement.

(0.37MB).

In equation 1, the comparison between the measured and the estimated VO2peak in the CPET was not statistically different (21.83±6.83 vs 21.68±2.73; p=0.867) (Fig. 1). Further, the bias was close to zero. However, a large data scatter was observed between the upper and lower limits of agreement, and there was also a linear relationship between the difference and the mean measurements (r=0.82; p<0.0001). In other words, despite the observed agreement, the bias was not constant and increased with the increase in VO2peak (Fig. 2).

This study showed that only in equation 1 was the estimated VO2peak similar to the direct VO2peak measured in the CPET in subjects with CHF. In addition, a lack of agreement was observed between the direct VO2peak and the estimated VO2peak in prediction equations 2–4, considering that to be consistent, the average difference between the measured and the estimated value (bias) should be close to zero and the LoA range (upper and lower limits of agreement) should be close to each other.

According to the Bland–Altman analysis, in equation 1 the mean differences (bias) between measured and estimated VO2peak were close to zero; however, the upper and lower LoA were elevated. This situation demonstrates that bias increases linearly with VO2peak improvement, and the large LoA range (∼3 MET) prevents estimation of maximal VO2peak. In equations 2–4, the correlation analyses showed associations of moderate magnitude despite the lack of agreement between the estimated and measured VO2peak. In this sense, it is necessary to use methods that establish the LoA to evaluate the agreement between two measures and the confidence intervals acceptable of clinical point of view, while the correlations only indicate the association between the variables, e.g., measured VO2peak and the VO2peak estimated by equations.

In all equations, the bias increased linearly with increasing VO2peak, indicating that the increase in NYHA class reduces the VO2peak prediction error. This finding may be due to reduced functional capacity of NYHA III patients, as their energy expenditure in the 6MWT was maximal or close to maximal.18 Moreover, the worst agreement was observed in NYHA I patients, probably because the 6MWT could be submaximal for the majority of the patients in this functional class. This could suggest that the equations might be good for predicting VO2peak from 6MWT in specific subgroups of the disease, i.e., the ones with lower maximal functional capacity.

The wide variability in the VO2peak estimated by the equations could be explained by the different methodologies used in the studies and the different variables used in the equations. Equations 2 and 4 were the ones that underestimated VO2peak the most (by 30% and 51.2%, respectively). In both equations, anthropometric variables, such as weight and height, were included. As these variables relate both to the distance covered and VO2peak, it is possible to include them in the equations, and although it increases the determination coefficient (R2), the estimate error19 can also increase. The samples that originated equations 1, 2, and 4 have clinical and anthropometric characteristics similar to our study sample, including young and middle-aged adults belonging to NYHA classes I–III.7,8 Only equation 3 was developed in patients with cardiopulmonary disease9 and not specifically CHF. None of the above equations were developed with NYHA IV patients. Left ventricular ejection fraction, direct VO2peak (CPET), and 6MWD were higher in the present study compared to the studies of equations 1–3. The equation 4 study did not report these values. The lack of agreement between the measured and the estimated VO2peak may have been influenced by the better functional capacity of our patients, but this result may also reflect the inability of indirect methods to predict the VO2peak of patients with CHF, which had already been observed in CHF patients in the literature.2,4,18

In patients with CHF, the assessment of maximal functional capacity also aims at risk stratification, prognostic evaluation, and indication for heart transplantation.1,2 Thus, the direct measurement of VO2peak by CPET is the gold standard for this purpose.3,4 On the other hand, the 6MWT has emerged as a viable and reproducible option of submaximal evaluation.20 Some studies6,21–24 have demonstrated its clinical utility in detecting the functional capacity of CHF subjects. Moreover, clinical applicability to 6MWT is available, technically feasible, low cost, and an indicator for activities of daily living (ADL).18,25 In this way, the 6MWT can be helpful for submaximal assessment because it represents ADL, but it is not appropriate to predict maximal functional capacity as expressed by VO2peak,18 because it has not proved accurate for predicting VO2peak.

The present study does have some limitations. First, a methodological limitation is the absence of patients classified as NYHA IV because they are the ones who would benefit the most from an evaluation such as the 6MWT. However, they are very critical, clinically unstable patients, with frequent decompensation and difficult therapeutic optimization. Thus, these individuals are very difficult to be sampled. Second, the relatively small sample size could be a limiting factor for the Bland–Altman analysis. In addition, it does not allow separate analyses of the NYHA classes or the formulation of a reference VO2peak equation from the 6MWT for this population.26 Equation 1 seems to be able to predict VO2peak in patients with lower maximal functional capacity. Future studies will be necessary to clarify these aspects.

Although all VO2peak values estimated by equations presented a moderate association with the VO2peak measured in the CPET, only equation 1 presented similar results. However, even for this equation, the VO2peak estimated by the 6MWD in the CHF patients was not a valid measure for predicting the maximal exercise capacity of this population because these values were not consistent with the gold standard method of CPET. This study emphasizes the need for validation of the proposed equations in specific populations.

The authors declare no conflicts of interest.