Measuring aerobic cardiorespiratory fitness (CRF) holds significant importance as it offers an objective assessment of the physical well-being of patients and their ability to cope with daily life activities. Furthermore, there is strong scientific evidence linking the level of aerobic fitness to survival and prognosis, particularly among individuals with heart conditions.1 The cardiopulmonary exercise test (CPET), which provides the measurement of maximum or peak oxygen consumption (VO2 max or VO2 peak, respectively), in mL kg−1 min−1, is currently recognized as the most accurate method of measurement. Yet, it requires specialized equipment and is not readily available in most cardiac centers. Indeed, despite being considered in the most relevant international guidelines,2 the use of CPET for quantifying CRF remains limited around the world,3 due to financial restraints, lack of training, and its complex interpretation.4 The conventional exercise test, which monitors the heart's electrical activity though electrocardiogram and evaluates blood pressure and heart rate (HR) during exercise, is widely available, but is a less accurate alternative.

Therefore, methods of assessing functional status without exercise testing, by means of self-rated physical fitness level questionnaires, may be useful in some cases. Still, little is known about the precision and validity of these questionnaires for estimating CRF in cardiac patients and the existing evidence is hampered by heterogeneous definitions and measurement methods.5

Araújo et al.3 developed and validated the CLINIMEX Aerobic Fitness Questionnaire (C-AFQ), which aims to predict an individual's CRF according to the responses to a series of simple questions regarding the maximum physical activity that the individual believes he/she would be able to undertake or perform.3 These activities have an attributed known metabolic equivalent task (METs), from which VO2 peak or VO2 max can be derived. Yet, as far as we know, its use in patients with cardiac disease has not been validated, in heart failure (HF) patients.3

The aim of this study was to evaluate the applicability and utility of the C-AFQ in an adult population of patients with cardiac disease referred for CPET at an outpatient clinic of a tertiary hospital.

The main findings of this study were: (1) C-AFQ does not accurately estimate the CRF in this study population (consisting mostly of patients with cardiac disease), despite good correlations; (2) in subgroup analysis, C-AFQ performance remains acceptable across different subgroups of patients, independently to age, sex, BMI, underlying cardiac disease and LV function, (3) C-AFQ performs better than the per protocol VO2 peak in estimating CRF and may be useful in guiding CPET protocol selection.

Cardiopulmonary exercise testing is the gold standard for assessing CRF and cardiopulmonary performance and prognostication based on VO2 peak and VE/VO2 slope among others, namely in HF. When performed in appropriate patients, maximal CPET provides a wealth of clinically useful information, including data on function, symptoms, ischemia, hemodynamic, and other diagnostic and prognostic information.12 However, its availability is limited at many centers. To address this, alternative strategies have been investigated. Questionnaires offer a user-friendly, cost-effective, reliable, and reproducible option, particularly in second line cardiology departments and developing countries where CPET is less available.13 Various surrogates, such as submaximal walking tests and non-exercise functional tools, besides symptom questionnaires, have been proposed.14 Currently, there are several other fitness questionnaires,4,12,13 which are primarily used in healthy populations and provide a quick, inexpensive, and safe way for physicians to gauge patients’ functional capacity. These questionnaires are based on the patient's physical self-perception, which correlates relatively well with measured physical fitness indicators.15 Assessment of C-AFQ accuracy in relation to other questionnaires is beyond the scope of our research.

The first questionnaire developed to assess physical perception was the Veterans Specific Activity Questionnaire (VSAQ), a 13-item self-administered questionnaire used to estimate aerobic fitness in METs.4 More recently, the C-AFQ, a questionnaire to assess aerobic fitness, was developed to surpass the unavailability of CPET and to overcome some VSAQ limitations, as previously reported.3 In a cohort of 1000 subjects, only 23.3% with known CAD and no report of patients with HF, Araújo et al.3 found a significant and very strong correlation coefficient of 0.91 between estimated (C-AFQ) and measured (CPET) VO2 peak, a correlation higher than that previous reported by Myers et al.4 for the VSAQ questionnaire (r=0.79). Indeed, the authors draw our attention to the difference in the range of exercise intensities and the scale covered in both questionnaires: VSAQ with intensities going from 1 to 13-METs compared to 1 to 20-METs in C-AFQ. Also, regarding the interval scales, in the VSAQ a 1 MET increment was used, while in the C-AFQ a 0.5 MET interval scale was adopted in the lower range of the scale. These adjustments enabled C-AFQ to be applied to both severely unfit and fitter subjects, with better discrimination and quantification.3 Also, C-AFQ has a two-step approach vs. the single-step approach in VSAQ, which enables the patients to be guided directly to a very simple and straightforward answer.

Yet, although subjective self-assessment questionnaires may be useful, they have some limitations, especially if used in complex populations, such as in patients with cardiac and pulmonary diseases. They may lead to inaccurate conclusions if applied to different populations without proper validation and may not generalize well to objective measures or broader contexts. Individuals might overestimate or underestimate their abilities or symptoms due to distorted self-perception or self-awareness, affecting areas like fitness, cognitive ability, or mental health.

These questionnaires provide an estimation of CRF, a crucial information for prognosis, although with some pitfalls. Compared to CPET, they lack the ability to offer objective measurements of various biometric parameters, many of them with prognostic implications, such as blood pressure values and ECG data regarding silent ischemia or arrhythmias, essential for comprehensive risk stratification. Additionally, the questionnaire's estimated CRF is insufficient for exercise training prescription in the context of cardiac rehabilitation programs since it does not provide HR corresponding to the first and second ventilatory threshold, which are crucial for identifying the optimal training HR range for cardiac patients, as well as HR chronotropic response and decay during recovery period.

In this prospective study, in which only cardiac patients were included, we found that C-AFQ was not sufficiently accurate to predict functional capacity, measured by VO2 peak, because even though there was a numerically strong and positive correlation (r=0.723, p<0.001) between measured VO2 peak and estimated by C-AFQ VO2 peak, there were unacceptably high levels of disagreement in the Bland–Altman plot analysis. Despite a minimal mean difference between the CPET and the C-AFQ VO2 peak values (0.62±6.93 mL kg−1 min−1), there was a wide confidence interval ranging from −13 to +14.2 mL kg−1 min−1. This 27 mL kg−1 min−1 interval is too broad, approximately equivalent to 8 METs, exactly 40% of the C-AFQ score range from 0 to 20 METs. Although the mean difference is small and negligible, the individual variation is very high, which makes the questionnaire a largely inadequate tool in our population. When interpretating Figure 2, it is worth noting an underestimation of VO2 peak in patients with lower VO2 peak and an overestimation in patients with higher VO2 peak, which confirms the inability of the C-AFQ to adequately predict VO2 peak in this population. At the individual level, this suggests that a patient could fall at either extreme end of this confidence interval. Therefore, our opinion is that C-AFQ has not yet proved to be sufficiently accurate to be used in such a complex population.

Our cohort was different from the original CLINIMEX cohort.3 Our cohort exhibited a lower mean VO2 peak values (mean 21.9±7.4 mL kg−1 min−1 vs. 25.7±0.4 mL kg−1 min−1), was older (61±12 vs. 55±16 years old), had a higher burden of major cardiovascular risk factors, had a higher proportion of patients with previously diagnosed cardiac disease and a higher number of patients medicated with antianginal drugs, mainly beta-blockers (86.3% vs. 25.6%) compared to the original CLINIMEX cohort. These results indicate there is a greater cardiovascular risk and a more pronounced limitation in functional capacity within our population.

Despite the previously reported limitations, in subgroup analysis, C-AQF VO2 peak strongly correlated with CPET VO2 peak independently of LVEF: the correlations were stronger in patients with LVEF >40% but they remain significant even in patients with LVEF ≤40%. Correlation between measurements remained significant across all subgroups of patients, regardless of age, sex, BMI, presence or absence of HF or after acute MI.

However, when analyzed individually, a moderate correlation was found in women compared to the strong correlation in male patients (r=0.580 vs. r=0.723; respectively). When age and sex were considered together in subgroup analysis, the performance of C-AFQ was better in men younger than 70 (r=0.709, p=0.001) and in women older than 70 years (r=0.711, p=0.009). The strength of the correlations found in women (below or over 70 years) must be analyzed with caution since the total number of female patients included was 31 (25% of the overall population), with only 12 older than 70 years. These findings are consistent with previous observations of female subjective insight into their functional capacity and real effort limitations.16

Additionally, the correlations were weaker (r=0.682) in obese patients, who had higher VO2 peak measured values when compared to the ones estimated by the questionnaire (18.4±5.4 vs. 17.3±7 mL kg−1 min−1; respectively). This phenomenon can be attributed to a trend for obese patients to underestimate their CRF levels, which is likely to stem from greater physical deconditioning and a lifestyle characterized by physical inactivity.

The circumstances in which a questionnaire-derived estimate of peak VO2 might be applicable are likely limited to healthy individuals who seek to gauge their functional capacity in terms of METs. C-AFQ may be useful in healthy individuals whose aim is to assess their functional capacity. However, in cardiac patients, where accurate measurement of CRF and biometric parameters are crucial for risk assessment, for managing their condition and prescribing exercise based on data derived from CPET, relying on biometric data such as HR at the level of first and second ventilatory thresholds becomes essential. Our findings highlight the wide confidence interval seen with C-AFQ measured VO2, meaning that it exhibits an unacceptable variation at the individual level, despite demonstrating relative accuracy at the population level with a minimal mean difference. Indeed, the questionnaire does not offer an objective measurement of real/true peak VO2 associated with CRF in a population where fatigue may have a multifactorial etiology. We infer that, even though this questionnaire may be useful in healthy individuals, it may have limited value in cardiac patients, where the objective definition of functional capacity and ventilatory thresholds are of major importance for risk assessment and cardiac rehabilitation programs, which require objective biometric measurements.

Nevertheless, the C-AFQ can aid in tailoring CPET protocols to individual patients, enabling physicians to select protocols that align with each patient's subjective perception of maximal physical effort. Indeed, the C-AFQ aims to objectively capture a subjective perception of maximal physical effort, thus offering potential utility in guiding CPET protocol selection.

We also assessed the association between measured, per protocol and C-AFQ VO2 peak values, an analysis not yet reported in the literature to our knowledge. We found that CPET VO2 peak strongly correlates to the exercise protocol-estimated VO2 peak (r=0.777, p<0.001). Albeit weaker, C-AFQ VO2 peak also had a strong correlation to the protocol-estimated (r=0.673, p<0.001), suggesting the potential role of C-AFQ in assisting the physician to adequately choose the appropriate protocol accordingly to patient's perception of their physical capabilities, maximizing CPET results, particularly regarding protocols with a 2–3 minute stage duration. Additionally, we also found that C-AFQ better estimates CPET VO2 peak values compared to VO2 peak estimated by the protocol in the population, since the latter has a wider range of estimated METs with subsequent lower discriminative power to predict true CRF.

To our knowledge, this is the first study to externally evaluate C-AFQ in a cohort of adult patients with heart disease, assessing the relationship between measured and perceived physical exertion, regarding heart disease etiology and across all spectra of LVEF.

Additionally, we used objective VO2 peak measured by a CPET and not estimated METs, compared to most studies evaluating the estimation of physical aerobic fitness using questionnaires.12

Physical activity and symptom questionnaires have been used as surrogates for exercise testing to estimate a patient's functional capacity and to individualize the choice of an exercise testing protocol in accordance with guidelines. In this prospective cohort study including exclusively cardiac patients, we demonstrated that the C-AFQ has limited utility in estimating CRF in this population, despite its good correlation with CPET-measured VO2 peak values. C-AFQ main utility may be to help to individualize CPET protocols.

All authors have contributed to this manuscript, reviewed, and approved the current form of the manuscript. MRL and RA were specifically responsible for manuscript conceptualization. MRL, JP, RA were responsible for data collection, analysis and first manuscript edition. MRL and JP were specifically responsible for statistical analysis. GC, LM, AD and MM specifically made the first revision and editing. AD, CGA and MM were responsible for the editing of the final version of the manuscript. All authors have contributed to this manuscript, reviewed, and approved the current and final form of the manuscript.

None declared.

The author involved in the development and publication of the C-AFQ Questionnaire also contributed as a co-author to this manuscript. This potential conflict of interest has been disclosed to ensure transparency and maintain the integrity of the research presented.

All data exposed in this article was acquired from our institution, after obtaining informed consent from patients. The data underlying this article are available in the article and in its online supplementary material.