Exercise Medicine Clinic (CLINIMEX) is a privately-owned medical enterprise founded at the beginning of 1994, located in Copacabana, Rio de Janeiro, Brazil. Its primary aim is to provide medical services through comprehensive evaluation protocols and advanced medically-supervised exercise programs. The CLINIMEX Exercise open cohort was established in 1994 incorporating all individuals evaluated since. Between January 1994 and February 2023, after excluding those formally classified as athletes,²² a total of 9055 individuals voluntarily underwent a medical-functional evaluation to assess their health and levels of physical fitness and to obtain advice regarding exercise and/or sports practice. The sample was largely white, and most of the individuals belonged to the country's upper socioeconomic and education strata. In general, individuals were referred by their attending physicians, although some were self-referred for the evaluations. Only data from their initial (baseline) evaluation were utilized for this study in those who underwent multiple evaluations.

For this analysis, a pre-defined sex and age range that included data from men aged 46 to 70 years were retrospectively reviewed (N = 2792). A final sample of 2201 individuals was obtained (78.9% of the initial sex- and age-selected sample) after applying the following inclusion criteria: a) having voluntarily agreed to be evaluated and given permission to have their data used for research purposes; b) absence of locomotor or musculoskeletal complaints or clinical restrictions to perform an CPX; c) having undergone a maximal CPX using a leg cycling ergometer; and d) having data available for censored vital status, including height (measured to the nearest 0.1 cm), weight (measured to the nearest 0.1 kg, without shoes and while wearing minimal clothing), BMI (kg/m²), and COP. In addition, information regarding relevant clinical variables were reviewed by the evaluating physician and included as a yes/no classification for: previous medical diagnosis of coronary artery disease; hypertension; dyslipidemia; obesity; diabetes mellitus; and history of myocardial infarction or coronary revascularization (either bypass graft or percutaneous interventions) (see dataset flowchart in Supplementary Figure 1).

All evaluations were carried out by only five well-trained exercise and sport physicians in a standard exercise lab facility, adequately prepared in terms of personnel and equipment to deal with medical emergencies.²³ Details regarding the identification and definition of COP are available in previously published studies.^12–20 In brief, during an individualized ramp-protocol cycling CPX (Cateye EC 1600, Japan, from 1994 to 2006 and Inbramed CG-04, Brazil from 2007 to 2023) designed to last between 8 and 12 minutes, the expired gases were collected by the use of a Prevent pneumograph (MedGraphics, United States) coupled to a mouthpiece, with concomitant use of a nose clip. The expired gases were measured and analyzed with a TEEM100 system (Aerosport, United States) from 1995 to 1999 and VO2000 metabolic analyzer (MedGraphics, United States) from 2000 to 2023. The systems were calibrated with a 2L-syringe and with gases of known concentrations before the first assessment of the day. Pulmonary VE and O₂ and CO₂ partial fractions were expressed every 10 seconds and averaged for each minute of CPX using a spreadsheet. The COP was obtained by identifying the minimal VE/VO₂ ratio at a given minute of CPX, thus being a non-dimensional value. Since COP is obtained at a relatively low submaximal exercise intensity, ^{12–14, 16, 21, 24} individuals who did not achieve formal criteria for a maximal CPX were not excluded from the study.

Vital status (alive or dead) was last censored on October 31, 2022, and date of death were obtained from the official records of the Health Secretary of Rio de Janeiro State. ICD-10 coding was used to identify the deaths with cardiovascular diseases as the main cause. The duration of follow-up was expressed in days, calculated as the time between the evaluation and last censored dates. All individuals read and signed a consent form allowing the evaluation to be undertaken and giving permission to use their de-identified results for research purposes. Both the evaluation and research protocols were formally registered with relevant governmental bodies and approved by an external research ethics committee.

Baseline characteristics were presented as mean ± standard deviation for continuous variables and count (percentage) for categorical variables. Depending on the type of variable analyzed, chi-square or one-way ANOVA were used to compare groups. Death rate (%) was separately computed for each decile of COP, enabling the identification of the most appropriate data grouping for survival analysis (see Supplementary Table 1), resulting in three numerically convenient round values of COP for the grouping: low (< 28), high (28 to 30) and very high (> 30). Kaplan–Meier curves were constructed, and log-rank (Mantel–Cox) tests were used to analyze survival times according to COP groups. Cox proportional hazard regression models were then used to estimate hazard ratios (HRs) with 95% confidence intervals (CI) for the association between COP and cardiovascular mortality risk, after confirming no major departure from the assumptions of the proportionality of hazards using Schoenfeld residuals. Data from other variables that were significantly associated with cardiovascular mortality were used as covariates to assess the independence of the association. HRs were calculated and progressively adjusted using the following four models: unadjusted (model 1); age (model 2); age, presence or absence of coronary artery disease, and history of myocardial infarction (model 3); age, coronary artery disease, history of myocardial infarction, and diabetes mellitus (model 4). Selection of covariates for adjustment was based on their previously established role as risk factors for cardiovascular mortality, according to evidence from previously published studies, or their potential as confounders based on known associations with cardiovascular mortality using the available data.

Additionally, a two-way ANOVA (COP groups and age groups) and Tukey's multiple comparisons were conducted to assess the differences in death rates (%) between three COP groups within five distinct 5-year age intervals: 46 to 50, 51 to 55, 56 to 60, 61 to 65, and 66 to 70 years. All statistical analyses were conducted using Prism 9.5.1. (Graphpad, United States), using a two-tailed p < 0.05 as a criterion for statistical significance. This written report adhered to STROBE recommendations for cohort studies.

To provide clinically relevant information, epidemiological studies should provide an adequate combination of event rates, follow-up period, sample size, and main characteristics of the sample. The minimal required sample size is inversely related to the absolute number of events during the follow-up. The follow-up period should also consider the characteristics of the events being censored as well as age and the typical life expectancy for the age of the individuals included in the sample. These aspects were considered when selecting the sex and the age range of the individuals for this current study. Since previous data have indicated that age-matched healthy men and women differ in terms of COP,¹² we selected only individuals of one sex (male). While data were available for younger and older individuals in the cohort, the death rates were too low or too high, respectively, during the available follow-up period to make meaningful and clinically relevant conclusions regarding the association between COP and cardiovascular mortality.

The CLINIMEX Exercise cohort started in 1994. Since then, it has been collecting medical and functional data among individuals of both sexes aged from 6 to 98 years who have voluntarily participated in the evaluation protocol. Notwithstanding, there are some characteristics of this cohort that deserve specific comments. The individuals evaluated in the cohort were primarily white, and they pertained to high educational and economic strata. Thus, they are not generally representative of the Brazilian population. Detailed information about the cohort is available in the supplemental materials of a previously published study.³⁴

In spite of the fact that most of the participants had a diagnosis of coronary artery disease or risk factors at their baseline evaluation, the annualized death rate was about half of that expected for the general Brazilian population for a similar mean age of 57 years.³⁵ This may be due to the demographic characteristics of this sample, as well as the possibility that these individuals had adhered to prescribed optimal medical treatment for their clinical conditions, including the chance to participate in a long-term medically-supervised exercise program.

COP has a very clear and logical physiological background, representing the minimal amount of ventilation required to consume one liter of oxygen and, very likely, reflecting the best possible interaction between circulation (heart and vessels) and respiration (lungs and cells). A previous study has shown that COP is equal or superior to several other CPX-related variables in terms of stability of measurement.³⁶ Given that it occurs at a submaximal exercise intensity, it does not require a maximal CPX, a potentially relevant advantage for several clinical conditions, as shown, for example, in patients experiencing panic disorders.²¹ In addition, unlike several other commonly used CPX variables such as VO₂ max, VO₂ peak, anaerobic or ventilatory thresholds, oxygen uptake efficiency slope, the VE/VCO₂ slope, and others, the determination of COP by the examiner is extremely easy to accomplish, and it is a practically observer-free error variable. COP has been safely determined in different populations, including patients with heart failure^{14, 31, 37} and also in individuals over 80 years of age,¹⁶ performing submaximal and maximal CPX. Indeed, it is possible for anyone retrieving data from a CPX performed in the past with access to the minute-to-minute listing of VE/VO₂ results to precisely identify the COP.

The clinical relevance of COP has been assessed in healthy adult men,¹⁵ in healthy middle-aged men,^17,18 in healthy and unhealthy middle-aged and older men and women,²⁰ in a small sample of adults with congenital heart disease,³³ and in male and female patients with heart failure.^{14, 31} In a previous study with the CLINIMEX Exercise cohort,²⁰ COP was found to be a strong predictor of all-cause mortality in a broad age range (41 to 85 years) among healthy and unhealthy individuals who were followed for a mean of 6 years. In the United States, Peterman et al.,¹⁵ found a significant association between COP and long-term mortality in 3160 apparently healthy individuals (46% women) between 18 and 85 years of age. Laukkanen et al.,^{17, 18} reported a significant role of COP in predicting all-cause and cardiovascular mortality and sudden cardiac death in a large sample of healthy Finnish men aged 42 to 61 years. Two different European cohorts of patients with heart failure were studied by Reis et al.³¹ and Kensen et al.,¹⁴ and high values for COP were similarly associated with poor survival. In a study conducted at a hospital in Lisbon,³¹ COP had the highest area under curve for 442 patients with heart failure for cardiac death and urgent heart transplantation over a 12-month period, and COP significantly added to the prognostic power when compared with other CPX ventilatory variables. More recently, Wenhart et al.,³³ showed that high COP values provided an objective feature to differentiate lesion severity in adults with congenital heart disease.

Our study confirms the available data from different populations and reinforces COP > 30 as associated with poor survival in men.^{14, 37} Preliminary evidence from the CLINIMEX Exercise cohort and from a Dutch cohort have suggested that attending exercise-based cardiac rehabilitation programs may reduce very high COP values.^{14, 38} In this sense, normalizing a very high COP by regular exercise training could be one of the pathophysiological mechanisms leading to better survival in patients enrolled in cardiac rehabilitation programs.³⁹

This study has several strengths, including its planning, design, data collection, and analysis. Data for this study came from a long-term open cohort that has been utilized in several other epidemiological studies.^{20, 40} This dataset has been carefully cleaned and organized over several years by the principal investigator and close collaborators, making the available information accurate and valid.³⁴ The quality of censoring was very high, being based on official sources reporting ICD-10 classified death certificates and complemented, when appropriate, by the patient's electronic records and by information obtained from attending physicians, relatives, and close friends. Over the 29-year study period, all CPX were conducted in a single center/exercise lab, using a cycle ramp protocol and metabolic analyzers for expired gases from the same manufacturer. All the pertinent medical information and CPX were conducted by only five specialized physicians, following strict protocols, and all data were entered into specifically-designed digital forms (Quattro Pro, Corel, Canada).

In addition, potential confounders were adequately controlled. Moreover, since differences in COP by mode of exercise (treadmill or cycle) have not been formally investigated, we selected only CPXs carried out using a leg cycle ergometer.

Due to the previously detailed demographic and clinical characteristics of the sample extracted from the CLINIMEX Exercise cohort, considerable caution should be applied in generalizing these results for other populations. Since this was a clinical study using a convenience sample of individuals referred to the CPX in our clinic for several distinct reasons, mixing healthy and unhealthy individuals, it was not possible to control for previous or regular use of medications. It has not been formally studied whether there is an influence of specific medications known to have cardiovascular and/or respiratory effects on COP. This should ideally be addressed in studies using repeated CPX with individuals on and off selected medications under well controlled conditions.

No maximal CPX data, such as maximum oxygen uptake or maximum heart rate, were reported, since both submaximal and maximal CPX tests were used to analyze the COP. In this context, no direct or add-on comparisons were made between COP and other hemodynamic and respiratory CPX variables. The feasibility and the validity of obtaining COP in a submaximal CPX was first shown in our study among patients with panic disorders²¹ and confirmed by data from a Portuguese study of patients with heart failure.³¹ Nevertheless, it remains to be properly addressed in future studies whether combining COP and VO₂ max results would improve risk stratification more than each of these variables separately.