Study design and ethical aspects
This was a prospective observational longitudinal study (registered at clinicaltrials.com—NCT04790643) and reported according to the “Strengthening the Reporting of Observational Studies in Epidemiology”28. At first, data were collected for all physical and clinical variables, after that the volunteers were followed up by independent researchers for 2 months during 24 routine sessions of CR to register the occurrence of minor AE. The data collection and follow-up were carried out between September 2021 and October 2022.
All procedures were approved by the Ethical Research Committee of São Paulo State University Campus of Presidente Prudente, São Paulo, Brazil (CAAE: 35831220.8.0000.5402), were under the Declaration of Helsinki and were performed according to the current guidelines regulating each physical test. All volunteers were informed about the objectives and procedures of the study and signed a consent form.
Participants
The population was composed of males and females included in the CR program carried out at an outpatient school clinic of physiotherapy located in Presidente Prudente, São Paulo, Brazil. Recruitment was made by convenience and all participants of the program were evaluated for eligibility by an independent researcher. Eligible patients were those with a main diagnosis of any CVD or with at least one cardiac risk factor (obesity, arterial hypertension, dyslipidemia, diabetes mellitus, and family history)29.
The inclusion criteria were not having orthopedics and/or neurologic disorders that could interfere with the physical evaluations, and not having a known lung disease, to minimize potential confounders in the pulmonary evaluations. Were excluded those who didn’t accomplish 24 sessions of exercise in the follow-up period, and those who had any alteration of the medication of daily use during follow-up. It is important to highlight that if the participants weren’t capable of correctly performing the respiratory maneuvers, and/or weren’t allowed to be submitted to bioelectrical impedance and/or presented errors greater than 5% in the RR intervals series were excluded from the analysis of these specific outcomes. Also, we performed an outlier filtering by outcome, thus values detected as outliers were excluded from the final analysis.
For the sample characterization was extracted from the patient’s medical record the presence of associated diseases, age, sex, period of treatment in CR, and medication of daily use. Also, for characterization purposes, the patient was asked the following question for assessment of stress: “Do you consider yourself a stressed person and/or that you suffer too much stress in your daily life?” if the response was affirmative, we considered that the risk factor of stress was present. The level of physical activity was evaluated by self-report according to the World Health Organization recommendation30.
Physical evaluation
Cardiovascular parameters
Resting heart rate (HR) was measured using the Polar RS800CX HR monitor (Polar Electro OY, Finland)31 in the supine position for 20 min. The average obtained between the 5th and 20th min was analyzed.
Blood pressure was measured using a stethoscope (Littmann, Saint Paul, USA) and an aneroid sphygmomanometer (Welch Allyn-Tycos, New York, USA). Three measurements were taken over 15 min in a sitting position. The average of the last two measurements was analyzed32.
Respiratory parameters
The peak expiratory flow (PEF), forced expiratory volume in the first second (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio, were obtained through spirometry33 (MIR-Spirobank II–Mini Spirometer, Italy).
Maximum expiratory (MEP) and inspiratory (MIP) pressures were assessed through a manuvacuometer (GerAr®, Brazil). The average of the 3 values obtained in maneuvers sustained for at least 1 s with no air leak was analyzed34.
All respiratory parameters were analyzed as raw values and the percentage achieved from the predicted, considering formulas validated for the Brazilian population21,35.
Cardiac autonomic modulation
Cardiac autonomic modulation was assessed using HR variability (HRV), recorded through the Polar RS800CX (Polar Electro OY, Finland)31.
Participants were instructed to abstain from stimulant substances and not to perform vigorous physical activities for 24 h. The HRV was assessed in a calm and thermoneutral environment, with the participant in the supine position, absolute rest, and spontaneous breathing for 30 min36. A series of 1000 consecutive RR intervals obtained between the 5th and 30th min were analyzed after digital (Polar ProTrainer 5 software, version 5.40) and visual filtering to eliminate premature beats and artifacts.
The indices analyzed were RMSSD; SDNN; LF (low-frequency spectral component ranging from 0.04 to 0.15 Hz) and HF (high-frequency spectral component ranging from 0.15 to 0.4 Hz) both expressed in milliseconds squared (ms2) and in normalized units (un); LF/HF ratio; SD1; SD2; SD1/SD2 ratio37. These indices were calculated at the Kubios HRV Standard software—version 3.0.038.
Quadriceps muscular strength
It was assessed using a digital dynamometer (Meditec, Brazil). The test consisted of three maximal isometric contractions from a 60º angle of knee flexion, with the body stabilized. The contractions lasted 5 s and had an interval of 30 s between them. The value considered was the highest peak strength measured among them39.
Submaximal functional capacity
Functional capacity was assessed by the 6 min walk test (6MWT). Two tests were performed on the same day with a minimum interval of 30 min, and the test with the longest distance was analyzed40.
Anthropometric measures
Were evaluated body mass index (BMI), waist-hip ratio (WHR)41, conicity index42, abdominal volume index43, and body adiposity index (BAI)44.
To obtain BMI, body mass was measured using a digital scale (Wiso, Digital Scale, W939 Body Analyzer, China) and height was assessed using a stadiometer (Sanny, Personal Caprice, Brazil). For the WHR and the conicity, abdominal volume and body adiposity indices were measured waist circumference at the smallest abdominal circumference; abdominal circumference at the umbilical line; and hip circumference at the greatest circumference in the gluteal region.
Body composition
Using a tetrapolar bioelectrical impedance45 (BIODYNAMICS, 310e, Seattle-USA) were evaluated the percentage of body fat, lean mass, and phase angle. During the exam, participants remained in the supine position with the electrodes positioned in the right hand and foot46.
Participants were instructed not to ingest any food or beverages for 4 h, not to perform physical exercises for 12 h, not to drink alcoholic beverages for 48 h, and to empty the bladder 30 min before the test. As a safety measure, the participants didn’t interrupt the diuretic medication46.
Cardiac rehabilitation program
The CR program was carried out at the Physiotherapy Study and Care Center located at Presidente Prudente, São Paulo, Brazil, an outpatient school clinic. The CR sessions were performed three times a week on alternate days in a room with controlled temperature (22–24 °C).
The sessions consisted of initial rest (5 min), warm-up (15 min), aerobic exercise (30 min) and cool-down (10 min). The initial rest period was dedicated to the evaluation of resting HR and blood pressure. The exercise period started with the warm-up, composed of stretching and varied free active exercises of progressive level of effort. After the warm-up, the participant performed a moderate-intensity continuous aerobic exercise on a treadmill or exercise bike. At the end of the aerobic exercise, the participants walked at a slow pace around the room and seated for cool-down.
The aerobic exercise intensity was set at 40–80% of the HR reserve, according to the ACSM recommendation29. The resting HR was extracted from the previous three sessions of CR before the baseline assessment. As for the maximum HR it was extracted from a stress test performed whiting the last year that was attached to the medical record, however, for those who did not have a recent stress test report, the formula of 220-age was used.
Register of adverse events
The AE were selected based on the study of Vanderlei et al. which identified the most prevalent AE in exercise-based CR8. We considered AE the: abnormal rise of systolic blood pressure during exercise, considering values ≥ 200 mmHg; muscle pain, considered as any discomfort referred in the skeletal muscles in any region of the body; fatigue, considered as a physical unpleasant sensation referred as tiredness that doesn’t ease with strategies to restore energy and accompanied by cognitive and emotional components; dizziness, referred as a sensation of altered body balance and spatial disorientation; and chest pain, considered as pain or discomfort of constrictive, compressive, or burning nature at the substernal or interscapular region, neck, mandible, arms, and fingers8, however, due to logistics was not possible to assess with an electrocardiogram if the report of chest pain was associated with actual ST alteration.
The presence of any of the symptoms tracked was asked by the physician at least one time in each session phase (rest, warm-up, aerobic exercise, and cool-down). Also, to ensure reliability, all participants were instructed regarding the definition of the symptoms and the importance of not omit any discomfort. Therefore, we considered as “adverse event” the report of any symptom during the CR session. Each report was counted as “one event”, however, events reported in the initial rest that were maintained or not during the exercise period were not included in the analyses (eFig. 3). The AE outcome consisted of the absolute value of events recorded throughout the follow-up period per participant, value used to develop the dichotomic models of analysis.
Due to the subjective characteristic of the AE tracked when reported, they were assessed concerning its intensity through a five-point Likert scale, being 1 very low and 5 maximum, to guide the conduct of care to ensure safety. When any symptom was reported the exercise load was reduced and the intensity of the discomfort was assessed again. If the symptom worsened or did not cease the exercise was interrupted and the participant would be referred to a medical center for further investigation. It is worth mentioning that during the study none patients had severe symptoms that required medical interventions.
Statistical analysis
The characterization variables were analyzed using descriptive statistics. The relationship between the dependent (occurrence of events) and independent (physical and clinical characteristics) variables was analyzed using uni and multivariable binary logistic regression.
The accuracy of the univariate models was quantified by the area under the ROC curve (AUC), and for the multivariable models the quality of the model fit was evaluated by the Hosmer–Lemeshow test, and for internal validity, simple bootstrap resampling analysis was performed based on the seed “2000000” for the generation of Mersenne Twister random numbers considering 100 samples with bias-corrected and accelerated 95% confidence interval.
The dependent variable, the occurrence of AE, was categorized into six models considering the occurrence of 1 to 6 AE, to this aim, the occurrence of AE in each model was defined as “positive” or “negative” accordingly to the following: model 1 (0 events = negative/1 or more = positive); model 2 (0 or 1 event = negative/2 or more = positive; model 3 (at most 2 events = negative/3 or more = positive); model 4 (at most 3 = negative/4 or more = positive), model 5 (at most 4 = negative/5 or more = positive) and model 6 (at most 5 = negative/6 or more = positive). After an outlier filtering based on the study of Hoaglin & Iglewicz47, physical variables were categorized into three models based on: (a) the median; (b) the 33% and 66% percentiles to obtain three groups with a similar sample size, and, (c) the first quartile cut-off for those variables in which a low value represents worse performance or third quartile for variables in which a high value represents worse performance. Clinical variables were categorized as “presence” or “absence” of the characteristic.
The analyses were performed using IBM SPSS Statistics—version 22.0 (IBM Corp, Armonk, New York) and MedCalc—version 14.8.1 (MedCalc Software, Ostend, Belgium) considering a statistical significance of less than 5%.
Missing data
There were missing data for the baseline characteristics evaluated. All missing data consisted of participants that for any reason, as described in the previous sections, did not perform some of the evaluations or had the obtained value considered an outlier. In these cases, we applied the deletion method, removing the participant only from the analysis of the specific outcome in which the data is missing. It is important to highlight that from all outcomes six participants were excluded due to a loss of follow-up, as pointed in Fig. 1.
The percentage of missing data per outcome, considering only the losses due to the absence of the evaluation and outlier filtering was: 27.8% for muscle strength, of which 21 participants did not perform the test and 1 was considered an outlier; 17.7% for 6MWT, of which 13 did not perform the test and one was considered an outlier; 21.5% for MIP due to incorrect maneuver and 27.8% for MEP, as additional five participants were considered as outliers; 16.4% for FVC, PEF, and FEV1 due to incorrect maneuver, and 17.7% for FEV1/FVC as one participant was identified as an outlier; 15.2% for the HRV indices of LF and HF in normalized units, SD1, SD2, and SD1/SD2 ratio as 11 participants did not perform the evaluation and one was an outlier; 17.7% for rMSSD, SDNN, and LF in ms2 that had three outliers excluded; 19% for LF/HF ratio due to four outliers; and 21.5% for HF ms2 due to six outliers; 1.3% for WHR and SBP due to one outlier and 2.5% for BAI and HR due to two outliers.
Sample size
The sample power was analyzed considering the analysis of AUC. The expected ratio between negative and positive cases was set at 0.6 estimated based on the previous study from our research group with a different set of participants from the same cardiac rehabilitation program6. The null hypothesis considered was an AUC of 0.5, the expected effect was set at an AUC of 0.79, alpha error at 0.05, and the power at 80%. The resulting sample was 66 participants, and accounting for 10% of sample loss the intended sample size was 73 participants.
link