The individuals included in this study were informed about all the procedures and provided written informed consent to participate. This study was approved by the Ethics Committee of the institution (CAAE: 62437816.4.0000.5515) and abides by the CONEP resolution 466/2012. The registration of this randomized clinical trial can be found at Clinicaltrials.gov (NCT03169387).

This is a parallel group randomized clinical trial conducted at the physical therapy clinic of the Oeste Paulista University (UNOESTE), in the city of Presidente Prudente, São Paulo, Brazil, from February to October 2017. The sample comprised 27 individuals divided into two groups: a conventional rehabilitation group (CRG) and a virtual reality rehabilitation group (VRG). Based on sample calculation, each group had at least 12 individuals, using the study by Pimenta et al. (2013)18 as reference. Fat-free mass was used, with standard deviation of 4.02, difference to be detected of 3.8 for the two-tail hypothesis test, power of 80%, and significance level of 5%. Figure 1 shows the flow diagram of the participants in every phase of the study, in accordance with the recommendations of the CONSORT Statement.19

Figure 1

The random allocation sequence was generated by a researcher without previous contact with the participants by use of the Microsoft Office Excel^® program, at an allocation ratio of 1:1.

The study included individuals over the age of 45 years, of both sexes, with CVD (coronary heart disease, postoperative period of coronary artery by-pass grafting, acute myocardial infarction, systemic arterial hypertension, diabetes mellitus). The inclusion criteria were as follows: hemodynamic stability (systolic blood pressure < 200 mmHg and diastolic blood pressure < 110 mmHg at rest, absence of angina, controlled arrhythmias, and resting heart rate < 120 beats per minute);20 absence of arteriopathy and muscle or orthopedic changes; and no supervised physical activity in the previous 30 days. The exclusion criteria were as follows: decompensations (circumstances posing a risk to individual integrity) during the training protocol; lack of adaptation to the protocol; and participation frequency lower than 75%.

Participants underwent an initial evaluation to detect comorbidities, to establish the diagnosis, to collect the clinical history and medications used, and to measure the anthropometric variables. The following parameters were assessed before the intervention and eight weeks after that: body composition; waist circumference; food frequency; functional capacity; and blood glucose levels.

Weight (in kilograms) was measured with a WELMY W300^® scale (accuracy to the nearest 100 g), with the individual barefoot wearing light and comfortable clothes. Height (in meters) was measured with a Sanny^® stadiometer (accuracy to the nearest 0.1 cm), with the individual barefoot, standing with his/her back to the height rule, feet together, head positioned in the Frankfurt plane, and the measuring rod lowered to the individual's head. Body mass index (BMI) was calculated based on weight and height (weight/height²).

Body composition was the study's primary outcome, and body fat percentage was defined as the primary variable. Body composition was assessed by use of bioimpedance analysis with a tetrapolar bioimpedance device (Biodynamics^®, model 310e), analyzing and quantifying the body fat percentage, body fat weight, lean weight, basal metabolic rate, and percent and total water. Bioimpedance analysis measurements were taken in the morning, in fasting condition, after voiding the urinary bladder, having suspended diuretics for 24 hours, no alcohol and caffeine consumption for 24 hours, and no intense physical activity for 72 hours in both the initial and final assessment.21^,22

Waist circumference was measured with a non-distensible measuring tape (accuracy to the nearest 1 mm) and the subject standing, with parallel feet and arms hanging freely by the sides of the trunk. The measuring tape was placed in a horizontal plane around the abdomen, at the level of the umbilicus, just above the uppermost lateral border of the right iliac crest, not compressing the skin. Measurement (in centimeters) was taken three times, and the lowest value obtained was considered for analysis.23

The Food Frequency Questionnaire (FFQ) was used to assess the frequency of food consumption. The FFQ consists of a list of foods and beverages with response categories to indicate the frequency of consumption over the time queried. The frequency is registered in units of time (days, weeks, months or years) according to the need.24

Functional capacity was assessed by use of the Six-Minute Walk Test (6MWT) performed according to the American Thoracic Society criteria.25 The 6MWT was performed along a corridor of the UNOESTE gymnasium.

Postprandial capillary blood glucose level was assessed (milligram per deciliter) by use of an Optium Xceed^® blood glucose meter that reads glucose in fresh capillary blood obtained by pricking the ring finger skin with a sterilized lancet. The measurements were taken before and after training in an individualized way.

Heart rate was measured by use of a Sigma^® cardiac frequency meter. Oxygen saturation was measured by use of a pulse oximeter (Choicemmed^®, model Md300c1). With the subject in the sitting position, blood pressure was measured from the dominant arm, using a Premium^® aneroid sphygmomanometer and a Littman^® stethoscope, following the recommendations of the Brazilian guidelines on arterial hypertension.26 The subjective perception of exertion was assessed by use of the Borg scale of perceived exertion.27

Both groups underwent 60-minute training sessions twice a week for eight weeks at the UNOESTE physical therapy clinic, adding up to 16 sessions. Four physical therapy students conducted the sessions, two for each group, and each student was responsible for two participants per session. In the initial and final 5 minutes, the following parameters were assessed: blood pressure, blood glucose level, oxygen saturation, and subjective perception of exertion by use of the Borg scale. During the entire session, the Borg scale was used and heart rate was assessed to ensure that the training heart rate calculated individually with the Karvonen formula would not be exceeded (50% to 80% of the reserve heart rate).27^,28

The VRG training comprised the use of the Microsoft Kinect(tm) sensor for Xbox 360^®, with a controller-free infrared camera that tracks user motion. Two games were used: Your Shape(tm) (Fitness Evolved) and Dance Central 3(tm). Your Shape(tm) (Fitness Evolved) is a program that allows users to exercise their upper and lower limbs in an isolated or combined way, and comprises mainly trunk rotation, diagonal flexion/adduction and extension/abduction movements, plantar flexion of ankle, hip flexion and squat. Dance Central 3(tm) is a music rhythm game involving performing given dance moves, which are tracked by Kinect. Both were performed for 25 minutes, the former in association with ankle Velcro weights and dumbbells as resistance. The increase in resistance was individualized and based on the Borg scale of perceived exertion (13: somewhat hard).27

The CRG training comprised the use of treadmills (Embreex^®) for aerobic training for 30 minutes and free weights, with 1-4-kg dumbbells and 2-4-kg ankle Velcro weights, for resisted training, performed in three sets of 10 repetitions, with 1-minute recovery interval. The exercises comprised shoulder abduction, elbow flexion, knee extension and flexion, all performed in the sitting position, except for knee flexion, performed in the standing up position.27 In the CRG training protocol, resistance was increased similarly to that in the VRG training protocol.

Data were assessed by use of the GraphPad Prism statistical software. Data normality of distribution was analyzed with Shapiro Wilk test. Proportion was compared by use of chi-square test. Paired analysis was carried out by use of paired Student t test in case of normal distribution or Wilcoxon test for non-normal distribution variables. Intergroup comparisons were performed by use of absolute variation before and after the interventions, and nonpaired Student t test or Mann Whitney test were used according to data distribution. The magnitude of the differences between groups was described by calculating the effect size, using Cohen's d. The effect sizes considered were as follows: small (d≤0.2), medium (d = 0.5), and large (d ≥ 0.8). Continuous variables were expressed as mean and standard deviation or median and interquartile range, according to normal data distribution. Categorical variables were expressed as absolute and percentage values with their respective confidence intervals. The significance level adopted was 5%.

Table 1 shows the clinical and anthropometric data of the individuals assessed, whose mean age was 63.46 ± 8.12 years, most of whom were of the male sex. There was no significant difference between groups regarding age and anthropometry. Systemic arterial hypertension was the major comorbidity. Anti-hypertensives were the most frequently used medication.

Table 2 shows the baseline and final body composition of both groups. The VRG showed a significant difference regarding the increase in body fat percentage and in fat weight, in addition to a decrease in basal metabolic rate and total water, while the CRG showed no significant difference. Waist circumference did not significantly differ between CRG and VRG.

Table 3 compares the absolute variation in body composition between the groups, evidencing significant differences regarding body fat percentage and fat weight, which were higher in the VRG, while total water was significantly lower in the VRG than in the CRG. The other variables showed no significant difference. The effect size of body fat percentage between groups was considered high (d = 0.96), as was the effect size of fat weight (d = 0.89). The effect size of lean weight was small (d = 0.06).

Regarding food frequency, the paired analysis showed no significant difference in most food groups (p > 0.05), but a significant increase in the intake of sweets and desserts was observed in the CRG (p = 0.0425), while the VRG showed a significant increase in the intake of vegetables (p = 0.0455), sauces and seasonings (p = 0.0245). Table 4 compares the food frequency variation between the two groups, showing a significantly higher intake of vegetables, sauces and seasonings in the VRG as compared to that of the CRG. No significant difference was observed in the consumption of other food groups.

Figure 2 shows the functional capacity values of the individuals assessed. The distance covered increased significantly in both groups after undergoing the training protocols.

Figure 2

CRG: conventional rehabilitation group; VRG: virtual reality rehabilitation group.

One individual reported dyspnea and intense fatigue when undergoing the VR protocol, which interrupted the session.