Virtual Reality-Assisted Exercise in Pediatric Hematopoietic Stem Cell Transplantation

Design: A randomized controlled, single-blind (assessor-blinded), parallel-group trial conducted at the Cardiopulmonary Rehabilitation Unit, Department of Physiotherapy and Rehabilitation, Faculty of Health Sciences, Istinye University. Pediatric patients who had undergone HSCT were referred from the Pediatric Bone Marrow Transplantation Unit of Medicalpark Bahçelievler Hospital, Istanbul. Eligible patients (5-18 years) who met clinical-stability criteria and had no history of orthopedic, neurological, or cardiac disease were enrolled after written parental consent and child assent.

Randomization: Stratified by sex using a computer-generated 1:1:1:1 sequence to ensure homogeneous sex distribution across the two arms. Concealment was achieved through computer allocation. Participants were assigned to (1) Control Group (CG): conventional pulmonary physiotherapy and rehabilitation, or (2) Experimental Group (EG): the same conventional program PLUS supervised, virtual reality-assisted, video game-based exercise. The outcomes assessor was blinded to group allocation; participants and treating physiotherapists could not be blinded due to the nature of the intervention.

Conventional pulmonary physiotherapy program (both groups): diaphragmatic breathing; thoracic expansion exercises (chest and bilateral segmental breathing); incentive spirometry (Triflo® or Voldyne®); postural drainage and percussion; cough technique training; breathing control; range-of-motion exercises. Each session lasted 25 minutes and consisted of 2 sets × 5 repetitions with 5-6 tidal-breath rest intervals to avoid respiratory-muscle fatigue. Exercise intensity was titrated using the Modified Borg Scale, targeting perceived dyspnea/fatigue between 4 and 6 (moderate intensity).

VR-assisted, video game-based program (Experimental Group only): delivered concurrently with the conventional program in 24 supervised face-to-face sessions over 8 weeks (3×/week). Two platforms were used in combination. (a) Breathing Labs Breathing Games (Breathing Labs, Slovenia): five games selected per patient preference - Balloon (level 1 weeks 1-4 → level 2 weeks 5-8), Plane (6 s → 8 s), Wolf (easy → moderate), Flowers (sunflower → rose), and Pluto (easy → moderate). Patients were seated on a backless chair to prevent falls; sessions were paused if dizziness or loss of control occurred. Visual score feedback was used to reinforce engagement. (b) Nintendo Wii Fit U with the Wii Balance Board: 8 balance/mobility-targeted games, with 4 of 8 played per session for 5 minutes each. Weight-shifting games (Balance Bubble, Table Tilt, Perfect 10, Super Hula Hoop) required mediolateral and anteroposterior weight shifts on the balance board; stepping/squatting games (Obstacle Course, Advanced Obstacle Course, Ski Jump, Basic Step) required marching, squatting, leg extension, or stepping. Participants stood with or without chair-back hand support depending on tolerance.

Assessments: A blinded assessor performed pre- and post-intervention testing in fixed order across separate days to prevent fatigue. The full battery comprised: sociodemographic and disease-related variables (diagnosis, transplant type, donor type, stem-cell source, time since transplant, conditioning regimen, HLA matching) extracted from the hospital electronic medical record; bioelectrical impedance analysis (Tanita) for body composition; waist and hip circumferences and waist-to-hip ratio; pulmonary function testing on a portable spirometer (Cosmed Pony FX, Italy) per ATS/ERS criteria with single-use mouthpieces and filters (FVC, FEV1, FEV1/FVC, PEF expressed in liters and as % predicted); respiratory muscle strength using a portable electronic mouth-pressure device (Micro Medical Micro RPM, Rochester, UK) recording MIP and MEP in cmH2O (best of three trials, expressed as % age- and sex-predicted); inspiratory muscle endurance with the Powerbreath Plus MR (HaB International, Southam, UK) at 60% MIP for up to 10 minutes; peripheral muscle strength using a handheld dynamometer (Lafayette Model 01165, IN, USA) for shoulder flexors/abductors and elbow flexors/extensors bilaterally; handgrip strength with a Jamar dynamometer (Nottinghamshire, UK) in seated position with elbow at 90° and wrist neutral, best of three trials, recorded in Newtons; respiratory muscle tone of sternocleidomastoid, upper trapezius, and pectoralis major using MyotonPro (Myoton AS, Tallinn, Estonia) with the device perpendicular to the marked muscle belly, recording frequency (Hz), stiffness (N/m), decrement, relaxation time (ms), and creep; diaphragm thickness measured at the 9th intercostal space (anterior axillary line) at end-expiration using B-mode 7 MHz ultrasound (Toshiba Aplio XG, Tokyo, Japan), and right-hemidiaphragm excursion during quiet breathing, deep breathing, and sniff testing using M-mode with a 3.5 MHz curved-array transducer (left hemidiaphragm not assessed due to image-quality limitations); pleural thickness assessed by the same ultrasound system. Functional fitness was assessed via the Modified Alpha-Fit Pediatric Test Battery: 6-minute walk test on a 30 m flat corridor (cardiovascular fitness); standing long jump and Jamar handgrip (musculoskeletal fitness); sit-and-reach test using a 35 × 45 × 32 cm box (flexibility); flamingo balance test on a 50 × 4 × 3 cm beam over 1 minute (balance); waist circumference and BMI (body composition; Turkish pediatric reference values per Neyzi et al., 2015).

Statistical plan: Sample size was calculated a priori with G*Power v3.1 (two-tailed, α=0.05, power=0.80, medium effect size d=0.40), targeting n=40 (20 per group). Given the achieved sample size and the resulting Type II error risk, all variables were analyzed with non-parametric tests. Within-group changes were tested with the Wilcoxon signed-rank test; between-group comparisons used the Mann-Whitney U test for continuous variables and the chi-square test for categorical variables. Descriptive statistics are reported as mean ± standard deviation. Post-hoc power analysis using the primary outcome (vertical jump test) within the Experimental Group yielded an effect size of 0.98 with α=0.05 and power of 0.80. Of 38 enrolled, 22 were withdrawn during the intervention (3 incomplete final assessments due to clinical instability or refusal; 1 scabies; 5 cutaneous GvHD with high-dose steroid therapy; 1 respiratory-tract infection; 6 systemic infections requiring isolation; 3 new psychiatric complications with poor prognosis; 3 relocation precluding follow-up). Sixteen participants completed the trial (CG n=8; EG n=8). Funding: Health Instıtutes of Turkiye (TÜSEB), Project No 31278, Program Code 2022-ACİL-11 Group A. The investigators declared no conflicts of interest.