Design and setting. The trial is designed according to the Declaration of Helsinki, Good Clinical Practice guidelines, and the CONSORT 2010 statement. All assessment, education, and exercise sessions will be conducted in the environment where participants provide care (the care recipient's home) and/or in the practice areas of the affiliated home-care/rehabilitation/support unit. Assessment and application areas will be standardized (standard chair, exercise mat, and a safe environment suitable for simulating patient transfer). The study will start only after ethics committee approval and written informed consent.
Sampling and randomization. Caregivers meeting the eligibility criteria will be enrolled by consecutive sampling. Participants will be allocated 1:1:1:1 to four groups by block randomization (block size 4 or 8) using a computer-based random number generator. The randomization list will be prepared by an independent researcher who takes no part in assessment or intervention delivery, and allocation will be concealed using sequentially numbered, opaque, sealed envelopes.
Sample size. Sample size was calculated a priori with G*Power 3.1.9.7 for a repeated-measures ANOVA (within-between interaction) testing the group × time interaction of the primary outcome (ZBI) in a 4 (group) × 3 (time) mixed design, with effect size f = 0.25, α = 0.05, power = 0.80, and correlation among repeated measures = 0.50. The required total sample was 40 (actual power 0.805). To allow balanced 1:1:1:1 block randomization, the minimum required sample was set at 48 participants (12 per group; power 88.7%), and with an approximately 20% allowance for expected drop-out, a total of 64 caregivers (16 per group) is planned (power 96.7%). Sensitivity analysis indicated that N = 48 detects f = 0.225 and N = 64 detects f = 0.193 at 80% power.
Blinding. Because of the nature of the educational and exercise interventions, neither participants nor the treating physiotherapist can be blinded. A single-blind (outcome assessor blinded) design will therefore be used: all T0, T1, and T2 assessments will be performed by an independent physiotherapist/researcher unaware of group allocation, and the researcher performing the statistical analysis will also be blinded to group codes (A/B/C/D).
Statistical analysis. Analyses will be performed with IBM SPSS Statistics 27.0 (and R 4.3, nlme/lme4, where required); two-sided α = 0.05. Primary analyses will follow the intention-to-treat principle on all randomized participants, with a confirmatory per-protocol analysis restricted to participants attending at least 75% of sessions. Normality (Shapiro-Wilk, skewness/kurtosis, Q-Q plots), homogeneity of variances (Levene), equality of covariance matrices (Box's M), and sphericity (Mauchly, with Greenhouse-Geisser or Huynh-Feldt correction) will be examined. The primary hypothesis will be tested through the group × time interaction of a 4 × 3 mixed-design (repeated measures) ANOVA. If clinically relevant baseline imbalance is present, ANCOVA adjusting for T0 scores will be used. Where normality is seriously violated, data transformation or non-parametric alternatives (Friedman; Kruskal-Wallis H with Dunn-Bonferroni post hoc) will be applied. Post hoc between-group comparisons will use Tukey HSD (or Games-Howell if variances are unequal) and within-group comparisons Bonferroni-adjusted pairwise tests; Benjamini-Hochberg false discovery rate correction will be applied to secondary outcomes. Dichotomous Nordic Musculoskeletal Questionnaire data will be analysed with Cochran Q and McNemar tests within groups and chi-square or Fisher's exact test between groups, with relative risks and odds ratios and 95% confidence intervals. Effect sizes will be reported as partial eta squared and Cohen's d with 95% confidence intervals; for ZBI, responder rates based on the minimal clinically important difference and the number needed to treat will also be presented.
Safety. All exercise and transfer practice will be supervised by a physiotherapist on safe flooring and according to biomechanical principles. Perceived exertion will be monitored (Borg scale). In case of any adverse event (increased musculoskeletal pain, excessive fatigue, dizziness), the session will be modified or terminated and the event recorded.
