Abstract
Abstract
Background
Digital health technologies are increasingly being used in scoliosis rehabilitation, but current evidence remains fragmented, and the effectiveness of these technologies across different categories, and their implementation in the real world remains unclear.
Objective
We systematically evaluated the efficacy, implementation factors, and potential mechanisms of action of digital health technologies across various categories for scoliosis rehabilitation.
Methods
We searched PubMed, IEEE Xplore, Embase, and Web of Science from inception to November 7, 2025, with an updated search on May 12, 2026. We included English-language controlled trials, cohort studies, and feasibility studies of digital interventions for any type of scoliosis that reported at least one quantitative outcome; nonoriginal publications, purely surgical studies, and those without extractable data were excluded. Risk of bias was assessed using the Cochrane Risk of Bias 2 tool (RoB 2) and the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I), and intervention reporting completeness was assessed using the Template for Intervention Description and Replication (TIDieR). Interventions were categorized into 5 technology types and further stratified by evidence maturity into 3 tiers. Evidence was synthesized using vote counting based on the direction of effect, with prespecified subgroup and sensitivity analyses, and certainty of evidence assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE).
Results
Thirteen studies (574 patients) from 5 countries were included. Among the randomized trials, 14.3% (n=1) were at high risk of bias, and among the nonrandomized studies, 33.3% (n=2) were at serious risk of bias. Across the 4 outcome domains, clinical outcomes showed a generally favorable direction for Cobb angle and flexibility but inconsistent evidence for the angle of trunk rotation (ATR); functional outcomes were consistently favorable for respiratory function and postural control; patient-reported quality-of-life outcomes were predominantly favorable, whereas evidence for pain and body image was inconsistent; implementation outcomes showed the least favorable overall direction of effect, with adherence and dropout more closely tied to supervision intensity than technology category. The main findings remained robust in sensitivity analyses; no serious adverse events were reported, and the overall certainty of evidence was low.
Conclusions
Digital health interventions showed a generally favorable direction of effect for spinal alignment, function, and quality of life; however, evidence certainty was low, and outcomes should be interpreted with caution. To our knowledge, this is the first review to integrate diverse digital interventions into a single framework of 5 technology categories and 3 maturity tiers and to synthesize effectiveness across technologies, unlike prior single-technology or single-function reviews. Its main contribution is to shift the focus from technology effectiveness toward sustained use and to identify human supervision as a key factor influencing implementation. Clinically, technologies may be selected by maturity tier and paired with an appropriate supervision strategy to extend access to rehabilitation. Larger, long-term multicenter trials are needed for confirmation.