Home-based digital exercise versus hospital physiotherapy for plantar fasciitis: a propensity-score-matched retrospective cohort study with dual-perspective cost-utility evaluation.

Patients chose pathway based on preference, geography and scheduling, not randomisation, so groups may differ in ways that affect pain recovery. Propensity matching did not meet balance target: only BMI had |SMD| below 0.10 after matching; digital patients remained younger, more urban, better educated, more likely to book by app and have e-prescription. Only 587 of 9,373 screened patients with M72.2 diagnosis (6.3%) had complete records, so selection bias is likely and results apply to patients with complete tertiary-hospital records. The matched analysis estimates effect among patients resembling digital adopters, not all hospital patients. Unmeasured confounders such as digital literacy, smartphone ownership, prior app use, self-efficacy, health literacy and motivation were not adjusted. Adherence was measured differently: app logged sessions averaged 71.1% versus clinic attendance 77.4%. Only 25.8% digital and 47.3% hospital completed at least 80% of sessions. The per-protocol analysis was underpowered. Loss to follow-up by 12 months was 10.9% digital and 27.0% hospital, which could bias results if dropout was related to outcome. Single-centre urban eastern China tertiary hospital; uncertain for primary care, rural settings, older populations and low-digital-literacy groups. The non-inferiority margin framework assumes randomisation; in observational data, residual confounding and selection bias can still affect the conclusion. The 12-month horizon does not establish long-term recurrence or cost offsets. An unmeasured confounder would need a strong association, with E-value 2.84 and lower confidence limit 2.03, to explain the difference, but joint confounding could be larger.

No conflicts of interest were declared. Funding was from the Ningbo Clinical Research Center for Orthopedics. The authors state no financial support was received from Joymotion or other commercial entity.

Do not conclude that the app alone caused lower pain and costs. The digital pathway included app videos, weekly therapist check-ins, adherence logs and symptom-triggered escalation rules; the hospital pathway included face-to-face sessions, manual therapy and clinic modalities, so the contribution of any single component cannot be separated.