Methodological Considerations in Comparing Digital and in‐Person Physical Therapy for Musculoskeletal Conditions

We read with great interest the study by Domingues et al. [1] examining surgical and low-value surgical utilization among adults with musculoskeletal conditions receiving a digital care program compared with in-person physical therapy. The authors should be commended for leveraging a large real-world dataset and for adopting a new-user, active-comparator design to address an important and timely question in digital musculoskeletal care. We would like to highlight several methodological considerations that may further inform interpretation of the findings. First, although index dates were defined as treatment initiation in both groups, enrollment in a digital physical therapy program and initiation of in-person physical therapy may represent different points along the care trajectory. Prior methodological work has emphasized the importance of aligning treatment entry points to ensure comparable baseline risk in observational studies, particularly when emulating causal effects in nonrandomized settings [2]. Differences in when patients engage with care could therefore contribute to non-equivalent baseline surgical risk between groups. Second, while the authors employed exact and propensity score matching, socioeconomic and caregiving factors relevant to uptake of digital interventions may not be fully captured. Although a regional Social Deprivation Index was included, individual-level health literacy, household support, and access to digital resources—factors known to influence both treatment selection and health outcomes—are not readily observable in claims data. As prior methodological literature has noted, propensity score methods can only adjust for measured confounders, and residual confounding remains a concern in voluntary, self-selected interventions [3]. Third, the study did not include direct measures of musculoskeletal disease severity at baseline. Clinical determinants such as pain intensity, functional limitation, imaging severity, and prior surgical recommendation are central to surgical decision-making in musculoskeletal care, as reflected in randomized comparisons of surgery and structured physical therapy [3]. In the absence of such measures, it remains challenging to distinguish whether observed differences in surgical utilization reflect treatment effects or underlying differences in baseline disease severity. Taken together, these considerations do not detract from the contribution of this important study but suggest that the observed reductions in surgical utilization should be interpreted with appropriate caution. Future studies incorporating more granular clinical severity measures and more closely aligned treatment entry points may help clarify the mechanisms underlying these findings. R.C. concept and edit. S.-Y.C. and W.-B.Y. writing. The authors have nothing to report. The authors declare no conflicts of interest. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.