Relative Efficacy of Virtual and In-Office Conservative Care for Musculoskeletal Conditions: A Propensity Score Matching Comparison Study (Preprint)

BACKGROUND Propensity Score Matching is a statistical technique which allows an observational study to control for differences in the study and comparison group, without the difficulty of establishing a long-term randomized trial. This method is very appropriate to the evolving field of telemedicine, due to the fast pace at which it is proceeding and the limitations on data imposed by its relatively recent popularization. These limitations prevent a randomized controlled trial from being applied to most questions, creating a strong case for the use of Propensity Score Matching. OBJECTIVE To evaluate the relative efficacy of digital and in-person physical therapy in a single clinic while controlling for the different patient circumstances that may affect the choice of treatment option. METHODS We used propensity score matching to construct a cohort of 890 patients split evenly between digital and onsite patients. We matched patients on injury location, instance of injury (first, second, third, etc.), reported pain level on presentation, age, presence or absence of prior treatment attempts, gender, and BMI (for weight-bearing injury locations only). We evaluated the matched cohorts in upper and lower body groups for differences in patient reported pain improvement, visit completion, avoidance of surgeries, and patient reported injury resolution. RESULTS We analyzed that data using a two-one-sided T-test. We found that upper body digital patients perform equivalent to within 0.5 standard deviations or superior to in-person patients on all metrics (p < 0.05 for all metrics). Lower body digital and in-person patients performed within 0.5 standard deviations on visit completion, avoidance of surgeries, and reported injury resolution. (p <0.05.) Results for pain improvement were inconclusive, with the 95% confidence range indicating a possibility of either equivalent performance (p = 0.48) or inferior performance for digital patients (p = 0.52). CONCLUSIONS The date indicates an equivalent or superior performance for remote treatment of upper body injuries, with data suggesting improved visit completion improves other metrics more than any disadvantages of digital treatment can reduce them. Data regarding lower body injuries is inconclusive, due to high varied level of pain reduction experienced by patients. In light of other data on lower body patients, this suggests that further research may be able to assess which lower body injuries can and cannot be successfully treated via remote physical therapy.