Preference of musculoskeletal pain treatment in middle-aged and elderly chinese people: a machine learning analysis of the China health and retirement longitudinal study

Abstract

Abstract Background Musculoskeletal pain is a major cause of physical disability, associated with huge socioeconomic burden. Patient preference for treatment is an important factor contributing to the choice of treatment strategies. However, effective measurements for evaluating the ongoing management of musculoskeletal pain are lacking. To help improve clinical decision making, it’s important to estimate the current state of musculoskeletal pain management and analyze the contribution of patient treatment preference. Methods A nationally representative sample for the Chinese population was derived from the China Health and Retirement Longitudinal Study (CHARLS). Information on the patients’ demographic characteristics, socioeconomic status, other health-related behavior, as well as history on musculoskeletal pain and treatment data were obtained. The data was used to estimate the status of musculoskeletal pain treatment in China in the year 2018. Univariate analysis and multivariate analysis were used to find the effect factors of treatment preference. XGBoost model and Shapley Additive exPlanations (SHAP) method were performed to analyze the contribution of each variable to different treatment preferences. Results Among 18,814 respondents, 10,346 respondents suffered from musculoskeletal pain. Approximately 50% of musculoskeletal pain patients preferred modern medicine, while about 20% chose traditional Chinese medicine and another 15% chose acupuncture or massage therapy. Differing preferences for musculoskeletal pain treatment was related to the respondents’ gender, age, place of residence, education level, insurance status, and health-related behavior such as smoking and drinking. Compared with upper or lower limb pain, neck pain and lower back pain were more likely to make respondents choose massage therapy (P < 0.05). A greater number of pain sites was associated with an increasing preference for respondents to seek medical care for musculoskeletal pain (P < 0.05), while different pain sites did not affect treatment preference. Conclusion Factors including gender, age, socioeconomic status, and health-related behavior may have potential effects on people’ s choice of treatment for musculoskeletal pain. The information derived from this study may be useful for helping to inform clinical decisions for orthopedic surgeons when devising treatment strategies for musculoskeletal pain.

Keywords

• Musculoskeletal pain
• Treatment preference
• CHARLS
• Machine learning