Classifying pain severity in middle-aged and older Europeans with back pain: a visual decision tree

Abstract Aim Back pain is one of the most prevalent and potentially disabling musculoskeletal conditions experienced by individuals, leading to reduced quality of life and long-term disability. The main objective of this study was to develop a simplified visual decision tree algorithm for the classification and prediction of severe or mild/moderate pain in the middle-aged and elderly European population with back pain, using as predictors sociodemographic variables and variables related to physical condition, pain symptom expansion, and mental health. Subject and methods This was a cross-sectional study of 15,314 middle-aged and older European individuals with back pain. A predictive algorithm for the level of pain was developed using classification and regression tree (CRT) analysis. A cross-validation study was conducted to assess the performance of the model. In addition, a multivariate logistic binary regression model was developed to predict the level of pain and compare its performance with the CRT model. Results The CRT found that physical inactivity, depression, and pain expansion were the combination of variables that predicted a high level of pain. The CRT model performed better than the logistic regression model in terms of accuracy (area under the curve [AUC] = 0.67 versus AUC = 0.66), and both identified physical inactivity, depression, and pain expansion as the main predictors of high pain levels. Conclusion Our findings provide useful information for clinicians and those involved in physical activity, exercise and sport programs for assessing the level of pain in people with back pain, offering an easy-to-interpret visual model with similar accuracy to traditional models.