Background/Objectives: Despite the effectiveness of exercise and nutritional interventions to improve aerobic capacity and quality of life in lung transplant (LT) recipients, their compliance is low. Strategies to reduce the high attrition rate (participants lost over time) is a major challenge. Artificial neural networks (ANN) may assist in the early identification of patients with high risk of attrition. The main objective of this study is to evaluate the usefulness of ANNs to identify prognostic factors for high attrition rate of a 10-week rehabilitation program after a LT. Methods: This prospective observational study included first-time LT recipients over 18 years of age. The main outcome for each patient was the attrition rate, which was estimated by the amount of missing data accumulated during the study. Clinical variables including malnutrition, sarcopenia, and their individual components were assessed at baseline. An ANN and regression analysis were used to identify the factors determining a high attrition rate. Results: Of the 41 participants, 17 (41.4%) had a high rate of attrition in the rehabilitation program. Only 23 baseline variables had no missing data and were included in the analysis, from which a low age-dependent body mass index (BMI) was the most important conditioning factor for a high attrition rate (p = 7.08 × 10−5), followed by end-stage respiratory disease requiring PT (p = 0.000111), low health-related quality-of-life (HRQoL) (p = 0.0009078), and low handgrip strength (p = 0.023). Conclusions: The prevalence of high attrition rate in LT recipients is high. The profile of patients with a high probability of attrition includes those with chronic obstructive pulmonary disease, low BMI and handgrip strength, and reduced HRQoL.
Read full abstract