Development and Validation of a Risk Prediction Model to Identify Women With Chronic Obstructive Pulmonary Disease for Proactive Palliative Care.

Abstract

Background and objective: Proactive palliative interventions can improve symptom control and quality of life in individuals with chronic obstructive pulmonary disease (COPD); however, they are often underutilised. This study aimed to develop and validate a prediction model to identify women with COPD in their last year of life to facilitate timely palliative care referrals and interventions.

Methods: Data from 1236 women diagnosed with COPD from the 1921-1926 Australian Longitudinal Study on Women's Health cohort, linked to administrative health records, were analysed. We employed Lasso regression and multivariable logistic regression to select predictors. To assess the predictive performance of the model, we used the area under the receiver operating characteristic (AUROC) curve, calibration plot, and calibration metrics. The Youden index was used to establish the optimal cutoff point for risk classification. The clinical utility of the model was evaluated using decision curve analysis (DCA).

Results: The final model to predict 1-year all-cause mortality included six predictors: smoking status, body mass index, needing regular assistance with daily activities, number of supplied medications, duration of illness, and number of hospital admissions. The model performed well, with AUROC of 0.82 (95% CI: 0.80-0.85) and showed excellent calibration. Using a cutoff of 56.6% predicted risk, the model achieved a sensitivity of 72.3%, specificity of 77.7%, and accuracy of 75.0%. The DCA indicated that the model provided a greater net benefit for clinical decision-making.

Conclusion: Our prediction model for identifying women with COPD who may benefit from palliative care has shown robust predictive performance and can be easily applied, but requires external validation.