Year 2021
October 2021


Abstract Title
Developing the SAD model for prognostication of mortality in End Stage COPD



Tan Tock Seng Hospital1

Background & Hypothesis

To develop a weighted index to prognosticate 12-month mortality in Chronic Obstructive Pulmonary Disease (COPD) patients.


GOLD stage 2-4 COPD clinic patients were recruited. Predictive variables for 12-month mortality were evaluated using receiver operator curve analysis. Variables with the highest area under curve (AUC) were selected to construct the prognostic model.  Weights were assigned to component variables based on likelihood of predicting 12-month mortality.


136 subjects (mean age 71.00 ± 8.69 years) with mean Charlson Comorbidity Index (CCI) 4.82 ± 1.97 completed 12-months follow-up. 70.5% had GOLD stage 3 and 4 COPD.

3 component variables were identified – (i) 4-metre Gait Speed (AUC 0.82; CI 0.66 – 0.97; p=0.001); (ii) Hospital Anxiety and Depression Scale – Anxiety score (AUC 0.78; CI 0.68-0.89; p=0.05); and (iii) modified Medical Research Council Dyspnea Scale (AUC 0.79; CI 0.67 – 0.92; p=0.005), to construct the SAD (Speed, Anxiety, Dyspnea) score. SAD had significantly higher AUC (0.87; CI 0.79 – 0.94; p<0.001) than its component variables and conventional prognostic indices.

Kaplan Meier survival analysis using stratified levels of SAD showed a significant difference in survival time (Log-Rank test p<0.001). Cox Proportional Hazard Regression and linear regression analysis respectively showed that higher SAD score was associated with higher mortality (OR 3.22; CI 1.62 – 6.39; p=0.001) and poorer quality-of-life at 12-months across all Chronic Respiratory Questionnaire domains.

Discussion & Conclusion

The SAD score demonstrates good prognostic potential to identify patients with higher mortality and poorer quality-of-life who can benefit from early palliative care.