Step | Risk factor analysis | Final model | ||
Model variable# | p-value¶ | AIC+ | HR (95% CI)§ | |
Investigator-reported acute exacerbations | FVC % predicted at baselineƒ | <0.0001 | 830.217 | 0.67 (0.55–0.80) |
Supplemental oxygen use at baseline | 0.0018 | 824.565 | 2.47 (1.37–4.47) | |
Antacid medication use at baseline | 0.0873 | 823.706 | 1.50 (0.91–2.47) | |
Randomisation to nintedanib | 0.1150 | 823.271 | 0.66 (0.40–1.08) | |
Adjudicated confirmed or suspected acute exacerbations | FVC % predicted at baseline | 0.0006 | 483.723 | 0.67 (0.53–0.86) |
Randomisation to nintedanib | 0.0010 | 475.169 | 0.33 (0.16–0.66) | |
Antacid medication use at baseline | 0.0676 | 473.938 | 1.78 (0.92–3.43) | |
Former or current smoker | 0.0938 | 472.805 | 2.13 (0.89–5.13) | |
Supplemental oxygen use at baseline | 0.1322 | 472.870 | 1.85 (0.80–4.29) |
FVC: forced vital capacity. #: covariates that achieved p<0.2 in the stepwise selection procedure are shown. The variables are shown in the order that they were selected into the model. ¶: p-value of the likelihood chi-squared test used in the decision to add the variable to the model. +: Akaike information criterion (AIC) was calculated for the model at the end of each selection step. Lower AIC values indicate better performance of the model. §: based on the final model, i.e. including all the variables that were selected during the stepwise selection procedure. ƒ: hazard ratio (95% CI) is calculated for FVC % predicted 10% higher at baseline.