Survival analysis in advanced lung cancer using Weibull survival regression model: estimation, interpretation, and clinical application

In this study, both Weibull and log-Weibull (Gumbel) survival regression models were applied to investigate the impact of key demographic and clinical factors on survival outcomes using a dataset of 228 patients from the North Central Cancer Treatment Group (NCCTG). The response variable was survival time (in days), while the covariates included age group, gender, ECOG performance score, Karnofsky performance score, calorie intake, and weight loss. The Weibull regression model demonstrated superior goodness-of-fit based on AIC (), BIC (), and log-likelihood values compared to the log-Weibull model. It revealed that ECOG performance status was among the most significant predictors of survival. Patients who were bedridden at least of the day had a hazard ratio of (95% CI: 1.426-89.448, p = 0.0217), indicating a markedly increased risk of mortality. Similarly, being symptomatic but ambulatory (HR = 1.62, p = 0.0168) and spending less than of the day in bed (HR = 3.342, p < 0.001) were strongly associated with poorer survival outcomes. Female gender was significantly linked to better survival (HR = 0.558, 95% CI: 0.398–0.783, p = 0.0007), suggesting a protective effect in this cohort. While the Karnofsky score did not show strong statistical significance in multivariate modelling (p = 0.0302), Kaplan-Meier curves indicated a trend toward worse survival for patients with poor functional status. Age group did not emerge as a statistically significant predictor after adjusting for other variables. Weight loss over six months exhibited borderline significance (HR = 0.988, p = 0.0666), highlighting its potential influence on prognosis.