A nomogram for suicide-specific mortality risk stratification in patients with tongue cancer: a Surveillance, Epidemiology, and End Results (SEER)-based study, 2004–2017

Abstract

Background Although suicide represents an important cause of mortality among cancer patients, prediction models specifically developed for tongue cancer patients remain unavailable. This study aimed to identify suicide-related risk factors and develop an individualized risk stratification model for tongue cancer patients. This study aimed to identify factors associated with suicide-specific mortality and develop an internally validated risk stratification model for patients with tongue cancer. Methods Patients diagnosed with tongue cancer between 2004 and 2017 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. Eligible patients were randomly allocated to training and validation cohorts at a ratio of 7:3. Cox proportional hazards regression was used to identify potential predictors and construct a nomogram. Model discrimination was evaluated using the concordance index (C-index) and time-dependent receiver operating characteristic (ROC) curves with corresponding area under the curve (AUC) values. Calibration was assessed by comparing predicted and observed suicide-specific mortality. Results Among the 25,024 eligible tongue cancer patients, 96 suicide-related deaths were identified during follow-up, corresponding to a cumulative suicide mortality rate of 0.38%. Multivariate Cox regression analysis revealed that sex and age (P<0.001) were independent predictors, and race, tumor (T) stage, and income (P≤0.2) were potential significant factors for suicide in tongue cancer patients. The nomogram had a good performance for the prediction of suicide-specific mortality both in the training {C-index =0.762 [95% confidence interval (CI): 0.710–0.815]} and validation sets [C-index =0.761 (95% CI: 0.673–0.849)]. ROC curve demonstrated acceptable performance in both the training and validation sets, with 3-year AUCs of 0.784 and 0.774, 5-year AUCs of 0.762 and 0.788, and 10-year AUCs of 0.729 and 0.783, respectively. Moreover, the calibration plot indicated moderate accuracy of the nomogram in both datasets. Conclusions Overall, this study successfully identified risk factors for suicide among tongue cancer patients and developed a nomogram to offer risk assessment to mitigate the risk of suicide in this patient population.