Objective To explore the risk factors and survival analysis of chemotherapy compliance in ≥65 years old patients with non-small cell lung cancer.

Methods From December 2018 to August 2020， 110 patients with non-small cell lung cancer ≥65 years old in our hospital were selected for study. According to their chemotherapy compliance， the patients were divided into untreated （25 cases）， partial chemotherapy （30 cases） and full chemotherapy （55 cases）. Multivariate Cox regression was used to analyze the independent risk factors affecting the compliance of chemotherapy among the three groups， and constructed a column chart prediction model and evaluated the diagnostic effectiveness of the model using receiver operating characteristic （ROC） curves， calibration curves， and clinical decision curves. The patients were followed up until August 2023. Kaplan-Meyer survival curve and Log-rank test were used to compare the differences in survival time between the groups.

Results There were significant differences among the three groups in gender， long-term smoking history， education level， Charlson complication index， toxic and side effects of chemotherapy drugs and dosage， pathological type， operation mode， place of residence， payment mode， chemotherapy stage， white blood cell count and neutrophils （P<0.05）. Multivariate Cox analysis showed that patients' education level， pathological TNM stage， toxic and side effects of chemotherapy drugs and dosage， operation mode， residence， payment mode and chemotherapy stage were independent risk factors affecting chemotherapy compliance of patients with non-small cell lung cancer （all P<0.05）. There were significant differences in the survival rates between the non-chemotherapy group and the full chemotherapy group， and between the partial chemotherapy group and the full chemotherapy group （P<0.05）， but there was no significant difference between the non-chemotherapy group and the partial chemotherapy group （P>0.05）. And based on this， a prediction model was constructed， and the ROC curve was drawn to show that the AUC of the model was 0.758 （95%CI： 0.743‒0.855）， the sensitivity was 0.788， and the specificity was 0.853， indicating that the model had high discrimination， the calibration curve indicated that the prediction model had good accuracy， the clinical decision curve indicated that the predictive model had strong clinical practicality.

Conclusion NSCLC tends to occur in the elderly. We should focus on the patients with low educational level， late pathological TNM stage， severe toxicity and side effects， high dose of chemotherapy drugs， having undergone open chest surgery， living in rural areas， without medical insurance， and in the period of chemotherapy consolidation and refractory relapse， so as to improve their compliance with chemotherapy.