Abstract: Objective To investigate the 3-year survival and prognostic factors of 550 patients with colon cancer, and to construct a regression model. 
Methods The clinical data of 550 patients with colon cancer who was discharged from our hospital after surgical treatment were retrospectively analyzed.  They were divided into the poor prognosis group (death, n=134) and the good prognosis group (survival, n=416) according to whether they survived at 3 years after follow-up. The 3-year survival was recorded, and the clinical data of the two groups were compared. Multivariate Cox regression analysis was used to analyze the risk factors affecting the prognosis of colon cancer patients, and a Cox regression model was constructed. 
Results At 3 years after follow-up, 416 of the 550 patients with colon cancer survived, and the 3-year survival rate was 75.64%. The proportion of patients with age >60 years, low differentiation, adenosquamous cell carcinoma, undifferentiated carcinoma, Dukes stage C and D, right colon cancer, family history, presence of complications, lymph node metastasis, the abnormal level of serum carcinoembryonic antigen, intermediate surgical approach, number of lymph node dissections <12, and intraoperative bleeding ≥ 200 mL in the good prognosis group were 79.85%, 48.25%, 14.18%, 10.45%, 33.58%, 65.67%, 85.82%, 22.39%, 95.52%, 59.96%, 71.64%, 67.16%, 58.96%, and 89.55%, respectively, which were higher than those (52.40%, 16.59%, 1.20%, 1.44%, 2.64%, 0.00%, 41.83%, 9.13%, 84.62%, 28.37%, 31.25%, 33.17%, 41.11%, and 9.86%) in the poor prognosis group; the proportion of married patients was 24.63%, which was lower than that (75.48%) in the poor prognosis group (P＜0.05). Multivariate Cox regression analysis showed that age >60 years, Dukes stage C and D, right colon cancer, lymph node metastasis and intraoperative bleeding ≥ 200 mL were independent risk factors for poor prognosis of patients with colon cancer (P＜0.05). The results of the prediction model constructed showed that the internal validation consistency index (C-inex) was 0.852 (95%CI: 0.819-0.885), and the correction curve showed that the predicted value was in good agreement with the observed value. 
Conclusion Age >60 years, Dukes stage C and D, right colon cancer, lymph node metastasis, number of lymph node dissections <12, and intraoperative bleeding ≥ 200 mL are risk factors for poor prognosis in patients with colon cancer. The Cox regression model is effective and has a good fitting effect. The patients with the above conditions can be given corresponding treatment and intervention measures to improve the prognosis.

Key words: colonic neoplasms, prognosis, model construction