Multinomial logistic regression for influence factors of esophageal lesions

  Objective   To investigate the associations of family history of esophageal cancer, smoking and drinking to esophagitis, esophageal hyperplasia and early cancer.

  Methods   In a high incidence community of esophageal cancer, a cohort of residents aged from 40 to 69 were examined for the esophageal lesions.The people with normal esophageal mucosa determined by gastric endoscopy served as controls, and the 3 disease groups respectively consisted of patients with esophagitis, esophageal hyperplasia and early cancer diagnosed by staining endoscopy and pathogenic examination.O R was used as the association indicator to analyze the multinomial regression result.

  Results   Seventy-one cases of esophageal early cancer, 266 of esophageal hyperplasia, 144 of esophagitis, and 2818 of normal esophageal mucosa were found in the mass screening.Family history of esophageal cancer was significantly associated with all the three kinds of esophageal lesions.The OR for esophageal cancer history of first-degree relatives was 2.6(95%CI=1.36-4.85), 2.1(95%CI=1.43-2.99)and 1.8(95%CI=1.13-3.03)in early esophageal cancer, hyperplasia, and esophagitis groups respectively, but no significant OR level was found in all the three groups for esophageal cancer history of second-degree relatives.Smoking or/and drinking increased the risk of having esophageal cancer to those whose relatives especially first-degree relatives had esophageal cancer histories.

  Conclusion   In high-incidence areas of esophageal cancer, abstinence from smoking and drinking alcohol may prevent esophageal lesions for persons with family history of esophageal cancer.