| Objectives This study aims to explore the risk factors of digestive cancers among residents in rural Anhui Province, China and to build a set of digestive cancer risk prediction models so as to facilite doctors distinguishing high risk groups of digestive cancer from the general population.Methods We searched CNKI, VIP, Chinese biomedical literature database(CBM), Medline, Elsevier, Science Diect, etc. to review and analysize all studies investigating the risk factors of digestive cancers(i.e., esophageal cancer, gastric cancer and colorectal cancer) in Chinese population so as to design the questionnaire. We adopted case-control design and used face to face webpage-questionnaire survey method to perform the study. All cases were inpatients from 5 hospitals in Hefei who were rural residents from Luan prefecture and had diagnosed with digestive cancer(primary esophageal cancer, gastric cancer and colorectal cancer patients) during July 2014 to February 2014. Controls were recruited from rural residents who were living in Jingan, Yuan and Shouxian of Luan and had not been diagnosed with any cancer before. The cases and controls were matched by gender and age(±2); and the ratio between the cases of esophageal and gastric cancers and the contol was 1:2, and the ratio between the cases of colorectal cancer and the controls was 1:3. We collected 4 aspects of data including socio-demographics, disease history and symptoms, lifestyles(diet, physical activities, smoking, drinking etc.) and psychological and emotional variables. We used SPSS 16.0 to perform statistical analyses including descriptive analysis of gender, age and educational levels, and multivariate conditional logistic regression analysis via Cox regression method, and constructed esophageal cancer, gastric cancer and colorectal cancer risk prediction models. Receiver-operating characteristic curves(ROC curves) were plotted to evaluate the predict value of models.Results For esophageal cancer, a total of 164 patients(133 males and 31 females) and 328 controls(266 males and 62 females) were included in the study. The average ages of esophageal cancer cases and controls were 61.00±6.49 and 61.01±6.45 respectively. No statistical significant differences were detected between the two groups in terms of gender, age and education level. Multiple variate analysis showed that first-degree relatives with cancer(OR: 1.91-2.15), swallowing pain(OR: 20.59-21.18), eating too full(OR: 2.25-3.33), irregular timing of diet(OR=1.96), drinking(OR: 1.73-2.10) and irritability(OR: 2.24-2.73) were risk factors of esophageal cancer. The AUC of esophageal cancer risk prediction model 1(without swallowing pain and negative life events), esophageal cancer risk prediction model 2(with swallowing pain, but negative life events) and esophageal cancer risk prediction model 3(with both swallowing pain and negative life events) were 0.732(95%CI: 0.683-0.780), 0.833(95%CI: 0.792-0.874) and 0.792(95%CI: 0.800-0.880) respectively.For gastric cancer, a total of 141 patients(107 males and 34 females) and 282 controls(214 males and 68 females) were included in the study. The average ages of gastric cancer cases and controls were 56.75±8.69 and 56.76±8.66 respectively. No statistical significant differences were detected between the two groups in terms of gender, age and education level. Multiple factors analysis showed that stomach symptoms(OR:3.05-3.55), helicobacter pylori infection(OR:61.49-67.07), spicy food(OR:2.35-2.61), eating too fast(OR:2.43-3.51), eating too full(OR:2.62-3.81), irregular diet(OR=2.43), introverted personality(OR:3.37-3.74) and irritability(OR:3.19-4.40) were risk factors of gastric cancer. The AUC of gastric cancer risk prediction model 1(without helicobacter pylori infection and negative life events), gastric cancer risk prediction model 2(with helicobacter pylori infection but negative life events) and gastric cancer risk prediction model 3(with both helicobacter pylori infection and negative life events) were 0.783(95%CI: 0.729-0.830), 0.793(95%CI: 0.741-0.844) and 0.782(95%CI: 0.731-0.834) respectively.For colorectal cancrer, a total of 124 patients(64 males and 60 females) and 372 controls(192 males and 180 females) were included in the study. The average ages of the colorectal cancer cases and controls were 54.26±9.66 and 54.26±9.63 respectively. No statistical significant differences were detected between the two groups in terms of gender, age and education level. Multiple variate analysis showed that first-degree relatives with cancer(OR=2.02), rectal bleeding(OR=4.16), constipation(OR: 2.76-3.31), heavy-fat diet(OR: 2.12-2.34), eating too fast(OR=1.97), eating too full(OR: 4.78-5.44), and short sleep duration(OR=2.41) were risk factors of colorectal cancer; while garlic(OR: 0.44) was protective factor of gastrc cancer. The AUC of colorectal cancer risk prediction model 1(without rectal bleeding and negative life events), colorectal cancer risk prediction model 2(with rectal bleeding but negative life events) and colorectal cancer risk prediction model 3(with both rectal bleeding and negative life events) were 0.784(95%CI: 0.731-0.837), 0.826(95%CI: 0.779-0.873) and 0.826(95%CI: 0.779-0.873) respectively.Conclusions The study found that first-degree relative(s) with cancer, swallowing pain, eating too full, irregular timing of diet, irritability and drinking were risk factors of esophageal cancer. Stomach symptoms, helicobacter pylori infection, spicy food, eating too fast, eating too full, irregular timing diet, irritability were risk factors of gastric cancer. First-degree relative(s) with cancer, constipation, rectal bleeding, heavy-fat diet, eating too fast and short sleep duration were risk factors of colorectal cancer, while garlic was a protect factor of colorectal cancer. Future studies should pay more attention to other factors such so as to perfect the risk factor "system" of the digestive cancers and construct more accurate digestive cancer risk prediction models. |
PDF Full Text Request