Community Based Endoscopy Screening Effect On Stage Distributions Of Onset Diagnosed Esophageal Cancer And Theoretical Research On Common Bias Control In Screening Effect Evaluation

ObjectivesBased on a multi-center cluster randomized cohort in esophageal cancer high risk areas of China,to evaluate the short-term efficacy of endoscopy screening by comparing the stage distributions of onset diagnosed esophageal cancer between intervention group and control group.Common bias and their control summary in screening effect evaluation to support the subsequent research.Materials and MethodsIn Linzhou and Cixian of esophageal cancer high risk areas,villages were randomly allocated to the intervention or control group to establish the population-based endoscopy screeing cohort.All 40-69 years old residents were included in the target population.The actually enrolled populations in both groups accepted the same baseline information questionnaire survey.Population enrolled in the intervention group accepted the endoscopy screening with iodine staining and biopy.Baseline information description and comparison to recognize the comparability of both groups.The esophageal cancer incidence data and pathologically diagnostic information from target population during 2012 to 2016 were collected,to evaluate the short-term efficacy of endoscopy screening through comparing TNM stage distribution and study-defined stage distributions.By reviewing previous references,to sum up and explore bias control methods.Results1.Baseline variables of the intervention group and control group were balanced except gender and education due to volunteer bias,so we think the two groups were overall comparable in enrolled population.Due to cluster randomization,the target population were also comparable between two groups.2.In the enrolled population,the stage distributions of control group and intervention group were significantly different.For TNM stage,the percentages from stage I to stage IV in intervention group were about 32%,41%,24%,3%versus 71%,19%,7%,3%in control group;For clinical stage,the percentages from early early,middle to advanced stage in intervention group were about 18%,49%,33%versus 59%,33%,8%in control group.While in the target population,the clinical stage distributions had significant statistical difference between two groups,the TNM stage distributions were merely borderline significant(P =0.093).3.Survival as the long-term evaluation endpoint in the subsequent study,the unavoidable lead time bias and length bias could be control by using statistical method or epidemiological method.The epidemiological method is more feasiable.ConclusionsMore early stage esophageal cancer cases and higher early stage proportion in the enrolled intervention group than the control group,the short-term efficacy of endoscopy screening for esophageal cancer was comfirmed,and the efficacy in target population depends on the intervention compliance.Based on the cluster randomization study design,epidemiological method is more recommendable to control lead time bias and length bias in the subsequent study.