Statistical methods in genetic epidemiology : An application for familial aggregation

Statistical Methods In Genetic Epidemiology: An Application For Familial Aggregation Many common diseases, including cancers, coronary heart disease, allergies and psychiatric disorders are known to be complex in etiology since genetics and environmental factors contribute to the disease process. First step in genetic epidemiology process, defining familial aggregation. Genetic epidemiology is a relatively new field which utilizes the conventional epidemiologic designs and methods to explore the role genetic factors play in determining disease. In this field statistical methods can diversifying. The aim of this study is to investigate if there is a familial aggregation in our data obtained from hematological cancers(Hem CA) and solid cancers(Solid CA) and non-cancer patient(Control) diseased study subjects by applying various statisticals models. With this study groups, four different statistical methods used for defining familial aggregation; binary logistic regression, regressive logistic regression, generalized estimation equation and exact inference. In modelling, SPSS 20.0 and R program packages will be used. All groups were compared by the demographics and family history features. Results of comparisons show that there is a significant difference between groups according to sex, age, education and marital status (p<0.001). For the familial aggregation, hematological cancer in siblings and family were different with the odds 4.3 and 2.6. Results of advanced analysis between hematolgical cancer and control group, in logistic regression, degree of kinship between pairs and environmental effects were significant. Scope of GEE methods, same results as logistic regression showed that person who has solid cancer in family will increase risk of hematological cancer (p=0.007). In regressive logistic regression, informations of all siblings calculated and the highest probability is in solid group and then hematological group and the control group has the lowest probability of cancer. Results of exact inference method, there is a familial aggregation in solid cancer group and no aggregation in solid cancer and control group. All statistical methods show that disease status of siblings has high risk, environmental effects like cigarette and the others has acceptable effect.