Using Unconditional Logistic Regression to Assess Prostate Cancer Risk Factors: Practical Messages for Public Health

Abstract

Abstract: Objectives: Risk assessment, early diagnosis and timely treatment of prostate cancer have a great role in survival and quality of life of patients. Assuming a different cancer epidemiology in developing countries, this study aimed to identify risk factors for prostate cancer in southern Iran to be implicated in Asian developing countries' public health systems. Methods: A case-control study of prostate cancer (PCa) was undertaken in two Iranian provinces. Ninety eight Cases with histologically confirmed incident disease and 98 control patients treated in the same hospitals for other minor diseases (2005-2010) were included. The data about demographic, socio-economic, sexual and biomedical variables were modeled through unconditional Logistic regression (using the EGRET and SPSS) as well as odds ratios and 95% confidence intervals. Results: There was no significant difference between the groups in terms of anthropometric variables (including height and BMI), marital, birth order of children, sexual function, history of STD and testosterone related phenotype between the two groups. However urban residency, smoking, patient's age and family history of prostate cancer were determined as significant risk factors for PCa (P= 0.006, P=0.03, P=0.008 and P=0.006, respectively). Among the reported past disease history, only the history of TB disease was a significant determinant (OR=2.08, P=0.01). Conclusion: The results are evidence against a relationship between anthropometric measures, marital status, sexual behavior, alcohol consumption and past medical history (including history of STD) and in favor of a relationship of age, family history of prostate cancer, urban residency and smoking with prostate cancer. Keywords: Unconditional logistic regression, Prostate cancer, Risk factors, age, smoking, family history. With the exception of variables such as smoking, age and history of prostate cancer in close relatives, there was not any evidence to show that other variables could be a risk factor of prostate cancer. Heavy smokers appear to be at higher risk than non-smokers and this increasing trend is statistically significant (P=0.03). Also the risk of disease increase with age and this trend is statistically significant (P=0.008). The odds ratio of developing prostate cancer in those who had a history of this disease in their close relatives was more than 3 times greater than those who hadn't such a family history of cancer in their close relatives (P=0.006).