Application of Parametric Models of Survival Analysis in Determining the Cancer Influencing Factors in Patients with Thyroid Nodules

One of the most common clinical problems among individuals is thyroid nodule diseases which are characterized by one or more nodules in the thyroid and are usually benign. It can be said that thyroid cancer is the most common endocrine cancer worldwide. This study aimed to determine the risk factors for cancer in patients with thyroid nodule in Mazandaran province،Iran، using parametric survival analysis. In the present historical cohort study، 26،730 patients with thyroid nodules who were referred to health care centers from July 2002 to March 2008 were identified. Parametric log-normal and log-logistic models were compared with and without taking frailty into account. The criterion for comparing models was Akaike''s criterion. All calculations were performed with the SPSS software and the significance level was considered 0. 05. The mean time of the conversion of thyroid nodules to cancer in patients was found to be 29. 32 months. Using Kaplan-Meier method، survival rates of one year، five years and ten years of nodule conversion to cancer was calculated 94. 6، 88. 6 and respectively. According to the log rank test age (p=0. 03)، hypothyroidism (p=0. 01)، bilateral nodules (p <0. 001)، a multi-nodular goiter (p <0. 001)، TSH hormone (p <0. 001)، T4 hormones (p = 0. 005)، cholesterol (p = 0. 03)، creatinin levels (p = 0. 001) a significant relationship was seen. Based on the Akaike''s criterion، the lognormal model which takes frailty into account best fits to the data. Based on the log-normal model with frailty، It can be concluded that the thyroid nodule patients with abnormal TSH hormone are 6. 55 times more likely to develop risk of thyroid cancer than patients who had normal TSH hormone overall. This model also indicated that patients who had heart palpitations are 5. 52 times more likely to develop risk of cancer than patients who did not have heart palpitations.