The objectives of this study are to develop a model for studying the relationship between urinary fluoride and skeletal fluorosis symptoms in elderly persons, and to analyze the factors affecting the skeletal fluorosis symptoms. In this study, we use the questionnaire for collecting the general information, health information, consumption of water and a check list for assessing the presence of skeletal fluorosis symptoms. The subjects are the elderly in Mae Hoi ngoen substrict in Doi Saket district and Suthep subdistrict in Muang district, Chiang Mai. There are 50 persons and 30 persons were selected from Mae Hoi Ngoen and Suthep, respectively.
The multinomial logistic regression is employed for studying the relationship between the urinary fluoride and the level of skeletal fluorosis which consists of four phases: normal bone, clinical phase I, clinical phase II, clinical phase III. The results of the study shows that only the urinary fluoride variable that had the effect on the skeletal fluorosis phases in the elderly at the significance level of 0.05. However, the model created by using multinomial logistic regression can describe the relationship between urinary fluoride and skeletal fluorosis phases of 24.8 percent (R2=0.248) and the classification accuracy is only 57.5%. Therefore, we groups the skeletal phase into binary variable where 1 indicates that the subject has skeletal fluorosis symptom and 0 is otherwise. And, we employs the binary logistic regression for studying the relationship between urinary fluoride and the binary variable. The model is log (odds) = - 0.222 +1.667 (urinary fluoride is higher than 1). This model give R2 of 12.3 percent and the classification accuracy of 71.3 percent.