FACTORS AFFECTING OSTEOPOROSIS USING BINARY LOGISTIC REGRESSION MODEL
This is a descriptive cross-sectional data based study. The data included Asian, Caucasian and African American races. The sample has been designed to accommodate a total number of 1958 individuals. The sample of this survey includes adults whose ages range between 18 to 90 years of both genders where males are (50.7%) and females are (49.3%). There is a statistically significant effect of the explanatory variables (age and corticosteroids medications) on the dependent variable (osteoporosis), and a significant value of chi-square with p-value (0.000) at 5% level of significance. The overall percentage of cases that are predicted correctly by the binary logistic regression model is 82.6%. This percentage has increased from 50.0% for the null model, with 49.5% of Cox and Snell R-square and 66% of Nagelkerke R-square. Depending on the observed groups and predicted probabilities, the model has shown its effectiveness in predicting a large number of cases correctly. The results clarify that different basic groups based on demographics and lifestyle factors are not statistically significant and have no effect on osteoporosis as well as medical conditions (rheumatoid arthritis, hyperthyroidism) and prior fractures are not statistically significant.
factors, effect, osteoporosis, binary, logistic, regression.
Received: June 16, 2024; Revised: August 9, 2024; Accepted: August 13, 2024
How to cite this article: Maysoon A. Sultan, Factors affecting osteoporosis using binary logistic regression model, JP Journal of Biostatistics 24(3) (2024), 439-448. https://doi.org/10.17654/0973514324024
This Open Access Article is Licensed under Creative Commons Attribution 4.0 International License