Master of Science
Epidemiology and Biostatistics
Sample size estimation is usually the first step in planning a research study. Too small a study cannot adequately address the objectives, while too large a study may waste resources or unethical. For binary outcomes, several sample size estimation methods are available based on logistic regression models, which focusing on odds ratios. In prospective studies, risk ratios are preferable for ease of interpretation and communication. In this thesis, we compared the power difference between the logistic regression model and the modified Poisson regression model via simulation studies. We then proposed sample size estimation formulas based on the modified Poisson regression model for estimating risk ratios. Simulation results suggested that both models have similar performance in terms of Type I error and power. The empirical evaluation indicated that the proposed sample size formulas are reliable in a wide range of scenarios. The sample size estimation procedure was illustrated using a subset of data from the Diabetes Control and Complications Trial.
Summary for Lay Audience
Medical and epidemiological research rests largely on assessment of risks. One key measure in such studies is the ratio of odds, which is commonly estimated using logistic regression models. However, ratio of odds has been commonly interpreted as ratio of risks. Since by definition odds is larger than risk numerically, this practice can exaggerate study results, especially when risk of event is not rare as in many prospectively studies. The modified Poisson regression model was proposed as a method to estimate risk ratio directly. The model has become increasingly applied in medical and epidemiological research. To facilitate its use, this thesis compares power of the modified Passion regression to that of the logistic regression using simulation studies. The results suggest equivalent power between the two models. The thesis further proposed and evaluated sample size formulas based on the modified Poisson model. Simulation results suggest the formulas performed well, providing an important tool for study planning.
Xue, Zhenni, "Sample Size Formulas for Estimating Risk Ratios with the Modified Poisson Model for Binary Outcomes" (2021). Electronic Thesis and Dissertation Repository. 7680.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.