Ms. Xinrui Zhang, a doctoral student in the department of biostatistics at the University of Florida College of Public Health and Health Professions and the College of Medicine, received the Jiann-Ping Hsu Pharmaceutical and Regulatory Sciences Student Paper Award at the 2016 Applied Statistics Symposium of the International Chinese Statistical Association held last month in Atlanta.
[Photo: Ms. Xinrui Zhang]
Ms. Zhang’s paper, entitled “Internal pilot design for repeated measures,” was motivated by the need to design a longitudinal study to investigate the effect of antiretroviral therapy for youth living with HIV. The cross-sectional study to which Ms. Zhang provided statistical support has identified potential contributing biomarkers whose profiles were associated with HIV treatments and immunological health, such as viral load. Planning a clinical trial with biomarkers measured over time requires power calculation for repeated measures and a proper choice of the error covariance matrix. Accurate determination of the nuisance parameters in the error covariance matrix is important to ensure adequate power analysis and sample size selection. The internal pilot design accounts for the uncertainty in specifying the covariance matrix by using a fraction of the data to re-estimate the nuisance parameters and adjust the final sample size.
In the paper, Ms. Zhang sought to extend the internal pilot design to studies with repeatedly measured outcomes, and derived the distribution of the Univariate Approach to Repeated Measures test statistic for final analysis. The methods she developed enable the power and sample size calculation for clinical trials or epidemiological studies with repeated measures of continuous outcome.