The purpose of the proposed research is to develop methods to better understand how variability of health measures may be predictive of future health outcomes of interest. Many statistical methods have been developed that treat within-subject correlation that accompanies the clustering of subjects in longitudinal data settings as a nuisance parameter, with the focus of analytic interest being on mean outcome or profiles over time. However, there is evidence that, at least in certain settings, the underlying variability in subject measures may also be important in predicting future health outcomes of interest. Hence we plan to develop methods that will better structure variability, decomposing it into short-term and long-term variance measures, and combining variance structures with mean structures such as mean longitudinal profile to more fully describe the information available in longitudinal datasets. In particular, we propose methods to jointly model mean profile and variance in continuous longitudinal data, including methods that treat variance as heteroscedastic within individuals as well as between individuals. We also propose methods to jointly model short-term and long- term variance in continuous longitudinal data. We will apply these methods to the analysis of within-woman trends and variability in reproductive hormones and time between menstrual cycles to predict the progression of health outcomes through the transition to menopause, and to the analysis of within-person trends and variability in cognitive testing to predict cognitive decline and progression of dementia in older adults.

Public Health Relevance

Public Health Relevance As more clinical and public health studies follow individuals through time, it becomes possible to consider whether variability in measures over time is an important predictor of disease. Most methods for studying such data focus on differences in averages and average trends across individuals. Our proposed study will consider whether adding information about variability over time will assist in predicting and ultimately better understanding the cause of various diseases, particularly chronic conditions associated with aging.

National Institute of Health (NIH)
National Institute on Aging (NIA)
Small Research Grants (R03)
Project #
Application #
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Hadley, Evan
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Michigan Ann Arbor
Biostatistics & Other Math Sci
Organized Research Units
Ann Arbor
United States
Zip Code
Jiang, Bei; Wang, Naisyin; Sammel, Mary D et al. (2015) Modeling Short- and Long-Term Characteristics of Follicle Stimulating Hormone as Predictors of Severe Hot Flashes in Penn Ovarian Aging Study. J R Stat Soc Ser C Appl Stat 64:731-753
Jiang, Bei; Elliott, Michael R; Sammel, Mary D et al. (2015) Joint modeling of cross-sectional health outcomes and longitudinal predictors via mixtures of means and variances. Biometrics 71:487-97
Jiang, Bei; Sammel, Mary D; Freeman, Ellen W et al. (2015) Bayesian estimation of associations between identified longitudinal hormone subgroups and age at final menstrual period. BMC Med Res Methodol 15:106
Huang, Xiaobi; Elliott, Michael R; Harlow, Siobán D (2014) Modeling Menstrual Cycle Length and Variability at the Approach of Menopause Using Hierarchical Change Point Models. J R Stat Soc Ser C Appl Stat 63:445-466
Elliott, Michael R; Sammel, Mary D; Faul, Jessica (2012) Associations between variability of risk factors and health outcomes in longitudinal studies. Stat Med 31:2745-56