Chengjie Xiong is professor of biostatistics at Washington University School of Medicine. He earned a bachelor’s in mathematics from Xiangtan University, master’s in applied mathematics from Peking University, and PhD in statistics from Kansas State University.

I started my career as a biostatistician in aging and dementia research by chance. In 2001, after a few years of teaching statistics in a mathematics department, I had to move due to my family’s relocation to St. Louis, Missouri.

I joined the research faculty in the division of biostatistics at Washington University School of Medicine in St. Louis to primarily support the research design and statistical analysis of the Knight Alzheimer Disease Research Center (ADRC). I anticipated a challenging transition from teaching to research, given my relatively limited training in biology and medicine, especially in the areas of aging and dementia.

On the other hand, I was attracted to the position on a few fronts.

First, it provided an opportunity for me to work with a huge real-world longitudinal database that could be used to examine the course of Alzheimer’s disease (AD) and dementia and aging from a variety of aspects, as well as to be part of a productive and creative research team.

Second, although I had sparse prior experience in biostatistics before taking the position, work in this area aligns with my long-term interest in the applications of statistical theories and models in the biological sciences and appealed to me as an area that could be approached from both methodological and medical/biological perspectives.

Third, many ongoing large longitudinal research projects in AD—both observational and interventional—pose a wealth of challenging and fascinating statistical methodological issues, including the appropriate use of longitudinal linear and nonlinear models, nonlinear mixed model, Markov chain–type of transition models, marginal means and marginal odds ratios models with generalized estimating equations, censored and truncated observations and survival endpoints with competing risks, receiver operating characteristic curves and surfaces, and missing data.

These methodological challenges are, in my opinion, the dream of statisticians who seek to make a difference in real-world applications of statistics. I was immediately immersed in the extensive interactions between cutting-edge dementia sciences and statistical design and analysis by working on a variety of statistical methodological problems associated with AD studies.

I have sought advice and been fortunate to have had three important mentors: Gerald van Belle from the University of Washington and Philip Miller from Washington University—both experts in statistical applications to aging and dementia research—and John Morris, a leading physician scientist in AD research. I met with these mentors on a regular basis.

As my interest in the statistical methodologies of aging and dementia research grew, I quickly realized I needed to understand and appreciate the clinical and biological aspects of aging and AD. Subsequently, as I gained more exposure to the variety of biomedical ways the natural history and interventions of AD are assessed, I came to appreciate the crucial role biostatistics plays in AD studies and the importance of interaction between AD investigators and biostatisticians. This cultivated my interest in assessing the status of statistical applications and quality of statistical methodologies in AD studies and in identifying possible weaknesses in the use of statistical methods in AD research.

A major driver for me are the many less-than-optimal statistical methods and reports in published AD studies and my desire to help identify the most appropriate analytic methods to deal with statistical challenges arising from AD and aging research. This wish eventually led to a successful application for a Mentored Quantitative Research Career Development Award (K25) from the NIH. The award allowed me protected time to not only receive necessary training and mentoring in biology and aging research, but also to develop ways to improve the quality of statistical applications in AD and aging research. It paved the way for me to grow professionally as a biostatistician by communicating statistical concepts and methods in the languages clinicians and physician scientists speak. Since then, I have gradually assumed more leadership in the biostatistics core facilities that support both the Washington University ADRC and the multi-center international Dominantly Inherited Alzheimer Network (DIAN) and its clinical trials unit.

My continued interactions with a wide range of biomedical investigators in AD and aging have led to several successful applications of independent research grant awards from the NIH to tackle the statistical challenges arising from cutting-edge AD and aging research.

Looking back over the past 17 years as a biostatistician in aging and dementia research, I have fully appreciated that statistics is a vital part of aging and dementia research and that a biostatistician must play a leading role in the design and analysis of these studies to safeguard the integrity of scientific findings. More importantly, to become a leader in a collaborative biomedical research setting, a biostatistician must understand the scientific content and articulate potentially complex statistical designs and methodologies in a relevant scientific context.

Whereas AD has been a high-profile disease for decades in this country and Congress has even passed the National Alzheimer’s Project Act to build upon and leverage federal efforts to help change the trajectory of AD and other dementia, I remain surprised about how few statisticians work in aging- and dementia-related statistics.