St. Jude Family of Websites
Explore our cutting edge research, world-class patient care, career opportunities and more.
St. Jude Children's Research Hospital Home
- Fundraising
St. Jude Family of Websites
Explore our cutting edge research, world-class patient care, career opportunities and more.
St. Jude Children's Research Hospital Home
- Fundraising
Kendrick Li Group
Providing statistical support in the design and analysis of pediatric cancer survivorship studies.
About the group
Observational studies frequently harbor both overt and hidden biases stemming from study design or from psychological and behavioral habits of study subjects. Hidden biases can impact data interpretation and lead to the generation of models or conclusions that do not accurately represent the true effects of a therapeutic intervention. Dr. Li’s group focuses on developing statistical methodologies and software to detect and correct biases in observational studies. While a primary focus rests on cancer survivorship studies conducted at St. Jude, the methodological and conceptual advances generated by the Li group are applicable across a wide range of healthcare disciplines.
Research summary
During study design, certain overt selection biases can be accounted for and mitigated, such as ethnicity, known cultural expectations, economic status, etc. However, other confounding factors stemming from human psychological and behavioral patterns are poorly defined or difficult to measure, including life habits and healthcare seeking behavior of individual study subjects. These and other difficult-to-define variables – known as unmeasured confounders – are unavoidable in epidemiological studies but can pose significant challenges to data interpretation and inference. Our group develops statistical causal inference methods and software to detect and correct hidden confounding biases. In accounting for unmeasured confounding effects, our work focuses on lending rigor to existing and ongoing studies and to provide design support for impending cohort studies.
While advanced treatment and healthcare practices have greatly improved the survival rate of childhood cancer patients in the US, childhood cancer survivors frequently experience long term physiological, psychological, and social effects due to the course of illness or treatment regimen. Our group’s focus at St. Jude is to better understand the challenges faced by childhood cancer survivors and identify therapies that can lead to higher long-term quality of life. We collaborate with both internal and external investigators in analyzing two ongoing childhood cancer survivorship studies at SJCRH: St. Jude Lifetime Cohort Study (SJLIFE) and Childhood Cancer Survivorship Study (CCSS). Our goal is to better identify potential biases during data collection and mitigate the effects of unreliable assumptions in data interpretation, to achieve the most accurate understanding of quality-of-life data surrounding cancer survivorship, and from which to inform future intervention and long-term healthcare practices for pediatric patients.
Insights gained from previous and ongoing work has culminated in the generation of software tools that are widely applicable to correct for overt and unmeasured confounding effects across a wide spectrum of epidemiological studies.
Software
pci2s: an R package that provides user-friendly statistical methods for unmeasured confounding bias adjustment using double negative control variables
Confounding factors are unavoidable in epidemiological studies. While many overt confounders are accounted for during study design, human behavior patterns, along with other factors that may not have been known at the time of study, may introduce hidden bias, resulting in inaccuracies in results interpretation. Negative control variables are routinely collected covariates which may assist with better definition of various types of hidden confounders and can be utilized to adjust for their impact on treatment, outcome, and known confounders. The software package pci2s leverages negative control variables to better adjust for hidden or unmeasured confounders, lending greater robustness and accuracy to data interpretation.
The pci2s software package can be found here.
Publications
About Kendrick Li
Dr. Kendrick Qijun Li is an Assistant Member of the St. Jude Faculty and received his PhD in Biostatistics from the University of Washington, Seattle. Dr. Li is interested in how confounding and selection biases impact study conclusions, and how to use statistical methods to correct for these hidden biases in observational studies. Before joining St. Jude, Dr. Li completed his postdoctoral training at the Department of Biostatistics, University of Michigan, during which he developed statistical methods to control for unmeasured confounding bias in epidemiological studies related to infectious diseases and post-market vaccine evaluation. Dr. Li now collaborates with investigators across the institution to lead data analysis and study design in childhood cancer survivorship research. He is passionate about pioneering novel methods to correct for bias due to study design and latent patient characteristics. Bringing statistical rigor to epidemiological studies, Dr. Li hopes to strengthen study inferences and uncover deeper insights on long-term chronic health burdens and responsiveness to interventions.
Contact us
Kendrick Qijun Li, PhD
Assistant Member
Department of Biostatistics
MS 768, Room S2041
St. Jude Children's Research Hospital
Memphis, TN, 38105-3678 USA GET DIRECTIONS