Fall 2017 Seminars

September 28th | Patenge Room C102 FeeHall 3:00 p.m.

Dale Sandler, MPH, PhD | View Seminar
Senior Investigator
Epidemiology Branch/Chronic Disease, NIEHS

“Prospective studies of environment and health”

Large prospective cohort studies typically focus on dietary and lifestyle factors in relation to future health outcomes. Using such studies to address questions related to the impact of environmental exposures poses some unique challenges. But, if done right, prospective studies can overcome many of the biases associated with retrospective studies of exposures that may have changed as a result of a disease diagnosis. Using as examples three distinct cohorts developed at the National Institute of Environmental Health Sciences, this presentation will illustrate the challenges and opportunities of prospective studies. The presentation will introduce two well-established cohort studies, The Agricultural Health Study and The Sister Study, and describe opportunities for collaborative research building on these cohorts. The presentation will focus primarily on the development of The GuLF STUDY, a prospective study designed to evaluate potential long term health effects associated with the April 2010 Deepwater Horizon oil spill disaster in the Gulf of Mexico.

October 12th | E4 Fee Hall 3:30 p.m.

Shayesteh Jahanfar, PhD | View Seminar
Reproductive epidemiologist
Assistant Professor, MPH Program School of Health Sciences
Central Michigan University

"Is birth weight discordance a good predictor of adverse perinatal outcomes? Does chorionicity matter?" 

Study objectives were 1) To establish the optimal threshold of birth weight discordance for prediction of stillbirth and perinatal mortality in twins born in British Columbia without access to chorionicity data. 2) To repeat the analysis in a twin cohort born in Children and Women hospital with chorionicity information 3) To compare the cut-off points between the two populations with and without the chorionicity data.

Using retrospective data on over 14,000 twins from both a hospital-based and a province-based cohort of twins from 14 hospitals over a 10-year period, intertwin birth weight discordance was analyzed in relation to the occurrence of stillbirth or neonatal mortality of twins of 20 weeks or more gestational age. Standard receiver operating characteristic curve analysis suggested that BWD of ≥30% and ≥35% are the optimal thresholds for stillbirth and perinatal mortality. Birth weight discordance is a good predictor for stillbirth and perinatal mortality irrespective of chorionicity, although more likely than not analysis without chorionicity data will overestimate the impact of BWD on perinatal loss.

October 19th | Patenge Room C102 Fee Hall 3:30 p.m

Ping-Shou Zhong, PhD | View Seminar
Assistant Professor
Department of Statistics and Probability

"Order-restricted inference for means with missing values"

Missing values appear very often in many applications, but the problem of missing values has not received much attention in testing order-restricted alternatives. Under the missing at random (MAR) assumption, we impute the missing values nonparametrically using kernel regression. For data with imputation, the classical likelihood ratio test designed for testing the order-restricted means is no longer applicable since the likelihood does not exist. This article proposes a novel method for constructing test statistics for assessing means with an increasing order or a decreasing order based on jackknife empirical likelihood (JEL) ratio. It is shown that the JEL ratio statistic evaluated under the null hypothesis converges to a chi-bar-square distribution, whose weights depend on missing probabilities and nonparametric imputation. Simulation study shows that the proposed test performs well under various missing scenarios and is robust for normally and nonnormally distributed data. The proposed method is applied to an Alzheimer's disease neuroimaging initiative data set for finding a biomarker for the diagnosis of the Alzheimer's disease. 

October 26th | E4 Fee Hall 3:30 p.m.

Albert Tenesa, PhD | View Seminar
Roslin Institute
MRC Human Genetics Unit
The University of Edinburgh

"An atlas of genetic associations in UK Biobank"

Genome-wide association studies have revealed many loci contributing to the variation of complex traits, yet the majority of loci that contribute to the heritability of complex traits remain elusive. Large study populations with sufficient statistical power are required to detect the small effect sizes of the yet unidentified genetic variants. However, the analysis of huge cohorts, like UK Biobank, is complicated by incidental structure present when collecting such large cohorts. For instance, UK Biobank comprises 107,162 third degree or closer related participants. Traditionally, GWAS have removed related individuals because they comprised an insignificant proportion of the overall sample size, however, removing related individuals in UK Biobank would entail a substantial loss of power. Furthermore, modelling such structure using linear mixed models is computationally expensive, which requires a computational infrastructure that may not be accessible to all researchers. Here we present an atlas of genetic associations for 118 non-binary and 660 binary traits of 408,455 related and unrelated UK Biobank participants of White-British descent. Results are compiled in a publicly accessible database that allows querying genome-wide association summary results for 623,944 genotyped and HapMap2 imputed SNPs, as well downloading whole GWAS summary statistics for over 30 million imputed SNPs from the Haplotype Reference Consortium panel (>23 billion phenotype-genotype pairs). Our atlas of associations (GeneATLAS, http://geneatlas.roslin.ed.ac.uk) will help researchers to query UK Biobank results in an easy way without the need to incur in high computational costs.

November 9th | Patenge Room C102 Fee Hall 3:30 p.m.

Omayma Alshaarawy, PhD 

Research Associate
Department of Epidemiology and Biostatistics

“Using Pre-clinical Biomarkers and Epidemiological Models to Assess the Effects of Cannabinoids on Disease”.

There are several scientific puzzles at the intersection of cannabis use, and obesity and diabetes mellitus. The 'munchies' reported after cannabis use might lead one to think there would be cannabis-attributable obesity, and also possibly cannabis attributable type 2 diabetes. Surprisingly rigorously gathered epidemiological evidence from large population based studies supports the idea that cannabis use is associated with a reduced occurrence of obesity, a reduced occurrence of diabetes, and lower levels of biomarkers indicative of altered glucose metabolism. Dr. Omayma Alshaarawy is a Research Associate at MSU Department of Epidemiology and Biostatistics. Her talk will focus on her NIH-funded research to investigate the immunomodulatory, and the metabolic effects of cannabis use.

November 16th | E4 Fee Hall 3:30 p.m.

Richard Chappell, PhD | View Seminar
Professor, Biostatistics & Medical Informatics
University of Wisconsin, Madison

"Response Adaptive Randomization: Dangers and Cures"

Response adaptive randomization (RAR) is the strategy of allowing subjects' randomization ratios in a clinical trial to depend on the outcomes of previous patients. It was originally proposed by Zelen (1979) as the "Randomized Play the Winner Rule" and subsequently criticized by Peto (1985) as unnecessarily prolonging a trial's conclusions and biasing results. The debate has not subsided since. This talk will give some background for RAR, describe the nature of the bias as resulting from confounding with time, and present some simple solutions.