Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

100 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsMonday, 25th August 2014 • 15:30-16:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust able randomization procedure can minimize the influence of present bias. There are many people suffering from rare diseases, thus there is an urgent need for new therapies and hence for statistical methods that can be used in small clinical trials. As most current statistical methods are unsuitable for the use in small populations, the EU launched several programs inves- tigation new methods for small population groups. The scope of this presentation is to investigate and compare several ran- domization procedures that minimize selection and accidental bias such as introduced by trends and unobserved covariates in small clinical trials. The theoretical results are illustrated by simulations based on an R pack- age the authors implemented for this purpose. Acknowledgement: The research is embedded in the IDeAl EU-FP7 proj- ect, Grant-Agreement No. 602 552. P1.2.178 Factors associated with fever control and identification of subgroups in sepsis trials: a regularization based approach AN Vidyashankar1 , S Katsahian1 , Y Dong2 , F Schortgen3 1 George Mason University, Fairfax, United States, 2 Temple University, Philadelphia, United States, 3 Réanimation Médicale Groupe Hospitalier Henri Mondor, Créteil, France   Aims: In a recent randomized controlled clinical trial, the effect of using external cooling in febrile ICU patients was evaluated and overall benefi- cial effects of external cooling was identified. However, the overall analysis did not take into account multiple sources of variation present in the data. The primary objectives of this study are: (a) to incorporate additional clini- cally relevant covariates and their interactions to identify subgroups of pa- tients benefitting from external cooling and (b) evaluate the statistical and clinical gains of the proposed alternative data analysis approach. Methods: When several clinically relevant covariates and their interac- tions with demographic variables are introduced into modeling, the num- ber of parameters far exceeds the sample size leading to large p small n problem. Additionally, accounting for variability amongst the centers and the patients lead to statistical models involving both fixed and random effects. We use the adaptive LASSO method, to perform variable selection and inference accounting for uncertainty in variable selection. Results: Incorporation of relevant clinical variables in the analysis of sepsis trials facilitates data based methods for identifying clinically ben- efitting subgroups within the patient population. Additionally, system- atic accounting for various heterogeneities into statistical modeling im- proves power for identifying effects of cooling between the subgroups. Importantly, the methodology facilitates construction of alternative clini- cal scores for monitoring of sepsis severity. Conclusions: Including several clinically relevant factors and their interac- tions into regression models lead to data-based methods for identifying clinically benefitting subgroups in sepsis trials. The methodology identi- fies variables for routine monitoring of sepsis progression.   P1.2.195 Benefit and cost of clinical trial operation quality monitoring - experiences and lessons from the ProTECT III trial W Zhao1 , S Yeatts1 , Y Palesch1 1 Medical University of South Carolina, Charleston, United States   Funded by NINDS, ProTECT III (NCT00822900) is a randomized controlled large multicenter phase III trial comparing the safety and efficacy of pro- gesterone vs. placebo for the treatment of traumatic brain injury. Due to variability in neurological emergency treatment settings, the heteroge- neous physiology of the disease, and the large number of clinical sites, this trial was vulnerable to the diversity of operation transgressions regarding the clinical standardization and guideline for patient care across sites - a serious concern for both sponsor and investigators. To quantify this variability, a detailed monitoring effort was designed and implemented to track transgressions in study patient care. Thirteen transgression Forms were defined and included in the study book; data on these physiological parameters were collected 24 hours a day for each patient, up to 30 days after injury. At the end of the study, a total of 882 subjects were enrolled from 38 clinical centers with a median enrolment of 18. For these subjects, 28,927 transgression CRFs were submitted with a total of 694,285 hourly records. To effectively monitor the trial operation quality, 116 data validation rules where created within the study database, and captured 34,794 transgression rule violations. The transgression monitoring protocol is a unique feature of the ProTECT III trial. It provides a rare opportunity for us to examine the benefit and cost of transgression monitoring, and the impact of trial operation trans- gressions on trial outcomes.   P1.3 Pharmacoepidemiology and drug development P1.3.51 Pharmacogenetic study of delayed hyperbilirubinemia in a cohort of 4,000 infants BL Fridley1 , A Wang1 , J Jackson2 , JS Leeder2 , SM Abdel-Rahman2 1 University of Kansas Medical Center, Kansas City, United States, 2 Children’s Mercy Hospital, Kansas City, United States   Despite recommendations by the American Academy of Pediatrics (AAP), screening for hyperbilirubinemia is not universally applied to newborns prior to their hospital discharge. As a result, hyperbilirubinemia remains the single most common reason for hospital readmission during the first 2 weeks of life. The goal of this study was to identify genetic biomarkers, in conjunction with clinical risk factors, associated with risk for progression to hyperbiliru- binemia. A cohort of 4,055 newborns (>34 weeks) were prospectively en- rolled. Demographic data and a cord blood sample were collected, along with known risk factors for hyperbilirubinemia. The primary phenotype of interest was progression to hyperbilirubinemia after discharge. Seven can- didate markers in genes related to conjugation, transport and clearance of bilirubin were genotyped (SLCO1B1, RXRA, SLCO1B3, SULT1B1, SLCO1A2, APOA2, LST3). Genetic association analysis of these markers found no sta- tistically significant associations when analyses were conducted separate- ly for white non-Hispanic (N = 2210) and African American babies (N=965). As breast feeding represents a risk factor for hyperbilirubinemia, we also investigated interaction of food source with genotype (gene-environment interaction) where we observed a modest association in African American newborns for rs7696239 (p = 0.045). Additional research is on-going to expand the pharmacogenomic assess- ment of potential genetic risk factors and develop a prediction model that will support a clinical nomogram to detect hyperbilirubinemia risk.  

Pages Overview