Developing Robust Subgroup Analysis Methods for Randomized Controlled Trials

PROJECT ABSTRACTThe impact of substance abuseincluding tobaccoalcoholillicit drugsand prescription opioidsin the U Sis immensewith the combined cost of its effects on health carerelated crimeand lost productivity exceeding $billion each yearCritical to addressing this national crisis is the need for improved data analysis and statistical tools that enable substance abuse researchers to more effectively exploit their datasetsSuch tools provide improved detection and estimation of statistical regularitiesthus offering the opportunity for new clinical findings that support the accelerated development of targeted treatments for individuals or groups suffering from substance use disordersSUDIn particularimproving statistical methods and tools for subgroup analysis on clinical trial data would be invaluable to SUD researchSubgroup analyses are utilized in efficacy and effectiveness studies and randomized controlled trials to identify heterogeneity in treatment effectsHTEby assessing how clinical factors impact estimates of treatment effect sizesThis allows researchers to better understand and explain how patients respond differently when clinicalpatientand system factors moderate treatment effect sizesSuch analyses are also utilizedespecially in the absence of valid clinical theoryby practitionersadministratorsand policy makers to make decisions that impact patientstheir quality of lifeand health care costsThis study will extend and evaluate the Best Approximating ModelBAMtechnology as an advanced statistical modeling approach that offers improvements over conventional subgroup data analysis methodsFirstBAM uses a systematic model search and selection strategy to incorporate known or posited factors while simultaneously identifying subgroups and estimating their impact on treatment effect sizesSecondBAM includes robust estimationspecification analysisstochastic exhaustive model searchand validation within the single model selection framework of a generalized additive modelThirda BAM is designed to handle common problems encountered in subgroup analyses including possible model misspecification and overfitting as well as multicollinearitysmall sample size biasand Type I error inflation due to multiple comparisonsPhase I research will investigate extending BAM technology as an advanced statistical modeling tool andquot BAMHTEandquotfor robust subgroup analysis byiincorporating new model selection functionalityiiimplementing novel subgroup specification testing and validation methodsiiievaluating its reliability and validity for robust subgroup analysisandivdemonstrating its robust subgroup analysis capability on a NIDA fundedCTNdatasetPhase I results will establish the essential feasibility for Phase II BAM HTE prototype developmentevaluationand findings disseminationwhich in turn will provide the foundation for subsequent Phase III commercialization PROJECT NARRATIVEThe impact of substance abuseincluding tobaccoalcoholillicit drugsand prescription opioidsin the U Sis immensewith the combined cost of its effects on health carerelated crimeand lost productivity exceeding $billion each yearCritical to addressing this national crisis is the need for improved data analysis and statistical tools that enable substance abuse researchers to more effectively exploit their datasetsSuch tools provide improved detection and estimation of statistical regularitiesthus offering the opportunity for new research findings that support the accelerated development of targeted treatments for individuals or groups suffering from substance use disordersSUDIn particularimproving statistical methods and tools for subgroup analysis on clinical trial data would be invaluable to SUD researchThis Phase I feasibility study will investigate extending Best Approximating ModelBAMtechnology for subgroup analysis byiincorporating new model selection functionalityiiimplementing novel subgroup specification testing and validation methodsiiievaluating its reliability and validity for robust subgroup analysisandivdemonstrating its robust subgroup analysis capability on a NIDA fundedCTNdataset