|
Activity Number:
|
242
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #302415 |
|
Title:
|
Risk Factors Associated with Young Adults Nonmedical Prescription Drug Use (NMPDU) Using a National Sample: A Comparison of Recursive Partitioning Trees and Logistic Regressions
|
|
Author(s):
|
Lirong Zhao*+ and Linda Simoni-Wastila and Zhenqiu Liu and Ming T. Tan
|
|
Companies:
|
University of Maryland School of Pharmacy and University of Maryland School of Pharmacy and University of Maryland School of Medicine and University of Maryland Greenebaum Cancer Center
|
|
Address:
|
220 Arch St, 12-606, Baltimore, MD, 21201,
|
|
Keywords:
|
Risk factors ; Outcome ; Recursive partitioning ; Logistic regressions
|
|
Abstract:
|
Objective: to compare the risk factors associated with NMPDU identified from logistic regression (LR) and recursive partitioning trees (RPT). Methods: 15,864 young adults (age 18-25) from the 2003 National Household Survey on Drug Abuse were used. 23 potential risk factors for social-demographic and behavior characteristics were extracted from data. LR (with stepwise variable selection) and RPT models were fit in SAS and R. Factors with P< 0.05 from LR and those that partitioned the sample in RPT were identified as risk factors. Results: 15.9% young adults used NMPDU in 2003. LR identified 12 risk factors including age, alcohol and cigar use, easy to obtain and use of several drugs, while RPT identified 6 of them. Conclusion: RPT is an alternative method to identify risk factors in health services and outcome research, especially when interactions among covariates are hard to determine.
|
