Cross-tabulations were used to calculate the prevalence of the 4 stages of alcohol and drug use by demographic characteristics. Conditional prevalence estimates for each alcohol or drug use stage were also calculated among those who had reached the earlier stage of use. Estimated projections of the cumulative probability of stages of alcohol or drug use as of the age of 18 years were obtained by the actuarial method implemented in PROC LIFETEST in SAS statistical software (version 9.2; SAS Institute, Inc). The associations of age, sex, and race/ethnicity with the stages of alcohol or drug use were examined using multiple logistic regression analysis. Because the NCS-A data are both clustered and weighted, the Taylor series linearization method implemented in SUDAAN (version 10; Research Triangle Institute) was used to estimate standard errors of logistic regression coefficients. The coefficients and standard errors were exponentiated to produce odds ratios and 95% CIs. Significance of predictor sets was evaluated using Wald F tests based on design-adjusted coefficient variance-covariance matrices. Statistical significance was consistently evaluated using 2-sided tests with a .05 significance level.
