We noted in Section 5.5 that an argument could be made for combining the previous and recent levels of IV drug use into one group yielding a dichotomous "never-ever" covariate. Evaluate the model using this covariate in place of history of IV drug use coded at three levels.

Is the dichotomous covariate significant? Do the coefficients for the other covariates in the model change? How does the fit of the model compare with the fit of the original model (Table 5.10)?