Variations in Methadone Treatment Practices

To the Editor. — Drs D'Aunno and Vaughn¹report that methadone treatment "units that have higher average dose levels are more likely to have clients who remain in treatment," which they interpret as supporting higher dosage levels. Unfortunately, serious defects in the data analysis render this interpretation highly questionable. Their Table 5 presents a set of multiple regression analyses relating characteristics of clients, units, and their environments (as predictors) to treatment practices (as dependent variables). But average dose level (one of their eight treatment practice variables) is then introduced suddenly, as an additional predictor variable in the multiple regression on average time in treatment (also a treatment practice variable). This choice appears arbitrary, because the bivariate correlation between dose level and time in treatment (r=.12) is not statistically significant (Table 4) and because the analysis excludes two other treatment practice variables—take-home dosage and decreasing dosage, which do exhibit