To the Editor.—
I enjoyed reading the article entitled "Treatment Experience With Pregnant Problem Drinkers" by Rosett and colleagues (1983; 249:2029). The drinking history questionnaire used was appropriate and the treatment plan followed was seemingly adequate, but where was the control group? Also, there was no randomized selection of the treatment group, which leaves one wondering about the reliability and validity of the treatment plan and also adds an element of measurement bias. Also, what happened to the social problem drinkers—symptom problem drinkers? Were there any statistically significant results when other social characteristics were considered?A final suggestion—what about using the pregnancy outcome of the 49 heavy drinkers as an outcome variable in the clinical trial?