Data attrition in follow-up studies of alcoholics.

Abstract

Drinking behavior questionnaires were mailed to 60 alcoholic patients (mean age 41; 25 women) as part of their posthospitalization follow-up at 3 or 6 mo. or 1 yr; only 26 responded. A comparison of staff ratings of responders and nonresponders showed that responders were doing significantly better with regard to their drinking problems (P < .005), as measured by months of abstinence (about 4 mo. of abstinence by responders vs. 2 mo. by nonresponders); and by the frequency of resumed drinking (about 1 bout/responder and more than 2/nonresponder). Nonresponders may seriously bias the results of follow-up studies; the extent of this bias was difficult to estimate precisely. The remedy for this problem may lie in increasing follow-up return. Specific procedures were outlined to achieve this end.