Drs. Bischof and Freyer-Adam did not correctly characterize the proportion of eligible participants who enrolled in our study. It was 65% (341 of 524 persons), which is remarkably high for an alcohol brief intervention trial. Readiness to change, AUDIT, and many other alcohol severity–related characteristics were similar in those enrolled, those eligible who did not enroll, and those with risky use who were ineligible (1).

Our trial had fewer exclusion criteria than most trials (2), and readiness was not among them. Nonetheless, many of our participants were considering change (as noted in Table 1 of our article). Furthermore, in medical patients, greater readiness is not predictive of less drinking or problems, calling into question approaches that select participants for intervention on the basis of stage of change (3, 4). We find lack of representativeness an unlikely explanation for our findings.

Dr. Merlo and colleagues write that our control participants received "assessment and feedback sessions." However, control participants received no feedback. At least 3 reasons argue against assessment effects as an explanation for our results: 1) mixed-model analyses suggested a lack of effect (similar when including participants with 1 or 2 follow-up assessments), 2) evidence on assessment effects in this population (particularly in those not seeking treatment) is not yet sufficient, and 3) brief intervention studies in other settings have been positive despite assessments.

Brief intervention has efficacy in selected patients with nondependent, unhealthy alcohol use in primary care. In other settings (such as emergency departments and inpatient medicine services), brief interventions may not have efficacy—as trials are beginning to demonstrate. Moreover, evidence does not support efficacy in non–treatment-seeking adults with alcohol dependence. This is not old news in medical inpatients, and it is relevant now because brief intervention is currently being widely disseminated to these populations. We agree that brief intervention may have efficacy in nondependent inpatients, but this remains to be proven and will be relevant to only a small proportion of screen-positive inpatients (about 20%). Our study shows that most medical inpatients with unhealthy alcohol use identified by screening are unlikely to benefit from brief intervention alone, and it is not the first negative study in this setting (5).

We do not agree that brief intervention is effective for inpatients who are considering change. Hospitalization may be a time for self-change, but evidence that brief intervention improves on this is limited. Our study and others are the best approaches to providing evidence to direct clinical efforts and for how to improve care where current brief interventions fall short.