Despite belief in a responsibility to help their patients with alcohol problems, primary care physicians are often hesitant to accept responsibility for the management of these disorders.30,31
This pilot study suggests that training and chart-based prompting can increase by nearly 3-fold the likelihood that clinicians will inquire about the alcohol history with patients who have changed their drinking behavior. The magnitude of our findings is similar to that of a primary care study of brief counseling for patients with current hazardous drinking in which providers who received 2.5 hours of training and were prompted to intervene were twice as likely as usual care providers to discuss alcohol use.27
The current study differed from prior studies in its exclusive focus on maintenance care for patients who have already made changes in their alcohol use (who are the majority of patients with alcohol issues in primary care),3,32
and we did not provide the clinicians with specific therapeutic recommendations.
For patients who were asked an alcohol history, the intervention also increased assessment of prior and planned alcohol treatment, and offers of prescriptions and/or referrals. These findings suggest that the intervention improved the appropriate content of alcohol-related discussions when they occurred. The greater satisfaction among patients of intervention clinicians who took an alcohol history bolsters this supposition. Other studies suggest that satisfied patients are more likely to return for follow-up and adhere to therapy.33,34
Patient satisfaction here might be a proxy for the quality of the therapeutic alliance, an important process marker and harbinger of better substance use outcomes.35
The intervention also tended to increase provider assistance through offers of prescriptions and/or referrals among patients with greater CAGE scores. This finding suggests that the intervention increased the appropriate counseling of patients with more severe alcohol histories.10
We cannot discern the independent contribution of training versus prompting. A randomized trial in a primary care setting found that prompting with alcohol screening results and individualized recommendations increased faculty physicians' alcohol-related discussions by 1.5-fold.36
In the current study, the prompt did not include screening results or recommendations. We sought to simulate the situation where a nurse or assistant might prescreen a patient then clip materials to the chart for the clinician. We decided to include such a cue because prompts are effective and educational interventions alone have limited effect on practice behaviors.27,36,37
The clipped materials appear to have piqued providers' interest in the alcohol history, and the limited decay in relation to the time since the training suggests the importance of an ongoing system of prompts. However, chart cues alone would be unlikely to induce medical clinicians to assess appropriately the patient's plans for treatment or to offer prescriptions and referrals. One can speculate that the latter findings represent an effect of the training, though further research is needed to discern these relationships.
Clinicians' confidence in managing problematic alcohol use influences whether they elicit an alcohol history.31,38
Clinicians' increased confidence might explain the findings, though small numbers of clinicians prevented detection of a direct effect. Patients did not report that intervention clinicians spent more time with them, though they did perceive that clinicians who inquired about the alcohol history spent on average 4 more minutes at the index visit. Although uncorroborated patient perceptions of time warrant caution, these findings suggest that training and prompting might not be enough to motivate clinicians to provide the 5 to 15 minutes of most brief counseling interventions without special visits.
This study has further limitations. Studies of unhealthy drinking in primary care commonly lose 45% to 75% of eligible subjects for logistical reasons or refusal.36,39
In the current study only 11% of patients refused participation, but a large proportion of those eligible could not be approached for logistical reasons. The arrangement with these busy private offices prohibited research staff from disrupting patient flow. Patients often completed the screener in the waiting room, but were commonly called for their medical appointment before study enrollment. Eligible persons were also missed while the research assistant was interviewing another subject. Although research staff screened individuals without regard to age or gender, the nonenrollee group was older and had more women than the enrolled group. We can speculate that older patients or women might have been more likely to have come for a sick visit, rather than a routine appointment, and were taken into the office more quickly. Selection bias, e.g., a healthy volunteer effect, might mean that the intervention would work more or less well if implemented for all patients with prior alcohol problems.
Exit interviews with the patients assessed the primary outcomes. Previous studies suggest that patients can provide valid reports of the content of clinical encounters.27,28,40,41
Despite their face validity, the current measures have not been validated against objective indicators or corroborative clinician reports. Resource limitations also prevented blinding of research interviewers to study condition. The study also examines a single visit with a single provider. “Cross-overs” to other clinicians over time limit our ability to detect small changes in practice behaviors. More definitive studies should consider randomization by natural practice groupings to minimize contamination across groups.
Despite these limitations, we conclude that training and prompting primary care clinicians in the maintenance care of patients with prior alcohol problems can modestly increase inquiry about the alcohol history and appropriate actions after elicitation of a suggestive history. These effects might last for a minimum of 18 months after the training. Alcohol dependent patients in primary care settings often have less severe alcohol problems than formal treatment populations.42
Many have remitted in response to medical problems or family concerns without formal treatment, and they commonly continue controlled drinking.32,43–45
Such patients, as well as those with treated alcohol use disorders no longer in treatment aftercare and those with less severe alcohol histories in remission, deserve monitoring, support and follow-up. With proper training, tools and logistical support, primary care clinicians might fulfill this role for alcohol use disorders as they do for other chronic disorders. Future research should determine whether primary care clinicians could effectively deliver maintenance care longitudinally and reduce relapse.