We assessed variations in readiness, importance and confidence regarding changing drinking after a single primary care physician visit and improvements in these constructs and drinking 6 months later. After the visit, we observed significant increases in readiness, importance, and confidence. The effects were small (i.e. 1 point for readiness, 0.16 for importance, 0.49 for confidence, on a 1 to 10 scale). However, a clinically significant change in these constructs has not yet been well-defined, and the impact of changes of any magnitude is not known. Based on the transtheoretical model and motivational interviewing, clinicians are encouraged to help patients increase motivation, which in turn is expected to lead eventually to behavior change [10
]. After a physician visit we can detect the beginning of such changes. Having a discussion about alcohol with the physician appeared to have an additional impact on readiness to change, an effect that was no longer detectable 6 months later. Like other measures of health states (e.g. blood pressure), "readiness to change" should be viewed as an instantaneous measure dependent on various internal and external influences.
In addition to these short-term changes, we studied the course of risky drinking and "readiness to change" in primary care patients. Six months after a physician visit, most subjects improved either their drinking or readiness. These improvements suggest that primary care physicians should be somewhat optimistic regarding the course of unhealthy alcohol use, with more than half of patients improving in a relatively short period of time.
We identified few predictors of changes in "readiness to change" and none that were consistent across measures or time. Not having a partner was a positive predictor of immediate changes in readiness but a negative predictor of improvement in two measures 6 months later. Being white was associated with worse readiness immediately after a physician visit and less improvement 6 months later but the finding was not confirmed in analyses with the RTCQ or the other two behavior change construct measures. Speculation regarding the mechanism for these hypothesis generating and inconsistent findings would be premature.
A discussion between the physician and the subject about alcohol predicted a positive change in readiness immediately after the visit (confirmed by the RTCQ), but paradoxically, was associated with less improvement in drinking or importance (but no change in VAS- or RTCQ-measured readiness). The fact that no association was found between discussion with the physician and drinking 6 months later may have been due to the use of ineffective counseling, but given the observed short-term effect, another explanation could be simply that this effect did not last.
Neither alcohol consumption, alcohol problems, nor illicit drug use significantly affected behavior change constructs or improvements. The fact that these markers of severity were not found to be negative predictors of improvement is of interest and should encourage physicians to address problems related to alcohol consumption even in the presence of concomitant illicit drug use, considering that most of their patients will have some improvement, independent of the severity of the alcohol problem.
A number of studies have assessed readiness to change and related constructs. In general, these studies have focused on characterizing specific populations [13
] or on studying readiness as a predictor of behavior change [12
]. Our study instead focused on how these constructs change over time. Improvement over time in untreated adults with unhealthy alcohol use and alcohol use disorders (alcohol abuse and dependence) has been previously reported [22
]. Alcohol abuse and alcohol dependence seem to be (especially the latter) chronic conditions characterized by recurrent episodes of disease activity [42
]. But the natural history of the spectrum of unhealthy alcohol use (risky drinking amounts through dependence) is not well described in the literature, nor is the natural history of readiness to change.
In addition to the aforementioned studies of the natural history of alcohol use disorders, studies of brief intervention for nondependent unhealthy alcohol use in primary care consistently report improvement over time in both treated and untreated individuals [43
]. For example, male heavy drinkers in primary care decreased drinking over 3 years by 25 to 53% (depending on the outcome measure) in both intervention and control groups [44
]. The improvement in drinking observed in our sample is in this range. Improvements such as these could be attributed to a regression to the mean [22
]. However, our study sample was not primarily composed of very heavy drinkers, and we also observed improvements in readiness, importance and confidence regarding changing drinking, which were in the opposite direction than any hypothesized regression to the mean, given the relatively high levels of readiness, importance and confidence in the study population at baseline. Assessment effects (improvements due to being asked questions about drinking and discussing answers to those questions) have also been suggested as causes of improvements in drinking [45
]. This exposure may have in part accounted for improvements in our sample. But if asking about alcohol and discussing drinking in primary care are in fact responsible for improvements, such effects should be viewed as favorable exposures in the primary care setting, and as part of the course in these patients, rather than as methodological nuisances.
This study has some strengths. To our knowledge, this is the first study to explore changes in readiness, importance and confidence during a single primary care visit. We described rapid changes in these constructs. We were also able to describe changes in readiness, importance or confidence and drinking over a 6 month period. Subjects studied were participants in a trial but there was no experimental brief counseling intervention nor a significant treatment effect on drinking amounts.
The study also has some limitations. First, the applicability of our findings may be limited to primary care patients with unhealthy alcohol use who agree to be screened and followed in a research study, and to those with similar characteristics as in our sample (e.g. 32% reporting no alcohol problems, a third with illicit drug use). Although participants differed little from those who did not participate, participants did drink more. Similarly, subjects lost to the analytic sample differed little from those studied except on gender. The effects of these differences can be considered in the interpretation of our results. Second, findings are from secondary data analyses. Causality (of predictors) cannot be inferred, and there could be many explanations for changes in the readiness constructs. However, observational studies such as this one are likely among the best ways to study the natural course of behavior change constructs and changes in drinking, particularly prospective studies. Third, in our attempt to explore changes over time, we had to combine actual behavior (changes in alcohol consumption) with cognitions about behavior change. It is likely that these two dimensions reflect different aspects of behavior. We presented data on the dimensions separately (and combined) but performed regression analyses on the combined outcome. From a clinical perspective, improvements in either drinking or readiness (combined) seem to be most relevant. Also, the interpretation of a 6 month change in readiness on a continuous scale for someone who continues to drink risky amounts is difficult, since it not clear how one would interpret a change in some number of points. Fourth, we did not adjust the level of significance for multiple comparisons. As such one should be cautious about interpreting associations, particularly those not in the hypothesized direction. Lastly, all data were obtained from interviews and are subject to recall and social desirability biases. But interviews with trained research associates and assurances of confidentiality took place immediately after the primary care visit to maximize accurate recall and minimize bias. Nonetheless, we do not know what patients meant by discussions about alcohol, which could have been brief or extensive and may or may not have included known effective components of brief interventions.