The current study adds to a growing body of literature that finds ED-based computerized alcohol screening programs are both acceptable to patients and effective in educating them about their alcohol risk level. Our feasibility measures revealed that the majority of patients approached for study enrollment were willing to participate (80%), that patients rated the program as easy to use, and that they were not bothered about completing the survey during an unrelated ED visit. The CASI program accurately identified a high number of at-risk drinkers (32%). In addition to these positive findings, research staff reported few study interruptions and logistical barriers to implementing the CASI protocol.
Our measures of comprehension showed that the majority of patients (90%) accurately reported their alcohol risk level after participating in CASI, and, in 97% of cases, staff provided patients with correct alcohol-risk take home materials. Overall, these findings suggest that CASI facilitates efficient and accurate identification of at-risk drinkers and provides accurate alcohol-risk education to patients seeking care in the ED.
Regarding feasibility of recruitment, some patients chose not to enroll because they did not want to participate in a computer-based study. However, it should be noted that the majority or participants, even those who reported “rarely” or “never” using computers, were able to participate without difficulty. We were able to enroll large numbers of patients in a very short time (three months) from a diverse demographic pool. Anecdotally, RA staff reported that even elderly patients (one participant was 85 years old) and another with a broken dominant arm were still willing and able to complete the study without a problem. Furthermore, the study was conducted in the second largest ED in the United States, demonstrating that, even within the most busy and crowded ED settings, CASI can be successfully implemented.
Our findings lend support to Ranney et al’s  recent study, which concluded that there is a high preference for technology-based behavioral interventions among ED patients. In their survey of over 600 ED patients, the authors found a high baseline level of computer usage (91%), and 90% of their sample expressed a preference for a technology-based (versus face-to-face) intervention for at least one risk-behavior topic. Our results are also consistent with a recent systematic review of 20 technology-based SBIRT studies conducted among ED patients over the past 10 years that focused on high-risk health behaviors in the areas of alcohol/substance abuse, alcohol and youth violence, interpersonal violence, unintentional injury, mental health, and HIV risk . These studies found high acceptance and feasibility for computer-based interventions among ED populations and supported their use in the ED to overcome the limitations of staff time, training, and resources, all of which hamper the feasibility of face-to-face ED interventions. They also highlighted the value of the sense of anonymity and privacy computer interventions can provide.
Our results are in contrast to Nilsen’s  recent study, however, which evaluated the implementation of a computerized alcohol-screening kiosk program among Swedish ED patients. In that study, only 41% of the target population completed their alcohol survey when directed to do so by the waiting-room triage nurse upon arrival in the ED. The authors cited various logistical barriers (staff, patient health, etc.) resulting in low participation. It should also be noted that the model used in the Nilsen study called for “minimal researcher involvement to test the concept’s viability under realistic conditions,” and the program was unattended by any dedicated staff. The lack of oversight of the project (beyond a nurse handing patients index cards instructing them to complete the test and someone checking the printer to make sure it had paper to print out results) was the most likely contributor to this study’s low acceptance and participation rates. Although laudable for attempting to implement a stand-alone program with little oversight, it is likely that computer-screening programs require some degree of staff involvement and a staff “champion” to be successful.
An additional topic highlighted by the current study is the need to focus on patient comprehension to ensure effective knowledge transfer within educational interventions. To measure the feasibility of implementing an ED-based SBIRT program, we felt it was critical to assess the patient’s comprehension of the educational materials CASI provided. We are not aware of any other ED studies that focused on comprehension as a key outcome. However, with the growing problem of health illiteracy in the US and the fact that many patients who use the ED have low health literacy , it is critical that interventions begin with this step during the developmental phase. This often-overlooked step must be done to ensure patients are absorbing the educational content provided during and ED intervention. This seems especially important when one considers that patients are often feeling very ill and are likely distracted by their current health status in the middle of an ED visit. Although it was only possible to use an immediate and cued recall test within the confines of this pilot study as a proxy for comprehension (and patients were presented with the computer-screening results within a few minutes of being asked to recall it), we feel it provides clear and objective evidence that, at a minimum, patients were able to read and retain the health information provided to them during the CASI intervention. Although this is an important first step, future studies would do well to use a more comprehensive measure of comprehension to be sure patients understand the meaning of their alcohol-risk levels.
Regarding comprehension, these feasibility data also provide objective criteria for crafting more comprehensible and comprehensive health education messages for future studies. For example, we found that patients in this study did not comprehend high or medium alcohol-risk education as well as the low-risk education (79%, 83%, and 94%, respectively). This may have been due in part to patients choosing not to attend to a “message they do not want to remember” versus true lack of recall or comprehension. However, with a comprehension discrepancy as high as 15% (low- versus high-risk categories), and based on close scrutiny of the length and grade level of the educational feedback provided, this finding also suggests the health message used for the medium- and high-risk groups may have been less well-understood, and thus may warrant revision.
Additionally, some feasibility/acceptance measures used in this study were endorsed below the 50% level. However, all endorsements would be considered clinically meaningful and desirable outcomes. For example, although only 28% of the patients said they had “some intention to consult a health-care professional about their alcohol use because of their study participation,” it is highly desirable to find 28% of patients recognizing the need to speak with a professional about their alcohol use. This percentage is also very close to the percentage of “at-risk” alcohol users (32%) found in the study.
A secondary goal of the project was to measure accurate deployment of alcohol risk-level education to patients. Our data reveal that, in a few cases, RAs did not provide patients with the correct take-home alcohol-risk materials. Nevertheless, the correct information did appear on the patient’s computer screen, and all patients received the same general alcohol education pamphlet, including NIAAA low-risk drinking guidelines and an 800 number to contact a treatment locator hotline. Therefore, although the correct take-home educational materials were not accurately deployed to every patient, no patients received “bad” take-home information, and the computer providing accurate information consistently and without fail.
Regarding limitations, the current study was limited by a cross-sectional design and use of a convenience sample. Additionally, we did not use a comparator group or follow-up with patients to see if they acted on the alcohol risk-level education provided. There were also fewer men and white patients in our sample, so gender and racial differences could not be fully explored. In addition, we were unable to modify the computer content because it was provided free to us by another website. As always, self-reporting bias is an inherent problem in studies of this nature. However, we feel that the anonymous and confidential manner in which patients were able to report their alcohol use likely minimized social undesirability and under-reporting of alcohol use in this study.