We were able to reach the majority of patients who attended their schedule appointments using a dedicated clinic staff member to identify patients and deliver two decision aids in clinic. A significant proportion (56% WLS, 32% PSA) completed viewing of their assigned decision aid, each of which was greater than 30 minutes in length. The mean viewing time of approximately 13 minutes for those who watched less than the complete video reflects the average waiting time in our clinic. The need to target decision aids to specific eligibility criteria limits the efficiency of the clinic staff when a significant portion of time is spent reviewing patient records. We were able to improve the efficiency by using an automated system; however, neither the manual or automated system resulted in perfect fidelity, with 10% of approached patients reporting that they were not eligible. We estimated that the average cost to implement decision aids using a designated clinic staff is $13 per decision aid viewed.
Previous studies have evaluated models of decision aid implementation in primary care practices. In general, the use of the materials when mailed has been limited. Previously, we found that 25% of respondents viewed a decision aid about Colorectal Cancer (CRC) screening when it was mailed unrelated to an office visit [5
]. Brackett and colleagues using questionnaire returns as a proxy for the number of decision aids viewed estimated that between 24-36% of patients viewed decision aids when mailed or offered to be mailed a decision aid before an office visit [4
]. Frosch and colleagues demonstrated a 41% viewing rate in private practices in underserved areas using similar methodology as we used in this study [3
]. Although we found a higher proportion of patients viewed the targeted decision aids than in our previous study, viewing differed by topic. It should also be noted that we did not directly compare the in clinic delivery strategy with mailing out decision aids; therefore, within the RE-AIM framework [11
], questions remain in regards to the reach and efficacy of these two models and additional research is needed to make this direct comparison.
In this study, we assessed both full and partial viewing of the decision aids. An ideal approach would provide sufficient time to complete viewing of decision aids so that patients would have the full benefit of the decision aid while interacting with their provider. While no formal data was collected on the reason for incomplete viewing, time was a clear barrier to identification and delivery in the current study. As the mean viewing time for partial completions was considerably shorter than the duration of the decision aid, it may be beneficial for the length of the decision aids to be condensed to a viewing time of 10-15 minutes. Limiting decision aid length to 10 to 15 minutes would likely have less effect on visit length. Our practice has worked diligently to decrease patient wait times, but found an increase of approximately 3-5 minutes over the duration of this project, which may be attributable to decision aid use. Efforts to modify patient expectations about decision support may be one approach to address time barriers and patient satisfaction. Interestingly, a greater proportion of the WLS decision aid population viewed the complete decision aid, despite it being the longer of the two videos, leading us to believe that topic may influence complete viewing. Patients in the present study may have been more willing to complete viewing of the weight loss surgery videos for several reasons including, but not limited to, lack of previous exposure to the information provided in the DA, and/or perhaps they found the information in the WLS DA more salient. Previously, Tingen, et. al, showed that, dependent on demographic factors, some men bypass education on PSA screening based on their previous screening practices and known familial risk factors [13
An additional barrier to implementation was the time invested in patient identification. This process of targeting identification and delivery to only those meeting specific eligibility criteria limits distribution efficiency; though it is necessary for specific decisions such as weight loss surgery. This study illustrates the variability in efficiency when comparing a chart review method to an automated medical records system. For clinics with limited resources or without electronic medical records, manual review of medical records is feasible with one staff member, if a singular decision aid is being distributed in-clinic. An automated system may improve efficiency when implementing multiple decision aids. The goal of this study was to test the feasibility of a staff member identifying and delivering targeted decision aids in a primary care clinic. We did not want to impede patient flow or viewing of decision aids with extensive data collection, which could have influenced our results. Consequently, desirable measures are lacking with this pragmatic approach which could be pursued in future studies. For example, we were not able to explore patient attributes or attitudes that may have been associated with complete viewing of DAs. In addition to patient attributes, we did not collect information on confounding variables of patient flow such as the type of patient visit, MD patient load, scheduling punctuality, patient wait time, appointment conflicts, etc. Although we did not formally assess physicians’ attitudes, we had almost complete provider participation from the 16 full time equivalents. Out of our 85 part time providers, one physician refused to allow his patients to participate. Anecdotally, the physicians concurred that the decision aids were worth a slight increase in patient wait time that occurred during while this project was ongoing.
The project was designed to test the feasibility of decision aid identification and delivery by a designated staff member for patients visiting an academic internal medicine clinic. Dependent on the identification and delivery method selected, the results may not be generalizable to other practice sites, given the diversity of primary care practices. For example, smaller practices may have less difficulty with this model based on structural design pragmatics of their clinical space. Furthermore, the time estimated to identify potentially eligible patients may not be relevant for other medical records systems.
Our findings suggest that using a dedicated clinic staff member to facilitate in-clinic viewing of decision aids may be feasible for identification and delivery of one or two decision aids in primary practice. Manual review of medical records may limit the efficiency of decision aid identification and delivery for targeted populations unless an automated identification system is used to improve efficiency. Average viewing time suggests that 10 to 15 minutes may be the maximum time available for in-clinic decision aids use in primary care practices.
In clinic distribution requires an electronic health information system to efficiently identify potentially eligible patients and a staff member dedicated to distribution. To facilitate in clinic viewing and reap the benefit of decision aid use prior to the clinic visit, brief decision aids (less than 10 minutes) are needed. Practices may want to consider coupling decision aid delivery with health coaching in order to increase the efficiency of the model using overlapping staff.