This study examined physicians' perceptions regarding AMIE's impact on health outcomes and healthcare costs. Physicians' opinions of AMIE varied in terms of its potential clinical benefits, AMIE's effect on workflow, and efficiency. Overall, study participants expressed satisfaction with AMIE and would recommend it to other providers. Emergency department (ED) physicians who participated in focus-group sessions were more likely to report that the HIE had positive clinical benefits. In contrast, non-ED providers generally did not perceive benefits in healthcare provision mainly because of the limited data available and their inability to find patients in the system. One possible explanation for ED physicians extolling benefits of AMIE is that ED physicians, in general, encounter a high volume of patients per day.12
This might also relate to a higher probability of finding patient information in the HIE as compared to other practitioners. Additionally, ED physicians commonly provide care and make decisions with significant information gaps.12
Thus, any improvement in data availability at the time of care is perceived as an important impact on healthcare quality and safety.
Results from this study are consistent with similar studies that have assessed emergency physicians' opinions of health information exchange.12
A survey of 216 ED clinicians found that 85% of respondents perceived that it is difficult or very difficult to obtain relevant outside clinical information at the point of care, and 97% expressed that HIE would benefit the delivery of clinical care. In addition, 88% of respondents perceived that HIE would increase the efficiency of clinical care, and 78% thought HIE would decrease the number of tests ordered.12
Another case study that examined the factors that motivated or prevented small primary care practices from participating in an HIE revealed that anticipated cost savings, quality, patient safety, and efficiency were motivators. In contrast, limited technical infrastructure and support were identified as one barrier to participating in HIE.15
There are several possible explanations why some physicians did not perceive that AMIE had a positive impact on their efficiency. One reason is that users were in a technology-learning phase.31
Thus, they were inexperienced with this new technology, and so more time would be needed to acquire the necessary computer skills and overcome this barrier. This was confirmed by the finding that participants who used other types of information technology, such as electronic medical records, felt that using AMIE was not a technological barrier. In fact, these providers felt AMIE was easily incorporated into their workflow. Also, additional time related to computer use can be perceived as a decrease in efficiency, as shown by Overhage et al
Another possible explanation for the difference in opinions of AMIE impact on efficiency may be that some users had the support of medical assistants to search for and retrieve patient information in AMIE. If information was found, the assistants printed and provided it to the physicians before he/she encountered the patient. Therefore, for many providers, the process of obtaining information did not appear to impact their routine directly. These results are consistent with results from a case study that evaluated the perspectives of clinical and administrative leaders in small ambulatory practices regarding HIE. This study found workflow issues were significant barriers to HIE adoption.14
Results from this study suggest that AMIE accomplished its objective of being a usable, user-friendly tool and that HIE functionality was not a technical barrier. Comments from the focus-group participants suggested that AMIE design and user support were outstanding. In contrast, the most frequently repeated negative feedback from focus-group participants was the limited data available. For example, the AMIE medication data included only Medicaid patients (from pharmacy benefit manager (PBM) claims data), laboratory results were included from one of the two large laboratory providers in the state, and discharge summaries were included from three participating hospital systems.
Results from the focus-group sessions suggested that patient type and practice environment impacted information retrieval and therefore clinical use of AMIE. As mentioned earlier, some respondents had more success than others finding their patients in the AMIE HIE system. In particular, the group of physician practicing in a homeless center had the most difficulty. Several factors could have contributed to this. First, the homeless center supplies only medication samples, and dispensing records for these medications were not included in the PBM claims data because AHCCCS did not pay for the medications. Since AMIE medication data were aggregated from pharmacy claims data, dispensing records for these medications were not included in AMIE. Second, the homeless patients rarely obtained prescriptions filled at a pharmacy, and therefore no other medication history information was supplied to the information exchange. In addition, focus-group participants from this homeless center suggested that their patients often purposely provide incomplete or inaccurate names for various reasons, thus inhibiting HIE patient match.
Focus groups provided information to help evaluators and designers to understand physicians' adoption behavior and attitudes, by identifying to whom the HIE was being most and least useful, and to explore possible reasons for this. During this study, focus-group discussions helped to clarify system misunderstandings and identify system deficiencies. Thus, opinions expressed by focus-group participants were valuable because they helped the AMIE team continually improve implementation of the utility. Moreover, focus groups allowed management of users' expectations. It is important to manage expectations during pilot studies because unrealistic expectations of system users can result in disappointment that may affect system adoption and use.33
The focus-group discussions allowed early adopters to share their experiences with less frequent HIE users, thus generating an environment of technology diffusion. In other words, the focus groups served as an educational space where the retelling of a respondent's successful experiences with the HIE may have helped improve the adoption of participants who exhibited less frequent HIE use. In addition, the continued contact with users during the focus groups motivated participants to use the HIE. Specifically, the medical director of the project maintained personal contact with users which motivated participants to use the HIE. Therefore, it is important to recognize that successful implementation of an HIE requires an ongoing organizational commitment to maintain system use.
Results from this study suggest that it may have been premature to assess user perceptions of the HIE's effect on dimensions such as clinical decision-making, quality of care and healthcare costs after only 3 months. Users presented examples of AMIE's impact on healthcare costs during the focus groups (eg, admissions prevented, decreased laboratory testing). Nevertheless, it is possible that impact on healthcare costs was underestimated because laboratory reports and discharge summaries were uploaded to AMIE progressively during the evaluation period. That is, not all data were available during the full evaluation period. Health information technology research suggests that it is best to wait a minimum of 1 year after a technology is in operation before evaluation.33
Therefore, future studies ascertaining the clinical impact of health information exchange should be conducted when the system has sufficient technical capabilities, and enough time has passed for users to overcome the technology-learning phase.
Another lesson learned from this investigation is that for a short-term evaluation such as this one, monthly focus groups were not necessary. After the second session, no new information was presented by participants, thus indicating that information saturation was achieved. Perhaps a pre- and postmeasurement design would have been sufficient to gather data.
It is important to recognize the limitations of this study. The respondents volunteered for participation, and this could bias the results. For example, the majority of respondents had previously used electronically medical records, and research suggests that clinicians are more likely to adopt a new technology if they have had previous experience adopting a similar technology such as electronic medical records.34
Additionally, the fact that most of the respondents in this study had an academic appointment may suggest that the respondents may be more likely to be engaged in research and take part in innovations. A bias toward socially desirable responses is also possible, given the organization director's endorsement of this study. The results of this study are limited by the response rate and to the clinical domain of the AMIE participants; and the population of this study was small, so the results cannot be generalized.