These results shed light on the current state of ambulatory safety evaluation. Most PSPs have few or even zero studies evaluating use in ambulatory care. Even for PSPs with a moderate evidence base, if the experience of hospital-based PSPs is any guide, there will still be a host of context and implementation issues that remain and require additional study. The combination of input from KIs and the literature scan demonstrates that, although there is some overlap in the hospital-based and ambulatory safety topics, the ambulatory environment has many distinct safety issues, most notably medication safety, safety culture, transitions among providers in ambulatory settings, and timely and accurate diagnosis, which includes issues arising from referrals from one provider to another, and management of test results. While the labels given these safety issues are similar or even identical to some hospital safety issues, the targets, the time course, and types of interventions may be substantially different.

In terms of medication errors and adverse drug events, the results of our literature scan showed a few PSPs, such as e-prescribing and pharmacist-led interventions, have a moderate evidence base. Practices such as pharmacist-led medication reconciliation and review of high-risk medication use^{42, 43} are two evidence-based solutions that have a persistent implementation gap; this is worthy of further study. However, KIs observed that current health IT solutions do not adequately support medication safety, and echoed earlier calls for large-scale studies in this area,⁹ particularly in real-world implementation and examining unintended consequences such as alert fatigue.

Patient safety culture seems to be an area of challenge for ambulatory safety. As an example, KIs described a general acceptance of sub-optimal results reporting and tracking. Reporting systems for errors are under-developed, and it is not clear what feedback results from such systems. It seems the fear of speaking up persists as well. Notably, KIs did not bring up or discuss widely used safety culture surveys or team training. There is a need to elucidate effective strategies to enhance ambulatory safety culture,⁴⁴ because the successful implementation of all ambulatory safety interventions requires a strong safety culture as a foundation.

While there was clear consensus about the importance of patient engagement, concrete best practices did not emerge from either the literature scan or interviews. Another key consideration in patient engagement is patient characteristics, such as educational attainment, health literacy, English proficiency, cognitive impairment/ disability, and health care access, as social determinants of health which are likely to affect ambulatory safety. However, there are few data to support these perceptions or inform ambulatory safety interventions.

The term “transitions in care” has come to imply post-hospital discharge, but the KIs identified many other unsafe transitions: among ambulatory providers, between ambulatory providers and the emergency department, between health care and social services, and managing pediatric to adult transitions for the chronically ill. Most of these transitions have not been the subject of a single PSP evaluation.

Interviews also emphasized the need for more research on diagnosis, including epidemiologic approaches to capture the incidence of diagnostic errors in the population, as well as in-depth behavioral and cognitive studies to improve the diagnostic process, as described in the 2015 IOM report on improving diagnosis.⁴⁵

In addition to the specific safety issues, the literature scan and KI interviews revealed both the possible safety advantages and many unintended consequences of health IT, as with a prior expert panel on ambulatory safety.⁹ Some advantages include safety improvements from computerized physician order entry in medication prescribing and medication list maintenance. KIs perceived advantages such as widespread information-sharing through health information exchanges as theoretical rather than actually functioning today. Many KIs mentioned struggles with poorly designed, expensive, cumbersome electronic health records as a source of physician burnout, which they see as a safety hazard. Health IT implementation emerged as a needed area of study, because of the concerns about alert fatigue and “workarounds” that may worsen safety. The entire workflow of ambulatory care is being reshaped by EMRs and health IT; we need more discussion of the negative and positive actual and potential impacts on ambulatory errors.

There are some limitations to our approach. We identified 8 KIs; although we felt we reached thematic saturation with this group, it is possible that results would have changed with inclusion of additional patient safety leaders, though this remains a small field. We performed a literature scan rather than a full systematic review, because of the sparse literature in this area and the desire to address a large number of applicable PSPs.

Both the literature scan and the KI interviews point to significant knowledge and implementation gaps. Current evidence does not permit the quantification of harms from ambulatory safety issues; the magnitude of problems remains unknown. Other than the medication-related and care transition practices mentioned above, few of the PSPs have significant evidence in ambulatory settings, and fewer still have been widely implemented. The KI interviews highlighted the lack of large-scale epidemiologic studies and multi-center interventions across all topics. Epidemiology using an injury prevention perspective rather than an error-based framework was also felt to be lacking.^{46, 47} We did not identify literature indicating specific organizational models of care to support ambulatory safety, although our KIs suggested that patient-centered medical home and team-based care models may hold promise. The PCMH model holds appeal in part because KIs felt it conceptually supports safety better than the current fee-for-service structures. In addition, care coordination with a multidisciplinary team was seen as an asset for the PCMH compared with traditional ambulatory practice.

These results inform a significant future research agenda. First, measurement development efforts are needed, directed at each of the safety topics the KIs focused on: medication safety, diagnosis, transitions, referrals, and testing. There should be multiple measures that can serve as outcomes for research, and there should be efforts made to support development of performance measures. Measures are critical for the quantification of harms. In turn, the quantification of harms will allow the prioritization of ambulatory safety issues. Second, research in patient safety needs to incorporate multiple disciplines with appropriately diverse methods. This would inform non-“error” based approaches to ambulatory safety. KIs felt that more rigor needs to be brought to the science of intervention development before those interventions are evaluated in well-designed hypothesis-testing studies. There should also be further emphasis on implementation studies to understand what promotes implementation, sustainment, and spread of successful ambulatory safety practices. Third, it is clear that there is a need to invest in improving the safety of the diagnostic process. The IOM report on diagnosis mentions several evidence-based strategies such as cognitive training and systematic feedback on diagnostic accuracy, which could be tested and implemented on a larger scale.⁴⁵ Several KIs emphasized the need for collection of primary, descriptive data in order to understand diagnostic accuracy. Fourth, epidemiology of adverse events in various types of ambulatory transitions warrants further study in preparation for developing effective patient, provider, and system-level interventions. Fifth, health IT is reshaping the workflow of ambulatory care, and research is needed on how this can enable PSP interventions and act as a barrier to safe practice; and ways to increase the former and decrease the latter. Finally, there are a host of safety culture measures, tools, leadership efforts, and interventions that have proliferated, but concerns with safety culture remain. This suggests the need for long-term, large-scale efforts not only to characterize, but improve safety culture. One approach to enhancing safety culture may be to develop interventions to treat and prevent health care provider burnout.

Because our results demonstrate multiple possible areas of focus in ambulatory safety, prioritization via a Delphi panel or process could help with a formal research agenda. Taken together, our results suggest the need for large-scale, prospective descriptive and intervention studies across multiple ambulatory environments in order to establish real-world evidence to support safer care in ambulatory settings.