Annotated Bibliography: An Update to “Understanding Ambulatory Care Practices in the Context of Patient Safety and Quality Improvement”

Overview of Safety Issues in the Ambulatory Setting

Despite the increase in attention to outpatient safety, recent studies have shown that significant areas for improvement remain. The following studies provide an overview of the key issues in ambulatory safety and give recommendations for improved practice or further areas of study.

Gardner examined and discussed (1) factors that contribute to errors and patient harm in ambulatory gynecology settings, (2) differences in outpatient versus inpatient errors, and (3) strategies to improve patient safety in the ambulatory setting. The author identified many human factors that could lead to issues in patient safety, including perception, teamwork, physical environment, physical demands, device/product design, process design, and mental workload/cognition. To combat human error, Gardner suggested collaboration with human factors professionals to help analyze barriers to safety and implement realistic solutions to improve office safety. With regard to differences between outpatient and inpatient error, the author identified diagnostic-related errors as being the most problematic in the outpatient setting. Multiple strategies were recommended for increasing patient safety in the ambulatory setting, including providing consistent leadership, prioritizing patient safety, creating a culture of patient safety, strongly committing to improvements in quality of care, and establishing structures to ensure office safety.

Hatoun et al conducted a systematic review of the literature on patient safety measures in primary care within the context of quality improvement (QI) and found a total of 21 articles that discussed 182 safety measures. These measures were further categorized into 6 dimensions of outpatient safety, including medication management, sentinel events, procedures and treatments, laboratory testing and monitoring, care coordination, and facility structures/resources. The authors concluded that outpatient safety measures are relatively few in comparison to inpatient safety measures. In addition, many outpatient safety measures need to be validated. They not only support conducting a large national epidemiological study of ambulatory safety, but also advocate for more research on how to define measures for tracking patient safety in the outpatient setting.

Incidence of Patient Safety Issues in the Ambulatory Setting

In the past decade, there have been considerable research efforts to understand the nature and occurrence of patient safety incidents in hospitals as well as potential interventions to reduce these incidents. However, significantly less research exists on the epidemiology of patient safety incidents in the ambulatory setting. The following references characterize the scope and nature of patient safety incidents in the ambulatory setting.

Panesar et al conducted a systematic review of the literature to explore how often safety incidents occurred in primary care and how many incidents were associated with patient harm. Ultimately, 9 systematic reviews and 100 primary studies with safety incidents between January 1, 1980, and July 31, 2014, were included in the review. The range of incidents in the included studies was from <1 to 24 patient safety incidents per 100 consultations. The median was 2 to 3 incidents for every 100 consultations. Among these incidents, approximately 4% were associated with severe harm. Severe harm was defined as incidents that produced permanent harm, including death, disability, or long-term physical or mental health consequences. Incidents that were related to diagnostic and prescribing errors were most likely to result in severe harm. Based on the results of the review, the authors surmised that although safety incidents are relatively common, most do not cause serious harm to patients.

Sarkar responded to Panesar et al in an editorial that praised the review’s contribution to the field while recognizing the limitations of its methods. The author notes that the limitations of the study exemplify the typical challenges faced when studying patient safety in ambulatory settings. One limitation of the Panesar et al review was its definition of incidents, which only included events that occurred when wrong things were done. Sarkar argues that incidents also should include omissions, events that occur when the right thing was not done. Because of this restricted definition, Panesar et al likely underestimated the incidence of events. Another limitation of the review was that it included primarily studies using record review to determine the number of incidents. Again, Sarkar argued that record review likely underestimated the number of incidents because of poor documentation. Finally, Sarkar noted that the definition of harm in outpatient settings is ambiguous, another limitation of the review. All these factors led Sarkar to note that the recorded incidents are just the “tip of the iceberg.” Sarkar then concludes by offering a number of recommendations, including that the field should reach consensus on defining events, harm, and preventability and a robust investment in primary care patient safety research and QI efforts.

Safety Culture

Development of and commitment to an institutional safety culture is one strategy that has been identified to improve patient safety in ambulatory settings. Miller et al demonstrated how building a safety culture encouraged the reporting of more events and near misses. The authors performed a retrospective observational cohort study across 35 different ambulatory practices to explore the relationship between safety culture and voluntary event reporting. Both overall and domain-specific safety culture were assessed at each of the different practices by a validated instrument (Safety Attitudes Questionnaire [SAQ]), which was administered to staff. Safety culture scores and event reporting rates were compared between the 4 months before and after survey administration. SAQ participation was 87%. The authors found a wide range of event reporting rates between practices, with rates between 0.00 and 6.99 events per 1000 total visits per month. Notably, overall and 4 domain-specific safety culture scores (teamwork climate, safety climate, working conditions, and perceptions of local management) were positively associated with practice event reporting rates. For instance, for every 1 percentage point increase in overall culture score, there was a 1.9% increase in monthly safety reports. The findings contribute to the literature by indicating that event reporting may be higher in practices with strong safety cultures. This debunks the misconception that fewer reports indicate fewer incidents and better overall safety.

Measures and Tools

Douglas et al conducted a cross-sectional survey study across 21 different pediatric and adult ambulatory practices to assess the relationship between items in the Model for Understanding Success in Quality (MUSIQ) questionnaire and defined objective measures of QI. The MUSIQ questionnaire is a commonly used tool to identify contextual factors that contribute to success in QI projects. In this study, the primary independent variables were each contextual factor’s MUSIQ item scores. The primary outcome (dependent) variable was “objective” QI project success. Objective success was defined as having 75% or more of patients with a glycated hemoglobin A1_C value <8% (for adult sites) or having 80% of patients who are 18 to 35 months old be up-to-date on all combination vaccines (for pediatric sites). The secondary outcome for the study was the difference in the mean percentage of patients meeting either definition of “objective” success between the current and last QI time periods. The authors found largely no associations between MUSIQ contextual factors and objective QI outcomes for both primary and secondary outcomes. Their findings conflict with prior studies that suggest a relationship between MUSIQ factors and subjective QI outcomes. Because of these findings, the authors suggest that there is no combination of MUSIQ factors that can be directly associated with QI success. Furthermore, contextual factors promoting QI success may be complex and dependent on the local environment. The authors recommend use of the MUSIQ tool before implementation of QI projects, so that sites may focus on specific contextual factors in order to increase their success.

Recent innovations in QI and safety have not only included tools to measure project success, but also models to foster successful QI and safety efforts. Kravet et al used the fractal model, an approach with vertical and horizontal accountability, to provide an oversight structure for quality and safety efforts at all Johns Hopkins Medicine ambulatory practices. In this model, the Ambulatory Quality Council, led by a physician and nurse dyad, defined expectations at all levels of ambulatory practice. The authors found that with implementation of the new governance structure, Johns Hopkins Medicine practices experienced improvements in patient safety/risk reduction, patient care/experience, externally reported quality measures, value, and even Medicaid value-based purchasing metrics. Based on their results, the authors suggest that the fractal model may help standardize quality and safety efforts at even broader levels (ie, nationally, internationally).

Team Training in Health Care

Team training is another strategy that has been used to improve patient safety in the ambulatory setting. Pitts et al used the Comprehensive Unit-based Safety Program (CUSP) in an academic primary care setting to identify safety concerns and set initial safety priorities within their practice. CUSP is a team-based approach that uses online trainings combined with staff surveys to identify and then prioritize safety concerns. The authors also used a validated SAQ that was administered pre and post CUSP intervention to assess whether CUSP implementation affected safety climate and teamwork. The 3 most common safety concerns were medication, diagnostic testing, and communication. As a result, the team set goals to improve communication and infection control. Although not statistically significant, the authors also found that there were improvements in safety climate on the post-CUSP survey, including attitudes toward reporting. The authors suggest that CUSP is therefore a good tool to improve safety climate as well as identify and prioritize safety concerns.

Patient-Centered Care: Engagement and Satisfaction

Growing evidence has shown patient engagement in delivery of care improves safety. Patient engagement also has been associated with lower costs, improved outcomes, and even better experiences of care. Multiple organizations, including The Joint Commission, AHRQ, and the World Health Organization have therefore recommended the assessment of and use of patient engagement in safety initiatives. ⁷ Recent patient-centered safety efforts have focused on health technology or novel communication strategies. In a study by Bell et al the impact of patient engagement with health data was assessed using OpenNotes, a reporting tool that allows patients to give feedback on notes, potentially increasing safety. The authors found that patients and care partners noted safety concerns in 23% of reports. These findings often would lead to changes in care or documentation. The authors conclude that patient engagement with reporting tools, such as OpenNotes, may help promote safety in ambulatory settings.

Similar to Bell et al, Khoong et al showed how technology is used to deliver patient-centered care. Khoong et al used a mixed-methods study to implement patient-centered medication labeling (PCL) at 4 ambulatory care pharmacies in a safety net health system. The authors assessed PCL adoption rates and explored factors contributing to implementation success by auditing prescription labels at each site and conducting interviews with key informants. Three pharmacies had success implementing a PCL format, with more than 85% of audited prescriptions converted to a PCL format. One site converted less than 25% of audited prescriptions to a PCL format. Factors contributing to success included close communication with prescribers, adaptable software, and/or automation for PCL conversion. Technology, especially adaptable software, was especially critical for PCL implementation. The authors noted that a major barrier to PCL implementation at one site was the inability of its software to automate changes. This result has implications for management of vulnerable populations. Patient-centered innovations that use health technology are often unevenly distributed across systems, which ironically may contribute to greater health inequities.

Patient-centered safety innovations also have focused on fostering provider partnerships with patients and their families. Recently, the National Patient Safety Foundation recommended that increased safety required a “systems approach”—one that involves partnering with patients and their families. In their commentary, Roter et al hypothesize mechanisms by which partnering with patients and families improves quality and safety in the ambulatory setting. They also summarize key findings from experience with a communication intervention for families and patients to partner with providers. Family members improved 6 different aspects of ambulatory safety and quality of care (as outlined in the Institute of Medicine’s report Crossing the Quality Chasm), including safety, effectiveness, timeliness, efficiency, equity, and patient-centeredness. The authors hypothesize that families increase safety by minimizing medication mismanagement and potential adverse drug events (pADEs) in a variety of ways, both at home and in office settings. Roter et al also reported success with a trial of a communication intervention, the Patient-Companion Partnership Tool. The intervention group had improved communication during office visits from all perspectives, including patients, their families, and providers. Based on the success from this intervention, Roter et al suggested that more communication-related interventions be tested in ambulatory settings.

Care Coordination and Continuity of Care

Poor care coordination and continuity of care between the inpatient and outpatient settings has been associated with reduced patient safety. A number of issues have been identified that contribute to reduced patient safety during transitions of care (TOCs). These include ineffective team-based care, inaccurate medication reconciliation, poor communication between inpatient and outpatient providers, and nonstandardized handoff protocols, among other issues. Armor et al focused on errors in medication reconciliation and conducted a study to characterize ADEs, pADEs, and medication discrepancies occurring during care transitions from the hospital to outpatient primary care follow-up. Study subjects were adults who were recently discharged from the hospital and seen by pharmacists for medication reconciliation. The authors found that 100% of participants in their sample experienced ADEs or pADEs and 81% of participants experienced at least 1 medication discrepancy. These results suggest a need for strategies to increase medication reconciliation during care transitions. The authors also emphasized the role of strong communication skills between patients and health care teams to prevent errors in medication reconciliation.

Rattray et al used qualitative research methods to examine communication barriers and facilitators between inpatient stroke/transient ischemic attack and primary outpatient care providers at a Veteran Affairs Medical Center. The study involved interviews with 21 providers and subsequent thematic analysis of interview content. Communication facilitators included thorough, consistent medication and treatment plans, standardized discharge documentation, and use of multimodal communication. Challenges to communication included rotating providers within a teaching environment and differences in local discharge practices. Because of pressures in acute inpatient settings and ambiguity with handoffs, inpatient providers may assign lower priority to discharge communication. In addition, the authors propose that the ease of electronic medical record templates produces a “checkbox mentality” for discharges, which further reduces the quality of communication with outpatient providers. Ultimately, they conclude that greater awareness of the challenges and opportunities in TOCs may help foster a culture that considers both inpatient and outpatient perspectives.

Rochester-Eyeguokan et al conducted a scoping review to summarize the different practice models that deliver TOCs in the United States. They also used the review to characterize best practice models. The authors found that 188 articles fit their inclusion criteria. Although best practice TOC models were typically multimodal interventions delivered by multidisciplinary teams, the authors noted that there was no one-size-fits-all model. Furthermore, despite an abundance of articles on TOC interventions, the authors identified multiple gaps in determining facilitators of effective TOC interventions. One such gap included more research on the role of medication reconciliation in TOCs. They encouraged more research on TOC interventions in a variety of settings as a means of improving the current status.

Medication Safety and Electronic Prescribing

Medication-related errors are the leading cause of ambulatory safety incidents, and health information-related technology, specifically electronic health records (EHRs), represent a potential tool to help mitigate these errors. Feng et al performed a systematic literature review to examine the methods and role of EHRs in identifying and evaluating ADEs in the ambulatory setting. The majority of the 30 articles that were included in the review used only EHRs as a data source to identify ADEs. Two articles used EHRs as a means of data and to give providers decision support when placing orders. The authors noted that the “gold standard” for many studies using EHRs was to use a retrospective, manual chart review. Although this method is helpful for measuring ADE incidence, it does not help capture preventable ADEs. Instead, the authors suggest that electronic-guided and triggered searches of the EHR in real time may better prevent and manage ADEs before they have a chance to occur.

Kjos and Bryant used a social network analysis to describe the process and methods of communication for medication management of patients on warfarin in the outpatient setting. ADEs can be prevented by improving communication between providers. As a result, this study uses a social network analysis that describes and quantifies communication patterns in the ambulatory setting to identify potential areas of error prevention. Electronic medical records were used for data collection and information collected included which providers were involved as well as their frequency and type of communication with the patient. Overall, the study analyzed data on 16 patients in one internal medicine clinic over 6 months. The study’s results showed that most communications were unidirectional and between players from different groups. A majority of communications were typically from nursing staff to patients. Furthermore, communications were characterized by their impact on patient safety. When this was done, moderate-impact communications were found to occur normally between providers, and pharmacists were found to be important gatekeepers in moderate-impact communication. The findings show that communication from pharmacists plays a crucial role in medication management, especially with regard to patient safety.

Although EHR use may have its advantages, its use in the ambulatory setting may adversely affect the quality of patient-physician interactions. Matta et al used a mixed-methods observational study to create a conceptual framework for how physicians balance EHR demands and patient communication during medication reconciliation in safety net primary and specialty outpatient practices. Patient-provider encounters were videotaped in 5 different academic public hospital clinics. The authors analyzed the proportion of medications addressed in the encounters and coded the time spent on different tasks during medication reconciliation as “multitasking EHR use,” “silent EHR use,” “non-EHR multitasking,” and “focused patient-clinician talk.” They also analyzed communication patterns to create a framework. Ultimately, the results were that nearly half of the time spent on medication reconciliation involved multitasking EHR use. The authors acknowledge that multitasking may be risky because of the cognitive demand it places on physicians. Physicians may turn to silent EHR use and focused patient interaction to reduce some of this risk. Finally, the authors point out further directions for research, including patient perspectives on EHR multitasking, clinical outcomes related to EHR multitasking, and potential safety interventions for EHR use during medication reconciliation.

Technology, such as clinical decision support tools (CDS) embedded in EHR systems, can play a role in reducing the number of patients who receive potentially inappropriate medications (PIMs). CDS interventions have been shown to reduce prescribing of PIMs in the elderly population, who are particularly vulnerable to adverse drug side effects. Ruhland et al conducted a quasi-experimental study to evaluate the effects of a CDS tool on the prescribing patterns for glyburide, a PIM in elderly patients >65 years of age. The CDS tool involved an alert to providers when glyburide was ordered for those aged >65 years because of greater risk of hypoglycemia in the elderly. Outcomes included glyburide orders, given as percentage of total oral diabetes medications ordered (primary outcome), and response to CDS tool alerts (secondary outcome). Outcomes were measured both pre and post CDS intervention to see if there were changes in glyburide prescribing patterns. Ultimately, postintervention glyburide prescriptions fell significantly compared to the preintervention period. From these results, the authors suggest that CDS tools can be used in ambulatory settings to improve medication safety, but also recommend testing them with provider education to optimize CDS tool use.

Although electronic prescribing (e-prescribing) has increased efficiency in ambulatory care settings, a variety of quality and safety issues with the technology remain. Yang et al identify prescribers’ directions to patients as one of the most “quality-sensitive” issues with e-prescribing and conducted a study to evaluate the quality and variability of free-text prescriber direction phrases in e-prescriptions. The authors performed a retrospective qualitative analysis of 25 000 e-prescriptions generated from different EHR applications from ambulatory prescribers across the United States. Content from prescriber directions was organized based on an expert-guided classification scheme. Prescriber directions also were analyzed for “quality-related events”—content that might create problems with accurate and unambiguous interpretation by receiving pharmacists. The study found extreme variation in prescriber directions despite conveying the same content. More than 50% of phrases were organized into just 25 distinct concepts. They also found that more than 10% of e-prescriptions contained quality-related events, indicating the need for greater quality control and standardization of prescriber directions. The authors conclude with a number of recommendations, including optimizing prescriber direction creation tools, implementing prescriber training, and greater standardization of formatting among prescribers and pharmacy systems.

Diagnostic Errors, Test Result Management, and Medical Error Reporting

Delayed or missed test result follow-up has been linked to poorer health outcomes and represents an important safety issue in the ambulatory setting. The process of test result follow-up is complex, resulting in difficulties with improvements in follow-up rates. Ai et al examined safety culture around test result follow-up with qualitative and quantitative patient safety and quality of care data at an academic center. In order to obtain multiple perspectives on test result follow-up, data were taken from 5 sources, which included patient and family complaints, 2 national surveys on safety, safety reports, and provider response times to test results in the EHR. The authors measured how providers and patients gauged the frequency of providing timely test results, patient satisfaction correlation with provider response time to test results, as well as themes in patient complaints and safety reports. One key finding was that as test result response time decreased, patient satisfaction increased. The authors also were able to use themes from qualitative data to find potential explanations for inappropriate test result follow-up.

Despite being one of the leading causes of medical error in ambulatory settings, diagnostic error remains largely understudied in part because of challenges measuring it. Al-Mutairi et al recognized the need for more tools to identify diagnostic error and created and piloted an instrument, The Safer Dx Instrument. This study built on the authors’ past work in which diagnostic error was measured subjectively by independent physician reviewers. The Safer Dx Instrument consisted of 11 questions on diagnostic processes in provider-patient visits and 1 main outcome question to identify diagnostic error; it was tested against samples of medical records with and without diagnostic errors. Overall accuracy of the instrument was found to be relatively high at 84%. Although the authors acknowledged that their tool did not capture all errors, they noted that it was a good “first step” to standardizing detection of diagnostic errors and recommend its use with trigger algorithms to potentially increase its effectiveness in practice.

The majority of research on diagnostic error in the ambulatory setting pertains to error caused by physician cognition or ergonomic or systems factors in a physician’s work environment. The role of patients in the diagnostic process is less investigated, but patient involvement may reduce error because patients have unique knowledge of themselves and their health care experiences. Blease and Bell used results from patient-reported surveys about patient-shared visit notes to recommend 3 ways that sharing visit notes may help reduce diagnostic error. The authors suggest that sharing visit notes helps reduce diagnostic error by increasing timely follow-up for tests, results, and referrals; by more frequently identifying errors in documentation; and by enhancing the doctor-patient relationship. The authors also identify areas for further research and propose potential disadvantages of sharing visit notes.

Ambulatory safety nets (ASN) are innovative organization-based interventions designed to improve patient safety with regard to diagnostic errors. ASNs take a multimodal approach by incorporating tools, reports, registries, and associated workflows to help manage abnormal results. Emani et al developed and evaluated 2 ASNs regarding delayed and missed cancer diagnoses at an academic medical center. The ASNs focused on creating safety nets for diagnostic errors related to colon and lung cancer, and each intervention had 4 parts, including report and registry development, collaboration between multidisciplinary teams, routine work group meetings to discuss progress, and patient surveillance and outreach. Effectiveness of each ASN was measured by EHR and registry data and was defined as the proportion of patients who were scheduled for or completed a colonoscopy after outreach (colon cancer) and the proportion of patients who were scheduled for or completed a chest computed tomography after appropriately being identified for follow-up (lung cancer). The authors found the effectiveness to be 44.0% and 56.9% for colon cancer and lung cancer ASNs, respectively. They noted that relative success with ASN implementation led to further funding for ASN maintenance and development of new ASNs at their institution. The authors also emphasized lessons learned and concluded by highlighting the need for more development of active electronic surveillance systems to reduce diagnostic errors.

Inadequate monitoring of high-risk patients is associated with greater patient harm. In particular, impoverished patients, similar to those served by safety net systems, are more vulnerable to lapses in monitoring and the resulting harms. McDonald et al used a novel integration of industrial and human factors methods to apply “journey mapping” and produce “design seeds” to improve patient monitoring, specifically diagnostic delays for cancer, in 5 subspecialty clinics at a large publicly funded health system. Clinic staff were interviewed to “journey map” or create a visual representation of each clinic’s process monitoring of high-risk patients, focusing on workflow and patient data tracking. Mapping allowed determination of common vulnerabilities across different clinics and development of potential solutions or “design seeds.” Ultimately, the authors found 45 vulnerabilities and 13 different design seeds to address these vulnerabilities, which were then ranked in order of importance by clinic staff. Highly ranked and therefore high-priority design seeds included keeping a list of patients who needed monitoring up to date, triggered notifications, customizing the patient list, and creation of patient registries.

Like McDonald et al, Patterson et al examined diagnostic delay from a human factors perspective. Specifically, they used a macrocognition theoretical framework adapted from engineering and design thinking to outline factors contributing to diagnostic delay and to propose methods to reduce these errors in practice. This work built on the concepts of patient safety termed Safety I and Safety II and, in particular, applied the concept of Safety II to improvements in diagnostic delays. ⁸ Although Safety I aims to reduce human error and variability by standardization, the concept of Safety II promotes increased “resilience” or human adaptability through a “supportive work system design.” From a macrocognition perspective, providers have 5 different complex tasks that may contribute to delays, including sensemaking, replanning, detecting problems, deciding, and coordinating. These tasks can be remedied by building a more supportive environment through the use of technology. The authors noted that technology could promote resilience strategies by alerting providers to follow-up activities in real time, decreasing documentation burdens, maintaining working diagnoses, and helping with pattern recognition in care delivery. They concluded by emphasizing that resilience building should not only refer to individual providers, but also involve larger interdisciplinary team efforts.

Diagnostic error remains a significant patient safety issue primarily because of difficulties measuring it. Singh et al recognized the need to develop more measures for diagnostic error and discussed both difficulties and opportunities for measuring it. The authors also offered an initial set of measurement concepts for future study. Although the authors discussed multiple factors that make diagnostic error difficult to measure, 2 key factors they highlight are definitional challenges with “accuracy” and “timeliness.” Making the appropriate diagnosis often relies on accuracy and timeliness. However, the authors argued that concepts of accuracy and timeliness are fluid, which in turn makes objective determination of diagnostic error difficult. They also offered 9 different measurement concepts for diagnostic error. Notable suggestions for measurement include organization-based root cause analyses of diagnostic error, training programs with use of simulations, judicious use of EHRs, and feedback mechanisms for providers. The authors concluded that these recommendations should be studied further and that organizations prioritize diagnostic error in order to see progress in this area.

Simulation Exercises in Patient Safety Efforts

Simulation training is a useful strategy to improve ambulatory safety as it has been shown to improve self-efficacy and objective procedural competence. Espey et al performed a pre- and postintervention study to assess the impact of a simulation-based curriculum on outpatient management of emergencies among residents and fellows in family medicine and obstetrics and gynecology. Trainees completed self-efficacy questionnaires and were videotaped during management of 3 different emergency scenarios both before and after implementation of a simulation-based curriculum. Outcomes measured were the difference between pre- and postintervention scores on an action checklist from 2 independent video raters (primary outcome) and the difference between pre- and postintervention self-efficacy scores (secondary outcome). All trainees experienced significant increases in self-efficacy and objective performance scores after simulation-based training. Specifically for self-efficacy, trainees improved in their confidence managing emergency scenarios and communication abilities during these situations. Importantly, the authors also observed that the 2 areas of improvement seen in this study—communication and systems-based practice—were initially highlighted as key factors to increasing patient safety in the NAM original report, To Err Is Human.

Innovative Future Strategies

Desai et al built and tested a multidisciplinary team-based ambulatory safety program at an academic center. The team comprised clinical, administrative, and patient safety members and developed a program to address safety reporting, safety culture measurement, medication safety, and test result management. Ultimately, the team was able to develop effective initiatives to address reporting, culture measurement, and medication safety. After testing different strategies, the authors reported feedback of up to 90% in the ambulatory setting. Safety culture was measured with the AHRQ Medical Office Survey, and the survey results were shared with different participating ambulatory sites. Rates of medication reconciliation in the ambulatory setting were found to increase significantly during the study. The authors were unable to assess test result management during the study and noted that they were expanding their program to include this.

As more care is delivered in outpatient settings, there is a greater need for graduate medical education in ambulatory quality and safety. Resident trainees have worse outcomes on ambulatory quality measures (AQMs), such as blood pressure control, than attending staff. Hwang et al acknowledged the need for trainee education in ambulatory safety and conducted a randomized trial among internal medicine residents to assess the effect of an audit and feedback intervention on AQMs. Residents were assigned to one of 3 different groups: a control group, a practice target group, or a peer comparison group. All groups received an email with the contact information of a population health coordinator and a list of AQMs (control). Ten AQMs were measured and based on typical issues in primary care, including measures related to diabetes care, hypertension, and cervical cancer screening, among others. The AQMs also included a composite quality score. The practice target group received individual AQM data compared to target AQMs for all primary care practices. The peer comparison group received individual AQM data compared to peer data from residents in the same year. AQM scores were then measured at baseline, 6 months, and 13 months. Hwang et al hypothesized that the peer comparison group would show the greatest improvement in AQMs but instead found significant improvement in the practice target group’s AQMs. Specifically, they found significant improvements in cervical cancer and colorectal cancer screening rates as well as composite quality scores from baseline to 13 months. Based on their results, Hwang et al recommended audit and feedback tools as a relatively simple means to improve ambulatory safety.

Despite a greater shift in focus to ambulatory safety, there are few validated and tested safety interventions, with a majority of interventions narrowly focused on specific safety domains. Schiff et al used a multimodal approach to determine the efficacy of a 15-month QI-based intervention randomized to 25 primary care practices. Practices receiving the intervention were exposed to learning networks, informational webinars, and in-person learning sessions and received coaching by improvement advisors. Coaching targeted improvements in “3+1” high-risk domains, including referral, test result, and medication management in addition to culture and communication issues. Chart review and patient and staff surveys at baseline and post intervention for control and intervention groups were used to assess intervention impact. Overall, the authors found modest but significant improvements post intervention for certain high-risk domains, including some measures of safety culture, referrals, and abnormal test result follow-up. They noted that the most salient result was improvement in follow-up of abnormal tests and argued the importance of this from a medicolegal perspective. The authors also noted multiple challenges in working with small and medium-sized practices and recommended additional resources, particularly improvement advisors, to potentially promote patient safety.