A resident-driven mortality case review innovation to teach and drive system-based practice improvements in the United States.

PURPOSE
Traditionally, Morbidity and Mortality Conference (M&MC) are forums where medical errors are discussed. Though M&MC can lead to identification of opportunities for system wide improvements, there is little in the literature to describe the use for this purpose, particularly in residency training programs. This paper describes the use of M&MC case review as a quality improvement activity that teaches systems-based practice and can engage residents in improving systems of care.


METHODS
Internal medicine residents at a tertiary care academic medical center reviewed 347 consecutive mortalities from March 2104 to September 2017. Residents used case review worksheets to categorize and track causes of mortality. The residents then debriefed with a faculty member. Selected cases were then presented at a larger interdepartmental meeting and action items were implemented. Descriptive statistics and thematic analysis were used to analyze the results.


RESULTS
The residents identified a diagnosis mismatch from admission to death in 54.5 % (n=189) of cases and possible need for improvement in management in 48.0% cases. Three 'management failure' themes were identified including failures to plan, failure to communicate and failure to rescue, consisting of 21.9%, 10.7 %, and 10.1% of cases respectively. Following the reviews, quality improvement initiatives proposed by residents lead to system-based changes.


CONCLUSION
A resident-driven mortality review curriculum can lead to improvement in systems of care. This type of novel curriculum can teach systems-based practice. The recruitment of teaching faculty with expertise in quality improvement and mortality case analyses is essential for such a project.


Introduction
The morbidity and mortality conference (M&MC) is traditionally utilized in academic medical centers as a forum to identify and discuss adverse events. Despite its long-standing tradition, the M&MC has 2 main flaws. First, the considerable variation in the structure of these conferences means that there is little consensus on the goal, method, and format [1]. Second, the literature regards the M&MC as a crude means of addressing errors since many adverse events are identified through other processes and are often not reported in M&MCs [2,3]. M&MCs have been further criticized for being a public forum for blame allocation [4]. Many institutions have restructured their M&MC format to identify and implement improvement initiatives and to sharpen a focus on mortality reduction through system improvements [5,6,7]. Nonetheless, the literature indicates that M&MCs rarely lead to system-based actions [1, 8,9]. A survey of US internal medicine residency programs revealed that most programs contained MM&Cs as a traditional forum to discuss medical errors and pathophysiology related to adverse events and deaths. However, only half of the residency programs had established a process for handling and following up on the errors identified in these conferences [1]. The Accreditation Council for Graduate Medical Education (ACGME) mandates as a core requirement that residents participate in quality improvement projects and 'understand common principles and techniques of quality improvement' [10]. Participation in clinical case review and peer review is a fundamental comhttps://jeehp.org J Educ Eval Health Prof 2018; 15: 31 • https://doi.org/10.3352/jeehp.2018. 15.31 ponent of the contributions made by physicians-in-training to quality improvement.
Although research has investigated how to involve residents in patient safety and peer review through the M&MC format, few studies have described the process of engaging residents in real-time patient case reviews focused on mortality. This paper describes the process of planning and implementing a resident-driven mortality case review curriculum to teach residents how to critically analyze factors associated with in-hospital mortality and to increase engagement in quality improvement initiatives related to mortality prevention. We also report results regarding the nature of potential system errors identified by residents and the actions taken by the department of medicine in response.

Ethical statement
The Institutional Review Board of University of Florida approved this study as a quality improvement project (QIPR Project ID 182).
After IRB approval to study the curriculum, (IRB201701579) the learners were surveyed about the curriculum anonymously and with informed consent.

Educational setting and participants
In 2010, to understand factors associated with preventable mortality, an inter-professional team of physicians, pharmacists, nurses, and quality analysts at the University of Florida Health Hospital adapted the Kaiser Permanente Mortality Review Tool to perform a pilot review of 50 inpatient deaths among patients with advanced illness [11,12]. The review sought to answer 3 questions: (1) Was there evidence of failure to rescue the patient?; (2) Was there evidence of failure to create contingency plans for patient decompensation?; and (3) Was there evidence of failure to communicate among the practitioners caring for the patient? Although most deaths occurred as a consequence of underlying progressive deterioration of comorbid conditions, some deaths were potentially preventable due to failure of rescue (i.e., failure of early transfer to the intensive care unit or to perform other escalations of diagnostic and therapeutic interventions). Prior studies noted that cases of failure to rescue can result from diagnostic errors, failure to coordinate care among physician specialties, and communication breakdowns [13]. Early warning systems were subsequently implemented and the department of medicine M&MC was restructured to focus on these cases. While our medicine residents attended these M&MCs, their experiential training on patient safety was limited to a 1-month quality improvement/patient safety curriculum modeled on the Institute of Health Improvement (IHI) Open School during their yearly ambulatory rotation. In our 2015 program survey, only 83% of medicine residents felt that they had participated in quality improvement activities. In response, departmental experts in patient safety, along with the residency pro-gram directors, designed and implemented a monthly resident-driven mortality case review curriculum. This curriculum takes place during the ambulatory rotation of the University of Florida Internal Medicine Residency Program in Gainesville, Florida.

Materials and subjects
Based on the above logic, participants reviewing cases were participating in a quality review project, and this was clearly stated prior to case review. We utilized the 'IHI Move Your Dot' framework to plan the new format of our M&MC and to design our mortality case review curriculum. This framework helps to inform a hospital's organizational performance related to mortality. The key principle of this framework involves case-based analysis to identify local system issues that can be improved [6,7].

Technical information
Development of the case review worksheet: Using the Move Your Dot framework, residents used a case review worksheet to categorize and track mortality cases into categories of management failure that were identified during case reviews ( Fig. 1). Longitudinal data from each worksheet were tracked in the online HIPPA-compliant Research Electronic Data Capture (REDCap) platform.
Mortality case review process and sessions: Cases were selected to include every instance of mortality from the general medicine teaching service. Residents reviewed every mortality case from the medicine teaching services and identified areas for improvement in diagnosis and management using the mortality worksheet during their independent review of the cases. Residents then met with a faculty mentor to discuss the cases and to analyze potential problems in medical management that impacted mortality risk. The debriefing session consisted of a brief case presentation followed by a discussion of whether the cases involved any of the categories of failure that had been noted. There was also a discussion of the categories of failure. During these debriefing sessions, residents proposed potential action items to prevent similar mortality issues. Mortality cases in need of further review were subsequently forwarded to an executive committee or interdepartmental conference specifically organized to discuss the case and to develop action items for improvement (Fig. 2). A synthesis of these in-depth discussions and analyses of mortality cases led to a consensus regarding action items to implement to address systemwide issues. Fig. 3 presents an actual example from the conference and actions taken.

Statistics and data analysis
Case review data on mortalities from the medicine teaching services were analyzed. Descriptive statistics were used to analyze the numeric data regarding mortality and management failure categories. To organize and index the data set, 2 of the authors (NSR and ZZ) initially coded the data. The Braun and Clarke [14] framework of qualitative thematic analysis was used to analyze the open-ended responses. Following the 6 phases described by Braun and Clarke [14], we independently analyzed the data and identified management failure themes, focusing on patterns and richness of responses rather than the number of responses, and assigned comments to themes. Another author peer-reviewed the themes for accuracy and an audit trail was maintained with comments.

Results
From March 2014 to September 2017, 347 consecutive mortality cases were reviewed by 107 internal medicine residents. The residents identified a possible diagnostic mismatch at some point from admission to death in 54.5% of cases (n= 189) and a possible need for improved management in 48.1% of cases (n = 167) (Table 1). Three possible management failure themes were identified, including failure to plan, failure to communicate, and failure to rescue, which were implicated in 21,9% (n = 76), 10.7% (n = 35), and 10.1% of cases (n = 35), respectively. Three additional themes were identified including possible overtreatment (9%, n= 32), documentation issues (12%, n = 41), and lack of involvement of the palliative care team (19%, n= 67). The categories of possible failures were not mutually exclusive and many of the cases had multiple categories that overlapped. A total of 108 patients had 1 or more possible failures (31.1%). Of these 108 cases, 73 (21.0%) had a single category of failure and 35 (10.1%) had more than 1 category of failure. Eleven cases (10.2%) had possible failures to plan and to communicate. Six cases (5.6%) had possible failure to communicate and to rescue. Thirteen cases (12.0%) had possible failures to plan and to rescue. In 5 cases (4.6%), all 3 categories of possible failure were identified (Fig. 4). The raw data are available in Supplement 1.
For theme 1, failure to plan, 2 sub-themes were identified, relating to possible delays in diagnosis and delays in treatment. The delay in diagnosis sub-theme identified by residents included possible delays in the diagnosis of evolving life-threatening conditions (e.g., septicemia, acute myocardial infarction, or stroke) and delays in ordering diagnostic tests such as computed tomography (CT) scans or lumbar punctures. Exemplary comments include: "Possible delay in diagnosis regarding the patient becoming septic. She was hypotensive." (R01) "I think he may have benefited from early lumbar puncture on the day of his admission." (R02) "Patient didn't get CT abdomen when he was becoming febrile abdominal pain and distention noted by consult note day 2, CT scan was done day 4, possible 2 day delay." (R03) A delay in treatment was identified in cases where additional treatment actions were needed but failed to occur: "Service was consulted for catheter placement was consulted for dialysis catheter placement however was unable to do the procedure on the day planned.  Fig. 3. Example of a mortality event and actions generated by the resident-led team.
As part of the resident mortality case reviews, a case was identified in which a patient with an intracranial hemorrhage (ICH) was admitted to the general medicine service for observation. The patient subsequently expired due to progression of hemorrhage and ultimately, aspiration, as patient had "Do not Resuscitate" status. While the resident and faculty review determined that the mortality may not have been avoidable, the case review revealed that there was no standardized process for consultation, coordination, or co-management between the general medicine, neurology, and neurosurgery services. This case, and the analysis thereof, was presented by the resident who performed the review in a combined conference with members from the leadership of all 3 services (medicine, neurology, neurosurgery). Based on this discussion, a memorandum of understanding (MOU) was drafted, delineating clear roles and expectations for the management of patients with ICH. This service agreement was approved by the leaders of the respective departments and implemented for patients with ICH. "Family and patient not informed of severity of condition on admission (or at least no documentation exists to state they were informed." (R08) Theme 3, possible failure to rescue, had 2 sub-themes, relating to delays in the recognition of deteriorating vital signs and inappropriate responses to changes in patients' clinical condition.
"Tachycardia not addressed entire hospitalization. Severe abdomi-nal/chest pain on HD#2... Although CT performed, may have missed underlying gastric perforation/ischemia which was noted on esophagogastroduodenoscopy HD#4." (R03) "Patient had a SWAT event with hypotension and syncope the day prior to decompensation and death. Twenty-four hours prior to death, the patient was found to have oliguria and lactic acidosis, but was not transferred to higher level of care. It is likely that the patient's blood pressure medication of metoprolol was blunting tachycardia response which may have warned the team." (R07) "Patient noted to be tachycardic to 130s and starting to become more hypotensive in the emergency department around 4 AM, no fluids given until 8 AM. At 8 AM patient is given 500 mL normal saline (NS) by day team, however per chart continues to be hypotensive with systolic blood pressure 70s-80 and mean arterial pressures 46-50s. No further fluids given or action taken until approximately 1:30 PM when another L NS given." (R08)

Discussion
Herein, we describe the educational value of our innovative mortality case review curriculum to teach residents and drive systembased improvements for mortality reduction. Through facilitated rehttps://jeehp.org views of real-life cases of mortality, residents were able to identify management failure themes. Approximately 50% of cases involved possible failure to plan care, communication issues with team members, and failure to rescue the patient prior to a terminal event. A ripple effect of the residents' critical analyses of real-time decisions involved in these cases of mortality was the proposal of quality improvement initiatives leading to system-based changes.
Our curriculum contributes to the current literature demonstrating that residents can be successfully involved throughout the quality process. Tad-Y et al. [5] described a process for a system-based M&MC with interprofessional and interdisciplinary participation that led to action items. Trainees attending the conference completed a modified fishbone diagram. A core team debriefed after the M&MC conference and followed up on suggestions and actionable items. Kwok et al., described a hospital-wide structured M&MC, with representation from 'high-impact' clinician groups and 'core specialty' representation [15]. Our curriculum is unique in that the residents were active participants in critically reviewing actual patient charts, documenting their findings regarding diagnostic errors, and proposing potential action plans. They then advanced to interdisciplinary teambased debriefings of the mortality cases including pharmacy, nursing, and subspecialty physicians from various departments. Examples of the action items created include a protocol for management of intra-cranial hemorrhage (Table 2), a process for early involvement of the palliative medicine team with terminally ill patients, and a protocol for early sepsis alerts. Since we found physician-to-physician and nurse-to-physician communication possible failures, we introduced a HIPAA-compliant secure text messaging application for smartphones.
During the implementation of our new curriculum, the mortality index (observed/expected [University Health Consortium/Vizient]) for the department of medicine improved from 1.21 to 0.7 (Fig. 5). However, as multiple other quality initiatives were undertaken in the hospital system, we cannot be sure to what extent the action items generated from the mortality reviews resulted in the decrease of our mortality index-enhanced.
A limitation of our curriculum was the lack of resident continuity from the initial case review to the proposal of action items and final system-based implementation. This limitation was unavoidable due to the residents' schedule and the lag time needed to implement system-based changes. Future directions may include an electronic learning management system that could be used to track which case was discussed and allow for direct feedback and discussion on a message board. This would obviate the need for the residents' physical presence to receive feedback. Given the experiential teaching focus of our curriculum, we did not directly assess the impact of the curriculum on residents' patient safety education (i.e., residents' perception or knowledge improvement on the M&MC process and principles). Nevertheless, our curriculum applied adult learning principles for residents to critically analyze and identify mortality causes and engage in change management.
The recruitment of teaching faculty with expertise in quality improvement and mortality case analyses is essential for such a project.
A key lesson we learned was the importance of identifying and recruiting a small group of core faculty with expertise in quality improvement and a strong interest in teaching learners. We also emphasize the importance of early buy-in and close partnership with  the residency program directors and hospital administration. To implement a similar mortality case review curriculum and M&MC redesign, we recommend early involvement and buy-in from the Office of Graduate Education, the Departmental Quality Improvement leaders, and the Senior Dean for Clinical Affairs. Future directions for this curriculum include having resident teams, ideally with representation from each post-graduate year, work on longitudinal action items. Each team should have a faculty mentor and designate 'resident quality champions' to work on action items identified through case reviews. These quality teams of faculty, residents, and interprofessional representatives could conduct longer quality projects.
In conclusion, our resident-led mortality case review curriculum not only satisfied the ACGME's requirement of experiential quality training for residents, but also promoted system-based improvements within the institution. We encourage residency programs to build similar curricula and have mechanisms in place to engage residents in the institutional mortality review process.