Using event reports in real-time to identify and mitigate patient safety concerns during the COVID-19 pandemic

Abstract

In the midst of a crisis, monitoring patient safety and quality concerns is of utmost importance, particularly if attention paid to monitoring is diverted by other operational needs. The emergence of novel COVID-19 virus emerged new patient safety concerns. In an effort to help organizations respond rapidly to safety concerns, we describe a close to real-time process used to monitor, escalate, and resolve concerns submitted to the patient safety event reporting system of an academic health system.

Introduction

Coronavirus disease 2019 (COVID-19) is a highly transmissible respiratory illness that spreads easily between individuals in close proximity via contact, droplet, or airborne transmission. ¹ COVID-19 was declared a global pandemic by the World Health Organization (WHO), with more than 15 million people infected worldwide and over 630,000 fatalities as of July 24, 2020. ² Already strained healthcare systems in the United States and around the world are working to manage the allocation of scarce medical resources, conserve and procure personal protective equipment (PPE), and manage healthcare personnel shortages, while maintaining communication and transparency until the infection stabilizes and ultimately declines.

As healthcare organizations work to quickly define new ways to manage care, it is of utmost importance to ensure robust processes are in place to monitor patient safety and quality concerns, particularly when attention may be diverted by other operational needs. Using Donabedian’s quality improvement framework, ³ we describe the process that Johns Hopkins Medicine (JHM) implemented to identify and mitigate COVID-19 related safety concerns in as close to real-time as possible.

Creating an enabling structure

In the U.S., JHM operates six academic and community hospitals, six surgery centers, numerous ambulatory patient care locations, and a home care organization. As the number of confirmed COVID-19 cases increased in our region, incidents and concerns related to COVID-19 were reported across these institutions using our voluntary event reporting system. For over twenty years, the healthcare industry has used voluntary event reporting as a valuable data source to learn about system defects from the frontline staff and used those insights to improve patient safety. ⁴ JHM activated a Unified Incident Command center (UIC) to coordinate and manage its response to COVID-19 across departments and entities. A COVID-19 Safety Officer role was established based on the Safety Officer role for the Federal Emergency Management Association (FEMA), which provides training on emergency crisis preparedness and response, and “identifies safety hazards and works to ensure personnel safety … during large scale public health disasters”. ⁵

The COVID-19 patient safety analyst

To support the COVID-19 Safety Officer with data on emerging concerns related to the pandemic, a process was needed to monitor events, in close to real-time as possible. In March 2020, the position of COVID-19 Patient Safety Analyst was created. This role rotates among members of the event reporting team, who are responsible seven days a week for reviewing event reports for COVID-19 related concerns. The COVID-19 analyst has primary responsibility for reading all events reported across the health system, and assigning a tag to any events directly related to COVID-19, or any events that may result from changes made due to COVID-19. Examples of the latter include concerns related to use of PPE, visitor restrictions, or staff exposures. Events and concerns do not necessarily include a COVID-19 positive patient. Specific issues may be given higher priority when expedited learning is needed to help manage dynamic situations. The analyst applies knowledge of previous themes, recent changes, and new information to determine which events need to be highlighted.

Once the event has been tagged, it populates a dashboard created using enterprise-wide data visualization platform (Tableau, Seattle, WA). The dashboard is updated automatically every two hours with key data fields from event reports. This helps patient safety staff across JHM monitor COVID-19 reports. Once populated in the dashboard, reports can be filtered on various fields to examine frequency and types of reports by location. This visualization makes it possible to identify when new events are submitted, and when additional follow-up from reviewing managers has been submitted to the electronic reporting system on previous events to help safety experts track event investigations.

Process for identifying concerns

Identifying COVID-19 specific concerns within event reports is challenging for several reasons. First, these concerns are embedded in free-text narrative descriptions. Keyword searches to identify reports and extract events could be made using terms like “COVID,” “COVID-19,” or “coronavirus”. However, reporters do not use those terms consistently when completing event reports, and most search functions require an identical match for them to be identified accurately. Misspellings or unanticipated variations in wording could cause event reports to be missed, or could cause unrelated concerns to be pulled in by a query. For example, the search term “corona” is shared by several other common terms such as “coronary.”

Second, there are limitations to creating a specific data capture form or event type for frontline staff to use when reporting events. Events often do not fit neatly into a single event category. Instead, events tend to be complex in nature with multiple contributing factors and concerns noted in a single report. Additionally, event reporting systems typically only allow frontline users to select one category as the main concern. Event reports also route to reviewing managers according to the location and event type selected. Creating a new event type could limit those who receive the reports. Similarly, this workflow would rely on frontline reporters to recognize that an event was related to COVID-19, and use the specific form or identify a specific event type, rather than the usual classification. Most frontline staff who report events are not trained to label or code an event using a specified taxonomy. Educating reporters is unlikely to be an efficient or effective strategy. Systems should be intuitive to use without training and most organizations lack the staff to educate frontline health care workers on changes in coding.

Outcomes: From event reports to improvement actions

Monitoring the system for new events and eventual follow-up is not the end goal of reporting. Event reports identify opportunities for making useful changes: to inform new policies and practices that may impact staff, or suggest workflow or other process changes to ensure patient and staff well-being. Concerns need to be shepherded through the review process so that they are evaluated and when necessary, escalated to key decision makers to ensure thoughtful and timely improvements. To facilitate learning and prioritization of concerns, the COVID-19 Patient Safety Analyst convenes a virtual daily huddle with event reporting leaders across the health system to discuss new events and emerging trends, and prepares a daily report summarizing the top concerns to be reported to safety leaders.

The daily summary reports are shared with the JHM Safety Officer in the UIC and staffed on a rotating basis by leaders in the Office of Patient Safety. The UIC serves as the top-level command center for all JHM institutions, performing real-time problem solving, communication, along with coordinating shared services across departments, including infection control, supply chain, occupational health, logistics, patient placement and transfer. In times of crisis, use of the command structure expedites follow-up of concerns, rather than relying on usual committee structures. In daily leadership briefings held at the UIC, the JHM Safety Officer shares emerging trends and notable events as part of their report. This provides an opportunity for specific events, themes, and interventions to be discussed with all of the relevant commands that have a seat in the UIC structure. An example of a change made as result of an event report involved inconsistencies with patients on a COVID-19 dedicated unit receiving disposable meal trays. Some providers were forgetting to order the disposable tray since it required them to submit a second order within the system. To remedy this, an interface was built between the electronic medical record and nutrition software to automatically notify teams ordering food when a disposable tray was indicated.

Conclusions

The detailed process of reviewing, tagging, and escalating events described here has proven useful for the emergency presented by the COVID-19 pandemic. A similar process can be applied to other larger organizational changes, when an organization may wish to monitor new safety concerns in real-time, such as implementing a new medical record software or opening a new clinical facility. The trained and dedicated team involved in this process helps to maintain continuity and provides a data-driven approach to system learning. While normal processes may capture some of the data, in times of crisis or substantial change, conventional review methods may be difficult to perform. The safety net provided by a dedicated analyst can ensure that concerns are dealt with in a timely way.

While concerns reported by staff are valuable, we have also explored other data sources to inform and enhance learning, such as patient relations data, concerns reported to our compliance call line, and information gathered during mortality reviews. In the future, we plan to investigate how safety concerns overlap among all of these data sources.

The free-text, qualitative nature of event reporting data makes analysis challenging and resource intensive. Conventional data analysis methods rely heavily on manual categorization of events and human memory to identify patterns. Most event reporting applications today do not offer advanced tools to automatically categorize and identify emerging issues. This kind of application is needed to handle rapidly changing situations like the COVID-19 crisis so safety experts can spend less time categorizing the data and more time implementing system changes that improve patient safety. Programs that facilitate sharing of patient safety data, and implementing changes in coordination with peer organizations could accelerate problem solving. In a pandemic, this could help organizations hit by later waves of infection be more proactive in their response. We hope that our experience will help other organizations learn more rapidly from novel safety concerns that emerge from this and future crises.