Supporting the VA’s high reliability organization journey through data
- By David LaBorde
- Jan 23, 2020
Early this year, the Department of Veterans Affairs officially began its journey to become a high reliability organization, putting procedures and protocols in place to maximize safety and minimize harm. The nuclear power and aviation industries have also applied HRO principles to achieve minimal errors, even in the face of highly hazardous and unpredictable conditions.
There are five main attributes of HROs:
- Sensitivity to operations (heightened awareness of the state of relevant systems and processes).
- Reluctance to simplify (the acceptance that work is complex, with the potential to fail in new and unexpected ways).
- Preoccupation with failure (viewing near misses as opportunities to improve, rather than proof of success).
- Deference to expertise (valuing insights from staff with the most pertinent safety knowledge over those with greater seniority).
- Practicing resilience (prioritizing emergency training for many unlikely, but possible, system failures).
With medical errors estimated to be the third-leading cause of death in the country, a growing number of health care systems are taking interest in adopting HRO principles. However, as part of the journey to becoming HROs, the VA and other health care provider organizations must empower their employees with advanced tools that provide the right data at the right time. To be quickly recognizable, data and analytics will need to be presented via data visualization tools that permit important trends in operating statistics and the status of a patient care to be easily monitored and tracked.
Organizations cannot rely on humans to carry out repetitive actions to identify out-of-control processes or abnormal lab or study results that require action. Humans simply cannot look at every patient and every care process every second. Software can.
Today in many health care systems, providers have no way of knowing which of their patients has a new result available without opening the patient’s chart. This workflow alone can lead to time delays between new insight and needed action. In other health care systems, it’s the opposite challenge -- providers are overwhelmed with alerts, but the alerts are not easily filterable and include both normal and abnormal results. This leads to alert fatigue and the drowning out of results that are actually important.
Software can help. Leveraging clinical business intelligence and access to real-time data, software can monitor for new results, assess the results, intelligently push out notifications only if needed and escalate those notifications if an expected follow-up action has not occurred. Software can do this very efficiently. In fact, this is what high performing HROs of the future will likely look like.
Humans can never be perfect, even the highest performing among us. But with the right tools, we can access the right data in real time so each of us will be able to perform our duties more efficiently and effectively, improving the reliability of the care provided.
Along the journey to becoming an HRO, mundane work that can be efficiently carried out by a machine with better accuracy is migrated to software solutions. This way, employees can spend most of their time operating at the top of their skill set. They can provide care and react to new insights, rather than spending hours every day gathering the data they need to actually do their jobs. Being equipped to quickly act on new data that requires intervention can be the difference between life and death for patients.
One high-risk patient population includes those who have been identified to be at high risk for suicide. Suicide is a major epidemic facing our nation at large and veterans in particular. According to data from the VA, on average 20 veterans commit suicide each day.
The VA has been a leader in the deployment of advanced technology that enables the identification of high-risk veterans. One example is a program called Recovery Engagement and Coordination for Health – Veterans Enhanced Treatment (REACH VET), which uses trained machine learning models to identify those veterans at a statistically elevated risk for suicide. The VA has done an excellent job ensuring that veterans have their care evaluated by the suicide prevention team and if needed, providing them with specialized suicide prevention focused pathways. It has also recently rolled out new, robust suicide-screening workflows that enable the identification of those veterans that may need closer follow-up and specialized services.
This identification brings its own challenges, however. Once a veteran is identified as being at elevated risk for suicide, it initiates a myriad number of new patient care workflows, so much so that the VA created the specific jobs of Suicide Prevention Coordinator (SPCs) and REACH VET coordinator. These suicide preventionists ensure the care of high-risk veterans is well coordinated so that no high-risk veteran’s care “falls through the cracks.” It is also an area where software powered by real-time data is critical to ensure care called for by VA policies is delivered efficiently and methodically.
Like most health care systems, the VA has developed performance measures that allow it to evaluate and benchmark quality and efficiency at medical centers. One such metric that maps to care pathways for veterans at high-risk for suicide measures whether each high-risk veteran has a safety plan on the chart within a certain window of time after having been identified to be at high risk for suicide. This is one aspect of care that the SPCs are responsible for monitoring and coordinating.
However, as the number of patients for which an SPC is responsible increases, the task of making sure each veteran has a timely safety plan can become increasingly difficult. To ensure highly reliable performance against this metric, SPCs need the ability to see the short list of patients whose care is out of compliance with the policy where the due date for action is approaching or elapsed.
This is much more actionable than having a spreadsheet or web page with a table enumerating hundreds of the high-risk patients. It’s not that the detailed report is not important, but rather that it is highly inefficient to look at all 100 plus high-risk patients to find the four that may require action today on the part of the SPC.
This is where presenting the right data -- not all the data -- at the right time, can provide major efficiency gains leading to better care, higher performance of care processes, and a lower probability that a required action gets missed. SPCs need to be able to very quickly identify the subset of high-risk patients that may require some action on their part on a given day and to receive pushed out reminders for actions that are critical and need attention sooner rather than later.
The VA is initially focusing on 18 facilities in its move toward HRO. Based on my interactions with caregivers at many of these locations, the initiative is gaining traction. VA personnel are adding HRO principles into their daily activities and using HRO vernacular to address their challenges. We must support them with better software to handle their caseloads.
When caregivers typically carry heavy caseloads of high-risk patients, a tiny error can have catastrophic consequences. Although humans might be able to review every patient chart weekly, the right software looking at the right indicators can monitor those charts every second. Intelligent software available today will help the VA achieve its HRO mission and save at-risk veterans.
David LaBorde, M.D., is CEO of Iconic Data Inc. and senior advisor with DSS Inc.