Today more than ever, healthcare has an increased focus on big data, predictive analytics and artificial intelligence. While these shiny objects may have some promise now and in the future, they are generally expensive, time-consuming and may or may not affect change in a positive manner.
Before beginning any analytics or data project, I ask one of two questions: “How are you going to use the results?” or “How does the information improve the health of patients?” Powerful software, terra gigs of data, and social determinants of health do not matter unless you have a plan to implement what you learn.
We are currently working on a project for which we’ve developed a framework for clinicians to assess and communicate the behavioral health or substance use risk of patients. The system, albeit simple, offers providers and their partners a common language for assessing and directing patient risks and need response. It’s entirely tier-based.
A tier 1patient can self-manage and advocate, providing insight not only into the behavioral health or substance use risk but also the ability to comply with a care plan, get to an appointment, and engagement. On the flip side, a tier 3 patient needs advanced access and wrap around support (transportation, housing or med assist). As the risk of the patient may (and will) change over time as they work through a behavior health crisis or episode, the support for that patient can adjust up or down by simply altering the tier to which they are assigned.
A partner organization can use this “language” to prioritize patients in terms of access and resources and design systems of care for patients to meet the needs of a patient population during a vulnerable time. The system is not data intensive nor overly sophisticated in terms of algorithms and predictive value. It is based on the clinical and experiential judgment of the clinicians who are interacting with each patient, and allows a team to implement better care across a care continuum and use limited resources wisely.
Another recent project focuses on segmenting pneumonia patients by risk for relapse, working with the hospital and post-acute care settings to design paths to meet each risk tier. By identifying simple observation points vital to the risk profile of pneumonia patients, Blaze worked across partners to develop communication protocols and differentiated care paths for those most at risk.
Our post-acute care client deployed respiratory therapists for the highest risk patients with increased monitoring and breathing exercises, while all patients received continued monitoring by all staff. If a patient’s condition worsens, staff has decision paths with trigger points on when to escalate and prevent a readmission or worse.
Data and Risk Segmentation
Finally, in a third project, Blaze brought together two large market players with expansive data assets on a portion of the patient profile, which including lab results and hospital encounters. By combining these data sets, Blaze had critical info on patients that allowed them to bypass a complex clinical data integration project. Blaze used a rigorous modeling technique called a random forest to create risk scores for multiple populations, creating a single scoring system for users.
Risk segmentation techniques have long been used in population health for transitional care, chronic care management, sub-populations (mental health, pneumonia), pre-contract and resource planning. When working to understanding the risk of a population, many factors need to be considered, including the available data, the population to be analyzed, and how the risk output will be used (transitional care, etc.). Using risk scores to implement differentiated care plans for the better outcomes, resource utilization and lower unnecessary utilization is the trademark of population health strategy.
Blaze has worked with many providers, populations, data sets and real-life scenarios in its population health practice. We have worked hand in hand with many providers using creative approaches and developing effective care paths to improve the health of each and every patient. Our collective knowledge in analytic techniques, process improvement, care management design and clinical intelligence position us to help turn data and observations into action.