The overarching concept of population health management continues taking shape amid an American healthcare system undergoing feverish digitization — and while some of the larger, IT-savvy health networks may more effectively be managing patient populations within two years, for most it will likely take six to eight years.
To that end, Daniel Newman, MD chief medical information officer of MEDfx, is slated to outline the triptych of population health pillars during a session next week at the Government Health IT Conference and Exhibition in Washington, DC.
Ahead of the show, Government Health IT Editor Tom Sullivan interviewed Newman via email about those puzzle pieces, how far the country is from widely employing such technologies and practices, and why true population health management looks at the whole person rather than merely their disease state.
Q: You list the 3 pillars of population health management as analytics, care improvement, and patient engagement. Are the technologies already available and, if so, what are the hard parts of each?
A: Analytics: Analytics can be thought of in two buckets: 1) Deductive analytics, which includes reporting and retrospective stratification and 2) Inductive analytics, which includes predictive modeling and stratifications. The technologies to do retrospective reporting and stratification are available and have been for some time. One of the major difficulties is getting the data into a form that can be easily used for reporting. Gathering data from multiple systems (PM, EMR, Lab, radiology, etc.) both within and across systems and then normalizing the data to standard terminologies so you can run queries is labor intensive and difficult to accomplish. Unfortunately, much of the data in healthcare applications is often unstructured, meaning not codified to a standard, due to usability issues for clinicians and/or not having a standard coding scheme available. The more accurate data you have, the more the outputs can be trusted. Predictive modeling and inductive analytics are being developed by several groups right now. Predictive modeling will be able to tell us where patients’ health will be given their current status. These models will need to be thoroughly tested and have continuous improvements to ensure their accuracy.
A: Analytics: Analytics can be thought of in two buckets: 1) Deductive analytics, which includes reporting and retrospective stratification and 2) Inductive analytics, which includes predictive modeling and stratifications. The technologies to do retrospective reporting and stratification are available and have been for some time. One of the major difficulties is getting the data into a form that can be easily used for reporting. Gathering data from multiple systems (PM, EMR, Lab, radiology, etc.) both within and across systems and then normalizing the data to standard terminologies so you can run queries is labor intensive and difficult to accomplish. Unfortunately, much of the data in healthcare applications is often unstructured, meaning not codified to a standard, due to usability issues for clinicians and/or not having a standard coding scheme available. The more accurate data you have, the more the outputs can be trusted. Predictive modeling and inductive analytics are being developed by several groups right now. Predictive modeling will be able to tell us where patients’ health will be given their current status. These models will need to be thoroughly tested and have continuous improvements to ensure their accuracy.
Care Improvement: The biggest issue with care improvement is that patients move around so frequently and to do care management well you need the data from all of these sources. While some systems have enough data to adequately perform well at this step, many patients care is too fractured between EMR systems to make this feasible. Technologies like private Health Information Exchanges coupled with the national eHeatlh Exchange will be able to begin closing this gap. Once you have the right data, the next step is developing rules and systems that will enable clinicians to use this data in real time. While there are some systems out there that do this, a comprehensive system has not yet been brought to market.
Patient engagement: Most people think of patient engagement and go right to the idea of a patient portal. Patient engagement is a set of actions one takes to improve patient activation. Patient activation is how educated, informed and involved patient are in managing their own health issues. Patient engagement is not just a patient portal issue. Every step in the care of a patient — be it inpatient, outpatient, care management etc. — has an opportunity for patient engagement. Making sure we consider patient engagement to not just be a technology but a series of activities with a goal is essential to having a good strategy. Portals can be excellent ways to engage a patient, but too many are flat views of a patient’s information or are not implemented with the full concept of an engagement strategy. Without understanding and measuring the effect of your patient engagement strategy, patients will be less likely to use your portal. From a technology perspective, portals are easy to find, however most sit on top of a single instance of a database and do not show comprehensive information. Expanding the data in these portals and making sure they are aligned with a strategy are the most difficult parts of patient engagement.
Q: Even optimistically here, how far is the US from being able to put those pillars into nearly ubiquitous practice?
A: Many years. I would expect some of the larger systems to have pieces of this in place in the next 1-2 years. I would not expect this to be ubiquitous for 6-8 years at best. These systems will need continuous cycles and measure and improvement to reach their optimal states.
A: Many years. I would expect some of the larger systems to have pieces of this in place in the next 1-2 years. I would not expect this to be ubiquitous for 6-8 years at best. These systems will need continuous cycles and measure and improvement to reach their optimal states.
Q: When the topic of population health management arises, many people point to diabetes, obesity and blood pressure as the sort of de facto standard examples of conditions we could, and should, be managing better in patient populations. What are some of the less well-known?
A: While the management of these diseases is essential to population health management, it’s important to not limit population health management to just disease management. True population health is looking at the entire person, not just their disease states. It’s important to realize that it isn’t just diseases and medical management, but social and environmental issues and barriers that affect care. For example, someone with asthma that is poorly controlled due to environmental factors in their house or patients that are not well controlled because they don’t have the money to pay for transport to the clinic. Population health is about looking at all these variables and making sure you know the right interventions to improve health. That being said, there are definitely diseases that are focused on because they have good evidence that optimal management leads to improved quality of life and lower costs. These include COPD, Asthma, CHF and Ischemic heart disease. While these are well known, there are less well-known examples like Rheumatoid Arthritis and HIV and mental health. Though these will obviously be a focus, hopefully we will be able to institute a global view of each patient and understand the many complex reasons people do poorly.
A: While the management of these diseases is essential to population health management, it’s important to not limit population health management to just disease management. True population health is looking at the entire person, not just their disease states. It’s important to realize that it isn’t just diseases and medical management, but social and environmental issues and barriers that affect care. For example, someone with asthma that is poorly controlled due to environmental factors in their house or patients that are not well controlled because they don’t have the money to pay for transport to the clinic. Population health is about looking at all these variables and making sure you know the right interventions to improve health. That being said, there are definitely diseases that are focused on because they have good evidence that optimal management leads to improved quality of life and lower costs. These include COPD, Asthma, CHF and Ischemic heart disease. While these are well known, there are less well-known examples like Rheumatoid Arthritis and HIV and mental health. Though these will obviously be a focus, hopefully we will be able to institute a global view of each patient and understand the many complex reasons people do poorly.
No comments:
Post a Comment