Don Morris
A point about how reimbursement impacts care
August 22, 2013
By Don Morris
“Humungous tonsils” was the verdict of the pediatrician who sent us off to see two ENT specialists with our twins. The first doctor explained that they were indeed humungous and that any tonsils larger than a certain size should be removed or they could lead to poor functioning at school, among other problems. He would be happy to schedule the surgery. The case seemed clear. But the second doctor needed to know more to make a decision: Were the twins tired and cranky in the morning? Did they breathe irregularly in their sleep? Could we send a video?
What happened here illustrates a point about how reimbursement impacts care, that I think is often missed amid dire forecasts that health care reform will lead to “cookie-cutter” medicine. It’s actually the other way around. The current fee-for-service system, represented by doctor number one, rewards a one-size-fits-all approach that increases volume; simple guidelines (e.g., the tonsil size rule) have been developed to make decisions easy so providers can focus their time on providing the reimbursed treatments. The second doctor, a salaried professor who had no financial interest, was much more thoughtful about her decision. As health care reform begins to make medical groups financially accountable for patients’ outcomes ranging from hospital readmissions to heart attacks, the industry is scrambling to collect individual health data and develop a new generation of analytics that that is focused on better understanding of impact of medical decisions on individuals. Because bad outcomes like readmissions will be penalized, and good outcomes will be rewarded, simple economics are driving the industry to personalized care.
Most medical decisions involve a tradeoff of risks of potentially fatal outcomes: a CT scan of your coronary arteries may help diagnose heart disease but could cause cancer; aspirin and other blood thinners reduce your risk of heart disease but could lead to serious bleeding. Simple cookie-cutter guidelines fail to take individual complexity into account, often recommending treatment to people who could on balance be harmed, or fail to recommend treatments to people who would benefit.
Before evidence-based medicine, medical decisions were all personalized, but because physicians didn’t have access to analytic evidence their decisions were based on intuition, or a limited number of personal experiences. Since that era, clinical trials have created a sound scientific foundation for decisions= and for national care guidelines, but the simplicity of these guidelines has led to less personalization of decisions. Now we’ve moved to a stage where there’s enough clinical evidence, analytic horsepower, and patient data in electronic records to guide medical decisions that are both personalized and evidence-based.
Looking forward, one of the biggest barriers to the development and use of personal analytics is data quality. Case in point: One group we worked with had been entering temperatures as heights in their electronic medical record until we pointed out how many people in their records were 98 inches tall. It was frightening that no one had noticed, but equally clear that this data had never been used for treatment decisions. With the increased use of patient data by computers to help guide decisions, methods will be needed to ensure that our individual health data is as accurate as our bank records. To anyone involved in health care this may sound like a Sisyphean task, but other industries have already led the way — wouldn’t you be shocked if you discovered that your bank had been recording your deposits as withdrawals?
Finally, the quality metrics chosen to determine health care reimbursement will be critical. They will need to accurately measure what is beneficial to patients — no easy challenge — or risk driving health care away from the mission of improving health to other parts unknown. Ultimately, if health care metrics are appropriately designed, economics and analytics should lead us down the path to the right treatment for the right person.
Our twins, by the way, still have their tonsils, and as of the last visit they were not quite so humungous.
About the author: Don Morris, PhD, is vice president of scientific product and technology development at Archimedes Inc., a healthcare modeling and analytics company based in San Francisco. He leads the development of IndiGO, Archimedes’ clinical-decision support tool, and other products for individualized risk prediction and decision support.
“Humungous tonsils” was the verdict of the pediatrician who sent us off to see two ENT specialists with our twins. The first doctor explained that they were indeed humungous and that any tonsils larger than a certain size should be removed or they could lead to poor functioning at school, among other problems. He would be happy to schedule the surgery. The case seemed clear. But the second doctor needed to know more to make a decision: Were the twins tired and cranky in the morning? Did they breathe irregularly in their sleep? Could we send a video?
What happened here illustrates a point about how reimbursement impacts care, that I think is often missed amid dire forecasts that health care reform will lead to “cookie-cutter” medicine. It’s actually the other way around. The current fee-for-service system, represented by doctor number one, rewards a one-size-fits-all approach that increases volume; simple guidelines (e.g., the tonsil size rule) have been developed to make decisions easy so providers can focus their time on providing the reimbursed treatments. The second doctor, a salaried professor who had no financial interest, was much more thoughtful about her decision. As health care reform begins to make medical groups financially accountable for patients’ outcomes ranging from hospital readmissions to heart attacks, the industry is scrambling to collect individual health data and develop a new generation of analytics that that is focused on better understanding of impact of medical decisions on individuals. Because bad outcomes like readmissions will be penalized, and good outcomes will be rewarded, simple economics are driving the industry to personalized care.
Most medical decisions involve a tradeoff of risks of potentially fatal outcomes: a CT scan of your coronary arteries may help diagnose heart disease but could cause cancer; aspirin and other blood thinners reduce your risk of heart disease but could lead to serious bleeding. Simple cookie-cutter guidelines fail to take individual complexity into account, often recommending treatment to people who could on balance be harmed, or fail to recommend treatments to people who would benefit.
Before evidence-based medicine, medical decisions were all personalized, but because physicians didn’t have access to analytic evidence their decisions were based on intuition, or a limited number of personal experiences. Since that era, clinical trials have created a sound scientific foundation for decisions= and for national care guidelines, but the simplicity of these guidelines has led to less personalization of decisions. Now we’ve moved to a stage where there’s enough clinical evidence, analytic horsepower, and patient data in electronic records to guide medical decisions that are both personalized and evidence-based.
Looking forward, one of the biggest barriers to the development and use of personal analytics is data quality. Case in point: One group we worked with had been entering temperatures as heights in their electronic medical record until we pointed out how many people in their records were 98 inches tall. It was frightening that no one had noticed, but equally clear that this data had never been used for treatment decisions. With the increased use of patient data by computers to help guide decisions, methods will be needed to ensure that our individual health data is as accurate as our bank records. To anyone involved in health care this may sound like a Sisyphean task, but other industries have already led the way — wouldn’t you be shocked if you discovered that your bank had been recording your deposits as withdrawals?
Finally, the quality metrics chosen to determine health care reimbursement will be critical. They will need to accurately measure what is beneficial to patients — no easy challenge — or risk driving health care away from the mission of improving health to other parts unknown. Ultimately, if health care metrics are appropriately designed, economics and analytics should lead us down the path to the right treatment for the right person.
Our twins, by the way, still have their tonsils, and as of the last visit they were not quite so humungous.
About the author: Don Morris, PhD, is vice president of scientific product and technology development at Archimedes Inc., a healthcare modeling and analytics company based in San Francisco. He leads the development of IndiGO, Archimedes’ clinical-decision support tool, and other products for individualized risk prediction and decision support.