How data can better address care gaps and health needs
For population health to really deliver on its promise, care organizations must use data to address all of a patient’s needs.
What is "population health?" Much has been made of the term – it often conjures up images of preventative care gaps, social services or other metrics that are not part of traditional fee-for-service. However, as I interact with people in the healthcare space, I have found that there are many variations of what people define as "population health."
The U.S. healthcare system's propensity to waste enormous piles of money and deliver horrible results is well established. Fee-for-service reimbursement mechanisms are rightly cited as the prime culprit. If you get paid for doing more, you will do more. If you get paid for doing more with very little consequence for bad outcomes, you might pay attention to certain quality metrics, but your primary concern is still doing more.
By contrast, the concept of population health is to take care of a population and make the "right" decisions for the patients who need care efficiently without the warped incentive to always do more (and document more) to create more billable codes. Patients who need more care will get more care. Patients who don't, won't. At least that is the theory.
The reality is that “population health” is much easier to say than it is to execute. In fact, if the concepts sound like the failed forays into the concept of health maintenance organizations in the 1990s, I think that is accurate. The HMO experiment 30 years ago failed because the capitation had to be applied across broad populations without nuance. It allowed physicians to cherry-pick healthy patients, collect the funds for those attributed patients and not have to worry about significant financial losses. It also created an incentive to withhold care and referrals. The whole system was a mess and collapsed under the weight of poorly aligned incentives.
The benefit of having data
The big difference now is that we have data and the technology and resources to fully leverage the data. Using such data, a physician can still cherry-pick healthy patients, but that physician will be paid less because the data will show that these patients are easier to care for. On the flip side, provider organizations will be incentivized to appropriately dedicate more resources to the patients who need them because cost, quality and other data will direct them.
Moreover, the technology has the potential to make healthcare delivery far less onerous. At least this is the theory – the reality is that it is much more difficult than simply sticking a flag in the sand and declaring, "we're paying for quality, not volume." Theoretically, we should focus on interventions that make the best use of the resources we have and also leverage resources that don't fall within the traditional walls of hospitals and other healthcare facilities.
For example, I heard about a diabetic who was repeatedly admitted to the hospital for uncontrolled blood sugars. While hospitalized, she was quite well controlled and seemed to understand how to take care of herself. It really puzzled the staff that she kept returning – that is, until someone probed a bit further. The staff found out that this patient’s refrigerator had broken and she couldn't afford to replace it. As a result, the insulin she was given at discharge became progressively less effective as it was being stored at room temperature.
The hospital system bought her a new refrigerator, and she wasn't admitted again. A $200 household appliance had saved tens of thousands, if not hundreds of thousands, of dollars of future expenditures and possibly saved this patient from future kidney failure, cardiac problems and a host of other diabetic complications.
This might be an unusual example, but it is emblematic of the information gaps that currently exist. There are enormous potential data resources we can tap to help deliver better care.
But how do we create more stories like this and do it at the scale that our system desperately needs? The reality is that we are hampered by our long-time focus on reimbursement codes. We do something, document the code and we get paid. Unfortunately, there is no reimbursement code for a refrigerator or even asking if the patient has problems at home that prevent her from taking care of herself (also known as social determinants of health).
Replicating “refrigerator moments”
How do we systematically create similar "refrigerator moments?" This is far more difficult than it sounds. For all its faults, at least fee-for-service is a concept that’s simple to understand and execute.
By contrast, how do you define whether a patient is getting good care or better outcomes? The refrigerator helped that one diabetic patient, but is it reasonable to ask every diabetic whether all their home appliances work?
We have a habit of carpet bombing our healthcare workers (and patients) with long streams of questions during screenings, and we can't continue to do this. Our front-line staff are already rebelling and burning out.
This is where advanced analytics and data science will come into play. We need to be able to use data to deliver precise opportunities that will help a particular patient or particular group of patients. However, this effort will require a lot of hard work and imagination. Where, then, are we currently missing opportunities?
There are certainly things that can fulfill the concept that we should intervene and do something early before a medical problem becomes more acute. For instance, some common items that fit into this category are drug therapy for lowering LDL in cardiac patients and widespread depression screening.
Whack a mole for care gaps
Unfortunately, the result of this exercise has manifest itself as what Dr. Carly Eckert has termed "care gap" whack a mole. We have replaced, or added to, chasing encounter and procedure codes with filling care gaps. Instead of measuring good quality and outcomes, we measure surrogates like LDL and depression screening. These "care gaps" have some scientific foundation and might reduce some downstream costs, but their main advantage is that they are easy to measure. Their value to the patient and the system as a whole, however, is rather nebulous.
How do we bridge this gap and evolve to a better data-driven picture of true quality and improved outcomes? We have a tidal wave of data but very little of it means anything. We will need to exert some significant data governance and cleansing efforts. We need to be able to create some signal from all the noise.
I have frequently used the analogy of oil in describing this. You may have heard the adage "data is the new oil." The reality is that our current data landscape is like crude oil. It is a raw material that may be useful, but it needs to be converted and refined to a usable product. You wouldn’t stick crude oil in your car and expect it to run – you need to refine it into gasoline. We need to do the same with our "crude" data – refine it into easily understandable and content-rich information.
Once data are actually usable, we need to move it around to all the various stakeholders and create real interoperability. Of course, the care I deliver in the emergency department needs to be effortlessly communicated to the primary physician, but those data also need to be seamlessly integrated into the larger corpus of data, information and insights of larger populations.
That level of interoperability then needs to have easy action arms with tools that are usable by patients and providers. One interesting observation I heard from a healthcare consultant with exposure to the top population health vendors is that when the vendors had to rate what they did the best, they listed their top skill as "aggregate data." However, when their provider clients were asked the same question, the top message was that the vendors in fact "aggregate data" quite poorly.
In our conversation, we think this was the vendors' interpretation of the aggregation effort as the technical task of pulling together digital bits from various sources. However, that process was not enough for the providers. Just pouring all the patient's data in a single bin is one thing – the insights from those data need to be synthesized in nuggets that are easily accessible and easily actionable. Delivering an impressive graph that can reveal a list of names or record IDs that need action isn't very helpful if someone has to then enter another system and manually enter these records to act on the insights.
Owning the transparency
Finally, as all these really meaningful data and insights become more accessible, we need to be comfortable with the radical cost and quality transparency that will result.
With this transparency, we need to be prepared to resist high-revenue providers, insurance companies and middlemen like insurance brokers and PBMs. Such transparency will make obvious where there is waste and will raise questions why low-value/high-cost products and processes exist in our healthcare system. This effort will catalyze opportunities to re-engineer the payment and delivery models.
Ultimately, these are the steps I would like to see as the end result of "population health." Admittedly, the EMR road has been bumpy, but at least we are now largely digital. We need to leverage this digital foundation and use 21st Century technologies to truly achieve the quadruple aims of improved quality, reduced cost, improved patient experience and improved provider experience.
Unfortunately, the process won't be as simple as a resolution to move to value-based approaches. All these components will require enormous amounts of effort, ingenuity and innovation, but we will need all those components if we are to save our badly broken healthcare system.
John Lee, MD, has been a clinical informaticist since 2006 and has been Chief Medical Information officer at two health systems. In 2019, he was honored with the prestigious HIMSS Physician Executive of the Year award. He is a firm believer that the key to solving the myriad problems we experience in our healthcare system is efficient and transparent delivery of information.