The Healthcare Analytics Summit was the first conference organized by the team at HealthCatalyst, and was one of the best run conferences that I’ve had the pleasure to attend. The venue was perfect, sessions ran on time, there was plenty of time between sessions, and the constant usage of audience feedback through a mobile app was an interesting way to roll the theme of analytics into the conference itself.
Today, Healthcare delivery organizations are under pressure to switch their operating model from fee-per-use to one where profitability will be dependent upon delivering positive patient outcomes. The conference focused on how healthcare analytics was a key tool to be able to achieve these objectives.
The opening keynote was delivered by Billy Beane, General Manager of the Oakland A’s. As Billy spoke about how data and analytics transformed baseball team management, it became clear that the same kinds of insights and challenges would also apply to healthcare organizations:
- Healthcare, like sports, is a very emotional business. A lot of decisions are made by using experience and intuition.
- You do not get partial credit for getting close – either you win or lose.
- Moving to an evidence-based model is really about culture change. People still attribute success to chemistry or some other intangible; even if the math proves otherwise.
Using data to run your business requires courage and discipline. Discipline requires you to ignore the noise from people that are guided by emotion instead of data. It is tempting to fall back on emotions. We tend to celebrate the serendipitous wins. Yet we hold data to a 100% accuracy standard. Card counters don’t win all the hands, but if they are disciplined, they can reliably beat the house.
The one fundamental difference between baseball and healthcare is that Billy Beane was able to remove the old guard that didn’t want to get with the program. In a hospital, the senior team tends to be composed of lead surgeons and care providers. As a result, Billy recommended that a good place to stimulate change is by working on the organizational culture to encourage the leadership team in place to embrace a new way to deliver care.
Lessons from the trenches
The Moneyball keynote was the perfect anchor for the rest of the conference; most of the other keynotes were from people on the ground, using analytics to improve the delivery of care in their organizations. Here are some of the key highlights from those stories:
- 50% of the care given is either too little, too much, or just wrong. Doesn’t benefit or actually hurts the people that they serve. Data is used to figure out how to extract the most value for the patient/member.
- There is an inverse relationship between the cost of care and its quality. The higher the cost, the worse the outcome. If you want to find the bad outcomes, look for high cost.
- Collecting data is easy. Usability of the data is what is important. This is how the feedback loop is constructed. Make the data usable and accessible.
- Don’t ignore qualitative data in addition to all the quantitative studies. After doing ethnographic research it becomes clear that patients don’t know to express their needs through surveys and questionnaires. Turns out patients want to be treated as individuals and with respect. They want a personal connection as opposed to dealing with an interactive bedside computer because they feel they will receive safer, high quality care.
- Patients also wanted to get all the members of the care team to communicate better. They pick up on the fact that the doctor and nurse are not communicating effectively.
- Linking clinical performance to reimbursements is the biggest transformative force in American healthcare now. This means going from asking how to fill the beds to asking how to keep as many beds empty as possible?
- EMR implementations failed to deliver increase in quality. Not only did it fail at increasing quality but it increased cost by multiples.
- Large health organizations already have all the pieces to become ACOs. (Payer to delivery) they just need to make the switch. Might miss the data analytics to understand the cost equation.
To conclude the conference Dale Sanders, SVP of Healthcare Quality Catalyst, delivered a well needed reality check about predictive analytics. According to Dale, predictive analytics is about managing risk. With healthcare the following challenges still stand in the way:
- Humans are hard to predict
- Risk root causes are socio-economic. Not part of healthcare delivery
- Missing data in the system
- Wisdom of crowds is mostly ignored
- Social controversy. Using PA in healthcare is controversial
The math still isn’t here yet in terms of predictive analytics and human behavior. Sampling rate for an experiment is directly related to the predictability of the next experiment. A 737 airliner generates 500Gb of telemetry for a 6 hour flight. By comparison, a patient might generate 100Mb of medical data per year. Healthcare is not big data.
Right now, we’re asking nurses and patients to be the digital samplers. That doesn’t scale. It takes 400 clicks in Epic to document one clinical note. Not counting the keystrokes. Each time we ask clinicians to click for more discrete data we are climbing a hill we can’t conquer.
Even if we were to solve the sampling problem and understand the math; we are likely to face a cultural issue with predictive analytics in healthcare. Society is ok with using analytics for dealing with the fate of criminals: 80% of parole boards use predictive analytics to assist in the decision to grant paroles. Will we be able to accept decisions made by data in the healthcare field? We’re not there yet but we’ll get there soon so we need to think about it now.
It’s made of people
Although healthcare analytics is a field that revolves around information technology, it was clear that the biggest challenges in implementing analytics in the healthcare field are not technology related. The challenges revolve around:
- The capacity of the patient and clinical staff to collect accurate, actionable data;
- The usability and transparency of the collected data;
- The willingness of the delivery organizations to transform their culture. Their ability to convince everyone in the organization to be aware of the cost of the service they provide and to be disciplined in using data to deliver care;
- The willingness of society to accept the consequences of a data-driven healthcare delivery system.