At the Hospital and Physician Relations Executive Summit in Scottsdale, AZ, we attended a lot of great sessions. There was one in particular that we thought would paint a very clear picture of how data analytics can improve healthcare and decrease costs. John Gillespie, MD, MMM of Palladian Health and Joseph A. Eberle, Managing Director of Data Analytics at Computer Task Group gave us an hour packed with great information about their analytics experience. We’ve summarized it in as small of a nutshell as we could muster, without leaving out any of the good parts.
The Healthcare Roadblock
If a patient has a chronic disease, they get tested, diagnosed and then they opt to receive the treatment. To a doctor, that is a logical flow of care but at Palladian Health it wasn’t occurring that way. Many patients were diagnosed and then didn’t comply with the treatment plan. The result was lower treatment rates, higher readmissions and increased healthcare expense.
To solve the mystery that seemed so logical to health care professionals, they took a closer look at the patient data.
The Data and Analysis
Through a grant, a multi-year project began to put in place a data analytics platform that improved patient outcomes and stem costs of chronic diseases, like Chronic Kidney Disease (CKD). The data gathered included 76.1 million claim records, 132.5 million lab observation results, 151.9 million diagnosis codes, 179.9 million procedure codes and 64.5 million medication claims.
Patient data comes from many sources. This is true in most healthcare organizations. The data mentioned above came from EMR, Rx claims, benefit plans, lab results, history, demographics, lifestyle and more. Funneling all of these data sources into one master database can be tricky. When building this data architecture, they found it important to architect the data flow for the future flood of big data from telemedicine like FitBits, embedded chips and other wearable devices.
They also struggled with “bad data”, referring to the data that is inconsistent and does not conform to industry standards. Should they keep it? Toss it out? Which is right? They ended up keeping the inconsistent data to have humans sift through it for important information that computers wouldn’t recognize.
Once the data analytics platform was built, their big data analysis capabilities increased to include:
– performing epidemiological studies and analyze population health
– viewing multiple aspects of a patient’s medical history on one screen
– following disease progression over time
– tracking effects of therapies over time
– connecting cost information to clinical data/outcomes, which is crucial to adapting to payment reform
– identifying gaps in care
– grouping patients in cohorts for more in-depth study/treatment
They found that in between the first visit and treatment for CKD, there were a whole host of roadblocks they hadn’t been considering.
– Before the diagnosis stage, the data showed that there was a lack of prevention efforts. Patients at risk for chronic kidney disease did not receive recommended testing.
– About half of patients who met diagnostic criteria for chronic kidney disease never had an official diagnosis.
– Delays in referral to a specialist caused many patients with late-stage CKD to not see a nephrologist at the appropriate point.
– There was also a lack of education and preparation for dialysis once a patient had been diagnosed.
An example of the lack of education was given about a woman who was diagnosed with CKD. When presented with the treatment option of implantation of a port for dialysis, she was stopped by an emotional roadblock. She was afraid that if she had a permanent or semi permanent implant, her husband would find her ugly and leave her.
The emotional toll that treatment would take on the patient was being over looked and left unaddressed. Patients were not being educated in the treatment process early enough and therefore felt scared and pushed into a path they weren’t ready for, reducing compliance.
These are pretty serious findings. Some are life changing for the patient and a burden for hospitals and clinics. Being able to look at the collected data in one spot and analyze it over time can produce clear findings like these. The results of data analysis can provide opportunities to improve patient care, health care operations and ultimately decrease everyone’s costs. The bonus of an analytics platform is that their analytic capability is now not limited to CKD but is able to expand to other diseases to optimize those processes and procedures as well.
Data Analytics Lessons Learned
– The data quality is going to be poor: accept it, embrace it,. And design for it.
– Leave data sources in disparate formats while allowing them to be integrated into a comprehensive view
– No data left behind: any piece of evidence can be useful
– Focus on a comprehensive security approach for sensitive data and protected health information. Lock it down from the very beginning
– Utilize a RIM to help maintain consistency in an every-changing world of codes
– Involve a medical oversight committee early and include doctors, nurses, social workers, case managers
– Involve patients early to ensure patient centered-ness
– Think BIG: know what you can do with this amazing data!
According to Infogroup’s latest report titled “Big Data’s Big Payday“, data analytics is still one of the biggest data related challenges facing marketers in 2015. Data analysis requires investment in either great tools or qualified resources, but as you can see from the story above, it is truly a vital piece of the marketing puzzle.