Innovations in analytics at the Phoenix VA Hospital

Fire departments arriving before an emergency happens, mobile medical units with precision targeting, dashboards that are actually useful. Find out how the Phoenix VA (Veteran’s Affairs) Hospital is using data analytics to transform care delivery in this interview with Hamed Abbaszadegan, MD, Chief Health Informatics Officer.

The VA manages an immense amount of data

First, it is important to understand the quantity of data the VA manages and how data is used to drive innovation. The VA implemented electronic health records in the mid 1980s, well ahead of their industry peers. Their systems have billions of health records, prescriptions, and notes across all their health centers nationwide. 100,000 notes are inputted into the system every day. The data feeds into a central data warehouse where it can be analyzed.

As Chief Clinical Informatics Officer, Dr Abbaszadegan leads the team that analyzes this data, creates dashboards for clinical decision support, and helps develop data-driven programs to improve patient outcomes. According to Dr Abbaszadegan:

“It’s the extraction of that data that leads to innovation. Our role is to figure out the relationships between clinical data that lead to predictions.

The purpose of the information is to get the patient in front of the doctor, ready for a decision. I see healthcare moving into a “pull” system, which is the opposite from where we are now.”

The Phoenix VA healthcare system itself is very large. It has a $650mm annual operating budget, a hospital with 212 beds, and outpatient clinics across Arizona. At this level of scale in terms of data and people, it is possible to create sophisticated predictive models and experiment with innovative interventions.

Putting the data into a clinical context

Healthcare providers often complain that the data is not actionable. Dr Abbaszadegan provided a great example of this disconnect in diabetes care and how his team fixed it.

“We used to give doctors a list of patients that had an A1C above 9.0 [a measure of blood sugar control] and doctors would throw it away. They would say, “I know who these patients are, they don’t show up for their appointments.” The list was useless because we didn’t provide any clinical context.

Now, we show when a patient was last seen, when their next appointment is, and whether their labs are trending up or down. For example, if a patient was a fifteen and now is a nine, then he is trending in the right direction. The person who was at a 7 and now is a 9 needs to be seen and evaluated.

We can also predict who might not benefit from another primary care appointment. If someone keeps falling behind then it may be time for behavioral counseling, life coaching, or some other intervention. We put clinical data into context to make it useful. Isolated numbers are almost always meaningless.

We’re the data and information people, so right now we have 60 dashboards that have been built to provide clinical decision support.

Informatics is about People, Process, and Platform. We can edit the platform, but the people involved in the care have to be engaged and the process has to be tailored to the use of the platform. All three go together. If you don’t have the right person looking at the platform and then the process by which the data is used, it won’t work.”

Working with the fire department to reduce ER admissions

One of the most interesting and innovative programs Dr Abbaszadegan discussed is the Phoenix VA’s partnership with the Chandler Fire Department.

This is how it works. When the fire department receives a 911 call for a medical emergency, they typically have only one algorithm: take the patient to the emergency room. Many 911 calls are not emergencies. Individuals without benefits or not knowing who to call will often use 911 in place of going to their doctor. This puts a significant burden on emergency services.

The Phoenix VA and the Chandler Fire Department put together a pilot project that combined telehealth + big data + predictive analytics.

The Phoenix VA created a Care Assessment Needs Score (CANS), based on frequency of ER visits, PCP visits, diagnoses, and numerous other factors. They found that patients in the 99th percentile had a 72% risk of readmission.

A scheduler uses a dashboard with a list of patients having CAN Scores of 90 or higher and calls those patients to schedule a visit from the fire department. The fire department comes in and connects that patient to healthcare providers at the VA via tele-health. This means that the patient is getting instant access to care right from their own home. The fire department can also check the patient’s medications and do a fall risk assessment.

Rather than just taking the patient to the ER, the fire department can now provide tele-health visits on the spot and even schedule follow up visits.

Dr Abbaszadegan showed me a picture of a large bucket of medications in a patient’s home and commented:

“This photo says it all. I order you medications, but if you don’t take them properly it doesn’t matter. Here the fire department and doctor are throwing away old pills and consolidating medications. In one case, the patient was down to one tablet of his key water pill. Without our visit, he would have run out, been off the med for 3–4 days, and ended up in the hospital due to backfilling of fluids. The big box with medicines is the reality of care.”

Mobile Health Units

Dr Abbaszadegan also spoke to me about his next innovation — mobile medical units that use data to precisely target the right patient population.

“The next big thing I am working on is a data-driven Mobile Medical Unit. I tell it where to park. I do geo-mapping. I look at low compliance of vaccines such as flu, pneumovax, etc. This van will get you properly vaccinated. I find areas in the city that have low compliance and then send a van with vaccines. We can also provide speciality care, primary care and offer telemedicine visits so the doctor can be anywhere. Tele-derm, tele-retinal, there are lots of different services we can offer.”

Can lifestyle interventions to reduce the cost of care?

Dr Abbaszadegan had mixed feelings on whether lifestyle interventions could reduce the cost of care.

“The vegan diet in general is anti-inflammatory . I have seen it in practice. But getting someone to do something is not at as easy as it sounds.

We have group visits, teams of nutritionists, social workers, etc. that work on these patient in aligned teams. Doctors, nurses, nutritionists, social workers, all work for that team to meet the needs of those veterans. We have nutrition counseling, weight loss programs, home PTSD, tele-healthcare. If you have mobility issues, we have modalities to provide care remotely. If you can’t make it to the pharmacy, I’ll mail you your meds. We are very rapid on all of these. Medications and getting that out to you. We have people in these positions in group visits. We do provide dashboards for ancillary services. Ancillary services are anything that are non-provider/physician based, such as lab work, social work, nutrition.

But we have not yet looked at the impact of nutrition, diabetes education, etc. on outcomes.”

Reflections

It was fascinating to learn about how data analytics and predictive models were being used to develop innovative interventions at the VA. Here are a few of my own reflections.

1. The bulk of the effort in population health is, not unreasonably, focused on managing the very high risk top 5% of patients that make up 50% of healthcare spending. Consequently, medication management rather than lifestyle interventions are a higher priority. However, once healthcare organizations get a handle on low hanging fruit with high risk patients, the next frontier has to be prevention and even disease reversal through lifestyle interventions. The current work being done in developing analytics, dashboards, and tele-health will be a great foundation for incredible innovation in the future.
2. If lifestyle interventions were viewed as a something of critical importance — not just an ancillary service — what would that look like? What kind of dashboards would help healthcare providers, nutritionists, trainers, and the like understand which patients were the most “impactable” or “ready for change” in addition to their risk of hospitalization? What kind of effort would be made at every level if this were a priority? Would the fire department have a look in the patient’s fridge as well as their medicine box?
3. Telehealth is a big part of the future. Both the fire department project and the future mobile health van are attempting to take care to the patient rather than the other way around. Telemedicine also makes it easier for healthcare organizations to be proactive because they are not asking patients to come into a doctor’s office on their own initiative. It removes many barriers. There will likely be more and more diagnostic technology that could allow the “boots on the ground”, such as the fire department paramedics, home health aides, nurses doing home visits and the like to gather diagnostic information on the spot and send it to the specialist to be analyzed.

Thank you for reading! I would love to hear your comments.

To learn more about me (the author), check out: my LinkedIn and Twitter accounts.

If you liked this, click the💚 below so other people will see this here on Medium.