zentilia - Fotolia
Integrating systems can be a daunting task, but, if a healthcare organization is dedicated to improving access to data, integration must be done. One way an organization can do that is by creating an efficient infrastructure.
Christopher Hutchins, associate vice president of healthcare analytics at Northwell Health in New York, spoke with SearchHealthIT at the Predictive Analytics in Health Care Summit in Boston about how his healthcare organization has been working to create a successful healthcare infrastructure to facilitate better access to data. This included the specific people that need to have a role, as well as the technologies his organization is using to create this healthcare infrastructure.
What is one of the first steps in creating an efficient healthcare infrastructure to improve access to data?
Christopher Hutchins: The key for me, in the first phase of this, was to have clinical informaticists embedded in my team because what we found is, as we're mapping data from the EMRs, is if you know one EMR data structure, you know approximately one [EMR data structure]. And if you don't know clinical workflows, you don't know what the user interfaces look like; it's just data structure to you.
As a technician, you could easily get the wrong iteration of something as simple as a discharge diagnosis. So what we found early on was we had a lot of smart folks that just couldn't get it done because they didn't understand this as components. When we put a physician into that lead role and ... those clinical informaticists, we started seeing some big value in producing the analytics that we're going to deliver, and there's a hard and fast rule ... if it's clinical in nature, it goes nowhere until those folks have seen it. It's got to be done.
What is another important step in this process?
Hutchins: Next thing I had to tackle was enabling self-service analytics because ... you've got folks that are trying to meet business needs right now, and they can't wait for their organization's enterprise maturity to get to a point where it supports what they're trying to do.
So we did some things early on to make sure that we're enabling folks to do analytics, [but] we're doing it in an organized way so that we're not just giving them access to the wide open. So they can do whatever they want to do, but using a tool like Tableau. We can give them access to some enterprise data models that we developed that are very targeted to very specific things, like looking at readmissions, length of stay, predicting readmissions, some basic things like that.
Some of the folks in our quality area ... provide analytics for provider performance; they're looking at how to perform risk contracts, things of that nature. So we're enabling them to continue to advance and keeping their report request queues from exploding while we're trying to lay a foundation.
What are some other technologies Northwell Health is using to achieve an efficient healthcare infrastructure?
Hutchins: We made the decision to put IBM InfoSphere back ... we're using massive data management, reference data management, ETL [extract, transform and load] tools, and we've also got the unified healthcare data model which, for me, was an important thing to do there because [Northwell Health is] at least 20 years behind where a lot of other healthcare organizations are in terms of having some data warehousing structure. There's a handful of data marks that they put together, but they did not have a platform to do the integration and to do the scale that we had to.
Christopher Hutchinsassociate vice president of healthcare analytics at Northwell Health
We also have a tool called Health Analytics Gateway we purchased from another vendor that was designed specifically to help substantiate data into that data model [that] has about 1,500 prebuilt ETLs that drive the data ... into the landing strip. So doing some things to accelerate that was really important because ... the goal of the health system is to triple their ambulatory footprint over the next three years.
That's pretty frightening from my perspective because that just means more data integrations that we're going to have to figure out how to do. And every time that happens, it causes great consternation for your clinical folks because they're the ones that have to do the really hard work of figuring out the data models and figuring out the clinical workflows because, again, clinical workflows in any one organization are oftentimes the differentiator; that's oftentimes what sets them apart.
The difference between converged and hyper-converged infrastructure
CIOs: EHR problems and single-solution EHR integration
Patients can access health data at their bedside