Do I really need a healthcare enterprise data warehouse?
One of the questions we often hear is, “Do we really need a healthcare enterprise data warehouse?” In this post, the first in a series, Health Catalyst Vice President Mike Doyle will share his thoughts on this important question from the perspective of both technologist and IT leader. One of the newest members of our team, Mike has led IT, data warehousing, and business intelligence teams at innovative healthcare provider organizations.
The Real Question
I’ve been hearing this question more and more frequently from CIOs and IT Directors following presentations and panel discussions. This is often followed-up with “My reporting application vendors are all telling me they have data integration capabilities. Isn’t that enough for what I want to do?” First, let me acknowledge, that is an important question. It indicates a dramatic and really positive shift in the way healthcare CIOs are thinking about the value they can provide. What we see happening now is physicians, nurses, executive leadership, state and federal agencies delivering the message that data are important, that measurement and key performance indicators are important, and that their organization must be more agile, facile, and fluent with their data. Implicit in the question, however, is a reluctance to employ a healthcare enterprise data warehouse solution to achieve those goals. Weird. Data warehouses used to be “cutting edge.” They were the original “Big Data,” and they spawned an entire industry of extract/transform/load or “ETL tools.” All the cool kids were getting them! What happened?
Here’s what I suspect happened:
- Many healthcare organizations tried. But most data warehousing projects within healthcare never delivered anywhere close to the value they had in other industries like retail, manufacturing, etc.
- Many other CIOs realized they had plenty of other work that needed to get done and a growing list of vendor partners ready to help them get that work done.
With 20/20 hindsight, we should have expected the lack of healthcare analytics tool adoption by many organizations. Very few individuals−and no vendors−had figured out a way then to reliably integrate the cornucopia of data sources in healthcare. In short, healthcare organizations lacked the right partner to supply people, process, and technology, until now. What the question, “Do I need a healthcare enterprise data warehouse?” means to me is that the first question organizations should ask, “Do we need better access to our own data?” has already been answered−and very strongly in the affirmative. As a result, healthcare IT leaders have been given much-needed support to begin exploring ways to build an infrastructure for performance measurement.
Go Beyond Data Acquisition
Hopefully, everything said to this point helps you understand that we feel “Do I need a healthcare enterprise data warehouse?” is an important and timely question. It also is important to acknowledge the claims that business intelligence (BI) tools provide a rudimentary level of data integration functionality are at least partially true. But consider this: To help answer the most critical questions in healthcare today, you will need to repeatedly and reliably produce data about your organization’s performance−often in comparison with other organizations−that combines clinical, financial, quality, cost, and patient experience data. Be prepared for your Chief Clinical Officer to request, if she hasn’t already, a monthly summary of your health system’s value, as defined as “outcomes per dollar spent.” The ability to bring one or more data sets together may help you demonstrate why a healthcare enterprise data warehouse is needed to answer the Chief Clinical Officer’s value question and other queries. However, we would not recommend that you build an analytics strategy around BI tool “integration-lite” functionality. Savvy healthcare IT leaders strive to make data acquisition repeatable. Your tools should reliably extract critical data that are currently locked in EMRs, claims, and billing systems. You also will want bullet-proof tools to automate the integration of these disparate data sources so your company’s most important asset−people−spend their valuable time analyzing, not acquiring, your second most important asset−data. When you empower your company to operate at this level of analytic maturity, you will be a hero. You have a choice–where in the technology “stack” should you put the integrated view of data? Your reporting vendor advertises this capability–should you put your data in the reporting layer? Well, what if you do, but then later want to switch BI tools because one of them is getting, as we say in Minnesota, a bit too “spendy”? You are, as they say, locked in. To migrate your integrated data to a different platform at this point likely means giving up access to those data, or at the very least, incurring a significant capital investment to extract and load your data into another tool. Our recommendation is to push your integrated data down to a lower level of the “stack.” In this case, the relational database: It’s tried and true technology that has been around for 30+ years. It provides a lot of flexibility and a long runway for you to work with:
- Want to migrate to a different relational database platform in about 15 years? Probably won’t be a big deal. The next database you choose will have tables and views, just like have had since the early 1980s.
- Want to invoke a Big Data tool in the future? No problem–most of the database vendors in the marketplace have that on their product road maps. Big Data plays well with relational databases–it’s in both approaches’ best interests to complement each other.
- Want to layer on a different, more capable, less-spendy BI tool in the future? You can do that too, if your integrated data are stored in a relational database.
Very likely, your data will outlive your choice of BI tool. Storing your integrated data in the database layer makes a ton of sense to us. It’s also a design pattern that has been proven in practice not only across many industries but also over several decades. What has a healthcare enterprise data warehouse done for your organization?