Health Information Technology: Why Point Solutions Strike Out
Health information technology has something to learn from the way statistics are used in sports. And there is probably no sport more obsessed with data and statistics than baseball. Practically since the game was invented, fans have obsessed (often heatedly) over the big stats, such as batting average (BA), runs-batted-in (RBIs), and earned-run averages (ERAs), as well as little things such as which left fielder hit the most late-inning doubles in the month of August.
Yet for all the attention paid to the most minute, obscure pieces of data, it took a guy named Bill James and the analysis he provided in his Baseball Abstract to turn all those little pieces of individual data from multiple sources into actionable information. That led to the approach made famous in the book and movie Moneyball, where Major League Baseball teams use the same core data in many different ways to determine where runs come from and which performance measures are important. All to help them determine which players will give them the most for their money.
Much of the health care industry today finds itself where baseball was pre-Bill James. Many hospitals and health systems use point solutions to look at and evaluate a single initiative or point of concern. Most of these point solutions draw data from a single system, such as an electronic medical record (EMR) or a separate financial system. Granted, some can bring in data from two or three sources.
While they generally do a good job for what they’re designed to do, ultimately the perspective they provide is very limited – like judging a ballplayer on batting average without taking the arguably more-important on-base percentage (OBP) into account.
A better approach is to use an enterprise data warehouse (EDW) specifically designed for healthcare to aggregate data from all the different source systems in the hospital or health system and make it available to analytics applications. With an EDW in place, analytics applications can draw from multiple source systems at once to form a more complete picture that will help organizations create meaningful care process improvements.
Why Healthcare Point Solutions Strike Out
Most organizations purchase a point solution because they’re feeling a particular pain, and they want it to stop. They may have other pains as well, but they don’t notice them at the time. Once they fix the first pain another may crop up, so they purchase a point solution for that. And so it continues until they have all these individual solutions. It’s like a physician treating individual symptoms instead of looking at the entire body to see if there is something bigger going on.
That’s only the beginning of the problems, though. Because point solutions draw data directly from the source systems while they operate, running a complex analysis of their data can bog those systems down. After all, those systems were designed to deliver small bits of data to individual users, not run complex queries coming out of multi-functional applications. If point solutions are drawing data from two or three sources, the workflow of the entire organization may be affected.
The time factor to complete the analysis is another issue. Take a process improvement committee that meets once a month. They determine an issue they want to investigate and put in a request to IT to run the analysis. When they meet the next month they look at the analysis, but of course the information is already weeks old and may not even apply to the current situation due to changes that have occurred since the last meeting. Even if it’s still applicable, they’ve already lost four weeks where they could have been making meaningful changes to improve care. And if they develop a new idea or want new information, the lengthy process repeats.
Finally, while some point solutions can draw from multiple sources they are very limited in what they can do. If you try to run a more complex query, the cross-connections get convoluted and you’ll run into problems with data integrity. At that point, you no longer have the single source of truth you were seeking; instead, you have a distorted viewpoint that can lead you to false conclusions. Which is how baseball teams end up paying millions of dollars to sign players they later can’t wait to unload on some other unsuspecting team a couple of years later.
Healthcare Enterprise Data Warehouses Deliver MVP Performance
Building a comprehensive EDW rather than purchasing individual point solutions overcomes these issues while improving the ability to get to a single version of the truth.
To begin with, because an EDW isn’t tied to any particular system or issue it provides a tremendous amount of flexibility in terms of what it can discover. In other words, an EDW won’t just solve the particular pain point you’re facing – it has the capability of allowing you to address all your pain points as they become important. Even the ones you don’t know about yet.
An EDW also has no impact on the performance of your source systems. It draws the data on a regularly scheduled basis (usually at a time when source system utilization is at its lowest), and keeps it in its own data stores. As a result, analytics tap into the EDW and its data marts rather than the source system, ensuring that source systems continue running at peak efficiency.
This ability to run the analytics against the EDW has a huge impact on speed-to-decision as well. Rather than putting in a request for analytics and receiving the data weeks later, performance improvement committees can use analytics tools to look at every ounce of data available to them, as broadly or narrowly as they need it to be, and get answers during the same meeting. Which means they can make decisions on the next step today instead of next month. The immediacy of these answers energizes the team and inspires them to dig further – as opposed to the momentum that’s lost when weeks or even months pass between questions and answers.
Here’s a concrete example of how that works. We have a client that spent years working with a company on a labor management system. It took many hours to obtain the data on procedures, time logs, etc. and send it to the company for analysis. Then they’d receive a report days or a week later. The problem was by that time there were new people in the hospital, the data was not reflective of the hospital’s current situation. In essence they were paying to find out what they didn’t do or could’ve done, neither of which affected what they should do now.
Using real-time labor data through the EDW, they were able to save two percent of their total salaries and benefits. That may not sound like much until you realize it was $10 to $15 million in the first year. The ROI on that application has paid for the entire investment in their EDW and then some. Every savings or process improvement they’ve made since then only improved the positive ROI they already have.
Drafting the Right Healthcare Data Warehouse
This is not to imply that any EDW will yield these types of results, such as those designed for other industries like financial or retail. The problem is that health care is much more complex than other industries, which makes it difficult to get to the data you need. There are more parameters to capture, and the data changes as rapidly as vital signs and patients themselves change.
For an EDW to work in a health care setting, it requires a late-binding schema rather than the typical star schema or early-binding models used for other industries. A Late-Binding™ Data Warehouse allows you to get data out of the transactional system at the most detailed, lowest level of granularity with the minimal amount of transformation. The data is available and ready to use, but it hasn’t been committed to any particular relationships, so you have a lot of flexibility in terms of what you can do with it. You can then transform and bind it only when and if you have an actual need for it.
For example, suppose a cardiovascular physician is interested in obtaining data in order to look for variations in care, with an ultimate goal of improving outcomes while getting costs under control. Because the data is interrelational, the physician can select whatever data is needed, from whatever source system, to get the answers he/she needs. It’s the equivalent of a baseball team being able to choose whomever they want in the draft whenever they want – without having to wait for their turn in a specific round.
Just as you don’t draft a singles hitter and expect him to bat cleanup, you want an EDW that is born out of healthcare methodologies and designed specifically to drive care process improvements in healthcare settings. Anything else will fall short of delivering what you need.
A Winning Strategy
Whether you’re analyzing baseball stats or healthcare data, the more data you can draw from, and the easier it is to manipulate without loss of fidelity, the more actionable information you can create.
While point solutions can solve immediate problems, they are too limiting to use as a basis for organization-wide change. And the more focused you become on process improvement, the more those limitations will get in your way.
An EDW designed specifically for the complexities inherent in healthcare data will give you the flexibility and immediate feedback you need to make meaningful, lasting changes and create a winning strategy for the future.
What have your experiences been with point solutions? Have they been flexible in helping you solve problems or are they limited? Are you able to get timely answers or are you constantly waiting for data to make decisions? How many are you currently running? And have you noticed any performance problems with source systems when you run queries?