By Russ Staheli, Health Catalyst
Twitter: @HealthCatalyst
When you invested in an analytics, I’m guessing you signed up for data-driven solutions, not additional healthcare analytics pitfalls and inefficiencies. Yet that’s exactly what your health system is dealing with — problems and inefficiencies that keep popping up when all you really want to focus on is improving quality and cost. I see this a lot when I’m out in the field helping health systems architect their analytics solutions.
Many of the pitfalls have to do with the types of healthcare analytics solutions organizations select. This is because not all analytics solutions can provide the deep insight health systems need. The waste and inefficiencies set in after the analytics purchase and implementation. Such inefficiencies are the result of how the organization chooses to use the analytics solution.
Finding a sustainable approach to using analytics solutions without problems and further waste can be a challenge. But it’s possible for health systems to avoid these concerns if they better understand how to choose a good analytics solution. Then, after implementing an appropriate solution, the health system can prevent the common inefficiencies that hinder cost and quality improvements by making adjustments to how the organization uses the new analytics system.
3 Common Healthcare Analytics Solutions and Their Pitfalls
Many health systems opt to implement one of the following three types of analytics solutions. Although the solutions may initially show promise, they inevitably fall short of expectations.
1. Point Solutions
When developing an analytics platform, some health systems deploy one or more best-of-breed or point solutions. These applications focus on a single goal and a single slicing of the data. For example, the solution might focus solely on reducing surgical site infections or central-line associated blood infections. One problem with this approach is something called sub-optimization. While the organization may be able to optimize the specific area of focus, these point solutions can have negative impacts both up and downstream. They also don’t offer much in the way of insight outside of the specific area of focus.
Another problem is what I call the “technology spaghetti bowl.” When a hospital or group practice has only a few point solutions in its dish, a small IT shop can provide adequate support. But with additional point solutions (consider them noodles, if you will) an IT department finds it all but impossible to unravel all the disparate noodles in the spaghetti bowl. Imagine there are 10 different point solutions, and it’s necessary to update coding standards from multiple source systems in each of these point solutions. You’ll end up with sauce on your face by the time you’re done.
Eventually, one or two of the senior IT employees may own this spaghetti bowl mess with a huge dependency placed on these individuals, creating an unstable house built out of playing cards. While this situation works well initially, if either of the individuals leaves the organization, then the house of cards will crumble, leaving a mess for someone else to clean up.
In addition, point solutions typically result in multiple contracts, multi-cost dependencies, and multiple interfaces. These, in turn, lead to complexity, confusion, and organizational chaos that impede improvement and continued success.
2. Electronic Health Record System
While implementing an EHR is clearly a necessary step towards data-driven care delivery, an EHR system alone is insufficient to enable an enterprise-wide, consistent view of data from multiple sources. The conversion of clinical data from paper to an electronic format is a necessary step; it allows for the use of data to improve care. However, without a way of organizing all sources — clinical, financial, patient satisfaction, and administrative data — into a single source of truth, a healthcare organization is unable to harness the analytic power of the data.
Some EHR vendors are beginning to come out with data warehouse offerings that run on top of the EHR’s transactional database. However, these data warehouses still have the limitation that they don’t aggregate data from a variety of external sources — either because the vendor can’t or won’t. Some EHR vendors are becoming willing to integrate some external data, but they are years behind analytics vendors in terms of doing it well.
3. Independent Data Marts in Different Databases
Independent data marts that live in different databases throughout a health system provide limited analytics capabilities because they can only deliver little sources of truth from the different siloed systems. Take, for example, the ADT (admission, discharge, transfer) information that lives in the EMR. When there’s a need to analyze the ADT information and the role it plays on costs, analysts move the data over to the costing system with the independent data mart model. Requests like this one can happen over and over for many different types of scenarios, which ends up becoming time consuming. It also slows down the entire system as analysts repetitiously bombard the system with requests for each new use case.
Avoid the 3 Most Common Healthcare Analytics Pitfalls with an EDW
For actionable clinical, financial, and operational insights that meet your needs across the enterprise, an enterprise data warehouse (EDW) is the best solution we’ve seen to date. An EDW captures, aggregates, and analyzes data in near real-time from the EHR and other internal and external systems that reside in silos. Only when clinical data is married with financial, administrative, and patient satisfaction data can breakthrough analytics be realized, the kind that lead to significant care improvement and efficiencies.
It’s rare to find hospitals or medical practices that have built a true enterprise-wide healthcare EDW because it’s very difficult to do without a flexible, agile architecture. Implementing an EDW that utilizes a late-binding architecture delivers the needed flexibility as well as a single source of truth across the entire organization.
2 Common Sources of Inefficiency
Even if you implement the best analytics technology in the market, the improvement efforts might fail if you get bogged down in the way you’re using the solution. These inefficiencies can be avoided, however, by making a few changes.
1. Report Factory Inefficiency
The report factory approach uses an analytics platform alone and assumes that if you build it, people will come.
When they do come, the first indication is a backlog of report requests to the IT department. Why? Data can be addictive, and once users get a taste, they want more and more data until a dependency is formed. It’s not uncommon to see queues of report requests into the thousands and growing. If the IT department can’t keep up, clinicians and department heads decide the analytics platform or the IT shop is too slow, and they hire their own dedicated analysts or architects. This situation creates its own kind of chaos that ultimately renders your chosen solution redundant.
2. Flavor of the Month Inefficiency
In order to avoid becoming a report factory, organizations sometimes opt for a different, more measured approach when developing their analytics platform. This, however, can result in a project-by-project or flavor-of-the-month approach to analytics.
Here’s how it happens: the hospital or medical practice may tackle projects based on some sort of prioritization, possibly responding to squeaky wheels or management’s pet projects. Initially, the organization may experience employee enthusiasm, improved care, and reports that really do help.
Inevitably though — especially once you move on from the original projects and tackle new ones — clinicians and staff grow dispirited. It becomes difficult to keep up on more than a handful of projects. Most importantly, quality and cost gains made in the initial projects are quickly lost.
How to Avoid Inefficiencies: A Robust Deployment System
A good deployment system addresses both the flavor of the month and report factory inefficiencies. There’s a great blog about this topic, but below are some of the highlights from the blog.
- A good analytics deployment system provides a methodology for effectively getting clinicians and other stakeholders throughout the organization to embrace your analytics solution and to use data themselves to drive decisions.
- A robust deployment system leverages existing personnel to form permanent, cross-functional teams within each area where the healthcare data analytics will be used. It’s still a one-project-at-a-time approach, but it’s sustainable since teams take ownership for their own projects in their own space — and are accountable for producing and sustaining results.
This article was originally published on Health Catalyst and is republished here with permission.