When it comes to data and analytics projects, Geisinger is about as complex as it gets – and has been for a long time. But even an environment as analytically mature as that one has its challenges with data strategies.

"One big challenge for us is moving from a culture of one-off implementations to one of data-driven operations done at scale," says Marko.

Marko adds that Geisinger has had a data-centric culture for a long time, including an EHR as well as data warehousing and storage technologies that have enabled employees to seek out data when making decisions.

"Whether it's a clinical question about care optimization, whether it's a population health question, or a business or operational question, a common, knee-jerk reaction is: Let's run some numbers and see what it shows," says Marko. 

But there's a flip-side to having such a data-centric culture.

"In the not-too-distant past, we'd say, 'Here's a problem we want to solve: Here's a ProvenCare system where we want to optimize a care pathway, here's a diabetic population where we want to standardize care,'" says Marko. "We'd build a whole system around that and leverage all the pieces of information we need to improve the quality of care for diabetic patients."

All that is great, of course. But on a large scale, over and over, it tends to become inefficient. With so many data-hungry clinicians all clambering for numbers and dashboards, the gains can sometimes get diluted.

"We found that we'd built a series of 100 different one-off tools and projects – each of which is working very well and is helping people, but every time you get a success, lots of people want to do that for their patients," he says. "The problem is, building things in a one-off fashion just doesn't scale.”

So recently Geisinger has been concentrating on cultural and technical changes to try to focus on analytics tools and strategies that are "very generalizable and scalable – so, if this approach works for this population or this problem, we've already built an infrastructure that, with minimal work, can be converted to one or two or five or ten other problems that are fundamentally very similar but just happen to apply in different domains," says Marko.

Granted, Geisinger's challenges are often of a different sort than those facing hospitals of more modest scale and analytic maturity. But Marko says there are some basic guiding data principles that everyone should keep in mind, regardless of size.

"The first thing you want to do, is figure out, what exactly is your goal, with using your data? You want to be fairly explicit about that," he says. "It's exciting to use data for things, and explore information. You can hire lots of people and buy lots of toys. But if you don't have a clear idea of where you want to want to be, you're never going to get there. You have to know why you're venturing down this pathway at all."

Second, "you want to define early what are going to be your markers of success," he says. "Data-related things are notoriously hard to develop metrics around. It's hard to monetize data related projects, sometimes it's hard to measure direct value from it.

The third key thing to keep in mind, the importance of which is hard to overstate, is that "data is more of a people business than it is a technology business," says Marko.

"Most of the technology you need to do just about anything with data in healthcare exists and is commoditized and is transactional – you can get it from lots of different places," he explains. "Where you really get the value, and what makes and breaks data and analytics programs, are the people you put around it. The people who build it, who maintain these things, who do the work every day.

Too many providers still think of data analytics as a "very techie thing," Marko says, but that’s not the right approach.

"You don't need a giant building full of servers to do stuff with your data,” he says. “But you do need a couple offices with smart people with talents who can construct a problem and come up with an answer."

This article is part of a series focusing on big data and analytics. Other pieces include: Do you need an enterprise analytics strategy? It depends. (But probably, yes.) and Analytics or BI? Centralized or federated data? Geisinger's CDO shares insights.

Twitter: @MikeMiliardHITN
Email the writer: mike.miliard@himssmedia.com

Like Healthcare IT News on Facebook and LinkedIn