The top of the heap in information technology readiness is the vaunted Stage 7 of the progression of IT capabilities as created by HIMSS Analytics. Cedars-Sinai Medical Center, Los Angeles, is one of those so vaunted, but in many ways it is just starting on the journey to make it pay off, said Darren Dworkin, its chief information officer.
While the designation is “a fabulous accomplishment” and proof of a fully deployed medical record, Dworkin said even a Stage 7 organization has to set up the analytical capabilities, data warehouse tools and, most of all, validity and trust in the data that will figure into significant assumptions about how the health system will operate.
In a Monday session during the HIMSS convention in Chicago, Dworkin presented a case for why Stage 7 is only the beginning of meeting the challenge of business intelligence.
“Being at the point we are helps wire us and enable us to get to even the next level, which is to be an organization that can truly leverage things like predictive analytics,” he said. “You can’t start off one day and say, ‘Hey, I’d really like to be an advanced user of analytics,’ without having the information underneath it.”
The initial benefit of having the IT pieces in place is “to better understand how our organization works, and how we can pivot to all the changes coming ahead--whether it’s risk-based payments, whether it’s new partnership models--and for us, the realization is that understanding that data underneath and being able to leverage that data is really going to help unlock the [means] to being able to do that.”
Where business intelligence meets Stage 7 is around requirements to, among other things, standardize data to a single standard such as Snomed or LOINC, which are then loaded into an analytics program for analysis and visualization; to do that requires a data warehouse, said Dave Garets, the health IT expert who helped create the seven-stage model 10 years ago. To gain the Stage 7 designation, executives have to use the information on a regular basis to improve operations, not just be able to perform analyses, he added.
Data governance also has to be addressed in the analytics strategy, said Garets. “We want to see how you’re ensuring the accuracy and the cleanliness of your data.” He sees data integrity as the next hurdle in making good use of data fed from a multitude of clinical and administrative sources into commercially available analysis tools.
Those tools are “pulling crappy data, and so they’re displaying crappy data, but in a very beautiful way.” That doesn’t work, Garets said, and tools without data accuracy can ultimately create a bad taste for the end users. “They can create gorgeous dashboards, but people don’t trust the data.”
Cedars-Sinai has embarked on a strategy to determine where data is unreliable and fix the feeds, partly through a set of “superusers” who can weigh the conclusions of reports based on warehouse data and advise on whether they make sense, said Dworkin.
“When you start your journey with analytics, the challenge is that everybody wants to make sure that it has rigor and all the controls of quality in place,” he said. IT professionals can’t spend forever validating every single data element, but it’s important to match the expectations of the organization as they begin to release the information. As people begin to see it as the single source of truth, it’s crucial for them to know they can rely on it, and that means “making sure that you’ve created this backbone of trust. You may find the occasional quality error in the data set . . . [but] there is general overall trust in the information.”