Data Warehouses: Overcoming Governance, Integration Issues

The CIOs who participated in a panel that tackled strategies for creating data warehouses at the Healthcare Informatics Executive Summit on May 7 largely agreed that the main challenges in doing so centered around issues of governance and integration.

“I think that governance for us is still the greatest problem,” said Patricia Skarulis, vice president, information systems and CIO, Memorial Sloan- Kettering Cancer Center. “I mean it in the sense that there are still some areas where [departments] withhold their data from putting it into the warehouse.”

Skarulis, who was part of the “Beyond the Data Warehouse: Strategizing the Use and Analysis of Clinical Data for Meaningful Use” panel, one of several discussions at the HCI Executive Summit last week in Orlando, Fla., said that Memorial Sloan- Kettering Cancer Center in New York City has had a robust data warehouse for more than 25 years, which contains 1.25 million patient records. “We have every piece of information about them that we can put our hands on,” said Skarulis. “We not only know what’s happened to them clinically, we know how long they waited to see a doctor after the initial request for a visit; we know [the amount of time] between a blood draw and when they saw a physician in the office itself, as well as all the financials—what we billed and received. So, we have a very thorough resource used by everyone at the institution.”

Skarulis emphasized the importance of having an institution-wide policy governing data warehouse usage that discourages departments from withholding their data in separate databases. She said that the greatest fear within her organization is that a researcher could make a faulty comparison from using the data when publishing research. She said to mollify opponents, certain departments have been allowed to maintain their own separate databases, which receive feeds from the central data warehouse.

At Children’s Hospital and Medical Center in Omaha, an issue addressed by its governance committee is data definitions, said George Reynolds, M.D., vice president, CMIO, and CIO. Reynolds said that at his 143-bed hospital the governance process mostly focuses on data quality and agreeing on the meaning of the data points like for instance, length of stay, which can be four calculated four different ways.  

Children’s, as a part of its clinical analytics and business intelligence project, and its broader journey towards becoming a data-driven, continuous improvement-focused organization, has a federated data warehouse that feeds more than 40 dashboards to support inpatient care, as well as the organization's affiliated pediatric physician practices, addressing financial performance and system-wide operational issues.

“Each of our dashboards has a champion or group of owners, and they’re developed in an interactive, iterative way, so we can determine what goes in them, and give [users] the rights management for what goes in them, [which] is pretty simple,” said Reynolds. “We’ve done all the front-end work [with definitions], so we don’t spend a lot of time after the fact worrying about what these data points mean.”
 
Phase two for Children’s will be moving toward a single vendor solution (from the Verona, Wis.-based Epic) for its clinical information systems to provide more integration for its data warehouse.

Rick Schooler, vice president and CIO, Orlando Health, said that his nine-facility regional health system’s executive governance committee creates priorities for next steps for its data warehouse that are based on the institution’s focus on quality, which includes reducing readmission rates and harm events. He said that integration, and the culture shift necessary to drive it, has been difficult for his organization. Orlando Health has invested millions of dollars in automating both its back office and point of care operations, but has yet to have build an integrated location for patient information, Schooler said.

“The problem has been driven by all the disparate technologies that we’ve invested in over the years in hopes that we could tie them all together,” said Schooler. “In our organization this is where the big challenge has been to culturally agree that we are going to begin consolidating all these pockets of business intelligence. We’re actually going to begin consolidating the teams, which means we’re going to have fewer people because we’re going to use common platforms. We’re going to let their focus be on using the information, instead of finding it.”

To bypass some governance issues, Bobbie Byrne, M.D., vice president of IT at Edward Hospital, took what she called a “stealth” approach to creating a data warehouse for her 350-bed hospital in the Chicago suburbs. “I’d seen a whole lot of data warehouses that were 10 years in the making before you ever got anything useful out of it,” said Byrne. “I was certain that we would under-promise and over-deliver.”

Byrne said she wanted to show her executive team what her department had done, instead of what could be done, so she made incremental investments and took what she called an “analyst to analyst” approach, having one of her team members reach out to an analyst in each department to pull its data into the data warehouse. Over the past two and a half years, she and her team have been knitting together the more than 300 software products, 10 of which are core to patient care, to create its data warehouse.

Byrne said the issue at her organization is not who puts data into the warehouse, but who takes it out. “We all talk about getting the data into the hands of users, but we felt like in order to do that in the future, we had to pull it back in tighter so we could have a little more process and a little bit more governance around the use of the data before we could start pushing it back,” said Byrne. Her team used a tech-heavy approach to get control by only turning on certain Microsoft Reporting Services access to those who had received training and were a part of the data integrity team.

Point of Care Usage
Another challenge that was aired during the panel discussion was how to use data from warehouses at the point of care.

“Now we’ve decided that [our data warehouse] has been pretty successful on the back end, but it’s not successful on front end,” said Reynolds. “It’s not successful at the point of care.” Children’s first step in this journey will be to align with its ambulatory facilities with the eventual goal to share data in a meaningful way without employing an “army of people” to do it, he said.

Memorial Sloan-Kettering’s data warehouse is updated every night at midnight. Even though its data warehouse can be currently used to track infectious disease spread within the cancer center (detailed patient information like what rooms patients visit and who the patient interacts can be accessed in minutes), Skarulis says her organization has been discussing moving to real-time updates to fuel alerts and further analysis.