Hospitals have only recently begun to invest in predictive analytics, but interest is accelerating. A 2016 Grand View Research report indicated the global healthcare predictive analytics market was worth $1.48 billion in 2015 and is expected to grow at a compound annual growth rate of 29.3 percent through 2025. Yet as many hospitals — particularly those that are small or independent — sign on with data analytics, they often find they are in over their heads.
This content is sponsored by GE Healthcare.
At roundtable discussion sponsored by GE Healthcare at the Becker’s 8th Annual Meeting in Chicago, a CEO of a small Midwestern hospital summed up the issue: “We don’t know what we don’t know.” His 25-bed hospital recently conducted a “rip and reinstall” of its EHR and switched vendors. However, the new system “is so robust we get a little bit lost,” he said.
Clinical, operational and financial data collection, normalization and storage are enough to make hospital executives’ heads spin. The complexity of these tasks can often overshadow or distract from the actual goal of data analytics: To query the data and produce an actionable answer that creates a return on investment.
During the roundtable discussion, 25 hospital executives discussed the biggest data analytics challenges they face and how they are beginning to meet those challenges. Here are the top four challenges they discussed.
1. Overwhelming amounts of healthcare data.
Many of the executives in the room lamented big data sets. A nurse CEO of a 100-bed rehabilitation hospital part of a large academic health system described her struggle: “We are deep-diving all the time into analytics. Sometimes I think there is just too much data to really plan from it. There is a lot of information and not enough time to segment it.”
The problem is exacerbated for small or independent hospitals that may be unable to rely on the infrastructure of a larger system to help with clinical and operational analytics projects. The president and CEO of a medical center in North Carolina said, “We are a small, independent community hospital, not part of a larger system. We have too much data and we don’t have the time, resources or manpower to dive into all that.”
2. Too much variation in clinical, operations and financial data.
Another common issue executives voiced was the lack of clinical, operations and financial data integrity and data integration. The first issue stems from data quality — data must be entered in a standardized format to be useful and trusted by clinicians. The CEO of a rural hospital in the Southwest confessed, “I’m embarrassed to say this … We did a data sort on origin point for falls in our EHR and we found that there were five different ways we were identifying [our city].” Sometimes staff entered the city name in full, other times it was abbreviated with spaces or abbreviated without spaces. “It became a training issue for the staff,” said the CEO.
Data integrity issues can also stem from attribution. A vice president of medical management from a large Midwestern health system described his system’s struggle with data attribution, particularly as physicians change shifts. “In large teams, someone is admitting the patient, someone is discharging the patient and someone is in between. So who are we attributing the data to?” he asked. The question is important for executives to grapple with as they roll out quality improvement initiatives, especially those that hold individual physicians responsible for patient outcomes.
The second part of data variation issues comes from integration of health data sets across an organization. This can be difficult as organizations grow up with different EHRs and develop different methods of storing data. It can also be difficult if patients leave the system for certain aspects of their care. The CEO from North Carolina said his medical center often struggles with patients occasionally going to other healthcare facilities. “Big picture, they are getting the care they need, but from our perspective, it doesn’t look like they are getting the care they need,” he said. If his system wants to measure on the percentage of patients receiving mammogram screenings, it’s actually a difficult question to answer because they don’t know if patients are going elsewhere.
“Everybody has to speak the same language,” said the CEO of the rural hospital in the Midwest. There’s no one standard “that says see it this way; taste it this way; smell it this way.”
3. Lack of actionable outcomes.
Lastly, perhaps as a result of the first two challenges, executives discussed their inability to do anything with their data. The CIO of a large healthcare system on the East Coast described his system’s work with data analytics as fairly robust. “We are trying to use data for a number of things: to drive care coordination; reduce cost on the readmissions side; we are using it to drive what service lines we grow and which ones we shrink; [to determine] where we should put our next outpatient center,” he said. However, the system still struggles with determining which data points are the important ones in the masses of healthcare data points the system has collected or bought. “We can slice and dice something 15 different ways and get locked into analysis paralysis.” The end result is a stalemate.
The system COO of a primary teaching hospital in the Midwest put it simply, “We have reams of reports. I need something that’s actionable.”
Many of the executives voiced that they were ready to move beyond simply recording and reporting. Asking pragmatic questions and using smaller data sets is the best place to start, particularly as hospitals and health systems look ahead to value-based care and a healthcare ecosystem that prioritizes outcomes.
“As a clinician, the world we still live in is a metric world as opposed to an outcomes world. It’s all about reporting,” the dean of a medical school in the South said. “The buy-in from clinicians is not huge in a metric world.”
4. High cost of connectivity.
Much of the frustration executives discussed stemmed from the fact that connecting disparate data systems and finding solutions to the aforementioned issues is downright expensive — and the costs of these projects often compete with more pressing needs.
For example, the CEO of a 25-bed hospital in the Midwest discussed the cost burden of connecting various medical devices to the EMR. “For a smaller facility, every time you hook up a machine to your EMR, you’ve just dropped $25,000. That is huge for a small hospital,” he said. After bringing in several products, “you have to interface them and they have to speak to each other and then you’ve got to have three CIOs in your building to make sure it all works for the first 60 days. Sometimes you just go, ‘Screw it.'” The CEO said he has found initial investments in software or hardware often end up costing 25 percent more than the sticker price to implement them and ensure they produce useful data. “We just don’t have it,” he said. “Nobody has that in a small hospital.”
The executives said they want to invest in data, but other, more immediate issues often take precedence. As the CEO of a rural community hospital in the Southwest put it, even if a vendor can offer a solution that will optimize staffing and save his hospital $5 million a year, that’s great — but first he needs to find a nurse to work a shift from 3 p.m. to 11 p.m. “It comes down to resource utilization at a macro level,” he said. “I would love to optimize my staffing and I would love to have right-time, right-size, right everything, but right now, for the next three weeks I have nobody to work nights.” He added, “It’s not so much a question of interest in using for me. I love data … It’s just about how do I get through the day and how do I get through the week?”
Start with “Small Data.”
The executives in the room agreed — healthcare data analytics are promising. But often the process of reporting, recording, standardizing, normalizing, storing and extracting useful information is daunting, especially at small or independent hospitals already strapped for resources.
While there may be no silver bullet for these four challenges, the consensus was hospitals need to start small. Implement staff training to standardize data entry. Focus on “small data,” or digestible data sets. These smaller, more cost-effective steps can produce modest wins and efficiencies from the reams of data hospitals already have on hand, which will start to move the needle, generate ROI and ultimately improve patient care.
More articles on health IT:
Senior US cybersecurity specialists’ Q1 salaries average $118K: 4 study insights
DHS investigates malware targeting critical infrastructure
Pew to Congress: Don’t cut funding to ONC
