Hospitals are increasingly focused on leveraging predictive analytics to streamline their services — but how many organizations have successfully met this goal?
Since 2014, Buffalo Grove, Ill.-based research and consulting organization Healthcare Center of Excellence has been tracking how the healthcare industry uses data analytics. J. Bryan Bennett, founder and executive director of HCOE, spoke with Becker's Hospital Review about the results of the organization's third annual State of Population Health Analytics report, which integrated interviews with healthcare executives, survey responses from hospital employees and findings from recent research on health IT.
Here's what he had to share:
1. Health IT is a spectrum, from EHRs to decision support. The State of Population Health Analytics report evaluates hospitals' evolving use of health IT 
based on a model developed by Mr. Bennett, called the "healthcare transformation change model."
The model predicts a technology continuum from non-analytic to analytic-focused, with the implementation of complex, analytical decision support services as the ultimate goal.
The suggested timeline begins with descriptive technology (characterized by EHR implementation), and moves through phases of diagnostic technology (characterized by integration of data sources), predictive technology (characterized by analysis and modeling) and prescriptive technology (characterized by real-time decision support).
In 2014, many healthcare facilities reported EHR implementation; in 2015, a substantial amount of healthcare facilities had begun integrating data, like clinical and billing reports.
"In the 2016 analysis, we determined that the needle moved just slightly from 2015. It's past integrating data sources, but not as close to analysis and modeling as people would like it to be," Mr. Bennett says. "It's because people are gathering data, but don't know what to do or how to perform the analytics yet."
2. High-level data analytics requires a broad foundation. Before approaching predictive or prescriptive analytics, successful healthcare organizations will need to lay the groundwork: collecting and integrating relevant descriptive data.
"People really want to do the 'sexy' stuff — the quantitative modeling and predictive analytics," he says. "But they don't want to go back and do the grunt work. If you don't have processes for data integration in place, the program is not going to succeed."
Mr. Bennett stresses that, outside of technological development, any new project also requires a well-developed foundation on the organizational level — including strong leadership, a talented team, well-defined processes and clear workflows.
"In 2015, we ran a study where we asked participants to write down their top five challenges for implementing healthcare analytics," he says. "Most people think the biggest challenge would be data management or technology, but the biggest challenge was leadership."
3. Teams need technical skillsets — but soft skills are important, too. "Leaders need to direct and inspire," he says. "You need to have both of those to be a good leader."
In the 2016 State of Population Health Analytics report, leadership is still an issue identified by respondents, according to Mr. Bennett.
"Healthcare has unique needs," he explains. "You have a lot of people who might not have the business experience to manage a really strong and opinionated group of followers, like physicians, when trying to implement new programs. You need to invest in leadership development."
Establishing a strong team of data scientists can also pose a barrier to creating a competitive data analytics program.
"Finding data scientists to work in other industries, like financial services or insurance, is hard enough. And then to find people who also have healthcare experience is even more challenging," he says, adding that small and mid-sized healthcare organizations are at an even greater disadvantage, since resources for new hires might be tight.
4. Prescriptive analytics are just one of many tools. As analytics move closer to the predictive and prescriptive phrases, Mr. Bennett cautions that these tools are decision support — not a replacement for traditional medical care.
"The system should not be trying to tell the doctor how to practice medicine," he says. "The system should be a support for the doctor, who makes that final judgement."