The healthcare industry is generally perceived to be behind the times when it comes to technology, and nowhere is that more apparent than in the analytics realm. National stakeholders are still having a debate about the best way to digitize patient records, and many organizations have yet to make the switch from paper to electronic health records (EHRs). As a result, data scientists and other analysts often have a hard time getting the data they need to run basic business intelligence applications, never mind more in-depth healthcare data analytics projects.
But some providers are forging ahead with cutting-edge analytics initiatives. For example, Kaiser Permanente, a large integrated health system serving mainly the western U.S., was one of the first medical organizations to implement an EHR system and is now using the data created and stored there to change the way it delivers patient care.
The EHR system, which Kaiser began implementing in 2003, has enabled the company's data analysts to focus on deeper questions about clinical care than they could before it was available, according to Terhilda Garrido, vice president of health IT transformation and analytics at Kaiser. Garrido is at the head of the organization's analytics efforts. She and her team have worked on a variety of projects, including one to alert doctors when patients with infections are likely to go into sepsis and develop potentially life-threatening complications. Others aim to identify hospital patients who are likely to be readmitted and smokers who might be receptive to help in quitting.
Garrido also is involved in the company's selection of analytics software and systems. Kaiser primarily uses SAS analytics tools and SAP's BusinessObjects business intelligence software to support its data analysis activities against the EHR system.
Clinical analytics waiting room
But it took Garrido a while to put her background in engineering and biostatistics to work in a clinical setting. She said that when she earned her graduate degree in biostatistics in 1983, the only statistical analysis being done in healthcare was in medical research. "You had to come in through the research arm," Garrido said. "I think it's wonderful now that people are much more aware of the importance of [clinical] data, but I would observe that that hasn't always been the case."
Because few healthcare providers were hiring clinical statisticians, she had to look for work elsewhere. Her first job, in 1986, was doing economic modeling for the European Economic Community, one of the predecessors of the European Union. After that she did internal statistical consulting at AT&T. It wasn't until 1995 that she found work at Kaiser in a new division called the Medical Economics and Statistics Department, a group that analyzed clinical data for meaningful correlations and was a forerunner of today's more expansive healthcare analytics teams. Garrido said the team's work was hampered by a lack of useful data, but it did help get the organization thinking about ways to use data analysis in the care delivery process.
Even though Kaiser adopted some technologies more quickly than other healthcare providers, it still has some room to grow on analytics, according to Garrido. She said outside factors will compel it to make analytics projects even more central to its clinical operations over the next few years, requiring an even more systematic approach to data analysis. Both private insurers and public payers, like Medicare and Medicaid, are pushing providers to adopt accountable care practices, which will result in organizations being paid for the quality of care they deliver rather than the quantity of services provided. Additionally, a provision of the federal Affordable Care Act penalizes hospitals with high readmission rates. Garrido said those two factors make it imperative for providers such as Kaiser to understand their operations at a deeper level and be better able to predict which patients will need extra attention.
"There's enormous pressure on the industry to do more with less," she said. "Healthcare reform is forcing the question; our national budgets are forcing the question. It's incumbent on those of us in the industry to really try to identify where we can leverage those resources we have to provide the best care we can for patients."
Healthcare analytics fights flu
Toward that end, Garrido is currently working on a Kaiser initiative to minimize readmission penalties through the use of analytics. Called KP Outpatient Safety Net, the program seeks to identify patients who haven't received appropriate follow-up attention after being discharged from the hospital so proper care can be arranged. For example, patients who undergo splenectomies have an increased risk of contracting the flu. Garrido's team will analyze the EHR data to make sure that those patients have received vaccinations and alert clinicians if they haven't.
More on the
growing role of
Read how BI can promote EHR adoption
See why data can sometimes go unused
Garrido said such initiatives are focused on detecting and filling gaps in patient care. Healthcare is complicated, and even the best clinicians miss things at times. But healthcare data analytics can help compensate for that, she said.
It remains unclear how quickly the rest of the healthcare industry will be able to adopt the same kind of analytical approach to patient care. Garrido said that about 50% of physicians work in one- or two-doctor practices. Their time tends to be consumed by day-to-day patient care activities, meaning they don't have the time -- or, often, the expertise -- to think about things like database architecture and other technical concerns that come with analytics projects. In addition, the pace of change in IT makes it hard for small practices to keep up.
Nonetheless, Garrido is hopeful that data analysis technology and know-how can be diffused more broadly throughout the industry so a larger number of organizations can take advantage of the benefits of healthcare data analytics. "There are many forces for the good that are improving opportunities for analysts and statisticians [in healthcare]," she said.
- A Guide to Predictive Analytics –TIBCO
- Deploying Predictive Analytics Models –DataRobot Singapore Pte Ltd
- Assessing the Impact of Predictive Analytics –Hewlett Packard Enterprise
- Predictive Analytics with Machine Learning –Estafet Ltd