Three Ways Health Organizations Can Improve Data Use

September 4, 2018
Donna Marbury
Donna Marbury

Volume 28, Issue 9

The amount of data in healthcare is growing, but that’s not necessarily translating to improved patient outcomes. Here’s why.

The amount of data in healthcare is growing, but that’s not always translating to improved patient outcomes. Why? The location, format, and structure of data continues to be siloed, which makes it hard for physicians to use data in meaningful ways. 

“A good example is social determinants of health, which might indicate issues with food insecurity or social isolation, or that a person was recently widowed,” says Simon Beaulah, senior director of healthcare for Linguamatics, a UK-based machine learning provider. If that information is not captured with other patient record information, a physician could miss it, he says.  “Social determinants of health can indicate a potential increase in patient risk, which is of particular importance in a value-based care environment.”

Here are three ways healthcare organizations can better utilize their data:

  • Combine clinical and claims data in a meaningful way

Too many organizations are still relying on separate data sets to make business decisions, says Zachary Blunt, manager of product management and population health at Greenway Health, an EHR and revenue cycle management software provider. “To have a complete view of the patients, this information has to be aggregated at the patient level.”

Many payers do not provide patient level data without a risk-based arrangement, Blunt adds. Organizations will need to consider this as they begin formulating their population health strategy.

But it’s not just clinical and claims data that needs to be better integrated.

“There are many slices of data that organizations will need to view,” Blunt says. “These include disease registries to get insight into the pervasive chronic conditions in your population, leakage tracking with office visits, wellness visits, and analyzing claims for care rendered outside of your organization, as well as identifying high risk patient populations with predictive tools.”

  2.  Incorporate unstructured data
Beaulah says 80% of EHR data is classified as unstructured, meaning it doesn’t fit into traditional EHR data fields. This unstructured data includes important information that can be valuable to patient diagnoses such, as clinical notes, images, and specialist reports like radiology and pathology documents.

Also, when physicians do have the opportunity to fill data into structured formats, they aren’t always doing so, says Beulah. “While many EHRs provide extensive predefined fields to enter structured information, there is growing evidence that indicates that the time required to fill in these fields is a major contributor to physician burnout. As a result, many of these fields are not completed, nor regularly updated.”

Clinician notes are a good example of EHR data that isn’t utilized to the extent it should be because the information is free text, so it’s not easily viewed and/or added to data analytics platforms, says Jay Anders, MD, chief medical officer of Medicomp Systems.

“As doctors are doing more, more of what they are saying and thinking is being lost.”

Some solutions are available. For example, Phoenix Children’s Hospital implemented Medicomp’s Quippe Clinical Documentation technology, and it has saved the hospital $1 million annually in transcription fees. The software works with existing EHRs to extract data from clinical notes and complete 86% of documentation, coding, and billing the day of the patient’s service.

The software also is being used to track juvenile rheumatoid arthritis and its effect on patients’ joints. “The ability to have intricate data collected and recorded every time the patient is seen, and trackable to the joint, is more valuable than just patient notes,” Anders says. “Those intricate notes can indicate how a disease is responding to treatment, and the patient can receive immediate care.”

Anders says more structured data is an important part of powering AI and machine learning technology that is coming quickly down the pipeline in healthcare.

“All of those systems are just as good as the data they have. The more accurate the data, the better they will operate,” Anders says of AI technology. “It’s a misnomer that doctors will be able to pick up a microphone and accurately create notes, and good data behind those notes. Good structured data that can be collected outside of speech and text have a long way to go.”

  3.  Launch an enterprise data warehouse (EDW)

An EDW serves as a unified patient registry, which can store clinical, claims, social, and other data from multiple sources, says Alexandra Gorman, assistant vice president of business development at Lightbeam Health Solutions, a healthcare technology company.
Healthcare organizations can use EDWs to populate reports; calculate clinical quality measures and risk scores; and view financial data that assist in achieving quality measures, she says. “A comprehensive view is critical for organizations and payers who strive to improve clinical outcomes, reduce costs, and steward healthcare resources wisely.”

Healthcare organizations searching for EDW vendors should examine the vendor’s ability to consume and transform large amounts of data from disparate systems, custom codes, and non-standard storing locations, Gorman says.

“Having a partner that has significant experience with data ingestion will eliminate trial
and error and will lead to a successful population health management project from the start,” she says. “Questioning your vendor will eliminate headaches and physician distrust in the future.”

 

Donna Marbury is a writer in Columbus, Ohio.

 

download issueDownload Issue : Vol 28 No 9