In healthcare, claims data is often described as a river that is a mile wide but only an inch deep, while clinical data—the analogy goes—is a mile deep and an inch wide. Yet, with the widespread adoption of electronic health record systems by providers, claims data often gets short shrift relative to EHRs.

However, understanding the distinctive benefits of claims and clinical data is critical if providers are to properly exploit their respective values.

Isaac Kohane, MD, chair of the Department of Biomedical Informatics at Harvard Medical School, knows better than anybody the value of different kinds of data and their respective limitations. As Kohane points out, EHR data is definitely more detailed than claims data. However, he believes there are circumstances when the latter is more useful than clinical data.

For instance, Kohane contends that research shows that claims data is more predictive than genomic tests for parents who have had a single autistic child looking to understand the probability of a second child having a diagnosis of autism.

While emphasizing that he leverages EHR data for a lot of analyses, Kohane points to an ongoing “sponsored research agreement” with health insurer Aetna that provides Harvard researchers with access to tens of millions of anonymous claims records from across the United States used to examine relationships among diseases, their consequences, and their heritability.

As it turns out, a detailed family history is often the most informative data for understanding inherited disease risk. However, very few EHRs provide the capability to easily enter a family history and link it to the broader genealogy. “What we’re trying to do is make medicine a much more data-driven discipline,” adds Kohane.

Yet, Karen DeSalvo, MD, former National Coordinator for Health IT, cautions that the research community has “years of experience in playing with claims data and creating well-fitting models that can help tell a story, and we have a lot less experience with the EHR data.” While these models which leverage claims data can answer very specific questions, DeSalvo points out that what’s needed are broader clinical models that are more sensitive and less specific.

“I’m not trying to minimize the importance to a family of being able to answer the question (of the probability of having a second autistic child). But, also recognizing that there are some bigger policy questions,” she says, at the same time adding that “we need to get everybody off the idea that all the answers are in the electronic health record—that’s just a source of data.”

Still, Micky Tripathi, president and CEO of the Massachusetts eHealth Collaborative, characterizes the difference between the two data types—claims and clinical—as breadth versus depth. He observes that EHRs give providers really deep data about patients for specific medical events and the overall context.

“If you only have claims data you would only be able to infer, for example, that a patient is diabetic or has pneumonia because there is nothing in the claim that tells you so,” says Tripathi. “All you know is that Blue Cross paid the provider to test for those conditions. You don’t actually have the test results. But, with clinical data, you find out what happened. In the EHR, you will have the actual lab results.”

At the same time, Tripathi believes claims data is generally “cleaner” than clinical data given that the former is simpler in design and that payers actively monitor medical billing in a way that EHR data is not.

“The problem with clinical data is there are lots of individuals who are documenting in their own unique ways that is not enforced on a day to day basis,” he adds. “One of the things about claims and financial data is that when it is submitted with the wrong codes it will get rejected right away because everyone one of those is sent through a central payer. That means providers will not get paid, so they care about getting the coding right.”

However, DeSalvo charges that both claims and clinical data are “super” dirty. “I’m not sure what’s worse,” she says.

Nonetheless, as the industry transitions from fee-for-service to value-based payment linked to quality, performance and reducing costs, DeSalvo contends that the more access providers have to clinical data about outcomes is going to be critical to ensure interventions are actually improving healthcare and not just decreasing utilization.

“The Holy Grail has always been the clinical data for the scientific, public health communities, and those interested in analytics, because it gives you some sense about outcomes more than process and help provides a window into complexity and control factors—particularly high-cost, high-need patients with chronic diseases—as opposed to just knowing what chronic diseases they have,” says DeSalvo.

When it comes to value-based purchasing and risk contracts, John Halamka, MD, chief information officer at Boston’s Beth Israel Deaconess Medical Center, contends that a repository of both clinical and financial data is needed by providers to succeed.

“That’s the only way to compare outcomes (clinical), quality (clinical), and total medical expense (financial) across providers,” says Halamka.

Also See: Lack of payer, provider info sharing complicates shift to value-based care

Sachin Kheterpal, MD, associate dean for research information technology at the University of Michigan Medical School, believes that access to claims and financial data is extremely important in evaluating the cost effectiveness of new technologies and therapies compared to current options.

“Clinical data and electronic health records are an important piece but by no means are they the only consideration,” says Kheterpal, who is focused on the use of IT and EHRs for patient care, quality improvement and research. “Claims data is not just financial data. It’s about value, disparities of care, and continuity of care.”

Kheterpal, who is also associate professor of anesthesiology at the U-M Medical School, notes that working with Blue Cross Blue Shield of Michigan and Medicare the institution has EHR data for tens of thousands of patients integrated with claims data enabling a “fuller perspective” in understanding the impact on quality and cost. “In 2017, it would be completely inappropriate for any researcher to say I would prefer to have one or the other type of data,” he says.

According to Kheterpal, one of the advantages of claims data is that it follows patients regardless of where they receive care.

“When you’ve got access to only clinical data, oftentimes it’s from only one health system,” he adds. “When you incorporate the claims data, wherever that patient may have gone—whether it’s an information system that you have access to inside your health system or not—the payer in a rather strange way is more connected to the patient than providers because they pay for care. ”

Bobbi Brown, vice president of financial engagement at analytics vendor Health Catalyst, says that although providers have made progress in aggregating their clinical data to create longitudinal health records, these efforts are often hamstrung by the unstructured nature of clinical documentation while clinical data aggregation alone does not incorporate financial, operational, and patient-experience data needed to fully take into account outcomes, quality, and cost measures.

“Providing better care for patients by getting all of this data together is ultimately what we’re trying to do,” adds Brown, who believes that both claims and clinical data is needed for meaningful population health management particularly as providers are “being asked to do more bundled payment, ACO, capitation kind of work.”

While Tripathi argues that providers can “go a long way” in using clinical data for measuring quality and outcomes, he concedes that getting claims data is a “hugely beneficial adjunct” to EHR data. However, he points out that provider access to claims data is restricted to patients included in risk-based contracts but is not available for other patient populations.

“That’s been a source of frustration for a lot of providers,” Tripathi laments. “If you’re a provider, why aren’t you able to get all the claims? Why is it limited to just risk contracts? That means nationwide only about 15 percent of patients are in a risk contract. So, for 85 percent of patients, providers are not able to get claims even if they wanted them because those are all under fee-for-service contracts and the payers won’t give them those claims. It’s a big issue.”

Likewise, Brown observes that “by far” clinical data is more accessible than claims data. “You can get into your EHR data much easier than claims data,” she comments. “The payer is not necessarily providing you with that unless you’re under some sort of risk arrangement contract.”

For her part, DeSalvo makes the case that both claims and clinical data are still “very highly blocked” in healthcare.

“People are so frustrated because they recognize not just the healthcare improvement value but the monetary value of the data—and, now that analytics and artificial intelligence are becoming the coin of the realm people are holding on to it even more tightly,” she says. “It’s pretty unbelievable to me how resistant the healthcare sector is to putting data about their performance out for people to look at and being willing to share it. There has to be much better blending of claims and clinical data. It’s really about understanding cost and outcomes.”

Kheterpal agrees that major claims data holders—like health systems with their clinical data—have business strategies built around restricting access to information. However, he mentions that not-for-profit claims data holders such as the Blue Cross Blue Shield Foundation and Blue Cross Blue Shield of Michigan “do not view that data as a revenue stream and a strategic differentiator” but as “an enabler of research and clinical quality improvement.”

Case in point: the Michigan Value Collaborative (MVC) is a Blue Cross-Blue Shield funded consortium that includes 75 participating acute care hospitals throughout the state who have joined together to help these hospitals achieve their best possible patient outcomes at the lowest reasonable cost. Coordinated out of the University of Michigan, MVC performs “analyses of claims data for hospitals so that they understand the impacts of their clinical choices,” according to Kheterpal.

Even when claims data is made available to providers, Tripathi adds that there is a significant time lag—three to six months—in getting the information. “That delay is best case if you’re getting the claims data directly from the health plan,” he concludes.

Another challenge to leveraging claims data is standardization, according to DeSalvo. “As much good work as has been done to standardize claims data, when you try to pull together an all-payer claims database it requires a lot of curation,” she notes.

Register or login for access to this item and much more

All Health Data Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access