Health claims data: Taking a 30,000-foot view on disease associations

Stethoscope and keyboard health insurance claims dataThe only time most of us ever look at an insurance claim is after a hospital or doctor visit, when we get a claim summary from our carrier. And then as far as we know, it gets filed away, never again to see the light of day.

But there’s a lot to be learned from these claims data.

As with electronic medical records (EMRs), behind every claim an insurer receives is a detailed record about symptoms, tests, diagnosis and treatment. Properly compiled and analyzed, claims data can be an excellent resource for taking population-level snapshots of disease, helping to identify trends and reveal or probe associations.

That’s why claims data recently caught the eye of Kenneth Mandl, MD, MPH, and Mei-Sing Ong, PhD, two researchers in Boston Children’s Informatics Program (CHIP). Using claims records for roughly 2.5 million Americans, they turned their attention to two conditions—epilepsy and asthma—with interesting results.

Associating asthma and antibiotics

Let’s start with asthma. In a paper published online last month in the Annals of Allergy, Asthma and Immunology, Ong and Mandl used their dataset to look for links between asthma and antibiotics. “There’s contradictory evidence in the literature,” explains Mandl, who directs the Intelligent Health Laboratory in CHIP. “Some studies suggest that there is an association, others say there isn’t.”

But if there were an association, it would be an important one to pin down. The Centers for Disease Control and Prevention estimate that 6.8 million children—nearly 10 percent of all children in the U.S.—currently have asthma. And U.S. pediatricians fill out upwards of 50 million antibiotic prescriptions for kids every year.

Looking at claims for the more than 62,500 children in their pool, Ong and Mandl found a strong association between antibiotic use before age 1 and early onset of asthma (before the age of 3). On average, children who were given antibiotics in the first year of life were 2.5 times more likely to develop asthma by age 3 than children who weren’t. And the more antibiotics a child received, the higher a child’s asthma risk.

Mandl thinks the results add weight to calls for more judicious use of antibiotics, calls that have gotten louder since a recent study documented an alarming rise in antibiotic-resistant infections among children in the U.S.

“These findings could have real implications for parents to consider when it comes to asking for antibiotics of minor childhood infections,” Mandl says. “The data suggest that even a single dose of antibiotics early in life can increase a child’s risk of asthma.”

Eyeing epilepsy and autoimmunity

Keyboard with large Health Insurance key claims dataClaims data also let Mandl and Ong delve into another possible disease association: epilepsy and autoimmune conditions.

About 470,000 children and 2.3 million adults in the U.S. have epilepsy, and while there are many options for treating it, the success of those treatments is mixed. Part of the issue is that about two-thirds of epilepsy cases are idiopathic—meaning the underlying cause is never found. As with asthma and antibiotics, previous studies linking epilepsy and autoimmunity have had inconsistent results.

Turning again to their cohort, Ong and Mandl found 10,000 patients with epilepsy and 137,000 with one of 12 autoimmune diseases, including lupus, rheumatoid arthritis, type 1 diabetes or psoriasis. Here again, they found a strong association: patients with an autoimmune disease had a nearly four-fold increase in their risk of developing epilepsy.

As they reported in JAMA Neurology, the magnitude of increase varied with the kind of autoimmunity, ranging from 1.9-fold for psoriasis to 9.4-fold for a condition called antiphospholipid syndrome.

What interested neurologist Mark Gorman, MD, a co-author on the study, were links between epilepsy and autoimmune disorders that have nothing to do with the brain.

“Psoriasis is thought to only affect the skin. While myasthenia gravis is neurologic, it’s thought to only affect neuromuscular junctions outside the central nervous system,” he says. But in the study, he notes, myasthenia gravis upped a patient’s epilepsy risk nearly five-fold.

Both rheumatologists and neurologists should consider screening their patients for signs of the associated condition, Gorman believes.

“There are subtle signs of epilepsy that may suggest to a non-neurologist that a patient should be assessed by a neurologist,” he says. “Similarly, if you are a neurologist, a referral to the appropriate specialist may be warranted if a patient shows subtle signs of autoimmunity.”

The autoimmune findings suggest that doctors should consider approaches to treating epilepsy that quiet the immune system, adds Mandl. “We may be overlooking a treatable mechanism in epilepsy patients.”

The claims for claims

Mandl thinks the population-level view that claims data brings has the potential to transform medical research when it comes to exploring disease associations.

“Claims data let you conduct investigations on huge datasets and identify things you wouldn’t otherwise find, because you wouldn’t have a sufficient number of patients,” he says. “Also, they have really good information on patients’ medications, much better than other sources.”

Claims data do have their limits, though linking claims with EMR data could help over come them. Mandl is investigating this now. “Claims don’t usually include patient outcomes or clinical notes, things that would allow one to go to the next level of depth,” he explains. “You need to work with other kinds of data to refine what you find and pursue biological mechanisms.”