Employers, researchers struggle to glean insights from deluge of insurer data. Can CMS help?

Everybody knew the storm was coming. Health insurance plans in July released a blizzard of data about what they charge employers and providers for covered services. Stakeholders knew that the volume could be overwhelming.

That’s why the data were not meant to be used by consumers—insurers must release that sort of patient-facing data next year under an Affordable Care Act mandate that’s enforced by the Centers for Medicare & Medicaid Services (CMS).

CMS encouraged intermediary computer experts to jump in and interpret the information for everybody else, saying that the situation would “offer third-party developers and innovators the ability to create private sector solutions to help drive additional price comparison and consumerism in the healthcare market.”

Yet, computer experts who followed that advice have yet to make sense of the insurer data.

“They expected that it would be for the app developers out there to take these files and turn them into consumer-oriented products,” Kosali Simon, Ph.D., said in an interview. “It would be for the large health systems and large employers to see what the prices are like.”

But even the computer whiz intermediaries aren’t having much success. “We have had a team of data science masters’ students analyzing the data here at Indiana University,” Simon explains. “Even to download one tiny, tiny part of it is a giant task, let alone open the file or understand the data.”

Simon is a nationally known healthcare economist at Indiana University. She’s generally considered to be the expert who’s taken the deepest dive into the insurance industry data. Simon believes that CMS should have fielded the insurer information rather than having them post it on their websites.

“CMS has been shown in the past to be really great at aggregating data,” Simon notes. “If CMS could build the setup for insurers to be submitting the data to them that would help a lot. Because they have to be doing that for auditing purposes anyway. That’s how the fines are going to be judged. ‘Did you put out the data or not?’ There has to be government involvement at some point in knowing: ‘Did you put out the data or not?’”

CMS uses the data processing company ResDAC out of the University of Minnesota to collect and aggregate data. “It would have been so much better if they had just hosted the database that would have kept these data,” Simon says. “CMS does this data hosting regularly, and researchers interact with their administrative data often through a well-run process.”

She adds that “there’s the in-network file and the out-of-network file. And all the action is really happening in the in-network file. The out-of-network file is pretty sparse.”

Insurers posted the negotiated rates it pays to physicians, hospitals, laboratories and imaging companies, among other entities. They can be fined $100 a day per affected enrollee for not posting data. That could be financially catastrophic. UnitedHealth Group, the largest health care insurer in the U.S. in terms of covered lives, insures about 26 million people. CMS could theoretically fine UnitedHealth $200 million a day for not posting data about a service it covers for 2 million of its enrollees.

“I have been in conversations with many healthcare experts, as well as large employers, and healthcare provider systems who have been trying to understand the data,” Simon explains. “When we talk to people who have amazing capabilities at data processing, they’ll say: ‘OK, we know how to open one or two of these files.' But to open all of these and make insights? That is pretty close to impossible.”

Sabrina Corlette, a researcher at Georgetown University’s Center on Health Insurance Reforms, tells Fierce Healthcare that “the intent behind these requirements was that the data would be used by employers who contract with health plans, insurance regulators, researchers and policymakers to help inform them about the workings of our healthcare system. But none of those people are able to access this data at this time. The data files that the insurance companies have posted are just too massive for anyone to use.”

Corlette’s interest in the as-yet unusable data stems from concern about how healthcare costs adversely affect Americans. More than 100 million people in the U.S. struggle with medical debt, according to a report issued in June by Kaiser Health News and NPR. Medical debt causes about 62% of bankruptcy filings each year, according to the National Consumer Law Center. 

“I believe that getting a picture of what is driving our healthcare costs is critical to devising and targeting solutions,” says Corlette. “Healthcare is increasingly unaffordable to American families and one of the reasons for that is the cost of medical services. That’s driving increased premiums and deductibles and that’s driving a lot of problems for people.”

Corlette says insurance companies may also be frustrated by the data problem. “My sense is that they actually have expended a lot of time and effort to comply with this requirement, although some went kicking and screaming and they were not happy about it. Now after spending all this time and resources posting the data, I’m sure it’s somewhat frustrating that it may not be that useful.”

America’s Health Insurance Plans, the largest lobbying organization for insurers, did not respond to a request for comment.

The data dump could stem from a systemic problem, according to Simon. It appears that insurers posted data for every conceivable condition, even if they don't cover that condition. “There are some services that are listed there that you think: ‘Wait. Is that place even licensed to do that?’ Is this just: ‘Well, here are all the possible healthcare services out there, and if one were to do those then here are the prices.’ But we don’t get a sense of whether you actually do those.”

It is not as if insurers intended to cause this confusion, Simon explains. Agency officials met with insurance company representatives several times leading up to the July 1 deadline for posting the data.

“I watched all five or so webinars that CMS did for developers of these files at healthcare payer companies,” she says. “There was a lot of back and forth in preparing these files. There was encouragement to try to reduce redundancy. They would say, for example, ‘If you use one file across all these different employers, indicate that rather than putting out all these separate files.' There was encouragement, but I don’t think there was a requirement for reducing redundancy. And what we see as a result are signs of a lot of redundancy.”

She stresses that her assessment isn’t a result of a comprehensive review, because “that would be impossible.”

The directive for insurers mirrors a similar one that went into effect last year for hospitals, but while insurers provided way too much information, hospitals supplied too little. They complained that compliance would be too cumbersome and costly, a point the American Hospital Association made in a lawsuit against CMS that the association eventually lost.

Simon believes the hard lessons learned from this experience should be applied next year when insurers have to post consumer-facing data. “Plans can take this time to think about how to make sure that when that consumer-oriented tool is out there, that it works.”