You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Exclusive: The Obamacare Error Rate Has Fallen Dramatically

Some perspective on the latest ACA freak-out

Getty Images: Joe Raedle

These days it seems like everybody following Obamacare is talking about the 834s. Those are the personal data files that sends insurance companies, in order to notify them of new enrollees. The data has been prone to errors and that's a real problem. You can't get new coverage if the insurer doesn't have some basic information about you—your full name, Social Security number, and so on. Prevalent errors could lead to disarray in the new year. On January 1, some people who completed the enrollment process through could show up at a pharmacy, physician office, or hospital only to discover they had no active coverage. Something like this happened in 2006, when the Bush Administration launched Medicare Part D.

But the 834 problem is fixable and, according to multiple sources in the public and private sectors, it is being fixed. In fact, one administration official tells The New Republic that preliminary estimates, just now becoming available, suggest the error rate has fallen from one in four during October to one in ten now. And most of those are files insurers received with errors, as opposed to files insurers never received. Plenty of work remains—namely, completing repairs that reduce the error rate further and dealing with the flawed data insurers have already received. But the administration is working with insurers and contractors on both issues. 

Some background: The figure 834 is actually shorthand for “834 EDI Transmissions" or "834 EDI Transactions," which consist of personal data organized in a way that insurance companies can decipher and process electronically. The 834s have been around for a long time, sent by employers, brokers, and government in order to arrange new coverage for individual employees, consumers, and beneficiaries. If you get private insurance through a large company, for example, somebody working in the human resources department once submitted an 834 with information about you, so that an insurer could add you to its rolls and send you an insurance card. communicates with insurers the same way. Once you’ve used the site and picked a plan, the system sends an 834 with your information to whatever insurer you've chosen. (Actually, as I understand it, the site sends the data once a day in batches.) The carrier then takes the information, signs you up, and sends you a bill for your first month’s premium. That last part is important. If you don’t pay that first premium, you won’t have insurance come January 1. That’s obviously a big deal, particularly for people with chronic health problems that require constant medication. 

Until a few weeks ago, almost nobody outside the insurance industry had even heard of 834s. Then opened for business—and the insurers discovered the 834s from the site contained errors at an alarming rate. Data was missing or in the wrong field: Some files had transposed parents and children, for example. Among the first to write about this problem was Robert Laszewski, the insurance industry consultant and widely read blogger. It also got attention in the New York Times and Wall Street Journal, and from Sarah Kliff in the Washington Post. If you want a fuller explanation of the forms and how companies use them, Kliff’s backgrounder at the Post’s Wonkblog remains the most accurate and thorough guide I’ve seen. 

Early on, administration officials confirmed that 834s were among the site's myriad failures and high on their “punch list” of tasks to get the site fully functional. But as the more visible parts of have started working better—from the looks of things, most consumers are now able to navigate the site easily and successfully—the lingering 834 issue has gotten more attention. On Monday, Amy Goldstein and Juliet Eilperin at the Post reported that about one-third of enrollments since October 1 contained errors. At that point, discussion of 834s entered full freak-out mode.

But this may not be freak-out material. Technical experts from the insurers have been working closely with staff at the Department of Health and Human Services (HHS) and with the private contractors. By the middle of November, the error rate had already declined significantly. Then, on Sunday, HHS introduced a key repair designed to fix the Social Security number problem. Laszewski, appearing on Fox News this week, said the error rate on Monday was as high as it had been three weeks ago. But other sources have said the latest batches of 834s insurers are, in fact, cleaner. “The enrollment files continue to get better and the new process they put in place this week to deal with the back end issues is making a difference,” says Robert Zirkelbach, spokesman for America’s Health Insurance Plans. Bryce Williams, who is Managing Director of Exchange Solutions for Towers Watson, says "We are hearing reports from insurers that the quality of the 834s being sent from is getting better—for example, fewer cases of children showing up as spouses." But, he adds, "there are still backend enrollment issues to be fixed."

The restraint in those quotes reflects some very understandable caution—and a grasp of the relevant math. While the rate of error is declining, the pace of enrollment is increasing, which means the system may still be creating too many flawed files. Even when the error problem finally falls to acceptable levels, the past errors will need attention. Nobody has said exactly how many people will end up in this category, but the administration source said a rough guess would put the eventual number in the tens of thousands.

Still, the significance of that figure is not clear. It includes files with relatively minor mistakes, such as an incorrect middle initial, and it includes instances where the problem may be simple duplication or a mistake on the insurer end. More important, Administration is confident that most of the 834 errors involve cases where insurers have the files, as opposed to cases in which insurers never received the files. That means tracking down the right information, though labor-intensive, is straightforward to do. The Administration is also conferring with insurers about how to find those never-sent 834s, perhaps through some kind of cross-referencing of applications finished at and new plans started by insurers. In the meantime, the Administration is reaching out to everybody who applied for coverage in October and November, via either email or phone, reminding them to confirm enrollment directly with insurers—and to pay the first month’s premium. 

Even if all of those efforts are successful, some people are bound to think they have coverage when they don’t. The Bush Administration, in 2006, eventually arranged with a private insurer to be a “payer of last resort” for Medicare Part D, so that seniors who needed medications could get them without delay. Sources I consulted said they had no knowledge of such a plan in the works right now. And, as Jeffrey Young wrote at Huffington Post on Thursday, the Affordable Care Act creates no funding stream for such a program. Presumably somebody at HHS or in the Old Executive Office Building is thinking about this—and maybe speaking to Mark McClellan, the former Bush Administration official who is now at Brookings, and who knows a thing or two about subject. 

One lingering question is why the Administration hasn’t publicly announced a figure, or even a rough approximation, of the 834 error rate. Julie Bataille, communications director for the Center for Medicare and Medicaid Services, has been getting multiple questions about this on conference calls with journalists. The back-and-forth has gotten tense, but the questions have come from some of the best (and understandably frustrated) reporters on the beat—including Phil Galewitz of Kaiser Health News, Noam Levey from the Los Angeles Times, and Louise Radnofsky from the Journal. 

Privately, Administration officials say the problem has been a genuine lack of knowledge. They say standard analytical tools for measuring traffic and speed, but no similar way to track this specific kind of data error. For a while, the administration was essentially depending on anecdotal feedback from insurers for knowledge of 834 problems. Only recently has new team working on 834 problems been able to make some calculations about error rates, using a combination of observation, more systematic feedback, and modeling. I can't vouch for this explanation or tell you why the administration has not offered it publicly. But if the improvement continues and the administration works with insurers on correcting the existing data errors, those questions will become a lot less important.