9 November 2013

A look at healthcare.gov and the Flathead

My background includes database construction and website design, so I poked around healthcare.gov for a couple of hours after midnight last week. I knew that even without political constraints and mistakes, and the generally sorry state of government information technology, building healthcare.gov was a daunting task because of the complexity of the health insurance system the Affordable Care Act created. And I knew that in addition to the system’s programming logic, there were the problems of data integrity and data compatibility.

What I found was sobering.

The array of plans differs not only from state to state, but from county to county. Just for Flathead County, I found 29 plans for individuals aged 49 and under, and 29 plans for individuals over 50. A platinum plan for a person over 50 costs $446.96 (not rounding these prices to the nearest dollar is stupid), but only $262.27 for a person 49 or younger (more failure to round stupidity). Add to that plans for couples, for families, and for small businesses probably brings the total for Montana to 3,000 – 5,000 plans. And that doesn’t count dental plans, which are (but should not be) separate from health insurance. I found at least 10 Dental plans for the Flathead (XLSX). Multiplying that by 35 states brings the number of plans healthcare.gov must handle to well over 100,000.

Those applying for a plan must pass through screens — state, county, age, income, etc. — after which the information supplied must be cross-checked with government databases to determine eligibility for subsidies. If those steps are completed successfully, the applicant’s information — what he submitted, and what the federal government added and calculated — must be converted to a format understood by the insurance companies computers and transmitted to the insurance company choses. If all goes as intended, the application is processed automatically, the health insurance policy is issued, the applicant is billed, and the applicant now has health insurance that meets ACA standards.

Obviously, an error in the system’s logic will lead to an applicant’s receiving the wrong insurance, or no insurance at all, and just as bad, pollute the system’s data, compounding the difficulty of getting an application straightened out.

But even if the logic is perfect, the data will not be perfect. Different agencies use different formats. Not all data are entered corrected. There are duplicate records. An applicant’s name or address may be spelled differently in different databases, making matching records more difficult. A one-digit difference in a Social Security number can produce a false negative. And so on. And that happens at all levels of government, as I can testify from personal experience.

None of this is a revealation. Both public and private sector IT experts know how difficult large scale data matching is. Compared to healthcare.gov, a voter registration database is childishly simple, yet there are many recent examples of how difficult cleaning voter registration lists is because of data errors and format incompatibilities.

Did the White House’s health care policy shop know this? Probably, but likely at a subconscious level. Obama’s policy experts believed that if they could get the ACA enacted, a system to administer it could be built (“if we pass it, they will build it”). Legislation first, administration later. Moreover, I suspect they made sure they didn’t let anyone tell them “the ACA is too damned complicated to be administered successfully.” That’s human nature, and not human nature at its best.

If both political parties work together to implement the ACA, healthcare.gov will become tolerably operable, and improve over time. But it never will become as user friendly or trouble-free as amazon.com. It’s just too complex. Still, for all its flaws, the ACA is a considerable improvement over the pre-existing conditions not covered system that it replaces. And that no-pre-existing-conditions system is the yardstick against which the ACA and healthcare.gov must be measured. By that measure, we’ve come a come a considerable distance and made considerable progress.

But the everyone covered for everything single-payer system yardstick reminds us how much farther we have to go.