McKinsey Study claims Previously Uninsured up to 27% from 11%...and tells us nothing.

Back in January, the McKinsey Center for U.S. Health Care Reform gave a lot of anti-ACA forces ammunition when they released a study which claimed that only 11% of the total exchange-based enrollees through the end of December had been previously uninsured.

This sounded awfully low to me, especially since New York was reporting that fully half of their enrollees were previously uninsured, but no one else seemed to be tracking this data point at the time, so it was all we had to go on.

Then, starting in early February, New York's previously uninsured rate jumped up to an impressive 66%, 6x the rate that the McKinsey study claimed. Now, obviously every state is different and New York is probably not typical of the country at large, but a 6x factor still seemed way off. In fact, since then, New York's "previously uninsured" percentage has actually gone up to over 70% previously uninsured. Even correcting for the inclusion of Medicaid enrollees in the mix, NY is still way higher than McKinsey has been claiming.

Anyway, today the Washington Post ran a story on an updated McKinsey study, in which they now claim the "previously uninsured" rate has gone up from 11% to a better-but-still-low 27%.

However, there was one line in the WaPo article which gave me pause:

McKinsey’s survey also includes people who bought insurance outside the new marketplaces. 

That's right...the McKinsey study (and I'm assuming the earlier one) doesn't distinguish between people who enrolled via the ACA exchanges and those who enrolled directly/off-exchange.

Sure enough, when you go to page 8 of the survey, you get this glossary definition:

Select new 2014 product: Either previously insured respondents who switched (either switching from one carrier to another, or switching policies but staying with the same carrier), or previously uninsured who enrolled; policies can be selected on or off exchange; includes those who paid their premium and those who hadn’t yet at time of survey.

This is important for a couple of reasons.

The new study was done in mid-February, when exchange QHP enrollments stood at roughly 3.6 million.

However, we have no idea how many off-exchange enrollments there have been since October. I know of at least 460,000, but it's almost certainly 2-3 million at a minimum, probably more.

So, to simplify, let's call it 3.6 million exchange-based and anotherh 3.6 million off-exchange enrollments as of mid-February.

27% of 7.2 million is around 1.94 million people previously uninsured and 5.26 million already covered.

However, without distinguishing between the two, it's entirely possible that practically all 3.6 million of the direct enrollments were people who were already covered, leaving only 1.66 million of the exchange-based QHPs already covered.

This would mean that instead of 27%, the actual number of ACA EXCHANGE QHPs that were previously uninsured is closer to 54%.

Of course, I have no way of knowing this. It could theoretically be closer to the opposite extreme--perhaps only 1% of exchange QHPs didn't have insurance before (well, except that NY accounts for around 7% of the total enrollments by itself, which means it's a minimum of around 4%).

Conversely, for all we know, there may have been 6 or 7 million off-exchange enrollments as of mid-February, which could mean that the exchange QHPs are closer to 100% previously uninsured (again, I doubt this as well).

The point is that by mixing the two groups together, McKinsey's study tells us pretty much...nothing.