In 2016, CBO started counting Medicaid enrollment differently

CORRECTION, 6/6/17: Somehow, I missed an email response to my query from CBO (and asked them again twice after receiving the email). In 2016, they did not count partial benefit enrollees in their Medicaid totals; rather, the difference was that they counted people with multiple sources of coverage, including Medicaid, as Medicaid enrolled, whereas in past years they counted only the coverage source deemed primary. I was correct that in 2016 they started counting dual eligibles in both Medicaid and Medicare.

In a US News article by Kimberly Leonard about why the ACA private plan marketplace is not meeting enrollment expectations, while the ACA as a whole is reducing the ranks of the uninsured more or less on schedule, this claim did not compute:

Under Obamacare, states were allowed to expand the program to many more people, but many states have opted not to. Despite that, 16 million people more than expected enrolled in Medicaid and the Children's Health Insurance Plan, including in states that did not expand the program. Many people, it turns out, realized that they could have qualified for Medicaid even before the passage of the health care law.

Similarly, last March, when the Congressional Budget Office published its latest forecast of the ACA's effects, The New York Times' Robert Pear reported:

when President Obama signed the health care law, the budget office estimated that 52 million people younger than 65 would have coverage this year through Medicaid or the Children’s Health Insurance Program.

It now estimates that an average of 68 million people under 65 will be covered by those programs in a given month this year

That's despite the fact that CBO's forecast of those who would gain Medicaid eligibility thanks to the ACA's expansion of eligibility dropped from 17 million in 2010 to 11 million this past March -- mainly because the Supreme Court rendered the Medicaid expansion optional for states in 2012, and 19 states have so far refused to enact the expansion.

The apparent overall increase in Medicaid enrollment stems from different estimates of those who would be enrolled in Medicaid and CHIP under pre-ACA eligibility criteria by 2016. In 2010  the estimate was 35 million; in this year's report, it's 57 million.

It's not true, though, that 16 million more people than expected in 2010 for this year have actually enrolled in Medicaid and CHIP in 2016.

The total Medicaid enrollment forecast jumped in the 2016 report because CBO changed who it includes in its estimates of nonelderly Medicaid enrollees. Before this year, the forecast of those who would be enrolled under prior law barely moved  In January 2015, CBO forecast that 34 million would be enrolled in Medicaid and CHIP "without the ACA" -- nearly the same as its 2010 estimate.*

Again, though, in 2016,  CBO's estimate of nonelderly Medicaid enrollment not attributable to the ACA expansion suddenly jumped to 51 million, with another 6 million in CHIP.  What gives?

While I puzzled over this for some time, I needed Loren Adler of the Brookings Institute to spotlight the smoking footnote for me. It's on page 3 of the March 2016 report:

5. Some enrollees receive only partial benefits from Medicaid. They include Medicare enrollees who receive only assistance with Medicare cost sharing and premiums, individuals who receive only family planning services, and unauthorized immigrants who receive only emergency services

There are a lot of people in these categories. I don't believe they were counted in the Medicaid totals previously.

In previous years, too, CBO reports specified that "people reporting multiple sources of coverage are assigned a primary source." This year's corresponding note instead warns that

The components do not sum to the total population because some people report multiple sources of coverage. CBO and JCT estimate that 12 million to 14 million people (or five to  percent of insured people) have multiple sources of coverage, such as both employment -based coverage and Medicaid. 

That would seem to imply that some people --perhaps 12 to 14 million** -- are double-counted, unlike in previous years.  And others in the Medicaid total have coverage, such as limited-benefit public plans, that is far short of minimal essential coverage as defined by the ACA -- and, I believe, would not have been counted in previous years.

To take one large category that may have been newly included in this year's Medicaid estimates: the 2016 report, for the first time, provides a specific breakout of people under 65 enrolled in disability Medicare. The estimate for 2016 is 9 million. Those of them who qualify for Supplemental Security Income as well are "dual eligibles," i.e. they also receive Medicaid help with cost-sharing. I do not know how many of the 9 million are dual-eligible and thus counted by CBO in the Medicaid totals. [Update, 9/1, courtesy of emergency doc Seth Trueger: according to the June 2016 MEDPAC data book, in 2012, 43% of dual eligibles were under age 65. If that percentage still holds, about 3.9 million of the 9 million dual-eligibles under 65 are also now counted by CBO as Medicaid beneficiaries.]

As for limited benefit programs [per correction above, not counted in CBO total], to take a one-state sampling: upon implementing the ACA Medicaid expansion this summer, Louisiana transferred about 190,000 enrollees from such programs into full-service Medicaid.  As I reported in June, these included 132,000 enrollees in Take Charge Plus, a program focused mainly on family planning, along with a few free office visits; and 56,000 from the Greater New Orleans Health Connection (GNOHC), a no-cost primary care program for low income people in the greater New Orleans area. GNOHC does not provide drug or hospital coverage.

Various states have established limited-benefit Medicaid family planning programs via Section 1115 waivers, which give states flexibility to design their own Medicaid programs. Nationally, in 2011, at least 3.5 million women were enrolled in such programs, according to a 2016 Kaiser brief citing their own and Urban Institute estimates. According to the brief, "Today [as of Jan. 2016], more than half of states have established programs that extended Medicaid eligibility for family planning services to people who would not otherwise qualify for Medicaid" [update, added 9/1].

With regard to emergency Medicaid, accessed mainly by undocumented immigrants, the only scrap I can find is a Kaiser estimate, cited in a PBS report, that more than 100,000 people received it in 2011.

According to a Kaiser analysis, Medicaid enrollment increased by 15.5 million nationally from summer 2013, prior to enactment of the ACA Medicaid expansion, to January 2016. Most of the new enrollees were rendered newly eligible by the ACA's expanded criteria.  According to Charles Gaba's tracking, approximately 4 million post-ACA enrollees were "woodworkers" -- i.e., would have been eligible under pre-ACA criteria.  New Medicaid enrollment has not wildly exceeded CBO expectations, if it all. [Update added 9/1]

I have asked CBO for clarification and confirmation of changes in the methodology of their Medicaid tallying. I will report further when I have more information.

*   In March 2015, CBO's updated forecast raised the forecast for Medicaid and CHIP  enrollment in 2016 "without the ACA" to 38 million. Perhaps that more modest bump-up does reflect an increased estimate of "woodwork" enrollment.

** As Loren Adler points out to me, if you add all the totals CBO lists as enrolled in each type of insurance, the sum exceeds the "number of insured people" by 15 million. [update added 9/1].