David Anderson

Ohio Mom in comments raised a good question in a touch situation:

The issue is, in a nutshell, he lost his job last Wednesday. HR is saying he is covered for rest of January (whoppee, four more days) and can get COBRA coverage but will not receive the sign-up information for up to two weeks…

I (Ohio Mom) turn 65 on March 13 and already have appointment with Medigap/Part D counselor to choose plans.

Ohio Dad turning 63 in March.

Maybe we should go with Obamacare instead of COBRA?

Fundamentally, the question is what are the trade-offs between COBRA and the ACA?

This is a damn good question.

COBRA is a continuation of group coverage where the beneficiary and any dependents have the option to pay 102% of premium to continue their coverage after a qualifying life event.  There is no underwriting and the coverage will retroactively start to the date of the termination of the regular coverage once the initial election and payment is received.  COBRA coverage keeps the already paid deductible and out of pocket cost sharing counted.  For instance if there is a $5,000 deductible for the calendar year and Ohio Dad had a $2500 test in January that hit the deductible under his base plan, he would still only owe $2500 in additional deductible on a COBRA policy between February 1 and December 31.  COBRA can be elected or turned down by each covered individual.

The loss of a job and loss of group insurance coverage is a qualifying life event for the ACA and can trigger a special enrollment period.  The ACA exchanges offers community rated, guarnateed issued insurance with premium subsidies for folks earning between 100-400% FPL if they are not Medicaid qualified. A new ACA plan will not recognize already spent deductibles and cost-sharing even if the insurer selling the ACA plan is the same insurer offering the work plan.

One last note; COBRA insurance is pure community rating; everyone, no matter their age, who purchases a given plan pays the same premium while the ACA is modified community rating where gross premiums increase by age.

So what is the best choice?

It depends!

The major variables are time of the year, amount of cost sharing left on the COBRA plan, hassle/transition costs, age and eligibility for ACA subsidies.

A 21 year old who has no claims in their work insurance who is COBRA eligible in February and is eligible for big ACA subsidies is likely better off in the ACA individual market as they are likely to see a low to no premium Bronze plan that they are unlikely to use as they are statistically likely to be as healthy as a horse anyways.

A 64 year old who has already maxed out their deductible when they became COBRA eligible in February, makes too much for ACA subsidies and has a knee replacement surgery scheduled in March will likely be better off in COBRA.

Why?

That 64 year old is paying full premium for either COBRA or the ACA.  They are likely to be one of the older and statistically likely to be more expensive people in either pool. The COBRA premium is the age blended premium of everyone in the company so the average age of premium payers might be late 30s or early 40s.  The ACA premium is at its maximum for a 64 year old.  For comparable plans, the ACA plan is likely to be more expensive in premiums plus the cost-sharing restarts at zero so it would be a double whammy.

Fast forward everyone to COBRA eligibility starting in December it is a one month calculation of : difference in COBRA and ACA premiums + difference in expected cost sharing — choose whichever is lower.

COBRA also minimizes learning costs.  You’re keeping the same plan with a new ID card and nothing else changes.  That can be valuable for folks who have spent a significant amount of time learning their insurance company due to high needs.  It is far less valuable for someone who has not seen a doctor in three years.

So, as always, it depends.

COBRA or the ExchangePost + Comments (20)

Later this week, the Trump Administration will be rolling out their guidance to states on how to apply for Medicaid block grants:

NEW: Trump admin will announce its Medicaid block grants on Thursday at event billed “Transforming Medicaid: A New Opportunity for Better Health.”https://t.co/FYv5SpvZ2V pic.twitter.com/9v3WXkFffg

— Dan Diamond (@ddiamond) January 25, 2020

The first guarantee is that any state that files for a block grant will be generating an amazing number of billable hours for lawyers. Nick Bagley explains:

Setting aside the dubious policy merits of block grants, however, I don’t think the proposal is legal. I don’t even think it’s close.

Under section 1903 of the Medicaid statute, the federal government must pay a fixed “match rate” (known in the statutory lingo as “the Federal medical assistance percentage”) to every state that participates in Medicaid. In Tennessee, the match rate is 65.21%. That means that, for every $1 that Tennessee spends on its Medicaid program, the federal government kicks in about $2….

As Tennessee recognizes, it’ll need a waiver from HHS to make these changes. And section 1115 of the Medicaid statute does allow HHS to waive lots of the law’s restrictions in connection with experimental projects that are likely to assist in promoting Medicaid’s objectives.

Now, I’ve written before that I’m not at all sure that block granting Medicaid counts as an experiment that serves Medicaid’s purposes. But there’s a more fundamental problem with Tennessee’s proposal. You can’t use section 1115 to waive section 1903. To the contrary, section 1903 is pointedly omitted from the list of statutory provisions that HHS is empowered to waive.

But let’s put aside the legal gymnastics of storming that castle for a moment. Let’s think about what happens to a state that chooses to block grant when faced with a shock?

• Major hurricane that destroys a lot of public health infrastructure and creates a lot of sudden, unanticipated morbidity and mortality?
• New virus that is readily transmittable with significant symptoms and care needs?
• New drugs/treatments that have high first year costs but are massive net present value savers under any reasonable discount rate after five or ten years?
• A family of clutzy triplets who all have  inhibitor resistant hemophilia moving across the state line ?

All of these are shocks to the payment system.  All of them could readily increase first year costs significantly.  Under the current system, the cost of a shock is split between the federal government and the state government.  The cost of the shock is split 3:1 in Mississippi and 1:1 in Massachusetts for legacy Medicaid and 9:1 federal/state for expansion.  The ratios differ as the federal government,which has the best capacity to eat risk and bad luck in the world, bears a good chunk of the risk.

Under a pure block grant, the federal government bears no risk of a shock. Instead that risk is shifted entirely to a state government whose risk bearing capacity is orders of magnitude less than the federal risk bearing capacity.

That is what a block grant will do when a state is faced with a shock.

Medicaid block grants and shocksPost + Comments (19)

CMS recently released the 2018 insurer/plan enrollment report.  I took a look at my former employer, UPMC Health Plan**, as their strategy still leaves me scratching my head.  They offer three distinctive networks in Western Pennsylvania.

The three networks can be quickly defined like this.  Each network is a subset of the network below ##  :

• Partners: Narrow network centered on UPMC owned facilities; offered in parts of Western PA.
• Select: Medium network centered on UPMC owned and close ally hospital facilities in Greater Pittsburgh
• Premium: Broad network that is basically the standard commercial network.   Offered in all of Western PA.

UPMC then attaches a set of five common benefit designs to each network to generate a fifteen unique plans.  People buy the cheap plan most of the time.

There are a lot of choices, but people are very consistent in buying the cheapest plan for a given network.  There is meaningful difference between networks once you get outside of Pittsburgh city limits so I can see people making distinctive valuations of the networks and choosing from there if they have multiple options.  The least popular plan is only getting 1% or 2% marketshare for a given network.  All three of the network submarkets would be considered very concentrated for silver plans if we were to apply an HHI standard.

I’ve always questioned the spread strategy as the UPMC strategy crushes the spread between the benchmark silver and cheapest silver which makes plans pretty expensive for healthy buyers.  I’ve been spending more time in the behavioral econ space over the past year, and cognitive load and sorting through options is costly and could lead to ineffecienct choice.  There is good evidence (McWilliams et al 2011) that choice overload crushes good choice in Medicare Advantage space, and I suspect this may be the same story going on in the Exchange space.

** Side note, these networks were what I was working on at UPMC Health Plan when I started writing for Balloon-Juice.

## Miniscule and idiosyncratic variations aside

Too many plans, too much confusion?Post + Comments (16)

Duke University sent out their W-2s early last week and mine arrived on Saturday.*  It contained all of the usual information.  It also contains Box 12-DD with a medium to big number in it.

Box 12-DD is the amount of money that Duke paid for my family’s health insurance in 2019.  Every employer that pays for some portion of the group  health insurance premium has to report on the W-2 how much they kicked in.  This is part of the ACA and the idea behind this is that total compensation is the sum of cash wages (the first couple of boxes on the W-2) plus benefits.  One of the biggest pieces of the benefit slice is health insurance.  The theory of change on this piece of the law is that if people realize that their health insurance premiums are replacing their cash wages, they would be more willing to make trade-offs between narrower networks, more restrictive plans, or lower actuarial values than they are currently receiving.

That has not worked out too well.  I think part of it is that the insurance benefit manager and decision maker for group insurance faces a much different decision problem with very different constraints than an individual buying an Exchange plan or Medicare Advantage.  The person in HR faces a budget constraint and a strong scream constraint.  They want to make a choice that minimizes the number of people with power yelling at them because their cardiologist or preferred hospital is out of network.

I also think that an annual report is in the odd spot of a lot of information with out it being timely.  I pay attention to the deductions on my paycheck.  I know why and what is happening when I see my take-home number change.  An annual report that has no direct impact on my monthly take-home is non-actionable information.  I think it is worthwhile to pay attention to Box 12-DD but if the objective is make people more sensitive to the fact that their employer’s portion of health insurance premiums is coming out of cash compensation, moving that number to the top-line of each pay-stub and then deducting it as a pre-tax deduction would be far more effective as I think that would make the number far more real to people.

** I am not sure when I transformed from doing my taxes in a mad scramble on April 12th type of person to doing my taxes by the Super Bowl type of person but that is who I have become.

Look at Box 12-DDPost + Comments (27)

Earlier this month, a powerhouse team of health economists looked at quality outcomes for hospital mergers. The results were impressive for policy purposes but quite depressing:

Kicking off the decade with a paper on hospital mergers in @NEJM https://t.co/cE7GvA1E20 We know mergers increase prices. What do we get in return? We find no evidence of quality improvement & worsening in patient experiences. Put that together: mergers=bad for patients on avg https://t.co/r6t1G6eRbD

— Michael McWilliams (@JMichaelMcW) January 2, 2020

We could conceivably be willing to pay more if we, as a society and patients, get better quality services and outcomes.

We could be willing to pay more if we get a better experience for the same outcomes.

Those are plausible improvements that may be worth paying more for.

However this study is not finding those types of improvements.  So what is happening?  Are these mergers mostly about market power and building moats to enable the collection of rent?  That is one hell of a plausible interpretation that has not been disproven yet.

Hospital mergers: what are they good for?Post + Comments (22)

The federal platform, Healthcare.gov released their final open enrollment numbers late last week.

Approximately 8.3 million people selected or were automatically re-enrolled in plans using the HealthCare.gov platform during the 2020 open enrollment period. …

Charles Gaba has a good summary of the meaning behind these numbers:

As a result, HC.gov ends the 2020 Open Enrollment Period down just 1.5% from last year…and it’s actually just 0.5% when you take into account Nevada splitting off onto their own state-based exchange.

It’s also worth noting that in the end, renewing enrollees dropped by 2.2% year over year (139,000 people), but new enrollees actually increased by 0.7% (14,000 people).

I was wrong.

I had thought that enrollment would decline due to pricing pressures. I expected declines in ACA enrollment for two mechanical reasons. First, Nevada transitioned off of Healthcare.gov to their own, state based marketplace for the 2020 open enrollment period. Secondly, several more states have expanded Medicaid or are in their second open enrollment period where on-marketplace enrollment declines are still expected.

I had thought pricing would matter more as the spreads compressed:

In 2019, the median premium spread between the benchmark and the least expensive bronze plan was $164 for a single 40-year-old. In 2020, this difference decreased to $148. For the same 40-year-old, the median premium spread from the least expensive silver plan to the benchmark decreased from $27 in 2019 to $14 in 2020.

The reduced silver spread, while small in dollars, may have a profound impact. At incomes up to 200 percent of poverty, the value of CSR—a free added benefit—exceeds the value of most bronze and gold plan discounts generated by silver loading. …Those discounts appear to have boosted enrollment, offsetting negative factors such as repeal of the individual mandate penalty and reduced federal spending on advertising and outreach. While enrollment at 200–400 percent of poverty has increased slightly from 2017 to 2019 (less than 1 percent), enrollment at 100–200 percent of poverty has fallen 10 percent in the same period.

 

I’m curious as to what is going on in the market as I thought the pricing environment was slightly worse in 2020 than it was in 2019 and the number of advertised outs was higher. I’m not sure what is going on.

ACA and Healthcare.gov Enrollment wrap-upPost + Comments (12)

Political rivals reveal deal to expand Medicaid to Kansas families earning up to 138% of federal poverty level. The cost would be offset by a surcharge to hospitals, and the plan would take effect by Jan. 1, 2021 #ksleg via @timvcarpenter @sherman_news https://t.co/d3uzvaYNyW

— CJOnline (@CJOnline) January 9, 2020

Details to come.

But the expansion looks to be fast with a 1/1/21 start date and it will be a full expansion to 138% FPL so it could potentially be a state plan amendment expansion with waivers either not needed or able to be sequentially added to the program. There is no work requirement. There is a work referral program, but that is common.

Elections matter, and here are some of the results of all the hard work Kansans put in over the 2018 campaign season.

Kansas to expand MedicaidPost + Comments (14)

Dr. Amy Finkelstein and team have a special report in the New England Journal of Medicine. They conducted a rigorous randomized control trial of the Camden Coalition of Healthcare Providers. The Camdem Coalition has gotten a ton of press about their “hot spotting” method of identifying folks who use a lot of medical services and then wrapping these folks in a bundle of care coordination and some social services. Prior assessments had found that using a simple pre/post analysis of the individuals included in the program that this approach saved a lot of money and reduced hospitalizations. Finkelstein and her team conducted a random assignment evaluation to see what is happening without intervention and therefore what the intervention is doing. And their initial findings of looking only at re-admission to hospital is that the hot spotting approach is not doing much:

RESULTS
The 180-day readmission rate was 62.3% in the intervention group and 61.7% in the control group. The adjusted between-group difference was not significant (0.82 percentage points; 95% confidence interval, −5.97 to 7.61). In contrast, a comparison of the intervention-group admissions during the 6 months before and after enrollment misleadingly suggested a 38-percentage-point decline in admissions related to the intervention because the comparison did not account for the similar decline in the control group.

WOW

So what is happening?

It all depends on the shape of the assumed counterfactual. Comparing people against themselves has an implicit assumption that the crisis that prompted their eligibility is a permanent phase change and the new steady state. Any change from that steady state could then be attributed to the intervention.

Another shape of trajectory could be reasonably assumed. We could hypothesize that a spike is often just a spike and it will naturally recede. A spike will trigger the intervention but at least some if not all of the future decline in utilization/re-admissions/costs could/would have occurred even if no intervention happened.

Figuring out the right counterfactual and therefore the shape of the response without an intervention is critical to evaluation. The Finkelstein study shows that the observed response of people with a huge spike is a natural decline in utilization with or without intervention. This change in the shape of the counterfactual is critical to this evaluation that shows that this program of hot spotting on at least this metric of re-admission is not doing so hot.

Good evidence is critical. Most experiments will either return null results or return slightly significant results. That is okay. My priors have changed significantly in the past sixteen hours. Hot spotting has a very plausible and facially coherent logic model of change but the evidence this morning is saying that the plausibility of actual results is far lower now than it was before the evaluation was published. This is a big deal.

Hot spotting ain’t so hotPost + Comments (16)

Scott in comments yesterday raised an interesting scenario that has been marinating in the back of my mind for a while now:

My son and wife had their daughter on 31 Dec by inducement a year ago (date chosen for purely financial reasons as the scheduled inducement was for 3 Jan until they realized that deductibles, etc reset at 1 Jan.

New Years’ Day is the 2nd least likely birthday after Christmas.

I am curious if researchers have been able to leverage the January 1 reset day as a means of evaluating the effectiveness of the Earned Income Tax Credit and other child related tax code features?

My now 11 year old daughter is a mid-afternoon New Years Day baby.  She was due on the 28th or the 29th.  She decided to take her time.  We had a low deductible plan that rolled over into a high deductible plan at 12:01 AM on January 1st.  She was ready to greet us well after that threshold was crossed.  Between the deductible resetting and our inability to claim her as a dependent and get child tax credits for a December birth, she cost us several thousand dollars in a year where we knew that we were going to be stretched thin.

The deductible hit was immediate.  I think we got the bill three weeks after birth.  We were able to claim her as a tax deduction fourteen months after birth instead of two months after birth if she had been born eighteen hours earlier.

I am curious if there is research that looks at long term outcomes using the January 1st birth eligibility discontinuity as there is a huge swing in very early life parental financial resources available.  Kids who were born on December 30th or 31st  are fundamentally similar to kids born on January 1 or January 2 except for the presence of a significant resource infusion due to a policy cut-off.  I would expect that we should see the late December kids to have better outcomes than the early January kids because their parents will have significantly more financial resources in the first three to six months of the baby’s life.  More resources might translate into less parental stress which could lead to better bonding and security for the infant.  Another pathway could be simply that the baby’s diaper gets changed more often as there is more cash to buy diapers and therefore the risks of abuse are lowered.  I’m speculating here but I am truly curious.

New Years babies and resourcesPost + Comments (21)

This month’s Health Affairs has a great article written by Dr. Michelle Moniz and colleagues that looks at the out of pocket expenses for the year before a live birth to the three months post-partum for commercially insured individuals who had a live birth at a hospital from 2008-2015.  The headline finding is scary.

Among women with out-of-pocket spending, mean total out-of-pocket spending for maternity care for all modes of delivery increased from $3,069 in 2008 to $4,569 in 2015

Having a baby is expensive if you have insurance through work.  About half of all births in the United States are by women covered by employer insurance.  Almost all of the rest are covered by Medicaid which has a completely different and far lower to non-existent cost-sharing profile.

However, there is something further in the article that piqued my interest from an insurance and value point of view.

mean total out-of-pocket spending for vaginal birth increased from $2,910 to $4,314, and for cesarean birth it increased from $3,364 to $5,161 (appendix exhibit A2). These trends were largely driven by a rise in deductible payments in the study period. Among women with deductible payments, the mean deductible for vaginal birth increased by 62.3 percent, from $1,617 in 2008 to $2,625 in 2015, and the mean deductible for cesarean birth increased by 72.3 percent, from $1,532 to $2,640.

Standardized costs for vaginal and cesarean births showed some variation but did not vary greatly over time (vaginal birth: $24,317 in 2008 [95% CI: $24,184, $24,451] versus $23,148 in 2015 [95% CI: $23,012, $23,283]; cesarean birth: $39,702 in 2008 [95% CI: $39,390, $40,015] versus $43,774 in 2015 [95% CI: $43,402, $44,145]).

We know that the US voluntary C-Section rate is high.  The high rate does not deliver better outcomes.  It is expensive in both direct reimbursed costs and recovery time for the mother and family.  We have seen in California efforts to drive down low risk C-section rates.  However, the implied insurance designs are not helping with this goal. Two things are happening here as I lay out in the chart I created from the data in the above paragraphs:

First the cost of labor and delivery through either pathway is too damn high.  The deductibles that are being paid in 2015 are roughly similar to each other  for either pathway.  In both cases, the deductibles are a tiny fraction of total costs.  In either case, as soon as a woman walks into her ObGyn’s office with a positive pregnancy test, there is no difference in the incentives being generated by the deductible.  A $2,600 deductible will influence decisions to go to urgent care or an ER, it will influence decisions to stick with a brand drug or a similar generic, it will influence low level decisions, but as soon as the cost bundle is big enough to actually drive a significant portion of national health care expenditures, the deductibles are irrelevant.

Secondly, the implied coinsurance rate (Non-deductible Cost Share/Non-Deductible Costs) is higher for the cheaper alternative.  That is messed up.  The total cost of non-deductible cost sharing is $800 less for vaginal delivery than a C-section but the implied coinsurance rate of each dollar of spending is 33% higher.  This is a weird set of incentives.

If we believe that there are too many low-risk, elective C-sections that don’t offer improved value over vaginal births, and if we believe that cost sharing influences decision making at the margin, we should structure our cost sharing so that vaginal births are much cheaper than elective, low risk C-sections.  Right now, the insurance structure that is being reported in this paper is removing strong incentives away from vaginal births over the marginal C-section.

Babies, deductibles and incentivesPost + Comments (15)