Earlier this week, JAMA Health Forum published a new study by Lovichikova et al on Covered California’s nudges to move people mid-year from bronze, gold and platinum plans to CSR 94 plans in the summer of 2021. As part of the the American Rescue Plan, two things happened mid-2021. First, subsidies increased dramatically so that anyone with incomes under 150% Federal Poverty Level could buy a Silver CSR 94 plan with a ~$500 to $1000 deductible with no net premium. Secondly, anyone collecting any unemployment income for 2021 would have their income deemed to be 150% FPL. Covered California, the California state based marketplace, reached out to their current enrollees who they knew had some unemployment income. This randomized control trial sees what the effect of an e-mail nudge is on enrollment and use.
The emails led to a statistically significant 3.1–percentage point (95% CI, 2.6-3.6 percentage points) increase in CSR silver 94 enrollment (a 74.8% relative increase) by July 31, 2021, and a 1.3–percentage point (95% CI, 0.2-2.4 percentage points) increase (a 2.3% relative increase) in practitioner visits by December 31, 2021. The emails had no detectable effect on prescription fills, emergency department visits, or hospitalizations.
What did I get from this study?
First, people don’t pay attention to policy changes. The switching rate without an e-mail was about 4%. The e-mail nudges moved the switching rate to 7% or about a 75% increase in the risk of switching. I can almost guarantee you that a lot of people left at least a substantial amount of premium on the table as 30% of the sample has enrolled in either Gold or Platinum plans. A switch within the same insurer to their Silver CSR 94 plan would be a substantial premium saver. Now there might be odd cases where someone had already MOOPED out in February so the change in six months of premium might not be worth the extra new cost-sharing, but that is a very rare case. Lots of money is being left on the table. The nudge helps but it is limited relative to cross-checking of automatic re-assignment and re-enrollment.
Secondly, actuarial value is not suppressing demand as there is very little demand:
People moved from lower actuarial value and higher deductible plans to higher actuarial value plans with substantially lower deductibles and cost-sharing:
in 2021, Covered California enrollees in a silver 94 plan paid $5 for a primary care visit, paid $3 for a generic prescription drug fill, and had an out-of-pocket maximum of $1000; by contrast, Covered California enrollees in a bronze plan paid $65 for a primary care visit, paid $18 for a generic prescription drug fill, and had an out-of-pocket maximum of $8200
Gold plans have a $35 PCP co-pay and $15 generic drug fill co-pay while Platinum plans would cost enrollees $15 to see their PCP and $5 to get a generic prescription. That is a big switch especially from Gold and Bronze plans to Silver CSR 94.
Yet there is no noticeable change in hospitalizations, ER use or prescription drugs and tiny changes in clinician use. Cost sharing is not doing much which might be a bit weird if we approach the switching problem as an ordeal mechanism. Ordeal mechanisms is an economics concept that says that burdens such as information gathering and compliance costs will sort the population into those who highly value a public benefit and those who don’t. It is an allocation mechanism aiming to give a benefit to the most deserving/needy when there is substantial private information. We would think that people who switch would have private information that their health is worse than average. But we’re not seeing much evidence here to support selection by severity of need.
What are the policy pull-outs?
First thoughts:
A) Nudges work but not too much.
B) Improving AV won’t change access to care that much
C) Even among a population that had a demonstrated value of insurance, complex policy changes are complex and opaque.
I like this paper. It asks a good question using a unique policy shock and good methods.
Ruckus
For most people there are 2 questions. What do I get for my money? What is the downside of spending this money?
My understanding of the situation is that costs were somewhat reduced and use went up slightly. In medicine better usage often lowers overall costs because some things can be caught earlier and reduce over all costs by better usage of healthcare. And some costs will go up because issues will likely be found and will cost something to fix them. Possibly cost less overall if caught earlier, possibly raise costs because not everything is cheap to fix in healthcare. IOW quite possibly a wash cost wise with an improvement in overall health. IOW betterment.
Look overall at the change in healthcare over the last 75 yrs. Yes we spend a lot more than we used to but the things we can do and fix and make better are stunning. 60-75 yrs ago it was stethoscopes, x-rays, crappy anesthetics and scalpels. Today we can look inside the human body with ease and see what is going on – MRI. And other machines that do things not dreamed about 75 yrs ago. We have to figure out how to get these services to ALL humans, not just the one’s who pay insurance and/or write big checks. If money is the only access to proper, reasonable healthcare – WE ARE DOING IT WRONG.
David Anderson
@Ruckus: There are very, very, very few things at the population level where lower costs lead to better outcomes — there are plenty of things that are very valuable to do even as the costs increase because the outcomes are great. The big counter-examples are early childhood vaccinations and cutting out extreme rent seeking behaviors
Martin
So we were one of the nudges – from gold to silver. Got the email, knew what they were doing thanks to you and thought ‘sure, let’s try it out’. Premiums dropped significantly for us (which I kind of feel bad about – I thought we were paying too little before) and because of Ms Martins diagnosis last month, we’re going to test out the MOOP for her. I thought our copay for the surgery was only $150, but there’s a bunch of other copays coming – we’ll hit the max by the time we’re done – so MOOPing out in April.
Overall, there’s no apparent change in what’s covered. Maybe that’s because so much of this is required by state law, not sure, but the higher MOOP is only about the cost of 2 months of premiums, so there’s no universe in which the silver plan for us could be more expensive. There’s a cost smoothing argument, but that barely holds either since we were paying half of the MOOP more per month before.
In Ms Martins case I’m going to assume that an early diagnosis combined with rapid treatment including an outpatient surgery is cheaper than a long course of chemo or radiation therapy. My HMO can control for the cost of the surgery – it’s their own staff doing it in their own facility, but when you get into drugs and medical equipment, etc. that the HMO doesn’t get to keep, those are externalities that they have to negotiate on price for, and that’s where I think the costs really get out of hand.
The biggest sins of the US healthcare system lie almost entirely between accounts payable and receivable.
Curtis
Have researchers in this space actually done a study, either via ethnography, or observation, or usability methods, to determine how different patient groups actually look at, utilize, and decide on policy plans, and their expressed reasons for their decisions? Even something as basic as using an eye-tracker when presented with the policy forms on a computer screen, do people more often present higher gaze or dwell times at “gold, silver, bronze” labels or the monthly premiums? I assume the different populations have at least been surveyed to determine how much they even look at health policy information unprompted, and what means they use to do the looking.
Or rather, how is the information about the policy plans even presented to individuals should they choose to access it? Is it understandable to them?
Looks like there’s at least a little work in that space, although it seems a little old and targeted to different populations than the one covered in the blog post: KS Furtado, KA Kaphingst, H Perkins, MC Politi (2016); A Schram, J Sonnemans (2011).
These results you describe seem fairly unsurprising from a cognition and decision-making perspective, or using a human factors / human-systems lens focusing on the individual and psychological level. People operate under multiple cognitive constraints, and demands on their attention, and they usually don’t have time to make complicated decisions, don’t want to do complex calculations, would prefer to use simple shortcuts and aren’t good at predicting their future behavior unless they’re using concrete information as a basis (and they’re still not that good, depending on the actual tractability of the system being analyzed). And a lot of people are simply poor at numerical reasoning.
I assume people do not make choices based on an accurate prediction of how often they’d use those options (e.g., hospitalizations), but on the apparent value of the overall package (low co-pay! low hospital visits!), specifically if the immediate costs like monthly premiums are tolerable and obtaining the true actuarial value requires both calculation and accurate prediction.
Think of someone who pays more up front for an expensive blender or another appliance because it has 7 extra features than its competitor (higher apparent value), even though they’ll never use those features and would have likely done just as well with the less expensive and more user-friendly gadget.
Let’s say the people who are going through long-term health conditions are the presumed individuals who are switching to the silver plan, do they even have the mental energy to process a lot of numbers (illness can be fatiguing), while being stressed about whether they’re getting good care (stress can increase the salience of negative information), and mentally inflating how often they may need to go to the hospital because they’re thinking about the topic or remember that they had to go in the past (availability heuristic, RIP Kahneman)? Would asking the question “how often did you want to go to the hospital in the last year” help them calibrate their predictions of hospital use prior to making these decisions?
Nudging is good and pretty valuable at the population health level, but I’d almost rather someone just design a wizard for people picking their health care policy to ask step-by-step questions (“if cost weren’t an issue, how often would you go to the doctor?”, and so on) and recommend or highlight 2 policies based on those, while allowing them to pick whatever policy they want. Or some other means to present complex information to support good decision-making in complex systems, as long as we had a good sense of how people in different patient groups (high need, low need, demographics, SES, etc.) even look at or consider health insurance policy information at the individual level.
Sorry about the rambling, something about this study set me off, and I was already highly caffeinated. It’s a pretty interesting result!
David Anderson
@Curtis: These are good thoughts.. and to the best of my knowledge, I have not seen the types of studies that you’re proposing (and we should be doing and publishing).