clock menu more-arrow no yes

This study is forcing economists to rethink high-deductible health insurance


In 2006, about one in 10 employees had a health insurance deductible over $1,000. Today? About half do.

To health economists, this sounded like good news; they've long theorized that higher deductibles would force down health-care costs. The idea was that higher deductibles would make patients become smarter shoppers: If they had to pay more of the cost, they'd likely choose something closer to the $1,529 appendectomy than the $186,955 appendectomy (yes, some hospitals really do charge that much). This would push the really expensive doctors to lower their prices so cheaper physicians didn't steal their business.

This was, however, just a theory. And a massive new study suggests it might have been all wrong.

Economists Zarek Brot-Goldberg, Amitabh Chandra, Benjamin Handel, and Jonathan Kolstad studied a firm that, in 2013, shifted tens of thousands of workers into high-deductible insurance plans. This was a perfect moment to look at how their patterns of care changed — whether they did, in fact, use the new shopping tools their employer gave them to compare prices.

Turns out they didn't. The new paper shows that when faced with a higher deductible, patients did not price shop for a better deal. Instead, both healthy and sick patients simply used way less health care.

"I am a little bit surprised at just how poorly patients were able to do when looking at very similar products, like MRI scans, and with a shopping tool," says Kolstad, an economist at University of California Berkeley and one of the study's co-author. "Two years in, and there's still no evidence they're price shopping."

This raises a scary possibility: Perhaps higher deductibles don't lead to smarter shoppers but rather, in the long run, sicker patients.

Higher deductibles mean sick people use less health care

drugs and money


Kolstad and his co-authors looked at the case of a large, unnamed company that shifted more than 75,000 workers and their dependents from a plan with no deductible to one with a $3,750 deductible. When the change happened, workers received a $3,750 subsidy to a health savings account — money they could spend freely on whatever health costs they incurred. The company also gave workers online tools to look up prices for doctor visits, tests, and other services they might need.

Workers' health spending dropped, and did so quickly. Average per-patient spending fell from $5,222.60 in 2012 to $4,446.08 in 2013. That's about a 15 percent decline in a single year — and it held true across all types of health services. Between 2012 and 2014, there was a 25 percent drop in emergency room spending, an 18 percent decline in physician office visits, and a 6 percent decrease in mental health services.

In one sense, then, the high-deductible plan did accomplish a key goal: lower health spending. But when the researchers looked at why spending dropped, they found it had nothing to do with smarter shopping. The average price of a doctor visit wasn't dropping.

Instead, under the high-deductible plan, workers just went to the doctor way less. The paper finds that "spending reductions are entirely due to outright reductions in quantity." Workers did use less "potentially wasteful care," like imaging services, but they also cut back on "potentially valuable care," like preventive visits.

Even more striking: The sickest workers were those who were most likely to reduce their use of care while still under the deductible — even though this is the group that needs lots of care and is most likely to blow through the deductible by the end of the year. Once these sick workers actually exceeded their deductible, though, use of medical services rebounded.

"People who are the most likely to go past the deductible also cut back by the most, and they did that entirely under the deductible," Kolstad says. "They respond to the spot pricing [the price of receiving care right then], and that leads to a very large reduction in care. We don't find any evidence they look for a lower cost. They just don't go."

Why does a deductible cause sick patients to forgo care?

This was the point, to me, that was most baffling in this new paper. Sick patients would likely have some sense that they needed a lot of medical care — and that they were probably going to hit their deductible. So why did they reduce the care they received in the start of the year instead of ponying up the costs, hitting the deductible early, and getting the care they thought they needed?

In some cases, you could chalk this up to a liquidity issue: A worker might not have enough money in her checking account to pay for all the care below the $3,750 deductible. But that explanation doesn't work here: In this case, the employer put a $3,750 subsidy in workers' health savings accounts.

Why wouldn't this group get the care they wanted, pay for it with the HSA, and just run through the out-of-pocket spending earlier in the year? Or, if they do want to reduce spending, why wouldn't they at least shop for a lower-cost provider instead of forgoing care altogether?

Kolstad doesn't have a definitive answer to this question, but he thinks it might have a lot to do with the difficulty all of us have, as patients, guessing how much we'll spend on health care in a certain year. This leads us to be more averse to upfront spending.

It's possible that even when we're sick, we tend to be optimistic. We might hope that our care costs less this year, and that maybe we'll even be able to roll some of our HSA account funds over to the next year.

"This is a difficult task for consumers to take on, and we now have very detailed data to show that's the case," he says. "When we've thought about the economics, we've generally thought this type of price change wouldn't be problematic, that sicker people would just spend their deductible and get the care they need. This research suggests that's not the case."

Americans aren't used to shopping for health care — and maybe we don't want to start

shopping cart


One study a few years ago, from the Altarum Institute, showed that Americans tend to spend more time shopping for dishwashers than for doctors — despite the latter being a rather more consequential decision.

For one thing, most of us don't have access to tools that would let us shop for doctors. I can go on Amazon and pull up prices for dozens of different dishwashers. But there's no website I can hop on, right now, to find out what different radiologists around Washington, DC, would charge me for an X-ray.

This study tried giving workers both the tools to compare costs and a financial incentive to go with the less expensive option. And, at least in this instance, those nudges weren't enough to encourage patients to choose cheaper doctors. Instead of looking for a lower-cost option, workers simply decided not to go to the doctor at all.

For Kolstad, this makes him skeptical of "demand-side" interventions in health care — those that rely on consumer demands for lower health prices to ultimately lead to less medical spending.

What's more, interventions that reduce demand could have the unintended consequence of actually raising long-term health-care costs. Think of the sick worker in a high-deductible plan who forgoes care in the early part of the year. It's possible that skipping preventive care or not filling some prescriptions could worsen health conditions that necessitate costly interventions a few years down the line.

This study only looks at two years of a high-deductible plan, and in that time period it doesn't show this theory bearing out. Still, Kolstad says it could be a long-term possibility that we just don't have enough data to know about yet.

"It's certainly plausible you'll see the cost changes later," he says. "That could manifest in lower productivity on the part of the worker, if you have people with worse health status. Those are long-run changes, but they are definitely a possibility."

Correction: An earlier version of this story misidentified hospital charge data as doctor charge data. This has been fixed.