How Many Health Services Are Actually Shoppable?

Since we’ve just been through Black Friday and Cyber Monday (Cyber Week now?), I thought this would be a fitting topic.

Let me start with the premise of why shopping for healthcare services is important: The only way to sustainably improve the value delivered by healthcare systems is to get people to start choosing higher-value providers and insurance plans. But how many health services are actually shoppable? This is not something I’ve written about before, and it’s because I am not sure of the answer. But here are some beginning thoughts on the topic.

First, I believe all insurance should be shoppable. No, people will not have perfect information about all the quality aspects of the insurance plans, nor will they be able to perfectly predict which cost-sharing arrangements will be cheapest for them, nor are all people mentally equipped to sift through this complex information, but major quality and price differences between different insurance plans should be clear for most people if the shopping experience is well designed.

Health services are another matter. And any effort to quantify the shoppability of health services should be broken down into two analyses:

  1. What percent of services are shoppable?
  2. What percent of spending is attributable to those shoppable services?

The second question is arguably more important if you are concerned with the fiscal stability of countries dealing with runaway healthcare spending. But you can only answer the second question after you answer the first question.

Since I like exhaustive, mutually exclusive categorizations, I enjoyed this breakdown of the first question:

NYC-MKT49701-032_Look into the Future Hospital Price _DTP-01

I’ll point out that they only considered hospital services. But one thing I like is how they considered how commoditized something is (how easy it is to shop for that service because quality is pretty standardized) and accounted for that when predicting how much prices will drop.

If I were to create my own categorization scheme, I would start with an emergent category. Not very many services are actually emergent requiring an ambulance ride straight to the nearest hospital that has capabilities to deal with the issue at hand. A quick thought to my residency days working in ICUs and EDs brings to mind the major things like acute coronary events, strokes, major trauma, sepsis, and acute respiratory failure.

What percent of spending is in this emergency category? It’s tough to calculate because emergency department charges alone don’t account for subsequent hospital care for those medical problems. But this report by the Health Care Cost Institute at least gives a few numbers that could help us ballpark it, and I’d put the total spending on emergencies at around 5-20% of total healthcare spending. Maybe there are better analyses out there on this very thing, but I haven’t seen them.

Once we’ve ruled out the emergent events, we’re left with the other category: everything else. These things could all theoretically be shoppable, but there are some with potentially insurmountable barriers. For example, some services have no easy-to-identify quality metrics (such as ambiguous symptoms requiring a diagnosis–how do you choose the right provider to diagnose that?). And other times, patients are not able to easily predict which services will be needed (such as when an elderly patient with multiple comorbidities goes to the emergency department with shortness of breath).

I could subdivide nonemergent health services into two categories: diagnostic services and treatment services. Per the examples in the prior paragraph, diagnostic services are harder to shop for. Treatment services are typically easier to shop for (like when you know you need a specific surgery). But there are times when the treatment of a condition is not easily separated from the diagnostic efforts. For example, if an orthopedic surgeon diagnoses you with knee osteoarthritis requiring a knee replacement, you typically won’t shop around for the highest-value knee surgeon–you’ll just get the surgery by the doctor who made the diagnosis (not that second opinions are impossible). Or, if you went to a specific hospital’s emergency department and were found to have a heart failure exacerbation requiring hospital admission, you don’t hop on your smartphone and start comparing which hospital can best treat that–you just get admitted to the hospital you’re already in.

So I guess I’m saying that diagnostic efforts are hard to shop for, and treatment efforts that are naturally connected to those diagnostic efforts are also hard to shop for.

How about some numbers? This analysis by the National Institute for Health Care Reform estimates that at least 70% of inpatient spending is shoppable, and 90% of outpatient services are shoppable. But then later they wrote that, when accounting for all potentially shoppable services in our healthcare system, those services account for only about 1/3 of total healthcare spending.

Clearly there is more to learn and discuss about this topic.

A couple parting thoughts: Keep in mind that, regardless of what percent of services or spending is not shoppable, it doesn’t mean we should give up on enabling patients to shop for as much as they are capable of. Getting them to choose higher-value options is still the only long-term solution to improving the value delivered by healthcare systems, as explained in my Healthcare Incentives Framework. And there will likely be a positive externality effect on non-shoppable services when providers up their game on shoppable services.

The Issue of Healthcare CEO Salaries

pnhporg
Image credit: pnhp.org

When I was an undergrad studying business strategy, people would ask me what business strategy is. I would tell them that someone who studies finance aspires to become a CFO, someone who studies operations aspires to become a COO, someone who studies marketing aspires to become a CMO, and someone who studies strategy aspires to become a CEO.

Defining the business strategy is the job of a CEO. This involves making decisions about what the company will and will not do. It’s deciding how the company will leverage or develop unique capabilities to offer a unique value proposition relative to competitors so that it can achieve a sustainable competitive advantage, which translates into earning greater profits in the long run.

A company’s board of directors is tasked with recruiting and hiring a CEO. This is a tough job. If they choose the wrong person, the company will end up with a losing strategy, and it could cost the company millions in profits. Therefore, they think of this person as an investment. Getting the right person may cost a few million dollars per year, but if spending a few million dollars per year on the right CEO results in the company making tens of millions of dollars more in profit each year, that was a pretty great investment! (FYI, usually the CEO’s base salary is much lower, but they are paid big bonuses for improving profitability.)

What if the board of directors decided that super high CEO salaries were just unethical in healthcare? What if they tried to offer a regular salary when they’re trying to recruit a new CEO? The most unselfish, ungreedy CEO candidate in the world is still motivated by financial incentives, and if they get an offer for $1 million from one board and an offer for $400,000 from another, that’s a hefty difference. They would have to have some very compelling reasons to give up an extra $600,000 per year. I’m sure sometimes that happens, but I doubt the majority of CEO candidates can find enough compelling reasons to forego hundreds of thousands of dollars per year. Heck, if taking a less desirable job means I can retire years earlier and spend more time with my future grandkids, I’d go for it.

So maybe all healthcare companies could decide together that they will pay their CEOs less? This would work out great until a board decides they want a specific candidate really bad because they believe that person will help their company more than any other candidate could. In the face of the decision to keep that truce for the good of everyone or to make the decision that they think is best for their company, they’re going to usually choose what’s best for their company. Therefore, a truce like that would never last. This is exactly what history shows us whenever companies have colluded over any other price–eventually you get defectors because the benefits of becoming the defector (especially the first defector) are so great.

How about having a new law that limits healthcare CEO pay, as was suggested by Steven Brill in his landmark Time article, Bitter Pill? It would be an interesting thing to try. We don’t know for sure what would happen, but we could infer quite a few things from seeing what has happened other times governments have enacted price controls. Without going into all of those, the biggest thing I suspect is that sophisticated companies would find a way around that limit by compensating them in different ways. This article does a good job of showing the complexities of CEO compensation, which I’m sure companies could leverage to find ways around any wage control.

So, with all that context, what do you think about healthcare CEO salaries? If you still say they are ridiculously high and that a world without such vast pay differences would be great, I agree. But let’s move past the ignorant complaints and comparisons that they seem to spark every time the cost of healthcare is brought up.

Yes, Quality Metrics Get Gamed

noahs-ark-zebra
Image credit: jewishpress.com

I focus a lot on the importance of getting people simple, relevant quality information to help them compare providers and insurers. But I realize this information will not be perfect. Let me give you an example as it relates to provider quality rankings.

Say there is a hospital that is doing pretty well on the Vizient rankings. They are near the top, but their mortality rate kept them from being the top spot. When the providers at this hospital have their meetings about what should be done to improve their mortality rating, they have two options for how to do that:

  1. Do extensive research into all the patients who died in the hospital in the last year and identify the variety of factors that may have contributed to those deaths, then launch initiatives and processes to eliminate as many of those contributors as possible. There will need to be a lot of research and implementation time spent on this, plus they will need to get buy-in (not easy) to get people to change how care is delivered, or they will possibly need to make changes to their EMR (even harder) to help protect against issues that were identified. The end result will likely be a modest reduction in some of those factors, which may or may not result in actual measurable differences in mortality the next year.
  2. Look at the billing codes providers are putting into the EMR and talk to them about making sure to get all those key codes in place (don’t forget acute metabolic encephalopathy, protein-calorie malnutrition, type II diabetes with complications, or hypokalemia!) so that their overall patient acuity is higher (i.e., more accurately reflects how sick their patients are), so then when their risk adjustment is done the next year to calculate their mortality rating, they will have a big improvement and snag that top spot.

You can see by how I described each option which one is the low-hanging fruit and is also much more likely to result in an improved ranking.

This is not just a theoretical issue–I see it first hand all the time. It happens everywhere. The sad thing to me is not that people who work in hospitals are trying to make their numbers look as good as possible, but that those conversations take up so much meeting time that it crowds out any discussion on how to actually improve the safety and quality of care for patients.

So, yes, quality metrics are not perfect. They have their issues. But I do feel that they generally are successful at revealing large differences in quality between different providers. And that is probably enough to motivate patients to avoid lower-quality options and shift market share to reward higher-value providers.

How I Figure 50% of Healthcare Spending Goes to Administrative Expenses

I got an email this week from my hospital system introducing the team of administrators who are managing the CME (continuing medical education) credit card they give to providers. You see, providers are given an annual stipend that they can only use for CME activities, such as trips to medical conferences, and this is the card we are supposed to use for those expenses. I read the email and looked around to see if anyone else thought it was ludicrous that there is a whole “team” of people dedicated to something like this.

The issue of unnecessarily high administrative expenses in the U.S. healthcare system has been on my mind ever since the recent article in JAMA detailed the “6 waste domains” in our healthcare system and pegged administrative expenses as the biggest one. So, today, I’m going to explain how I calculate administrative expenses in our healthcare system. I believe it’s more intuitive than other approaches because it follows the flow of money through the healthcare system and shows each point where some of the money is siphoned off for administrative expenses. Disclaimer: I’m using nicely rounded numbers and lots of assumptions for the sake of clarity.

Money enters our healthcare system when the patient pays insurance premiums. Let’s say the amount of each dollar paid to an insurer that gets spent on actual care is around 85 cents (their “medical loss”), which means 15 cents goes to administrative expenses in the form of insurance overhead:

 

 

Picture1

And then what about the 85 cents that is being paid to healthcare providers? Studies have used reports submitted to Medicare to estimate the percent of hospital revenue that goes to administrative expenses, and they come up with about 25% in the 1990s, but it has been rising. So if we guess that about 29% of those 85 cents paid to hospitals and other facilities goes to administrative expenses (like teams of people managing CME credit cards, plus armies of coders, “revenue cycle managers,” and so many more), that works out to be 25 cents in additional administrative expenses:

Picture2

 

Of the 60 cents left, it’s going to actual care, such as paying caregivers and buying other care-related things like drugs and devices, right? Not so fast. First off, let’s assume about half of that 60 cents goes to drugs and devices (both of which are generally overpriced, but that’s a different subject I’ve written about before). This leaves 30 cents of every healthcare dollar for caregivers. And if 100% of caregivers’ time were devoted to care, we would be dancing and singing every day we go to work. Instead, we spend a great portion of our time on “the system,” dealing with insurance issues, federally mandated trainings, documentation time ensuring billing and diagnostic codes are accurate and thorough, etc. etc. etc. When accounting for all of those factors, it adds up to much more than the estimated 15% of my time dedicated to administrative tasks. I’d peg it closer to 30%. Others would probably say 50%. Things seem to be worse here in the U.S. compared to in the motherland. Anyway, that means roughly 15 cents of every dollar can be attributed to caregiver overhead:

Picture3

I will not include in here the additional money providers have to be paid to compensate for the high cost of medical training and malpractice insurance. So, overall, I’m guessing that 50 cents of every dollar that enters the healthcare system goes to non-care things, which we call administrative expenses. Now, this number is very different from the number researchers came up with in the study cited above, which estimated that the overall administrative cost was around 30%, but they’re adding things up differently. That’s fine. We’re counting differently. But the important thing is looking at each of those siphoning points and understanding why they have higher-than-necessary admin costs. I won’t go into detail on that today, but I believe the main factor is the complexity of the system. So. Much. Complexity. We need a simpler healthcare system. I have described a few options here and here. The simplicity we could achieve may come partly by direct efforts to simplify, but I believe it will be more a side-benefit of making the other changes that are necessary in our system.

 

Are Patients Smart Enough to Choose Higher-value Providers and Insurers?

I’ve been hearing the argument lately that healthcare is so complex that patients are incapable of making good decisions. I (mostly) disagree. Here’s how I break this argument down:

Treatment decisions: This is when the patient weighs the different treatment options’ risks and benefits and decides on the one that is right for them. This certainly cannot be done without the assistance of a physician, who needs to simplify those risks and benefits and help clarify which ones are most relevant to the patient. Shared decision making! These decisions are important and determine many aspects of healthcare spending and patient outcomes and satisfaction, but they are not integral to fixing the healthcare system.

Insurance plan decisions: This refers to patients choosing which insurance plan they will get. There are actually usually two rounds of decisions for this. The first-round decision is done by the employer, which narrows down the many options in the region to just a few. Then the employee does the second-round decision of choosing one of those options. A similar process happens with Medicaid in many states–Medicaid contracts with a few different insurers, and the enrollee chooses one. And even Medicare Advantage has a similar two-round decision. Patients need to understand what their likely care expenses will be that year (super tough to predict for many people!) and then choose the plan that seems most likely to cover those care needs with the lowest out-of-pocket spending. And since out-of-pocket spending is a combination of many things (deductible, copays, coinsurance, other alternative payment schemes like reference pricing or multi-tiered networks), this can get pretty complex, and would likely be impossible for a majority of people to easily identify the plan that would be best for them. The two-round decision most people face simplifies this quite a bit by narrowing down the options (at the risk of agency costs), but sometimes there are still too many options. And for those dealing with the open market without a first-round decision narrowing down their options, ways to simplify their selection are an absolute necessity. That’s why I like the idea of dividing plans into different standardized tiers according to their coverage, networks, and out-of-pocket requirements. Software-encoded decision algorithms, which have a patient put in their information and then identify the plan best suited to them, are also great.

Provider decisions: Patients get to choose which providers they go to for care. Lumped in this category is also imaging options (like my experience shopping for an MRI), lab options, and any other medical services they may need. These provider decisions are subject to two rounds of decisions just like insurance plans. The insurer does the first-round decision by narrowing down the provider options (the “in-network providers”), and then the patient typically chooses one of those. What do you need to understand to be able to choose the best provider? I’ve talked about this elsewhere. Patients need simple, salient quality metrics, and they need to know their expected overall out-of-pocket costs. Relatively well-educated people can use well-structured information like that and 80% of them will make the right decision. I don’t know what percent of people with less education can make the right decision when there’s an obvious right answer, but certainly it will be much lower.

So, can patients make good provider and insurer decisions? I believe most of them can if they have the right information and the incentives to consider both cost and quality of the options available to them. It will never be 100%, but that’s okay, because we don’t need 100% of people to make a perfectly rational high-value decision in order to get the changes necessary to start fixing healthcare.

The Impact of Single Payer on Healthcare Spending

warren sanders
Image credit: politico.com

One of the main reasons people are pushing for a single-payer system in the U.S. (termed “Medicare for All” these days) is to achieve universal access to insurance coverage. But what about its impact on total healthcare spending? This is how I think about that question.

First, remember that there are two aspects of spending that we need to be thinking about. The level of spending is how much we’re currently spending, and the trend of spending is how much that amount is increasing over time.

Level of Spending. Even though a single-payer system will increase the number of people with insurance, it’s possible it could still reduce the level of spending overall (although estimates vary widely). How could it save money? The overall cost to administer health insurance would probably go down simply due to sheer economies of scale (of course, this is debated too). Plus, the amount of advertising done by a single government-run insurer will be less than the amount currently being spent by private insurance companies. Single payer allows greater price control as well, so that could be a huge savings (although, if it means paying all providers current Medicare rates, that could be catastrophic).

Trend of Spending. Healthcare spending continues to increase faster than inflation, mostly due to medical innovation, rising prices, and an older and sicker population. Would single payer affect our trend of spending? Well, sort of. It can refuse to increase prices, which is a very tempting spending reduction mechanism that our government has tried in healthcare before (ahem, SGR). But with the relative administrative simplicity single payer would bring to providers, there would be at least some wiggle room for price reductions. Or it could refuse to cover low-yield or really expensive drugs and treatments (rationing), which we generally don’t take well to in this country. Unfortunately, there just aren’t a lot of effective ways single payer can directly impact the trend of spending, which is ultimately the most important one.

I know I’m leaving out lots of other impacts Medicare for All would have on the level and trend of spending, but I think I’ve covered most of the major ones.

So does this mean we are doomed then? Even if we get universal coverage, there’s no way to avoid total fiscal collapse secondary to runaway healthcare spending? No! Implemented correctly, single payer could do a great job helping to avoid many episodes of care, and it could also enable the costs of delivering the care that cannot be avoided to go down immensely. I’ve explained how already.

So let’s just keep all this in mind as Medicare for All is being debated. People can quibble about the projected savings/costs and various other numbers all day long, but the important part in the long run is how it’s implemented.

Why Isn’t Price Transparency Working in Healthcare?

transparent
Image credit: wish.com

I strongly believe that getting people the information and incentives necessary to choose higher-value providers and insurers is the solution to improving value in healthcare (see my Healthcare Incentives Framework). But, you say, we’ve tried that and it doesn’t work, and current efforts are a waste of time!

Here’s an example of some great research that you might use to support your opinion:

Examining a Health Care Price Transparency Tool: Who Uses It, and How They Shop for Care (Sinaiko and Rosenthal, Health Affairs, April 2016)

The news media would see this and report the main findings–that only 3% of enrollees used Aetna’s price comparison tool–and argue that even people who have the opportunity to shop for care will not do it, which they will interpret to mean any “consumer-driven” healthcare effort is proven through evidence not to work. People can wrest information to prove whatever they want.

But what if you actually read the study?

Sinaiko and Rosenthal found that only about 60% of enrollees even had a claim during their study period. And of those 60%, I’m guessing a large percentage of those were outpatient visits (primary care or specialty) with established providers, which are claim types that people historically do not shop for. Think about it, if you have your favorite hairdresser who knows you best, you have a relationship with that person, and you like how they cut your hair, are you going to price shop every single time you need to get your hair cut?

Now take out all the non-shoppable services (the subject of a future post), and we’re left with a relatively small percentage of enrollees who may have actually had a reason to shop for care. But, wait, what about the people whose insurance plan required them to pay the same amount regardless of which provider they chose? (Remember, people need an incentive to choose higher-value options.) I wish I had the numbers to know what percent of enrollees would be left after all those exclusions. 30%? 10%? Those enrollees would be the target audience who we want to use the price comparison tool.

I assume Aetna did a good job notifying people about this price comparison tool, so maybe 75% of people read the mailer and then half of them remembered it when it came time to shop for care. Why only half? Because people aren’t yet accustomed to shopping around for value in healthcare. They’re used to going to the lab their doctor tells them to go to, for example. It doesn’t even cross their mind that there are cheaper options out there. I think there’s also an assumption that you have to go where your doctor tells you to otherwise your doctor can’t get the results. Thanks, non-interoperable EMRs.

So what are we left with? 3% of enrollees seems about right.

What about shoppable services, specifically? Considering all the factors above, the percent of people getting those services who actually shopped first blew me away. Tonsillectomy: 54%. Total knee replacement: 48%. Inguinal hernia repair: 27%. Cararact or lens procedure: 18%. Vaginal delivery or C section: 16%. Carpal tunnel release: 12%. These lower-percentage ones strike me as the ones that would more commonly be performed by a doctor you’re already established with (again, making you less likely to shop around), but I could be wrong here.

Anyway, you know what this proves to me? That price comparison tools can work! And I believe they will be used more and more as people start getting insurance plans that require them to pay more for more expensive options, and as they remember they can shop for price. The younger generation will probably drive a lot of this because they will be more used to using these tools and shopping for care and will eventually get older and start needing more services.

Here’s another great paper on this topic:

Americans Support Price Shopping for Health Care, But Few Actually Seek Out Price Information (Mehrotra, Dean, Sinaiko, and Sood, Health Affairs, August 2017)

This study supports what I’ve written above. Here are some takeaways from it. It’s looking specifically at people who have the type of insurance plans that would give them an incentive to compare prices:

  • 72% of people think it’s really important to shop for value in healthcare
  • 93% of people know prices vary greatly among providers
  • Only 22% of people think higher prices in healthcare equate to better quality
  • 75% of people said they don’t know of a resource they can use to compare costs among providers
  • 77% of people who didn’t price shop for their last healthcare service said it was because they were seeing a provider with whom they were already established
  • Only 1% said they didn’t shop because it was emergency care