Over the past two decades, a lot of low-hanging fruit has been lopped off health care costs. Inpatient days per thousand have come down by about one third. Generic drug utilization has reached 90% in some classes. And yet, there has been no appreciable bend in the health care spending curve.
In the search for savings, what’s been more or less overlooked is care that is of little or no value. A few years ago, the Institute of Medicine labeled more than a third of health care spending as wasteful. Subtract out the subset that represents fraud and abuse, and you still wind up with about 30 cents on the dollar that goes to low-value or no-value care.
What do those terms really mean? It’s not just unnecessary tests or overuse of antibiotics. It’s unwarranted variation in care—a culprit studied for years by the folks who produce the Dartmouth Atlas. The classic example is low-back pain: Regardless of whether most patients try stretching and weights, visit a pain management specialist, or choose surgery, outcomes are roughly the same—though the cost of each differs dramatically.
“We’ve always known there’s a ton of variation, but under fee for service, nobody cared,” says Joshua Rosenthal, chief scientific officer at RowdMap. The shift to value-based payment and risk-bearing schemes has changed all that, he says. “Being able to get rid of that 30 cents is really, really important. Essentially, as population increases and expenditure decreases, you have two choices: start rationing needed care or get rid of the unneeded care.”
With offices in the heart of Louisville’s Whiskey Row, RowdMap is a young company full of data scientists and smart hipsters who are exploiting CMS’s so-called data liberation movement, one of the largest releases of government-held data ever. CMS is also making much more provider-level information public through the Physician Compare and Hospital Compare websites and nearly a half dozen other databases. As payers divert compensation from fee-for-service to risk-bearing models, providers will need all the practice data they can get.
Having the data is one thing. Knowing how to use it is another. Applying its computational power to the data, RowdMap puts providers into high-, medium-, and low-value buckets compared with peers in their markets, using specific benchmarks to show why outliers differ from the norm. RowdMap has developed “no-value care” and population-health profiles for every physician and hospital in the country.
RowdMap, a Louisville, Ky., health information company, uses publicly available data to create reports on physician efficiency and readiness to take on risk. The chart below is an illustrative example. Color-coded dots signify values relative to their peers, and the dotted line in the middle is the midpoint and benchmark. The red and orange dots represent performance below the benchmark and the green and blue dots, performance above it.
Physician risk-readiness report
“It’s not just summary benchmarks; it’s literally about how they practice. How they spend their time. What they bill for and associated costs,” says Rosenthal. “That allows you to do some pretty interesting things.”
Like working with payers to help them develop high-value provider networks. For one Northern California health plan, the average provider in its network cost the plan $274 per member per year (PMPY). Using RowdMap’s provider profiles, the plan removed from its network 101 physicians whose profiles suggested that they provided an excess of low-value care, replacing them with 121 high-value physicians. Over 12 months, the average PMPY cost across the plan’s provider panel fell to $179, according to RowdMap.
“We can say ‘Per patient, per interaction, the difference in the dollars you pay to this ortho specialist compared with the average ortho specialist could be a thousand dollars,’” says Rosenthal.
As you might imagine, that kind of information can be explosive—which is why the Northern California plan will remain anonymous—so a lot of RowdMap’s work is intentionally low profile. However, at the Health Datapalooza conference in May, one payer openly shared how it works with RowdMap to pool publicly available information with its own internal data and analyze it to accomplish two very different ends.
Stephen Ondra, MD, who oversees medical policy for Health Care Service Corp. (HCSC), the parent company of Blues plans in Illinois, Montana, New Mexico, Oklahoma, and Texas, uses that information to create networks that are narrow—not by discount but by the value providers bring to patient care.
“Who are the desirable providers you want in your network? Who aren’t? How many do you need and where do you need them?” Ondra told a session at Health Datapalooza. “Is there a mismatch of disease burden and provider availability?”
That’s not only an issue of access, it’s one of unnecessary costs, said Ondra, a neurosurgeon by training. “As a surgeon, when you’re busy, your criteria for surgery get really narrow. When you hit those lulls, your criteria broaden. Not inappropriately, but enough to fill out the OR schedule. … When you have too many providers in an area, they’re going to do things on the margin—not inappropriate, not abusing people, but low-value things that fill their time.”
In contrast to HCSC, CareFirst, the Blues affiliate serving Maryland, Washington, D.C., and two counties in Northern Virginia, is building an expansive provider network. CareFirst says it is using publicly available cost-efficiency information to help primary care physicians understand the cost implications of their referral patterns.
CareFirst leaves the judgment of a specialist’s quality up to the referring physician, says Jonathan Blum, CareFirst’s executive vice president for medical affairs. “We’re using this concept to help primary care physicians understand who are the best partners in order to manage care.”
That’s critical information for any physician or group looking to evaluate its own readiness for risk, says Rosenthal. “Providers ask, ‘How much low-value care am I creating? Why am I creating it? And which partners would be best for my value chain?’” Similarly, data can help physicians understand which flavor of risk matches their style: “Should I pick an MSSP or a Next Gen ACO?”
Using Dartmouth Atlas definitions and public data, RowdMap benchmarks physicians against their peers in four areas: visit intensity, procedural efficiency, pharmacy efficiency, and referral intensity. For each of these measures, a physician is given color-coded dots, signifying high, medium, or low value relative to peers. Physicians are also graded on their overall efficiency.
“Green or red doesn’t mean you have better or worse clinical outcomes; it means you practice higher-intensity treatment,” says Rosenthal. “So, what is the cause of that? Is it your referral pattern? Prescriptions? A particular drug? Is it variations within your practice?”
Specifically, visit intensity refers to the number of patient encounters for an episode, procedural efficiency is indicative of the intensity of a procedure, and pharmacy efficiency is suggestive of the cost of prescriptions written for a patient’s condition. The fourth measure, referral intensity, can be a big cost driver—a concern especially for primary care physicians in risk-bearing networks.
“If you come in with low-back pain, do I refer you to a higher-intensity service like a pain management specialist or an ortho? And if I refer you to an ortho, do I refer you to a surgeon or someone else? Do I refer you to a green dot or a red dot surgeon? Do I refer you to someone who is low or high value?” asks Rosenthal. “Docs’ eyes pop out of their heads when they see all that red.”
The beauty of it all is that as fee for service fades into the rearview mirror, the road ahead is paved not only with good intentions but with data that are publicly available, waiting to be harnessed and not locked away in a proprietary database. Rosenthal sees American health care as preparing to leave fee for service behind and readying itself for a value-based present and future.
“The world that was is about fee for service. There is a data architecture and a business mindset associated with fee for service. It was really about claims and reconciliation. Maximize as much stuff as you can get, lock your data away, don’t share it, sell the secret sauce. That’s the playbook for how to do it.
“In the world that is, where it’s all about pay for value, there’s a data architecture that matches it and a business mindset that matches it. The data architecture is about taxonomy and interconnected meta-data, and the question is, ‘How do I statistically normalize a hospital referral region to Medicare contracts?’ So our data architecture looks different, and five years ago, it would have been impossible to do what we are doing now in a meaningful way.”