One thing that you will hear about constantly in healthcare circles is that we need to replace fee-for-service with value-based care.
If you are not familiar, fee-for-service (FFS) is a payment model that involves a provider (i.e. doctor) getting paid for every service they do. The unfortunate side effect of this payment model is that it incentivizes providers to give way more treatment than is necessary. Value-based care is the emerging alternative where providers get paid based on patient healthcare outcomes rather than sheer volume of treatments. It’s like paying a carpenter based on how well they redid your kitchen instead of how many nails they hammered.
So surely we should just replace fee-for-service with value-based care and we should be all good right? Well, yes but, as usual, the devil is the details. We need to talk about what value-based care is, the nuances of how it’s implemented, and what challenges we are going to have to overcome if we are going to implement it successfully.
Value-Based Care Approaches
There are several broad approaches to value-based care, each with different levels of provider risk. The broad approaches are:
Pay for Performance: Providers are paid more if they hit certain performance metrics
Shared Savings/Shared Risk: If providers hit their assigned financial and quality metrics, they get a percentage of the savings; else, they have to pay a percentage of the losses
Bundled Payments: Providers are paid a fixed amount to treat a patient for a specific illness, condition, or medical event (e.g. a hip replacement)
Capitation Model: Providers are paid a fixed amount per patient per unit of time to treat a patient for all their conditions
Full Risk Insurance Model: Providers and payers integrate
In this article, we will delve into the details of the two value-based care approaches that are rapidly gaining traction and attention: bundled payments and the capitation model.
Under a bundled payments structure, providers are paid a fixed amount to treat a patient for a clinical episode, e.g. a specific illness, condition, or medical event. For example, a provider may be given $32,000 to take care of everything related to a patient’s hip replacement. Payers give providers a target price for each clinical episode. If providers are able to treat the patient with high quality for less than the target price, the providers can keep a portion of the savings. If not, the providers have to pay a portion of the losses.
The actual implementation of bundled payments differs based on the specific payment model. One example is the Center for Medicare and Medicaid Services’ (CMS) Bundled Payments for Care Improvement Advanced (BPCI Advanced) Model. Here’s how it works:
1. Every year, CMS calculates the target price for each clinical episode.
To calculate the target price, CMS first calculates the benchmark price for each clinical episode by looking at:
The hospital’s historical expenditures
The hospital’s patient demographics and health risks
The hospital’s peer group expenditure trends
The hospital’s peer group patient demographics and health risks
Then, CMS calculates the target price by applying a 3% discount to the benchmark price for each clinical episode. The 3% discount is applied to force providers to reduce costs over time.
2. Providers treat patients and get paid under the FFS system as per usual.
For the next 6 months, providers treat patients as they normally do. Whenever a patient comes in needing treatment that is covered under the BCPI Advanced Model, a clinical episode is triggered. Providers will treat the patient and get paid under the fee-for-service system.
3. Every 6 months, CMS will retrospectively compare costs and quality of clinical episodes against targets. Depending on performance, CMS will send providers additional payments or providers will refund some of the money back to CMS.
CMS collects seven administrative quality measures;
Unplanned, all-cause readmissions within 30 days of hospital discharge
Risk-standardized complication rate 90 days after total hip and/or knee surgery
All-cause, risk-standardized mortality rate 30 days following Coronary Artery Bypass Graft Surgery (CABG)
Days spent in acute care within 30 days of discharge from a hospitalization for Acute Myocardial Infarction
Use of a particular type of antibiotics for certain treatments
Starting from 2021, hospitals have the option to be measured against more clinical episode-specific alternative quality measures.
For each clinical episode and quality measure, CMS compares how each hospital performs compared to all the other hospitals across the country. It then rolls up these scores into a composite score per hospital and assigns them a composite quality score (CQS) adjustment amount.
CMS calculates the reconciliation amount by taking the difference between how much the hospital spent on the clinical episodes vs. the target prices. It multiplies the reconciliation amount by the CQS adjustment amount to get the gross savings/losses. If the hospital has saved money, it will get paid a portion of those savings from CMS. Otherwise, the hospital will have to repay a portion of those losses to CMS.
Under a capitation payment structure, providers are paid a fixed amount per patient per unit of time to treat a patient for all their conditions, with financial incentives and/or penalties based on quality. For example, a provider might be paid $40/month to take care of any problems a patient may have.
There are two main tracks within capitation: global or partial capitation. In global capitation arrangements, providers are paid a single, fixed amount for all healthcare services given to a patient, including primary care, specialty care, and hospitalizations. In partial capitation arrangements, providers are responsible for a set of healthcare services provided to a patient, such as primary care and lab tests, but the rest of the services are reimbursed with a fee-for-service model.
Similar to bundled payments, the actual implementation of capitation payments differs based on the specific payment model. One example is CMS’s Direct Contracting Model. It has two payment mechanisms: Primary Care Capitation, where providers are paid monthly capitated payments to provide primary care services to patients, and Total Care Capitation, where providers are paid monthly capitated payments to provide all services to patients. Here, we will focus on the Total Care Capitation. Here’s how it works:
1. In the beginning of the year, CMS calculates the total care capitation payment for each direct contracting organization.
CMS first calculates a benchmark price based on:
The organization’s historical expenditures
The organization’s patient demographics and health risks
The organization’s peer group expenditure trends
The organization’s peer group patient demographics and health risks
From this benchmark, CMS:
Applies a discount to force organizations to reduce costs over time
Withholds 5% that can be earned back by the organization based on their quality performance
Withholds for any fee-for-service claims that come from a provider outside of the direct contracting organization treating the patient
2. Providers treat patients as usual. Every month, CMS sends the direct contracting organization a total care capitation payment.
Direct contracting organizations get sent a monthly check from CMS per patient. Any provider from outside of the direct contracting organization that treats the patient bills CMS fee-for-service as per usual.
3. At the end of the year, CMS will retrospectively compare costs and quality of clinical episodes against targets. Depending on performance, CMS will send providers additional payments or providers will refund some of the money back to CMS.
CMS will calculate the quality score for the organization by looking at:
Patient survey that measures the patient and caregiver experience
Rate of all-cause, unplanned admissions for patients with multiple chronic conditions
Rate of risk-standardized, all-condition readmissions
Days spent at home (for high needs patients)
Based on the quality score, CMS will determine how much of the quality withholding the organization gets back.
Then, it will calculate the organization’s performance year expenditures by adding together the capitation payments and total FFS payments to get the gross savings/losses. If the organization has saved money, it will get paid a portion of those savings from CMS. Otherwise, the organization will have to repay a portion of those losses to CMS.
Thoughts on Payment Models
Shifting to value-based care is hard. Aligning the incentives between all the players is difficult. Radically changing how a multi-trillion dollar industry with so many moving parts operates is difficult. Ensuring that this system can’t be gamed is difficult. Given all this, the two value-based payment models we’ve seen so far are extremely well-designed.
The performance of the value-based payment models, however, has been mixed — some of the payment models have generated savings while others haven’t. The reasons for this are multifold: it’s still early days for the value-based payment models, and the transition period is rough for all parties involved. Specifically:
Providers are shifting to a radically different way of delivering care, where they are on hook for care coordination, quality, and more. This requires massive investments in infrastructure as well as changing workflows and culture.
Payers are shifting to a radically different way of paying for care, which also requires massive investment of infrastructure as well as experimentation on how to structure payment models, determine the right benchmark prices, collect the right quality metrics, etc.
Furthermore, participating in these new payment models has been voluntary for providers up till now. This has created an adverse selection problem where providers can choose to participate in the payment models that have the highest financial ROI for themselves but may not be the best for the healthcare ecosystem overall.
Given all of this, I’m optimistic that we’ll really be seeing the impact of value-based care on improving quality and lowering costs in a couple of years. Below are a look at the kinks we need to iron out before then!
Under most value-based care models, providers are incentivized to keep costs as low as possible since their revenues are essentially fixed. One way that providers may be tempted to do this is by only taking on healthy patients, who barely need any medical care. To counteract this, payers will usually risk-adjust their payments, paying providers more to take care of sicker patients.
An unfortunate side effect of this, however, is that one of the easiest ways for providers to make more money through the value-based care system is to upcode patients and make them look sicker than they really are. This way, providers can get more money from the payers to take care of the patient but don’t actually have to spend that much money on the patient (since the patient is actually pretty healthy). Many startups have also popped up in this space, promising to help providers get more money by finding all of the health risks their patients have, showing just how lucrative it is.
CMS has tried to counteract the upcoding of patients by looking at the demographics and health risks of the region’s population, not just the organization’s. This has definitely quelled the impact of upcoding a bit but the incentive still exists.
Some strategies to solve this problem might be to have 3rd parties — such as nonprofits — weigh in on the riskiness of the organization’s patient population (like they do in the UK) and have zero-sum risk adjustments.
Restricting Access to Care
Another method providers can use to keep costs low is to restrict access to care. If patients can’t access the care they need, providers don’t need to spend any money taking care of them.
Two strategies to combat this are: have more quality metrics that measure access to care and increase the impact of the quality metrics on how much providers get paid (right now, in the Direct Contracting TCC model, only 5% of the provider’s payment is dependent on quality).
Measuring the Wrong Metrics
One of the fundamental questions in value-based care is defining what “value” is. How do you determine whether a treatment was successful or not? There are so many factors that go into someone’s health and well-being so how do you capture all of that in one number?
In the absence of this holy grail metric, payers have chosen metrics — such as readmission rates and complication rates — that are easy to gather and tightly correlate with their bottom lines. These metrics, though valuable, still leave a lot on the table. Just because you didn’t have to go to the hospital doesn’t mean that you were in perfect health.
Another problem with having a few metrics is that people fixate on them and they become a goal in and of themselves. This is a classic example of Goodhart’s law: “when a measure becomes a target, it ceases to be a good measure.” For example, when pain scores were measured to make sure the hospital staff were doing everything to help patients’ pain, providers were given a perverse incentive to prescribe more opioids, which solved the pain problem but introduced a host of addiction problems.
No metric is perfect but the first step is to devise what quality means for different procedures and populations. These metrics should focus on actions that cause significant harm or waste by extreme outliers. Then, if applicable, design counter metrics to ensure that the metric can’t be gamed (e.g. a counter metric to lowering costs is ensuring that access to care is not restricted). Lastly, if applicable, leverage tech, especially remote patient monitoring tools, to collect measures that more closely correspond to an individual’s improved quality of life.
Overtreatment in Bundled Payments
Under the value-based bundled payments model, providers are paid a fixed amount to treat a patient for a clinical episode. This means that the more episodes the provider treats a patient for (e.g. the more hip replacements a provider does), the more they get paid. This is better than fee-for-service, where providers get paid for each action they take rather than for a course of treatment, but still incentivizes overtreatment. Given that a whopping 21% of medical costs are attributed to overtreatment, this is an important problem to solve.
In fact, the quality metrics payers collect to ensure that the providers effectively treated the patient can actually incentivize overtreatment. Who has a better chance of doing better after a surgery: a sick patient who really needed the surgery or a healthy patient who didn’t need the surgery at all? By increasing the pool of healthy candidates getting a procedure they don’t need, providers can get paid more and score better on their quality metrics.
Marty Makary, in his book The Price We Pay, tells the story of an obstetrician who had a habit of performing a C-section on any woman in labor, whether or not she needed it. This behavior would never be caught by measuring infection and readmission rates. In fact, this obstetrician’s complication rates were probably impeccable since operating on a healthy person who doesn’t need surgery tends to have extremely low complication rates.
One strategy to combat this overtreatment is to include Appropriate Use Criteria into the quality metrics to see if the treatment was truly needed.
Not Actionable Metrics
Currently, the quality metrics being collected — such as mortality and readmission rates — are not actionable to help providers improve. First, the metrics are collected at a hospital level, which obscure individual variability. Second, so many factors affect these metrics that they don’t give providers direct insight into what they can do to change the way they practice.
Two strategies to combat this are to collect metrics at the provider level and develop metrics in known problem areas per specialty. For example, in Mohs surgery (used to treat some types of skin cancer), surgeons are paid per layer of tissue they remove. This creates a strong financial incentive to remove more skin than necessary. To fix this, surgeons who were consistently removing more layers of tissue than what was considered appropriate were notified. 83% of the notified outliers changed their behavior for the better, which led to $18M in direct savings to Medicare in the first 18 months alone.
One of the main goals of value-based care is for all of the patient’s providers to work together to provide better care for the patient. This means, for example, a patient’s primary care physician would work closely with a patient’s specialist on care plans.
Unfortunately, larger hospitals are better poised to take on this care coordination. This leads to more competitive pressure on smaller clinics to be acquired by larger healthcare systems.
ACOs help push back against this trend a bit by pooling together different provider groups and helping them coordinate care and resources together. Providing more support to small provider clinics trying to do value-based care should help push back a bit against all of the consolidation.
It’s essential that we deal with all the aforementioned problems in order to see the real value of value-based care. Without that, providers can make much more money by gaming the system instead of by taking better care of patients. Once that’s dealt with, value-based care will really change the way healthcare is delivered, for the better.
Since we are in the early days of value-based care, there is a lot of high impact, low-hanging fruit to help decrease costs. One example is identifying and helping patients that are most likely to get sick since these patients are the costliest to take care of. We are currently seeing a lot of innovation tackling this problem through population health management and care coordination tools.
Once this low-hanging fruit is solved, I imagine that value-based care, by putting providers on the hook for ensuring that their patients stay healthy while keeping costs low, will drive more innovation towards precision medicine. Care pathways and treatments will be personalized for each patient. Value-based pricing for drugs will put pharmaceutical companies on the hook for improving health outcomes as well.
I also imagine that there will be a lot of critical examination of long-held beliefs. For example, residents at Johns Hopkins used to regularly prescribe patients 30-60 opioid pills, following what the last resident had taught them. For most patients, 0 opioid pills were actually necessary. The more data that we collect and the more incentive we have to drive better and lower cost care, the more of these long-held assumptions we can test.
As you can see, value-based care has a lot of potential. But first, we have to overcome the challenges of designing a value-based care system that has the right structure, incentives, and metrics. With the combination of the government, private payers, providers, and startups innovating on these challenges, I think we are going to see a lot of progress in the next 5 years!
Was this article helpful to you? Do you have any feedback or particular topics that you want me to cover? If so, I would love to hear it! Please comment below or email me at firstname.lastname@example.org!