Background
The U.S. health system, like other aspects of the economy, was pretty much free from excessive control by government policy and suppliers of service and organizations (hospitals, clinics, office-based professionals, drug companies, etc.) were able to do their own thing. The “American way” philosophy of entitlement prevailed: “if you could pay for it, you were entitled to have whatever you wanted. But, if you couldnt afford it, then you needed to work harder.”
Like cars, fresh vegetables, and everything else, health care services were bought by households directly out of their pocket. But after WWII employer based insurance grew very rapidly, as the IRS permitted a huge tax breaks for the insurance premiums paid by employers on behalf of their employees. And the private health insurance industry as we know it, arose to become the primary source of health care financing in America, replacing out of pocket payments by households (and vastly reducing medical bankruptcy in America). In the 1960s the New Deal’s Social Security Act of 1935 was amended to provide medical insurance for some of the persons not able to participate in employer based plans: public insurance for elders (Medicare, title XVIII) and for the poor (Medicaid title XIX).
Up to 1983 health care providers like hospitals, nursing homes and doctors were growing in numbers and flourishing with the rapid growth of insurance for their services. Patients were able to use professional and institutional services for a fraction of the full cost of those services as a consequence of insurance. And this fueled access to care and the volume of services and overall spending on health services. During this period, health care providers were being paid by the public and private insurors on the basis of bills generated after the services were rendered in one of two ways: (1) provider-set-their-own-prices (billed charges) or (2) on the basis of incurred costs (used by Medicare and Medicaid to pay hospitals and other institutional providers). These two payment approaches did nothing to encourage efficiency or slow spending growth. Indeed, in the decade after Medicaid/Medicare was passed in 1966, rates of increase in health care spending grew rapidly (from 1965 to 1975 hospital payments rose by an average annual rate of 14.1%). Much of the growth in spending occurred, in part, because of all the newly insured retirees and poor moms and kids, who were now able to better access care with their new insurance. But also evident in the spending growth that the professional health care provider community had no “check” on giving themselves a raise (higher billed charges), nor any check on providing unnecessary services, or hiring unnecessary staff or paying them too much, or buying marginally necessary equipment (since in hospitals and nursing homes they were usually paid incurred costs, so if they spent more, they would just get more revenue) — a nearly unconstrained health system. I have been told by hospital executives that when they were being paid on the basis of incurred costs, they “never had to tell a doctor “no” to a request for new equipment, or pretty much anything else”.
So, as this was happening, officials in States, Congress and the Medicare/Medicaid agency (then HCFA), became alarmed about the unexpectedly high rates of hospital spending growth and the associated outflows from state and federal budgets and the Medicare trust fund. Nobody had planned for such spending growth. Health care programs were crowding out other government services: roads, schools, defense department, etc. States were under extreme budget duress caused by Medicaid spending (states finance about half of the program’s cost, the feds pay about half too) and had already begun to act by creating approaches to better control payment to hospitals through payment rate regulation (New York & Maryland for example were among the first). Others states developed “hospital payment regulations” in the 1970s too, often in concert with pilots encouraged by HCFA (by granting state waivers of the payment policies of the Medicare and Medicaid programs). To understand whether these state programs were being effective to check the rapid increase in hospital spending a Congressionally mandated study was ordered for 15 underway state pilot programs. I began to work on that study in 1978 soon after it began, as a novice health economist.
These early experimental payment programs in the states were all very different. The approaches used to set regulated payment rates, or budgets included
(1) government regulation of any increases in hospital prices (full billed charges),
2) industry self regulation of full billed charges (eg voluntary programs),
(3) administratively set rates per diem (per day),
(4) administratively set rates for the entire stay per type of patient (in NJ–the DRG system),
(5) government sets annual budgets for each hospitals (Maryland – Global budget system which is used in many developed countries),
(6) and other variations on these themes.
But, all share a common feature: setting prospective payment rates (or budgets) or freezing existing rates rather than waiting to determine the payment amount due after the patient was discharged.
This “prospectivity” creates an incentive for “cost control” stemming from worry about fiscal sustainability of the organization if they don’t take steps to be as efficient and to “manage” the course of treatment in order to prevent “surprises” when seeing what services were provided at discharge and what the costs were. While the particular incentives are somewhat unique to the way payment is calculated (eg the “unit of payment” drives the incentives), generally the ‘prospective’ determination of the payment amount puts all hospitals at fiscal risk for overspending and suffering a operating loss (paying more than planned for the drugs being used, hiring too many staff, or paying too much, for using unexpected ICU services, or keeping the patients too long, or having a community flu epidemic). Some of this “overspending” can occur if a surprise cost event occurs in the hospital (nursing union gets a pay raise, or Rx prices go up), and some of this risk stems from the fact that the hospital may have been long been very inefficient relative to peers and the prospective rate is possibly below costs in the beginning. For whatever reason, if costs exceed payment at the end of the day, the organization is “at risk” for becoming unsustainable. But, if they are paid more than they spend, then they can pocket the surplus as higher operating profit. And, as managers take steps to cut unnecessary costs, and more strongly control practice patterns of previously autonomous physicians, there are always corollary risks to “underserve patients” as a way of preventing any excessive spending. This, again, is a threat to sustainability.
As suggested by managers that lived through the introduction of these programs: “The problem of managing in this environment is complex. Where do I push? How hard to push on overall efficiency. How hard to push on clinicians to cut LOS and ICU use? How much do i worry about pushing this year’s problem? How much is the issue a longer term problem to be solved over time? It is hard.”
Under the traditional payment approaches (being paid based on “costs” or by setting their own charges) the insurers had to worry about the fiscal impacts of these sorts of cost surprises. So, the change in payment approach represents a shift in cost risk from payors to the providers. Said another way, the introduction of prospective payment is a way of enlisting providers in the war on cost control in American health care, as contrasted with the earlier payment approaches–where the provider was often the source of the problem.
The importance of this experimentation period before Congress finally acted to alter Medicare in 1983 by establishing the DRG system (the NJ state pilot program) was three fold:
- The incentives worked to slow cost growth. And the pattern of consistent with the expectations based on the types of incentives created by the the unit of payment. The earliest studies of impacts of such prospective payment approaches were generally that they led to savings in operating costs over time (relative to expected levels of costs), access was not adversely affected, profit levels of prospective rate control were lower, and the quality related effects were mixed, somewhat inconsistent and judged to be insignificant to policy (National Hospital Rate-Setting Study, Final Report, HCFA 500-78-0036, Feb 1, 1988).
- Even before Congress acted, the Hospital Industry and other pundits about health policy matters could see that the days of “cost reimbursement” and “set-your-own-price” payment services were over for hospitals. States were acting, and the Feds were poised to do something to save the Medicare trust fund (which they did in 1983). This “end of cost reimbursement and self pricing” was in the air, and hospital CEOs could see that the good old days of doing whatever they wanted was over. Individual hospitals needed (and would be forced by future government payment policy) to become more accountable for efficiency by keeping their payment levels above their costs, or they would not be able to survive. The industry changed, became more businesslike, worried more about competitiveness, about introducing new services, about getting bigger, about about making profit targets, about employing doctors on staff (eg control). The good old days in hospitals ended around 1980.
3. Payment policy for providers began to be, and has continued to be a popular and nearly universal tool of insurers (private and public insurers here and all over the world). Payment incentives work predictably to change provider behavior in safely limiting unnecessary health spending. Here, the 1983 legislation for hospital payment of inpatients was followed in Medicare and Medicaid by prospective payment approaches for paying physicians (RBRVS), home health care (PPS), nursing homes and rehab, hospital outpatient services, and others. Mass BCBS introduced value purchasing to create incentives for both cost control and meeting population quality targets, and Medicare has adopted similar programs of quality targets. Both Medicaid and Medicare and the ACA have opened more aggressive ways to shift even more risks to providers by moving to using capitation arrangements to pay health plans or ACOs so as to shift all risks to provider organizations.
Payment methods for providers of all types have become the key policy instrument for adjusting the performance of the health system– to improve efficiency (any prospective fixed rate payment system does this, particularly if the rate is bundled), to improve optimal integration across services (again, bundling is the key here–the bigger the bundle the more providers have to worry about efficiency integration), and even to improve quality (through P4P programs like the AQC by Mass BCBS)
More on Risk and Risk Shifting in Provider Payment Policy
Investors, business owners, insurance companies, even household decision makers all seek to balance the expected future return they’ll get from some action, with the expected risk they are exposing themselves to. What is this “risk” stuff, anyway? It is certainly not expected costs. It is the uncontrollable variation in the future return that they are expecting. They expect a return of, say, 10%, but they know that it maybe higher, or lower. They don’t know. This is risk. The risk factors may end up creating a higher return than we expected, or they may swing in the other way and we’ll get less. We just never know. Sometimes the +- swings can be expected to be bigger (a riskier investment) or sometimes the swings may be expected to be smaller (a safer, or more secure or conservative investment). Normally, the decision makers who are putting money into an investment (which might be a degree program at Simmons, or a particular mutual fund, or choosing a particular house to buy) want to decide based on two things: (1) the expected (average) return they should expect based on historical data, and (2) the level of risk (amount of uncontrollable variation in the return) that they should expect. Investment options that have a lot of risk (uncontrollable variation) require a high return, or they won’t be attractive. Low risk alternatives, on the other hand, can be successful even with a low return (because it is a secure one, without much risk). A guaranteed return has no risk, and such things offer very low returns.
Risk is typically measured by the average or standard deviation in the returns in the past. If swings in returns are often and large, then the standard deviation around the average is going to be large. If the returns are pretty stable, then the average or standard deviation is low.
Health care application of this risk-return theory.
One of the primary motivations for changing the way providers (hospitals and doctors and nursing homes) are paid by insurers (payors) is to get providers to absorb more of the risks of uncontrollable events that might increase the costs 0f care.
What is risk? Risk is uncontrollable variation, in this case uncontrollable variation in the costs of health care. When we think about a population of people, we have two metrics for describing what volume of health care services that population is going to use in the upcoming year: (1) the average or expected amount we think they’ll use (based on historical data), and (2) the variation we can reasonably expect (the standards deviation), also based on whatever historical data we have. A mean, and a standard deviation are used to describe the expected payout per capita for the coming year. We don’t know of course what the exact amount will be. But, from years past we can calculate the average or expected payout, maybe add some inflation to it, and we know how much variation there has been over the years— the average deviation (standard deviation) is the average amount of variation we have seen—sometimes +, sometimes – relative to the average payout. Know body knows exactly what to expect. In some populations the mean and standard deviation of expenditures will be bigger, or sometimes smaller. Nobody knows.
What are the things that contribute to the uncertainty about what health care needs are going to be for that population? Well, some of the things that the insurance company (Medicare, BCBS) doesn’t know are:
- Whether the flu season will be easy or hard this year
- Whether the price of the drugs they will need for patients will favor the really expensive branded ones, or whether it will turn out that their will be a run on less expensive ones
- Whether the mix of patients being admitted (or treated) will take lots of therapy, or very little
- Whether the costs of certain things we have to buy (xray film, IV solutions, replacement monitors, new IT systems for the Pharmacy, etc) will be a lot more expensive in the year ahead, or whether some of these things will actually cost us less
- Whether important staff will stay, or leave—which may have a noticeable impact on the efficiency of our operations and the need to hire (or not hire) supplemental staff to make it all work
- Whether the mass nurses association will be successful in negotiating high salaries for nurses this year, or not
We just don’t know what the year will bring. Costs may be higher than our expectation, or they may be lower. And, we can’t control all of the sources of variation in spending. So if we are a payor (BCBS, Medicare) and we have a contract requiring us buy health services for our population from providers, then we worry. We got paid a “premium” at the beginning of the year to provide health services for 200,000 people. While the annual premium we agreed to was a reflection of the population characteristics, the usage patterns we have seen historically, and the likely prices we will pay to our providers, we still have no idea whether the “uncontrollable risk factors” will break in our favor, or against us. We can add some margin to our “premium” to help compensate for the fact that we just don’t know about what we’ll have to pay (this is called the risk premium, and it will be somewhat higher for small groups than larger employer clients—do you know why?). But, in the end, the payors are still subject to variations in what they’ll have to pay to hold up their end of the bargain that they’ve made with those people who paid premiums.
So, what they can do to “manage” their risks is to pay providers in different ways— ways that help shift some of the uncontrollable risks to the provider (who has a means to control it better than we do). If they begin pay hospitals, for example, based on a fixed price per admission (like DRGs) then all risk factors that might affect the costs of providing care for and admission next year will be the responsibility of the provider, not the payor. Changes in the price of resources (xray film, drug prices, nurses salaries) and things like how long patients stay become the worry of the provider not the payor. The provider may experience unanticipated increases in these costs. Or the provider may benefit by lower costs in some of these items. It is not possible to say whether the provider will win or lose by being forced to absorb more risk.
Risk shifting is much different than shifting more costs to providers. Shifting cost could be done by paying lower payment rates to providers. THIS IS NOT THE SAME AS SHIFTING RISK. When risk is shifted it may end up being good for provider, or bad, we just don’t know. That is what risk is about. We just don’t know.
Different provider payment methods shift more or less risk to providers. Under Fee For Service (FFS), providers bear risks associated with the input prices and efficiency with which they produce procedures or visits. Bundling the payment approach into visits, days, episodes, or even capitation puts the provider at more and more risk (the results of more and more uncontrollable risk facts are falling on the provider). This doesn’t mean that providers do less well when they accept more risk. Not at all. It means that they are subject to more of the uncontrollable forces that bear on how well they will do. The maybe better off than they’d thought they would be at the beginning of the year (the flu season was wimpy) or they may be worse off.
What’s the point of shifting economic risk to providers by the insurance companies. From society’s standpoint, there are a couple of reasons why provider-bearing-risk is good. (1) Providers are in a better position to assess the “need” for services by the patient, and can best be able to balance the potential benefits of a service, with the consequences of not providing that service. This kind of decision making, patient by patient, day after day, drives the over health spending situation. By making them more concerned about “economy” in a general sense by giving them more risk, we believe that providers will do a more careful job of assessing who needs what. (2) Providers may be in a better position to “control” some of the other controllable drivers of health spending: patient compliance, early screening and secondary prevention, and even primary prevention. They know the patients, they can even do assessments of the health risks that some patients face and focus their efforts in a way that can avoid certain problems. Insurers just can’t do these things.
Why use provider payment to try to get spending under control? In these big insurance funds with excessive growth in spending what are the options?
— cut the number of eligible persons?
–cut the coverage
— just cut the rates paid to providers (by contracting only with the least cost providers)
–trimming wasted excessive administrate costs out of the program
Unfortunately, the last option may be a good one, except the bulk of administrative costs are in provider organizations and policy has no direct control over them. Only the provider can do it. The other three options all cut access to the program benefits by restricting access to services. Incentivizing providers to be more efficient is a much easier solution politically.
Downside to Incentive Payment Rates or Budgets. When care is bundled and the provider paid according to the “average” amount required to treat patients in that category there are two problems.
(1) an incentive for under-service is always created. The provider benefits by providing less service, or erring on the side of not doing a service. This contains cost, but the saving goes into the provider’s pocket. Cost containment yes, but society doesn’t ever see the benefit, and may experience the harm from worse outcomes. Paying for performance (P4P) is a logical extension here, where providers are paid with bundles, augmented by “bonuses” for achieving quality of care objectives (to counter the tendencies for under-service).
(2) an inequity is created for small providers. The nature of bundled payment is that if the “average level of reimbursement” for a category of patients is paid, then with large numbers of patients, this will be adequate. Sure, some patients will cost more (and providers will lose money on them), some will cost less. But, over a large group these will even out, and the provider will do OK. But, if the provider is small (a rural hospital for example) then the ‘law of large numbers’ will not protect the provider from the inevitable variation in ‘need’ that will occur in patients representing the ‘bundle’ being paid for. If they see only 3 patients, for example, in a DRG, then they may b=get very unlucky (or very lucky). This is unfair to small providers. Either they shouldn’t be paid this way, or they should receive an added compensation for the added risks they bear. Note that this risk of being small is different than, but additive to, the risk uncontrollable cost variations.
It is no accident that when DRGs were implemented in hospitals in 1983 about 1000 hospitals failed—they were mainly the very small community hospitals in urban and suburban areas.
More About the Unit of Payment and What Providers are at Risk for
The amount of risk borne by providers is a function of the size of the bundle being utilized in designing the unit of payment. To illustrate the relationship between unit o0f payment and how much risk is being transferred to the provider we look at hospitals. We consider the total budget for inpatient hospital care for a population. The amount of care and cost will depend on things like how big the population is, how much disease and trauma there is, how many services are used when people go to the hospital, how long they stay, and how much hospitals have to pay for the resources they need. So, we look at the drivers of inpatient hospital cost per capita in a population (or if you wish, on the drivers of changes in that cost per year).
Cost per capita per year can directly be decomposed into four multiplicative factors—each one corresponding to a type of risk.
per capita cost for = cost per/service x services/per day x days per /admit x admits/per person hospital care
Risk factor efficiency risk scope of care intensity risk epidemiology risk risk
The first factor is total cost / number of specific services rendered per patient. There are hundred of types of services, and a total cost for each type of service. This ‘cost per service” is about efficiency. And, the second factor is about breadth of services delivered per day of patient care. We call this the Intensity risk. The third factor is about how long patients stay. We cal this the LOS risk. The last factor is about how many patients there are per the number of people in the population. We call that the epidemiological risk.
So if we wanted to use this model to decide what kind of risk (and incentives) is borne by the hospital and payer for a particular unit of payment, we could use the following chart. XXXXXX is hospital borne risk.
| unit of payment method |
efficiency risk |
intensity risk |
LOS risk |
Epi Risk |
| billed fee for service |
XXXXXXXXXXX |
payer |
payer |
payer |
| per diem |
XXXXXXXXXXX |
XXXXXXXXXXX |
payer |
payer |
| per admission |
XXXXXXXXXXX |
XXXXXXXXXXX |
XXXXXXXXXXX |
payer |
| capitation |
XXXXXXXXXXX |
XXXXXXXXXXX |
XXXXXXXXXXX |
XXXXXXXXXXX |
| cost reimburse |
payer |
payer |
payer |
payer |
| Global Budget for a hospital |
XXXXXXXXXXX |
XXXXXXXXXXX |
payer (if ex post severity adjustment) |
payer (if ex post severity and volume adjustments) |
So, different kinds of payment systems are not just different in the kinds of costs that have incentives to avoid— and they also have different ways of incentivizing providers. Capitation (paying providers one annual fee to provide all types care) clearly shifts all these cost risks to the provider –they are at risk for efficiency (or becoming more inefficient) of producing all services, they are at risk for how many services their patient use, they are at risk for how long patients stay, and they are also at risk for changes in how many admissions happen in the population they are being paid to care for. Other units of payment shown in the chart (FFS, per diem payment rates, per admission rates (like DRGs), etc.) have different cost risks being shifted from insurors to the provider. And, each unit of payment is also different in terms of what providers must do to get more revenue (income)
a. FFS — do more procedures and visits
b. bundled procedure (including pre op and post op service) — do more procedures
b. fixed rate per bundled patient day —– produce more patient days
c. fixed rate per bundled episode — produce more episodes
d. capitation rate per person per year– enroll more people if you want more revenue
All fixed rate systems (regardless of the unit of payment) have strong incentives to economize on expenses, including incentives for underserving patients. But, all payment systems do not have the same sized unit of payment (procedure, day, episode, per patient year). The bigger the bundle of services included in the “unit of payment” the broader are the opportunities for providers to make changes in the care pattern in order to produce a more economical result.
bundling
Medicare Prospective Rate Policies
What started in the lat 1970s as a move to think creatively about paying providers (for Hospitals) — Congress eventually adopted the model used in a New Jersey state pilot program that used a fixed rate per hospital episode for each of 383 DRG categories, has led to fixed rate systems for home health, hospice, physicians, outpatient care, nursing homes, and others. It is a veritable landslide of payment reform, aiming to put providers in positions where they have to manage their expenses to be as low as possible, and to use the most efficient staff to deliver services.
The following chart shows the primary payment methods and Unit of payment used to pay for Medicare health services.
m pay sys
The choices Medicare made in setting up these systems for payment needs to be reviewed. These prospective systems are of a somewhat common type. They all pay a single national rate set by the government for each type of patient. The bundles of services are different in hospitals (a hospital stay) and home care (a 60 day period of care) and outpatient (a day of service). Each type of service being paid has a different rate for each type of patient in that provider group.This is a way of recognizing that different types of patients have different service needs, which cost more or less to deal with. So, each type of provider has a unique scheme for defining patient type: there were originally 383 groups of inpatients (DRGs) in hospitals, hundreds of types of outpatient care categories of patients (APCs), there are now 153 types of home care patients (HHRGs), there are many 1000s of types of slightly bundled doctor service groups of activities ( the RBRVS categories used for physician payment), etc.
Second, all the Medicare payment approaches have a rate-per-category of patient computed and paid by the government. Providers, of course, have software allowing them to to know, on admission, how much they will be paid. At the year onset, Medicare calculates and publishes a base rate for each type of care (an average hospital inpatient episode, or a base rate for an HHA episode, etc). That rate is converted into a “payment amount for a particular patient” by multiplying that base rate by a relative resource use weight for that “patient’s category of service” (say patient is in DRG 381). The relative weights are calculated by the Government for each of the 383 categories of hospital inpatients. This is done by dividing the average costs for a patient in each DRG (say, the average DRG 381 patient = $14,000) by the average cost of all types of inpatients (say,12,000) to get a relative weight for DRG 381 = 1.17. The payment rate for this DRG 381 patient would be computed by the government as the base rate * 1.17. Sometimes there is a regional, or type of facility adjustment also factored into the formula to pay for a particular patient.
This general approach to “pricing” the amount to be paid by the government is roughly the same for hospitals, HHAs, outpatient care (base rate, patient groups each with a relative weight, sometimes facility level adjustments). This is described below. A set of some of the DRG weights are also illustrated below.
drg rate
relative weights
Medicare and medicaid payment systems for all providers are under review and change nearly every year. One reason there are changes is because changes occur in how patients of some type are treated—often these changes occur because of new treatments and technologies (laser procedures moving many patients to outpatient settings). These kinds of refinements cause new patient categories to be created or eliminated, and relative weight to be recalculated (this is called recalibration). In 1983 there were 383 patient categories of DRGs; now there are well over 500 DRGs. The changes are generally small improvements, but provider payment reform is now rather continuous, and not as subject to the scrutiny of the media (and even by the Congress) to the extent of sexy policy matters of universal coverage, new drug regulations, etc. MedPAC is the congressional agency that reviews Medicare performance and makes new policy recommendations every year to the Congress (not to the Executive branch). They spend a lot of time proposing payment reforms. And, Medicare is definitely the “trend setter” in payment reform in the U.S. Other payors tend to look carefully at (and sometimes follow) new things that are adopted by Medicare.
A final point about Medicare prospective payment systems. Capitation is the nuclear weapon of provider payment reform. It make the provider organization getting the payment responsible for coordinating and integrating all services in a way that allows expenses no larger than the revenue of the plan. If the provider cant manage the integration, and the efficiency of all covered services, they are in trouble and will fail. The plan will also be strongly incentivized to keep their enrollees well, and out of the hospital. That is, capitation is the only provider payment method that provides incentives for keeping members as well as possible, and keeping utilization of services as low as possible. All other payment methods encourage more provider utilization in one fashion or another — whatever the “unit of payment” is, there will be incentives to do more of those things in order to increase revenue.
Practical issues in Designing Prospective Payment Systems
I used to sometimes design payment systems for states (Montana, Alaska, Massachusetts, others), trying to utilize lessons from the big national applications for Medicare to smaller situations and markets. There are other design choices that are often feasible in smaller markets.
Generosity. The first issue is understanding the importance of the “generosity of the rates”? If the rates set by government are too high, does it mean that the providers don’t pay any attention to the marginal incentives? And, the corollary, that the only providers who worry about the attempting to comply with incentives are the inefficient ones? No, this isn’t exactly what happens. The incentives, for example, to pay as little as possible for the resources the provider buys, do make the provider better off, no matter if their base costs are high or low. The marginal incentives are important for all facilities. But, it is also true that for the group of high cost providers the marginal incentives may be critical to survival. So, the impacts of incentives may be somewhat higher for some providers (the high cost ones) than others.
Provider specific Rates. Some of the pilot programs by states set rates differently: sometimes each provider organization got its own rate based on it’s cost or payment history. Sometimes the rate paid was a blend of provider baseline data and the history of the larger peer group. This choice by Medicare, to set the national rates and not the so called facility specific rates, was a deliberate attempt to make implementation of prospective payment in 1983 and thereafter administratively feasible. It was simply too much work to make blended rates , or negotiated rates (like the all payer budget control system in Maryland). Those of us who worked on the evaluation and Congressional report in the 1980s generally thought the all payer budget setting program in Maryland outperformed the other approaches being used to set prospective rates. But, when Medicare (and Congress) chose which of the state pilots to choose, they chose to model the national Medicare hospital program after New Jersey, not Maryland. The Maryland model required negotiating with each hospital to set a budget.And, it required too much political capital to get all payors (Medicaid, BCBS, private) to participate. So, they chose to go with a Medicare only model, one that did not require negotiation, and was therefore much easier to administer. Calculating national rates, and weights and a rule that explained them was far far easier and feasible than negotiating or computing rates for 4000 hospitals. Medicare simply didn’t have the resources to start a Maryland-type program.
Dynamic Compliance. One important issue in states concerns how (or if) to adjust the next year’s rate for this year’s performance for a facility. Do we just apply an overall COLA or % increase to adjust last year’s rate for the changes in the price level of inputs the provider must buy regardless to the amount of surplus or deficit the provider achieved last year? Again, for national programs, this is the obvious way to it because the provider is forced to absorb full risk for making or losing profit at the national payment rate schedule. Full risk has stronger incentives than some sort of “shared risk” between the payor and the provider. This would be an approach that sets the next year’s hospital specific rate for the amount of profit or loss that was made by the hospital. In watching the results of these Medicare programs over time it has become obvious that some providers earn surpluses year after year, and other lose money year after year. These chronic winners and losers (as it is referred to) occur for some facilities. And, MedPAC has always made comparisons of performance of the chronic winners compared to other providers (treating them as some form of “best practice” facilities. But, the Medicare programs rely on National rates, and expose providers to absorbing the full impact of surpluses and losses. They do not have the administrative wherewithal to set and monitor facility specific rates.
Fairness Issues. State programs, and Medicare as well, have other problems setting up prospective pay systems. First, there are problems with using national DRG-type programs because of inter-institutional unfairness. The idea of paying all providers an “average amount” for all providers of DRG 381 services is unfair because there will naturally be a variation in severity/costs for these patients. The small rural facilities may see 2-3 of these cases per year, the large community facilities might see 50-60, and the large medical centers might see 80-100. One one hand, small (low volume) institutions unfairly run higher risks under such programs than do larger volume institutions. If the small facilities get a high cost patient , it may do much more significant harm, than the same hi cost patient will do to a larger volume organization. On the other hand, the medical (referral) centers can also suffer from the fact that difficult patients (across and within DRGs) will be much more likely to be found in their mix of admissions. So, the average rate may be systematically too low for these referral institutions because of their case complexity as a referral organization.These problems (or paying everyone the same rate, does not assure that every institution is being faced with absorbing the same level of risk. The use of a “teaching adjustment” for teaching hospitals is the main way that some equity is provided to referral centers.
A different remedy sometimes is available from a situation in rural areas, where access to care depends on institutional survival of small providers. If a provider is to important (for access) that policy people will not allow the provider to fail, then what does it mean to have a payment system designed to impose failure risk unless they act to contain costs? If a hospital is failing for whatever reason, and the politicians stand ready to write them a supplemental check to keep them in business, then why are we opting for this kind of cost control program for this kind of situation? Montana and Alaska both had many of these small, rural and essential providers. In both cases, I set up Medicaid DRG programs statewide. In both cases the Medicaid DRG systems were only implemented in the larger organizations. In both cases the rural (and small) facilities were paid their incurred costs rather than DRG payment rates. A DRG program was simply unfair to small providers, and unfair to the residents of rural areas, who were at more risk than their urban peers.
Outliers. Almost all prospective rate systems employ a risk mediation technique call “outlier payments “. In order to mitigate the effects of a catastrophic level of care for an occasional patient, the technique of “outlier payments” was introduces in some state programs and by the subsequent Medicare pricing programs. An outlier “threshold” was established arbitrarily based on incurred costs, or days of care, and for patients that exceeded the threshold, supplemental payments to the provider were made, usually based on marginal costs, not full costs. These outlier policies are always unique to the program. But the idea is for the thresholds to be set so that very few patients will exceed the threshold (maybe 1-2%) and estimates of marginal costs from the research literature are applied to the addition total costs incurred by the patient (maybe 50-60% of total outlier costs). The supplementary payments offer some catastrophic risk protection for the provider.
In the HHA program of prospective payment for a 60 day episode of home care the outlier policy in more complex. The program has both HIGH outliers (patients who use an extremely high volume of service) and LOW outliers (patients that use a very low volume of services (usually because the die of are discharged after a few days). This approach protects both the provider and the payor from unfairness stemming from the extreme patients on both ends of the distribution of episodic costs.
The Best Way to Design Payment Policy? From a policy making perspective prospectively paying providers (or setting budgets) has become an important way to “control” provider behavior, and keep a “check” on provider spending. (and to encourage desired levels of quality metrics–which is discussed below). But, there are choices to be made about how to pay providers (among them are choice of unit of payment, choice of setting rates based on peer groups, or based on provider-specific data, and whether to share risks with provider, whether to include small providers, and how). And, there is no obvious, or best way to pay providers (from the vantage point of the payor, or from the vantage point of policy). We also know that prospective or incentive payment methods create powerful incentives on provider decision- making and utilization (as well as drive financial outcomes). And they make Managers in provider organizations and health facilities work hard to balance financial tensions with clinical decision- making. Never forget that the methods that are used to pay Providers will drive their behavior. Every single payment approach tells providers what they must do to get more money—and they will do these things to the extent that they have autonomy to do so. Therefore, payment incentives are a powerful tool of health sector reform.
What do we know about the effects of payment reforms on Patient Care
There is a large literature demonstrating that providers (all types) choose what they do for patients based on how they are paid. There really is not any literature that providers pay no attention in their practice to how they are paid. We watched the onset of insurance insurance and fee for service (and cost based) payment spawn huge increases in utilization in America, and can also document how the risk of litigation has created more increases in unnecessary utilization. We see below, that when facing risks of economic boom or bust providers (individual and institutional) have systematically and predictably responded by altering practice patters to avoid bad results. Essentially, providers responding to the incentives offered by how they are paid traces the history of America’s health system, and its public policy. The response of providers to changes in payment has been predictable, and often fast:
• Rand HIE — Patients in HMO’s (fixed payment per enrollee per year) used over 1/3 less inpatient hospital care than other insured patients
• Following the implementation of DRGs in 1983 the average LOS dropped > 10% in one year in US hospitals
• HHAs dropped usage of Aids for Medicare patients by 1/3 when PPS was implemented
We provide evaluation evidence of the impacts of payment policy changes of various kinds below:
We review briefly some of these evaluation activities of prospective payment programs to give a flavor of the way payment policies create impacts on patient care.
DRG system Patient Care Impacts
The prospective payment approach that started it all, in 1982 has been discussed at length. Hospitals responded to the fixed rate per stay for 383 patient groups (defined by primary diagnosis, procedure, age, gender, and whether there were complicating diagnoses). Hospitals we at risk for this fully bundled episode rate for everything that happened during the stay (pay raises for nurses, ICU use, redoing tests, how long the patient is in the hospital, price of xray film. And, patient care was indeed changed, The American Hospital Association fought the fight to try to stop Congress by claiming that:
- patient care was efficient and effective and things like length of stay could not be safely shortened, and would not be reduced (in spite of economic incentives to do so)
- costs would not be reduced because they were both “necessary and reasonable”.
Well, there were immediate and often large changes in patient care in response to the incentive of the new payment system. In the first year, length of stay nationwide fell by 10% or so. There was no mortality of readmission disaster, as projected by the AHA. There were problems with the post acute capabilities of VNAs all over the country. In the beginning they didn’t have many nurses who had experience with the levels of severity they began to see coming from earlier discharged patients (patients with tubes and pumps and more post op pain than they were used to seeing). The VNAs had to start hiring more nurses with recent acute hospital experience, and take on the added workload referred to them from hospitals. The HHA industry grew much larger over the next several years.
And. rates of growth in hospital costs were sharply reduced. It turns out the AHA was wrong again—hospital costs before DRGs were not reasonable, nor necessary. And the cost impacts of the DRG program were almost all the direct result of changes that were made in patient care. Here are the major consequences of the DRG prospective payment program in the first few years:
DRG consequences
HHA PPS Impacts
A second program of Medicare payment rates was initiated 17 years later (2000) in home care (HHAs). The rapid growth of the industry following the DRG system, which encouraged shortened LOS, led to this policy change. The rate was for a 60 day episode of home care (two months, not 60 days of provided services of some sort). It covered HHA covered services including RN visits, aids visits, and PT, RT, and OT in the home. There were initially 60 types of patients and a fixed rate for each (growing to 153 types today) supplemented by both HIGH and LOW outlier policies.
hha pps desc
The impacts of the HHA service volumes (indicated by the changes in staffing mix shown here) were large. Two main effects are shown here in the differences before the program and after (the highlighted data are after the program was in affect after 2000). (1) therapy visits were increased. This was a consequence of the fact that the way the original 60 patient groups and rates were designed, a patient receiving therapy was in a group that received a much higher rate— so agencies, mysteriously, upcoded — provided the PT, OT, RT services to get the higher rate! This is bad system design. (2) The RNs/Aid mix of staff has changed. A richer mix of RNs is an important consequence of the incentivized system. What this suggests is that when agencies were tightening their belt, the discovered that the wage gap between RNs and Aids was smaller than the differential in productivity (they discovered that given the wage gap, hiring more RNs and replacing Aids improved agency profitability) . Yes, Aids mad less money per hour, but productivity of RNs was so much higher that it mire than made up for the higher wage. THIS SAME CONSEQUENCE WAS OBSERVED WHEN HOSPITALS WERE SUBJECTED TO DRGS, AND A RICHER MIX OF RNs / LPNS WAS STUDIED.
hhh conseq
More broadly, the HHS pricing program allowed many agencies to become more efficient had make surpluses of revenue over costs. The impacts of quality haven’t been studied comprehensively.
hha 2
Capitation Impacts
Capitation is the biggest bundled payment approach their is; paying an annual (or monthly) fee for all covered services per month (PPPM) or per year (PPPY). Providers are essentially being paid the entire premium, and are at every kind of financial risk; a disease epidemic, introduction of new technologies of care, inflation in the economy, prices they pay for resources, availability of sufficient qualified staff, inefficiency, intensity and other risks. Every recent proposal for health system reform (Nixon, Clinton, Obama-ACOs) has sought to fix the “bang for the buck” (or another term: flat of the curve) problems with the U.S. Health System … has chosen to use capitation.
We show some hypothetical data on a population’s use of health services , and costs, and the calculation of a capitation rate per capita.
cap rate
Many health systems around the world use something called partial capitation as a way of paying for primary care, and to make these primary providers at some financial risk for their decision to hospitalize patients. The idea of partial cap rates in the next slide is to pay the primary physician for 100% of the costs of primary care +10% of the hospital episode premiums (eg DRG) –for all her patients. And then when a patient of hers is admitted, she pays 10% of the hospital payment to the hospital. This provides an incentive for the physician to profit by making the decision to hospitalize only when absolutely necessary.
partial cap
The power of the incentives of full capitation are quite evident in the randomized trial called the Health Insurance Experiment (HIE) done in the 1970s by Rand Corp. One arm of the trial enrolled a randomized group of patients in the Puget Sound HMO, which was a capitated health plan. The health and health care used by these persons were compared to another cohort of the study, enrolled in an insurance program with free care where providers were paid fee for service. The HMO enrollees had 39% fewer admissions than the FFS group, and less total health spending (by 25%), most of which was the result of fewer admissions to hospitals. Satisfaction with care was somewhat lower for the HMO group. And, other aspects of quality including outcomes were not materially different for the two samples.
Global Budget Impacts
Setting an all payor budget cap is a very direct way of shifting risks to hospital managers about efficiency, intensity, and all other risks of hospital spending. Many systems make ex post volume of care adjustments (allowing recoup in the next year for the marginal costs of exceeding volume targets), which tend to mitigate risks of an epidemiological nature. But hard budget caps are an unambiguous and effective way of limiting hospital spending, forcing managers to make hard choices about how to allocate the budget. And, sometimes this causes delays in getting elective care, like Canada (and like the VA for that matter). The data suggest (but don’t prove) that Global Budgets are possibly the most effective way to limit health care system spending.
global budget success
Canada is a summer place to go for me for many years. It has taught be lessons about Global budgeting. My neighbor for many years was the Lt. Governor of the island province of Prince Edward Island (population about 120,000). A long time politician, appointed to this job by the Queen. He had a son who drove a motorcycle. Some years ago there was an accident, he had a head injury, but there was no CT scanner at either of the 2 smallish hospitals on the island so they couldn’t diagnose. They took him to Halifax, Nova Scotia by helicopter. He survived with some residual issues. Given his fathers position of influence, I just figured it would be no more than a few months before the province had a CT scanner of its own. Years later (to this day) the lady’s auxilliary of the hospital and other groups are still selling pies, scones and other Irish breads to raise money to pay for a CT scanner— because the budget limits haven’t allowed it— there are still not enough people on the island to justify budget approval by the federal & provincial government budget setters. Global budgets work this way.
There have been a few evaluation projects dealing with Global budgets— but not as much as you would expect. France switched to to global budgets, and there was a study. Netherlands had a study. And, here in the states Maryland has had (since the 1970s) a all payer global budget system. They all show more effective hospital spending control than other (preceding systems of hospital payment.
Value Purchasing or P4P (Pay for Performance)
This approach to provider payment methodology is directly aimed at providing incentives for meeting quality targets (bonuses and/or penalties). Medicare now uses in for hospitals, MassBCBS invented an approach for physician groups that combines capitation-type cost targets, which if met, allow bonuses for meeting quality targets (the AQC). P4P or Value Based payment schemes of various types are cropping up all over the world (i evaluated on program for hospitals in China a few years ago). But in general,
- P4P=Payment algorithms designed to encourage achievement of performance goals
- Use P4P together with other cost incentivized payment methods like DRGs, FFS, Per Diem Rates, capitation
- Mass BCBS Alternative Quality Contract uses it, along with capitation-like Budget Targets
- P4P metrics of performance are for things like:
Access improvements
Readmission rates
Quality of Service indicators
Patient Satisfaction scores (HCAPS)
Wellness/preventative service rates
These schemes are very easy to set up. You create the metrics, a verifiable way to capture provider performance data, and create a pool to finance the high achievers. Since these programs must sit along side some other payment scheme (like DRGs, FFS, Capitation, etc.) the financing for the bonus pool is easy to finance out of the funds used to pay provides (DRGs, FFS, Capitation). This is done by paying providers at a fraction of the payment rates (say 95% of the DRG rate) and use the 5% to finance the P4P pool.
p4p carve out
At the end of the year, the providers that meeting the target indicators would be paid their bonuses. Those providers not meeting quality targets would receive no supplements to the 95% they got for serving patients. The providers earning bonuses would earn bonuses to bring their payments to, say, 105% of the basic payment rates. Obviously the relationships between the bonus pool tax rate, the quality threshold to be met for a bonus, and the size of the bonus need to worked out to make the system work. But, these systems can be easily set up without raising public (payor) budgets in this fashion. And, they are being established all over the world, giving credence to the idea that “we only pay if providers demonstrate meeting some threshold of access, or quality, or patient satisfaction or whatever target we want. If no provider makes the targets, then they all get paid 95% of the payment rate. Seems like a prudent way to set up a performance based payment system. They appear to be working. Why didn’t we think of these before?
The BCBS of Mass approach of Alternative Quality Contracting (AQC) is a form of payment for physician groups ( serving as HMO Blue providers). They are paid a global budget) along with financial incentive for making quality targets— this places providers at risk for excessive spending and rewards them for quality. Several evaluations have been done (see Song’s most recent work at https://www.nejm.org/doi/full/10.1056/NEJMsa1404026.
ACQ contracted groups have saved money relative to what might have been expected otherwise (the control group).
song aqc spending
Quality of care indicators also improved (see chart).
aqc quality
The results are impressive for a young program.
Somewhat following the AQC project in Massachusetts (the quality bonusing aspect of it, not the global budget part of it) Medicare also implemented a P4P or VBP program for hospitals to provide bonuses for achieving quality markers and patient satisfaction thresholds. The program is still being phased in, where the amount of bonuses for each of the 3 components of the program are changing in importance over time, and the amount of hospital revenue at risk keeps going up. This rising percentage is indicated by the bold percentage shown for each year foir 2013 to 2018. The components of the program are shown on the pie charts for each year, and the percent of the pie being devoted to metrics for those components. A large portion of the bonus is based upon the discharged patient survey (CAHPS) done by medicare for each hospital.
medicare VBP
What Did Incentive Payment do to the health System?
As said earlier, prospective payment approaches, value based payment, and other changes in the incentives of paying providers have been effective in changing the behaviors of providers—changing the way resources are used. This is true, as far as we can tell from the research evidence, for all types of providers. And, the incentives (to cut costs, and underserve patients) get stronger the bigger is the bundle of services as defined by the “unit of payment”. There are very few good tools for policymakers to contain costs of covered benefits (less coverage, cost sharing with beneficiaries, less generous payments, lower administrative costs). Shifting more cost risks onto providers via prospective payment policies is a way of providing “cost containment” and avoiding these other alternatives that are often more painful.
What has the effect of this movement beginning in 1980 or so to the entire US health System. Hard to say. But, this chart below helps crystalize the international comparison.
what happened in 1980
This is pretty alarming data. The rest of the world is certainly getting more bang-for-the-buck in their health system than we are. The divergence in this metric from 1980 forward is evident. What happened here? Several hypotheses occur.
- prospective payment turned out to be bad for the quality of the health system, reducing quality and outcomes of care
- the other countries (for the most part) utilize global budgets (a fixed budget per year, not unlike what is done in Maryland to pay hospitals). And, they are possibly more effective than DRGs and other U.S. payment policies.
The early U.S. evaluation of the first generation prospective rate payment programs in states found that Maryland’s system (along with New York) was much more effective in checking costs (in Maryland a 25% cost reduction over 10 years, as contrasted with about half that in other mandatory models of prospective payment). So, budget mechanisms, may indeed be more constraining across the globe than has been our DRG system.
3. a third hypothesis is that something else happened in the U.S. around 1980. This may indeed be true. Case study work suggested that the switch from the pre 1983 payment environment (hospitals were basically paid their incurred costs, or and set their own prices). And then every thing changed in 1983. The party was over for hospitals—and they all could see it coming with all the state pilot programs and the unrest in Congress at the 14+% annual rate of inflation in hospitals. SO hospitals woke up to the reality that what they were doing was unaffordable, and reforms were being piloted by some payers. CEOs and Boards of Directors had to wake up and realize they could fail like any other business if they didnt start acting like a business. Indeed about 1000 small hospitals did fail in the U.S. during the 1980s. And the vast majority of the others replaced their CEO. Hospital managers didnt know how to cope with the prospective rate environment. They needs new products, new competitive strategy, better ways to get necessary changes in practice patters framed and implemented (eg less autonomy for MDs). Managers needed to start managing resources and start making their profit targets. The chart may well reflect the fact that about 1980 the U.S. hospital industry “woke up” to the business realities (after being protected from cost reimbursement previously) and began to develop new products, and new markets, and sought to expand off site locations, urgent care offerings, and general scale of operations in ways never seen before.
4. people in Europe are different. They are accepting of limits imposed by “society” in order that everybody can have access to health care. They are not handicapped by the “american way” of fairness (if you can afford it, you can have it and if you can’t afford it, you should work harder). Americans dont like limits, and they certainly don’t like the idea that wants should be supressed in favor of equality. This is a fundamental difference in the path of health systems in developed countries other than the U.S.
I personally think the answer to the question “what happened to U.S. healthcare after 1980” that drove it in a different direction than Europe and other advance countries are #2 and #3 and #4: the more effective budget control policies–a more disciplined way to control hospital spending; the willingness of people in other countries are more accepting of limits, and the “wake up call” for U.S. hospitals that they and and will fail if they fail to cope well keeping costs less than the DRG payment rates (and other payor models of control).After 1980 U.S. hospitals realized, for the first time, that they could fail to survive if they couldnt keep their head above water financially. The started to behave like businesses, not just doctor’s workshops. They got rid of Dr. CEOs, and other incompetent managers, they got concerned about fighting waste and inefficiency, about market and competitive strategy and about putting patients first, and about making investments in new services and new locations to enhance financial strength. They became just like other financially oriented businesses. It changed the culture of the organizations.
professorgaumer 