econ- ct7

S Y M P O S I U M : A B J S C A R L T . B R I G H T O N W O R K S H O P O N H E A L T H P O L I C Y I S S U E S

I N O R T H O P A E D I C S U R G E R Y

Economic Incentives to Promote Innovation in Healthcare Delivery

Harold S. Luft PhD

Published online: 19 June 2009

� The Author(s) 2009. This article is published with open access at Springerlink.com

Abstract Economics influences how medical care is

delivered, organized, and progresses. Fee-for-service pay-

ment encourages delivery of services. Fee-for-individual-

service, however, offers no incentives for clinicians to

efficiently organize the care their patients need. Global

capitation provides such incentives; it works well in highly

integrated practices but not for independent practitioners.

The failures of utilization management in the 1990s dem-

onstrated the need for a third alternative to better align

incentives, such as bundling payment for an episode of

care. Building on Medicare’s approach to hospital pay-

ment, one can define expanded diagnosis-related groups

that include all hospital, physician, and other costs during

the stay and appropriate preadmission and postdischarge

periods. Physicians and hospitals voluntarily forming a

new entity (a care delivery team) would receive such

bundled payments along with complete flexibility in allo-

cating the funds. Modifications to gainsharing and

antikickback rules, as well as reforms to malpractice lia-

bility laws, will facilitate the functioning of the care

delivery teams. The implicit financial incentives encourage

efficient care for the patient; the episode focus will facili-

tate measuring patient outcomes. Payment can be based on

the resources used by those care delivery teams achieving

superior outcomes, thereby fostering innovation improving

outcomes and reducing waste.

Introduction

Economics, in addition to medical research, influences how

medical care is delivered, organized, and progresses. Fee-

for-service (FFS) payment encourages the delivery of ser-

vices; given the current payment structure, time spent

doing procedures is rewarded more highly than time with

patients or coordinating care. Some argue these relative

payments should be adjusted; shifting that balance, how-

ever, will differentially affect various specialties. The

concern of this discussion, however, is different: regardless

of the relative values for evaluation and management

versus procedural services, the prevailing system of fee-

for-individual-service offers no economic incentives for

clinicians to efficiently organize the care delivered by other

providers.

This paper discusses the role of payment incentives and

how realignment of payment can improve both efficiency

and quality at any point in time and, furthermore, reduce

the rate of growth in expenditures. The focus here is not on

payment levels—providers typically complain fees are too

low and payors complain they are too high—but rather on

the way that payment is structured. One alternative to fee-

for-service, global capitation, provides cost-containing

incentives because it gives decision makers a fixed budget

for all the care their enrollees need. It works well in

highly integrated practices but not for independent practi-

tioners without an external entity to coordinate their care.

The failures of utilization management in the managed

care efforts of the 1990s demonstrated the need for an

alternative approach. Episode-based payment is such an

The author certifies that he has no commercial associations

(eg, consultancies, stock ownership, equity interest, patent/licensing

arrangements, etc) that might pose a conflict of interest in connection

with the submitted article.

H. S. Luft (&) Palo Alto Medical Foundation Research Institute, Health Policy

Research, 795 El Camino Real, Ames Building, Palo Alto,

CA 94301, USA

e-mail: [email protected]; [email protected]

123

Clin Orthop Relat Res (2009) 467:2497–2505

DOI 10.1007/s11999-009-0930-7

alternative to FFS on the one hand and global capitation on

the other.

Bundling together all the payments associated with an

episode of care can better align incentives. Building on

Medicare’s model for hospital payment, one can define

new expanded diagnosis-related groups (EDRGs) includ-

ing not just hospital costs, but also physician and other

costs during the stay and during appropriate preadmission

and postdischarge periods. Physicians and hospitals vol-

untarily forming an entity I will call a care delivery team

(CDT) would receive a bundled payment, along with

complete flexibility in allocating the funds. The new

financial incentives will encourage efficient care for the

patient; the episode focus will facilitate measuring patient

outcomes. Payment levels, moreover, should be based on

the resources used by CDTs achieving superior outcomes,

thereby fostering innovation improving outcomes and

reducing waste in such care. Medicare has recently

begun a demonstration of selected aspects of this concept

for certain cardiovascular and joint replacement proce-

dures [2].

Restructuring payment for the ongoing management of

chronic illnesses is more complex than the much shorter

and more concentrated care associated with an inpatient

episode. In brief, for the former I propose a comprehensive

approach with monthly chronic illness management pay-

ments to accounts associated with primary care physicians.

These payments will be risk-adjusted to reflect the chronic

conditions of the physicians’ patients, and reflect the

resources used by those clinicians achieving better than

average outcomes for their patients. A major risk pool will

pay the CDTs directly for hospital and similar episodes, so

the chronic illness management payments need not include

those costs. A payment intermediary working on behalf of

the primary care physicians will monitor the data and, for

those whose patients have lower use of inpatient care but

comparable or better outcomes, argue that they should

receive higher monthly payments for their superior man-

agement of care. The monthly payments serve as the basis

for paying not just the primary care physician, but also

specialists and other ambulatory care services. The com-

prehensive approach includes methods to ensure enroll-

ment, the sharing of data, and creation of information

useful to both patients and clinicians [5].

The first section of this paper outlines several ways

payment can be structured, from the perspectives of how

dollars are disbursed by payors and how they are received

by providers. The second expands on one of these options,

payment from a payor bundled around an inpatient episode,

and discusses how funds may be disbursed among the

providers involved in the episode. The third section illus-

trates this concept with data concerning knee arthroplasty

procedures. A final section discusses the likely implications

of such a payment scheme on efficiency, quality, and long-

term cost growth.

Payment and Incentives

Healthcare providers (meaning both physicians and hos-

pitals) often use the term ‘‘reimbursement’’ for their

services, but this is a misnomer. Reimbursement actually

refers to situations in which one is paid back for money

spent on an official or approved purpose. For example, I

trusted the sponsors of this workshop would reimburse my

travel expenses upon presentation of the appropriate

receipts. They would, however, impose certain constraints,

such as reimbursing only up to full coach airfare if I chose

to fly first class. On the other hand, they would not pay me

the price of a full coach ticket if I purchased a nonre-

fundable, discounted ticket using a roundabout route. That

would be akin to offering me a fixed allowance to cover all

travel expenses, thereby shifting the risk and potential

benefits to me. A fixed travel allowance would give me

incentives to search for cheaper airfares (perhaps freeing

up funds for an honorarium), but it would also force me to

bear the risk of occasionally absorbing the expenses of

nonrefundable tickets for trips that had to be cancelled.

Providers prefer to talk about ‘‘reimbursement’’ for at

least two reasons, one historical and the other political. The

historical reason is that hospitals used to have contracts

with Blue Cross plans (and for a number of years, Medi-

care) in which they kept track of the proportion of their

patients attributed to these payors and at the end of the year

were literally reimbursed for that share of their allowable

costs. The political reason is that ‘‘reimbursement’’ sug-

gests merely being paid back for expenses incurred—the

‘‘language’’ implies no one is making a profit.

In fact, there is very little true ‘‘reimbursement’’ these

days. Hospitals set charges and negotiate rates with private

insurers. Payments are often above actual costs incurred; if

not, the facility would eventually go bankrupt. Medicare

sets the payments it offers hospitals and typically does not

care whether these payments cover the costs of its

patients—it is like a fixed travel allowance based on the

destination of the traveler. Medicare payments to physi-

cians are also set unilaterally, but those payments are based

on each unit of service, as if I were allowed a fixed amount

for each airline trip, each taxi ride, and each meal. In both

instances, there are vigorous arguments over the adequacy

of the payments, but little discussion as to the effects of

how the payments are structured.

FFS payment to providers is the most common approach

to structuring payment. FFS generates some revenue for

nearly every service provided; depending on the costs of

delivering those services, providers may make or lose

2498 Luft Clinical Orthopaedics and Related Research 1

123

money. There are some variable costs involved (such as

supplies), but medical practice has substantial fixed costs,

reflecting rent and depreciation, as well as malpractice

insurance, office staff, and operations. The individual

physician is usually the residual claimant; his or her

income is based on what is left after all the costs are

subtracted from the revenue received. High fixed costs

mean additional services rendered are likely to be profit-

able even if on average the revenue received is less than the

average cost. In economic terms, this means the marginal

(or incremental) revenue exceeds marginal (or incremental)

costs associated with one additional service rendered.

Incentive effects, or the additional effort one will expend in

a given situation, usually reflect such incremental revenues

and costs. Thus, FFS creates incentives to provide more

services. This is not necessarily problematic; the vast

majority of businesses have incentives to offer more of

whatever they sell.

Applying these incentives in the context of medical care,

however, is problematic because of what is being ‘‘sold’’ or

‘‘purchased.’’ Physicians sell specific highly skilled ser-

vices, hospitals sell operating room time and bed days,

pharmacies sell drugs, and device manufacturers sell

devices. Most patients, however, want a treatment for their

health problem or ongoing care to keep them healthy. In

this context, the specific services and items provided by

practitioners are simply inputs. While patients often have

(or should have) substantial say in decisions about their

care, they rarely have the technical expertise to assess the

value of specific inputs in achieving a desired outcome.

If a patient needs a surgical procedure, such as for a

knee arthroplasty to improve mobility, the patient typically

selects the surgeon who decides on the best technical

approach. The hospital provides the inpatient services and

arranges for the anesthesiologist and the implant. If the

system works reasonably well, nothing goes wrong and

the procedure is a success. There is nothing, however, in

the simple incentives of FFS payment that drives the

overall process to become more efficient or produce better

quality. For example, the instruments and approach a par-

ticular surgeon prefers may be optimal for him or her, but if

each surgeon uses a different approach, the operating room

team does not have the opportunity to standardize proce-

dures to reduce waste and errors.

A useful, although not perfect, analogy is what happens

when undertaking a kitchen remodel. The initial decision

as to whether to go forward with the project is based on the

homeowner’s assessment of the benefits and costs

involved. In a kitchen remodel situation, some people

decide to serve as their own general contractors, hiring

specialists to do specific jobs and coordinating the sched-

uling. Unless very experienced however, the do-it-yourself

general contractor often runs into problems: the floor

refinisher wants to be last, but the painter wants to do the

baseboards last. General contractors earn their fees both for

ensuring seamless coordination and for their expertise in

choosing subcontractors who will perform as expected. The

subcontractors know failure to perform will affect repeat

business, not from the current client, but from the general

contractor who needs reliable subcontractors and can offer

many jobs in the future.

The problem with the general contractor-subcontractor

model is since subcontractors are often working for several

general contactors, scheduling cannot be fully integrated

and each subcontractor’s slightly different work style may

not mesh well with others. A few companies, however, are

fully integrated; their own crews do all the electrical,

plumbing, countertop, and other work. Being able to con-

trol everything, they prepare all the materials well in

advance in the warehouse and then gut the kitchen and

install everything in a week of on-site work. They reduce

labor costs for initial steps by having workers move from

one ‘‘job’’ to another in the warehouse in the same day. The

firm, however, has to bear the risk of having staff on the

payroll even when demand slacks off. From the perspective

of anyone who has lived through a remodel, the promise of

a job completed in a week and on budget is very attractive

and this may offset a somewhat higher bid price.

We rarely see either of these contracting arrangements

in medical care. Although a hospital may hire nursing and

other staff and facilitate access to certain hospital-based

physicians, such as radiologists, anesthesiologists, and

pathologists, each physician typically acts as an indepen-

dent provider. One physician may ‘‘manage’’ the clinical

aspects of the case, but he or she cannot pay the others to

change their schedules, agree on a practice approach, or

otherwise coordinate their activities. Various laws and

regulations preclude such financial interconnections. Even

in the unusual situations in which physicians are employed

by the same entity, such as a medical school, a history of

specialty independence and prerogatives usually trumps

economic integration.

Bundling Payment for Inpatient Care Episodes

What might a new model of integration for inpatient care

look like? The new system would pay for an episode of

care, such as a knee arthroplasty or care for a heart attack.

These are events a patient can understand, rather than the

long list of services, disposables, and devices used in

delivering the care. Although I refer to inpatient care, I

include similar episodes addressed in an outpatient setting;

a procedure involving a 23-hour stay is not much different

to the patient from one that extends to 25 hours. The epi-

sode should include closely related preadmission services,

Volume 467, Number 10, October 2009 Economic Incentives Promoting Innovation 2499

123

such as an MRI performed 1 day before a procedure or the

emergency room visit ending in the heart attack admission.

Likewise, reasonable postdischarge care should be

included, including costs due to complications.

One’s initial reaction may be that such a system could

not work because of various legal, political, and economic

risk-related reasons. These are discussed in turn. The legal

and regulatory barriers, such as those precluding fee-

splitting and gainsharing are changeable by new legisla-

tion; I will return later to legitimate concerns about the

troublesome conflicts of interest that could arise from

simply removing existing prohibitions.

The political barriers are less the ‘‘big P’’ political

issues any proposal might face in Congress, but those

arising between hospitals and their medical staffs. There

is often little trust between hospitals and their affiliated

physicians. Although the hospital accounts for the largest

share of expenses within most episodes, there is no reason

for it to be the recipient of the payment. Instead, the

bundled payment could be offered only to a new entity

able and willing to provide (or assure the provision of) all

the services needed for the episode. Legislation altering

existing gainsharing regulations could require that any

such new entity establish its own forms of governance

acceptable to its members. A hospital could threaten to

replace individual physicians, but it will not be able to

establish such a new entity on its own. Because most

patients seek out physicians and surgeons, not hospitals,

this will encourage win-win arrangements. The new

governance rules could be written by the participants with

an appropriate balance of power between the hospital on

the one hand and the participating physicians on the other.

Not all physicians on a medical staff will want (or need)

to participate in the entity; just enough are required for

the new entity to reasonably accept episode-based pay-

ments. Avoiding an ‘‘all-or-none’’ situation reduces the

power of a single physician to block the deal. CDTs

might initially be encouraged under an expansion of

Medicare’s demonstration authority to offer bundled

payments to the voluntarily formed new entities. The

amount might be no less than that paid under existing Part

A and B coverage and be guaranteed not to decline even

if there are fee schedule reductions in the future. Teach-

ing hospitals creating CDTs will avoid the cumbersome

reporting and documentation requirements of the house

staff and attending roles in each patient’s care. The new

CDT acts as the general contactor accepting a fixed price

for the remodel and arranging for and paying the various

subcontractors needed to complete the job. A hospital

might participate in several separate CDTs, such as for

cardiovascular or orthopaedic care. Alternatively, there

may just be one CDT with varying arrangements with

participating physicians.

The economic risk concerns arise from the nature of a

bundled payment to the CDT. Even within a specific

diagnosis-related group (DRG), costs vary across patients

in the same hospital. Including responsibility for compli-

cations within a reasonable period of time substantially

increases that variability; ignoring complications, however,

eliminates economic incentives to avoid them. The CDT’s

financial risk can be addressed through reinsurance. The

post-discharge responsibility period may be DRG-specific

and should maximize the likelihood that readmissions are

associated with the CDT’s care, rather than the patient’s

underlying medical problems or poor care in the ambula-

tory setting. For example, a longer window may be

appropriate for admissions of basically healthy people

undergoing a joint replacement or delivering a baby, while

only a short period would be used for a person with a

naturally recurring problem, such as CHF or emphysema.

In some instances a readmission is clearly a complication

of care, such as a wound infection with a hospital-specific

strain of MRSA. In other instances, the readmission is

almost certainly unrelated, such as for a post-discharge

gunshot wound. Most readmissions have less clear causes.

Across many patients with a given initial reason for

admission, however, one can readily estimate the percent-

age readmitted within a specific period of time for any

reason. That percentage times the average cost of those

readmissions is the amount that would be added to the

payment to the CDT to cover, on average, the cost of

readmissions.

Medicare already absorbs some similar episode-based

risk with outlier payments for individual hospital admis-

sions that are substantially more costly than the specific

DRG payment. An alternative approach can offer more

direct incentives for process improvements. The payment

to the CDT would include the average cost of a reinsurance

policy protecting the CDT from the risk associated with

unusually expensive cases. For example, if outlier cases

and readmissions currently account for 10% of all costs

associated with a specific DRG, the payment could be

increased by a bit over 10% (to allow for the legitimate

costs of the reinsurer) and CDTs could purchase coverage

on their own. Reinsurers will have an interest in keeping

down such costs and will seek to learn the techniques used

by CDTs able to reduce occurrences of complications, or

manage them very well once they occur. Reinsurers can go

beyond merely selecting CDTs with low complication

rates; some will learn best practice techniques from their

clients and pass them on to those CDTs wanting to improve

their processes. In contrast, the current Medicare system

provides neither incentives nor mechanisms to learn and

disseminate better techniques.

The notion of bringing together hospitals and physicians

in an economically aligned unit is not new; it is part of

2500 Luft Clinical Orthopaedics and Related Research 1

123

what underlies the concept of ‘‘focused factories’’ for

orthopaedic and cardiovascular services [3]. Such specialty

hospitals are very attractive in some circles but strongly

opposed in others. The interest in them indicates the

technical and internal political issues to make them work

can be overcome. The opposition arises from two general

concerns. The first is, by specializing in certain highly

profitable categories of care and by drawing those patients

out of other hospitals, they impair the ability of the latter to

cross-subsidize patients who are less profitable. The second

is they may encourage admissions and procedures that are

not necessary.

The first problem is merely a reflection of the current

Medicare DRG rates; they are too high for some proce-

dures and too low for many other categories. This is a

problem of how those prices are set rather than the

underlying concept of bundling; the new severity-adjusted

rates implemented by Medicare are designed to address this

problem. In a reformed system, prices for each inpatient

episode (including all professional and facility costs, plus

the appropriate preadmission and postdischarge care)

should yield roughly the same profit (or net revenue)

margin for each EDRG category. With the need for cross-

subsidies eliminated, the first argument against specialized

facilities will disappear. Some will object that simply

having appropriate pricing of EDRGs, however, does not

eliminate the need for cross-subsidies to cover the unin-

sured. That is a legitimate point if the discussion is

focusing only on a specific payor, such as Medicare, but

can be addressed by universal coverage for such care, as

outlined in my proposal for comprehensive reform [5].

Appropriately setting prices for bundled payment is not

trivial, but doing it well can yield several benefits. Medi-

care currently sets relative DRG weights based largely on

the costs nationally of caring for patients in each DRG

category with adjustments to reflect local input prices. The

final figures are then subject to across-the-board adjust-

ments to meet certain budgetary goals and hence reflect

high-level political lobbying. A different approach would

use not the national average as the standard, but the costs

incurred by those CDTs achieving better than average

outcomes for their patients, again adjusting for local wage

rates. (The focus here is on relatively proximate outcomes,

eg, unexpected complications or achieving anticipated

functional levels within 6 months. Long-term outcomes,

such as graft patency or device longevity, probably reflect

more the choice of intervention than the skill of the CDT.)

In contrast to most current attempts to reward quality that

focus on either hard endpoints, such as mortality, or

adherence to professionally determined processes, the

choice of outcome measures should be more patient-ori-

ented. That is, panels should be established with substantial

patient input to identify the outcomes patients desire most.

Clinicians will then be able to experiment with new pro-

cesses to improve outcomes and reduce costs. Total

payments, moreover, should not be constrained to meet an

overall budget. We should instead rely on the incentives

inherent in bundled payments to reduce the inputs used in

an episode; other market-based incentives to address the

prices for physician time are discussed elsewhere [5].

A commitment to set payments based on what it takes to

achieve superior outcomes will increase demand by phy-

sicians for collection of better data on outcomes. It will

also lead to improving risk adjustment techniques to

account for underlying patient comorbidities and under-

standing what techniques lead to the best outcomes. This

will foster innovation in workflow and processes that

increase quality. The new system should also explicitly

cover the additional costs of new treatments as they are

being assessed, much as Medicare offers such coverage

with evidence determination. Better outcomes at specific

CDTs will result in not just professional and public rec-

ognition (thereby attracting referrals) but also in an

increased payment level for all, as the new interventions

are included in the costs reported by the CDTs achieving

superior outcomes.

The problem of demand creation—the classic example

being that of Redding Hospital (Redding, CA) and its car-

diovascular team who performed procedures on patients

who did not need them—must also be addressed [4]. Phy-

sicians choosing to be part of a CDT cannot be in the

position to self-refer by deciding that a certain patient needs

a procedure or admission. Clinicians financially involved

with the CDT should not be assessing the need for an

admission, but in many instances primary care physicians

will not have the technical expertise to make such decisions.

Independent specialists doing those assessments can be

compensated based on their time and expenses, without

rewards for decisions to either intervene or withhold care.

Once it has been determined a procedure or admission is

warranted, patients will have free choice of CDT teams and

physicians within CDTs. The CDT members can certainly

decide not to undertake a procedure on a given patient.

Physicians not part of a CDT, however, can be involved in

the outpatient management of care and in the decisions of

whether an admission or a procedure is needed. Primary

care physicians and independent medical evaluators outside

the CDT may have important insights on the care of patients

and the CDT would be able to pay them a fee comparable to

their time in the office for consultations.

An Illustration

To illustrate the potential of bundled payments, we use data

for patients undergoing knee arthroplasties during 2003 to

Volume 467, Number 10, October 2009 Economic Incentives Promoting Innovation 2501

123

2004 drawn from enrollees in a set of large commercial

health plans across the nation. These unpublished data

came from the proprietary Ingenix Impact National

Managed Care Database which is a large administrative

claims database containing inpatient confinement, medical

claims, pharmacy claims, and member eligibility infor-

mation. For this illustration, we used a 10% random sample

(over 800,000 members from numerous national health

plans during 2003–04). Of these, 284,446 members had

24 months of continuous commercial (non-Medicare)

coverage and 4,910 had knee surgery. (The database is

fully deidentified in compliance with HIPAA Privacy

Rules.) The database includes some people aged 65 years

and over covered primarily through employer-based plans

rather than Medicare. The dollar amounts reflect ‘‘allowed

charges’’ as paid by the plans and patients. We used the

Ingenix Symmetry Episode Grouper 6.5 to create episodes

of care, but focus only on the services during and proxi-

mate to a hospital admission. Patients were selected if they

had a knee arthroplasty (CPT-4 code 27447) with a phy-

sician identified as an orthopaedic surgeon. We excluded a

small number (95) of persons with more than one knee

arthroplasty in a year because we could not determine

whether the code reflected complications of the index knee

operation rather than surgery on the other knee. We only

included patients grouped into ETG 0721 (joint degenera-

tion, localized, with surgery-knee and lower leg) for the

knee arthroplasty group. We examined costs by various

types of service and those delivered either during the

inpatient stay or in the period before or after admission.

The mean episode cost across all 4910 patients was

$22,454, with an average of $18,596 incurred during the

inpatient stay and the remainder during preadmission and

postdischarge periods, including rehabilitation stays. Not

all patients incur charges in all categories, and some

charges are seen for only a small fraction of all patients

(Table 1). The total cost is rather stable, with a coefficient

of variation of 37%, indicating the standard deviation is a

little more than 1 . 3 of the mean. Nearly all episodes include

charges for hospital, anesthesiologist, and ancillary costs,

along with surgeon fees. Hospital costs averaged $13,189,

nearly 60% of the total, whereas surgeon fees during the

episode averaged $2948, or about 13% of the total. Total

ancillary fees exceed surgeon fees but with a coefficient of

variation nearly three times as large.

Without detailed clinical information, it is impossible to

know whether the variability in resource use across patient

episodes is due simply to underlying patient factors. The

widely documented variability in resource use across

geographic areas and individual medical centers without

any clear evidence in similar differences in patient risk,

however, suggests patient factors may not fully account for

what we see. (See, for example, discussions by Jon Lurie

[6] in this issue.) There is little evidence that areas or

centers with consistently higher resource use achieve better

patient outcomes. In the absence of compelling new

information, we should modify the usual baseline in which

it is assumed (1) most variation is due to patient risk fac-

tors, and (2) increased resource use controlling for patient

risk factors is indicative of higher quality. Instead, our

Table 1. Costs for various services incurred during knee arthroplasty episodes of care commercially insured patients, 2003 to 2004

Costs during inpatient stays (acute only) Cost preadmission and postdischarge

(including rehabilitation stays)

Costs for the whole episode

Services/Fees Number

of

patients

Mean

cost for

users

Coefficient

of variation

Mean for

all

patients

Number

of

patients

Mean

cost for

users

Coefficient

of variation

Mean for

all

patients

Mean

cost

Percent

of total

Coefficient

of variation

Surgeon fees 4910 $2685 56.3% $2685 4273 $302 146.1% $263 $2948 13.1% 53.4%

Inpatient facility

costs

4910 $13,189 38.6% $13,189 0 $0 $0 $13,189 58.7% 38.6%

Anesthesiologist

fees

4587 $699 65.3% $653 480 $232 115.9% $23 $676 3% 78.4%

Other

professional

fees

2213 $395 133.9% $178 3713 $636 154.1% $481 $659 2.9% 169%

Ancillary costs 4496 $2065 212.7% $1891 4867 $1440 113.4% $1428 $3319 14.8% 141.2%

Rehabilitation

facility costs

0 $0 $0 848 $7852 51.9% $1356 $1356 6.0% 300.4%

Outpatient

pharmacy

costs

0 $0 $0 3509 $431 151.8% $308 $308 1.4% 212.4%

Total costs 4910 $18,596 4899 $3866 111.3% $3858 $22,454 100% 37%

2502 Luft Clinical Orthopaedics and Related Research 1

123

baseline should be (1) the extent of patient-specific need

must be measured with better clinical data, and (2) out-

comes should be continually assessed to determine what

mixes of inputs and services are best for specific patient

problems.

The potential for creative resource allocation is appar-

ent; a relatively small reduction in hospital costs could

allow substantially increased compensation for other

members of a CDT (Table 1). Such a saving in hospital

costs, however, is likely to require adaptations by nursing

staff, anesthesiologists, and others involved in the episode.

These solutions cannot be easily imposed by external

bodies but are likely to be achieved voluntarily by the new

CDTs with the flexibility to allocate funds. Ancillary costs

exceed surgeon fees but are usually ordered by the sur-

geons; they could also be a source of savings. CDTs

choosing to share some of their savings with price-sensitive

patients may increase their volume, and hence total reve-

nue, at the expense of other, less efficient CDTs.

Discussion

There is much concern about how much we spend on

medical care and how uneven is the quality of care we get,

but little attention has been paid to the implications of how

we pay medical care providers. The current system of fees

for individual services is at the root of both our cost and

quality problems. The incentives implicit in a bundled

payment covering all the relevant providers involved in an

inpatient episode can encourage far more efficiency. If

designed appropriately, such a payment approach can also

facilitate learning what works best in practice and then

changing processes to improve quality. In the long run, this

new focus will impact the translation of new research into

practice, emphasizing increased efficiency and improved

value.

The conceptual arguments in favor of episode-based

payments are clear; the details and transition are more

challenging. Episode-based payments and bundling are far

more complex in the outpatient setting for numerous rea-

sons, so this discussion has focused on inpatient (and

similarly intensive and short duration) care. It is easy for

payors to ‘‘bundle payment’’ but much harder for inde-

pendent practitioners and a hospital to figure out how to

share risk, rewards, and responsibilities among themselves.

With the fear of malpractice in the background, few pro-

viders will leap at the opportunity to redesign care in ways

that may appear to be ‘‘doing less,’’ so a reform of the

liability system must be considered. For a few categories of

conditions, such as acute myocardial infarction, there is a

reasonable consensus on outcomes, such as death within

30 days, that are reasonably well measured. For most

patients, such as those having elective surgery, procedure-

associated death should be a rare or never event. From the

patient perspective, outcomes matter most; not only should

each professional do his or her job excellently, but the

overall processes of the team members involved in the care

must facilitate ‘‘hand-offs,’’ optimal choices of treatments,

and continuous improvement. Developing agreed-upon

outcome measures incorporating both professional opinion

on what can be measured and patient values on what is

important will take time and resources. The model pro-

posed here focuses on quality measures across groups of

similar patients, with ample opportunity for increased risk

adjustment as we learn about relevant risk factors. Without

having to score each case as a ‘‘success or failure’’ one

can use robust statistical approaches that account for ran-

dom variation while seeking to detect consistently better

than expected performance. None of these issues are triv-

ial, but the best way to address them is to begin with

volunteering organizations, feasible target conditions, and

demonstrations.

Episode-based payment creates incentives for physi-

cians to focus on the best ways to manage resources during

an episode of illness or care; the CDT creates the economic

unit in which such incentives can be turned into action.

(The issues are somewhat more complicated in managing

an ongoing chronic illness involving a wide range of pro-

viders who may not normally work together and often

requiring patient cooperation in adherence to various

treatment regimens. Thus, different incentives and struc-

tures are needed for a relatively brief episode of care

focused on an inpatient stay and for the ongoing manage-

ment of illness. The latter are discussed in much more

detail in my comprehensive proposal [5].)

Episode-based payment allows the appropriate com-

pensation for the time professionals need to effectively

coordinate services. Payment approaches within the new

CDT may vary widely. Surgeons may still be paid per

procedure and anesthesiologists by the time in the operat-

ing room, while the hospitalists involved in the post-

operative care may be paid by the shift. The CDT may hire

patient navigators and technical support personnel to effi-

ciently coordinate all the members of the team or optimize

the use of various implants and instrumentation. Payment

rates, moreover, might be adjusted by workload to better

use scarce resources, such as encouraging the use of

operating room time during weekends.

The bundled payment that includes not just professional

fees, moreover, creates incentives to redesign workflow

within the inpatient stay and at other times during the

episode. This can range from relatively simple things,

such as standard preoperative protocols and drug regimens,

to agreeing on a smaller number of implants or a pre-

ferred formulary. While one cannot assume all imaging

Volume 467, Number 10, October 2009 Economic Incentives Promoting Innovation 2503

123

performed by referring clinicians meets the CDT’s stan-

dards, many may be perfectly adequate. CDTs might even

train referring clinicians in the techniques needed to reduce

the preadmission workup process—a savings not only in

resources but also in time and risk to the patient.

Episode-based payment must be designed to avoid

incentives to skimp on care to increase profits to the CDT.

Quality concerns, however, are not particularly associated

with episode-based payment, as illustrated by the tainted

medications derived from Chinese manufacturers [1].

There are always incentives for producers to skimp on

quality if it is not well monitored and if the losses due to

discovery are small relative to the gains for the badly

behaving producer. If we no longer assume more is nec-

essarily better, then paying for episodes means attention

will shift to the outcomes of episodes, a measure about

which patients care, rather than to the processes and inputs,

about which they have little technical knowledge. Using

independent medical evaluators outside the CDTs to assess

whether surgery is needed offers a useful approach. The

independent medical evaluators may also recommend

which CDTs have the best outcomes based on both risk-

adjusted outcome measures and reports gathered from the

patients seen.

Turning our focus to outcomes will initially be a chal-

lenge. Few meaningful measures are routinely captured in

administrative data other than events that should rarely

occur, such as death or serious complications. In most

instances, medical and surgical interventions are not

intended to postpone death but to enhance function and the

quality of life. These often require direct assessment by the

patients themselves. Such assessments should be viewed as

evaluating not the specific professionals involved but the

ability of the whole team to achieve the outcomes desired

by the patient and should recognize patients may differ in

what they think is important. Automobiles are not evalu-

ated in the market solely based on ‘‘hard end points,’’ such

as their crashworthiness and miles per gallon; customers

have varying preferences for style, reliability, and perfor-

mance. Highly unsafe cars should be kept off the road, but

there is room for a great deal of individuality in many other

aspects. Just as Road and Track and Consumer Reports

may rate the same cars differently, patients should be able

to choose among assessment schemes that match their own

preferences for certainty, performance, and processes.

Most healthcare professionals simply want to achieve

the best outcomes for their patients, but the fear of mal-

practice liability may lead some to perform unnecessary

tests and to be uneasy about the prospect of experimenting

with new ways of organizing and delivering care. Episode-

based payments to CDTs should be combined with reform

of the malpractice liability system that benefits patients,

providers, and even attorneys.

In brief, if an injury occurs due to an error in the care of

a patient, there should be a quick resolution with an

apology by members of a CDT and compensation paid to

the patient. Negligence will not need to be proven; the

payment is set through an administrative determination of a

preventable adverse event and the extent of the injury

sustained. (Medical care often involves inherent risks; no

compensation would be provided for unfortunate outcomes

without evidence of an error.) Plaintiffs’ attorneys would

simply collect and document the data supporting the claim

and probably be paid by the hour. Most CDTs would self-

insure for these common and relatively small claims. A

series of similar claims paid by the CDT without evidence

it attempted to improve its processes, however, could result

in a second-tier suit proving ‘‘corporate negligence’’ with a

much larger potential award. That award would be shared

with those patients (and their attorneys) involved in the

earlier claims who implicitly granted access to their data.

In this reformed system, patients would receive just com-

pensation for errors and CDTs would have incentives to

learn from previous mistakes, two legitimate arguments of

those defending the current system. Physicians would not

be charged with negligence for what was merely an error,

even if an avoidable one, and far less would be spent on

litigation, legitimate arguments of those opposing the

current system.

Episode-based payment creates immediate economic

incentives for teams to use resources more efficiently.

More important for the long term, it changes the incentives

for biomedical innovators. The current incentive is to

develop new medications, devices, and procedures with

some promise of improved quality of care, irrespective of

cost. For some innovations, their attractiveness is even

enhanced by additional professional fees associated with

their use; if physicians can earn an additional fee for the

procedure accompanying the biologic or device, they

effectively become part of the marketing team of the

manufacturer. In contrast, episode-based payment will lead

clinicians to consider all costs in relation to the enhance-

ment in quality. The most attractive innovations will be

those improving quality and reducing cost, even if this

means reducing what are now billable services. Other

innovations may have no measurable effect on quality but

will reduce costs; for instance, new technologies may allow

lower-skilled workers to substitute for some of the tasks

highly paid clinicians now do. With the cost of the device

bundled into the payment, surgeons will demand evidence

that ‘‘innovative’’ devices truly are better. Device manu-

facturers will find, unless they have breakthrough

technology with demonstrably better results, they will be

financially better off lowering the costs of tried-and-true

devices and gaining market share. The new incentives will,

therefore, alter the mix of technological innovations

2504 Luft Clinical Orthopaedics and Related Research 1

123

entering the healthcare system, slowing the long-term rate

of growth in expenditures.

Reform of the healthcare system needs to go beyond

strategies that simply provide more people with insurance

coverage that fosters inefficiency, poor quality through

disincentives for coordination, and mindless cost-contain-

ment efforts through fee reductions. Instead, reforms

should focus on improving the payment system to appro-

priately align incentives and facilitate clinicians in being

the best professionals they can be.

The charge for the Association of Bone and Joint Sur-

geons/C.T. Brighton Workshop is to discuss where we are

now, where we need to go, and how to get there. Nearly all

observers of the US healthcare system agree it fails to

cover a substantial fraction of the population and yet is

more expensive than all other systems in the developed

world. Most agree, while the quality of care in some

instances is outstanding, it is not as good as it can be. The

rate of growth in expenditures is unsustainable in the long

run and something needs to be done to slow this. Explicit

rationing of services is politically unacceptable. Reductions

in provider payments yield short-term cost containment,

but many believe the current system merely adapts by

increasing the use of services and procedures, further

fueling expenditure growth while devaluing physician

effort and driving many clinicians out of practice.

Fragmented payment to individual providers for their

own services is one of the causes of this sorry situation. A

solution for inpatient and similar care that can work in the

US environment is to bundle payments around an episode of

care. This new payment will be made to a CDT composed of

physicians, hospitals, and other professionals who will

provide the necessary services. They will decide how to pay

themselves and organize care to both increase quality and

efficiency. The CDTs will demand better information on

patient outcomes and learn from other CDTs how to con-

stantly improve. Biomedical innovation will shift from

being cost increasing to being value enhancing.

The changes needed to move to bundled payments are

within reach. Some existing federal and state legislation

will need to be modified to allow the creation of CDTs.

Bundled payment could be done without reform of liability

laws, but enacting the type of malpractice reform suggested

above can help reassure physicians that well-intended

innovation will not unduly expose them to liability claims.

The underlying data to begin the implementation process

are available, but incentives are needed to pool the data

from Medicare and other payors. Medicare has begun some

relevant demonstration projects and these should be

expanded. Creating the new teams will require substantial

effort, so there should be clear guidance from national

leaders that the payment system will be changing and

incentives provided for voluntary early adopters. For

example, new CDTs might be guaranteed payments no less

than their present Medicare Part A and B payments with

exemptions from any future fee reductions, but the ability

for their payments to be adjusted upwards along with

everyone else. Academic medical centers might avoid their

troublesome service-by-service documentation require-

ments if they transformed their practice plans into CDTs.

A shift to episode-based payment will require a different

way of thinking about how care is provided and profes-

sionals are paid. It will require the development of

functioning teams that replace totally independent eco-

nomic entities. It offers, however, an economic basis for

creative approaches to medical care delivery that are based

on what works, yet adaptive to patient needs and scientific

advances. Not all the details of such a fundamental change

can be worked out in advance, but we cannot wait for the

perfect solution to our rising healthcare costs.

Acknowledgments I thank Laura J. Eaton, MD, MPH, for her help with the data analysis included in this paper.

Open Access This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which per-

mits any noncommercial use, distribution, and reproduction in any

medium, provided the original author(s) and source are credited.

References

1. Barboza D. China orders new oversight of heparin, with tainted

batches tied to U.S. deaths. New York Times. March 22, 2008. Available at: http://www.nytimes.com/2008/03/22/world/asia/22

heparin.html. Accessed November 29, 2008.

2. Centers for Medicare and Medicaid Services. CMS announces sites

for demonstration to encourage greater collaboration and improve

quality using bundled hospital payment, January 7, 2009. Avail-

able at: http://www.cms.hhs.gov/apps/media/press/release.asp?

Counter=3394&intNumPerPage=10&checkDate=&checkKey=&

srchType=1&numDays=3500&srchOpt=0&srchData=&keyword

Type=All&chkNewsType=1%2C+2%2C+3%2C+4%2C+5&int

Page=&showAll=&pYear=&year=&desc=false&cboOrder=date.

Accessed May 6, 2009.

3. Herzlinger R. Market Driven Health Care: Who Wins, Who Loses in the Transformation of America’s Largest Service Industry. New York, NY: Basic Books; 1999.

4. Lenzer J. Healthcare group agrees $500 m settlement for unnec-

essary surgery. BMJ. 2006;333:59. 5. Luft HS. Total Cure: The Antidote to the Health Care Crisis.

Cambridge, MA: Harvard University Press; 2008.

6. Lurie JD, Bell JE, Weinstein J. What rate of utilization is

appropriate in musculoskeletal care? Clin Orthop Relat Res. 2009 May 19. [ePub ahead of print].

Volume 467, Number 10, October 2009 Economic Incentives Promoting Innovation 2505

123