Readmission report

8/4/2019 Readmission report

1/51

Reducing Hospital Readmissionsin New York State:A Simulation Analysis

of AlternativePayment Incentives

Sb 2011

Deborah Chollet

Allison BarrettTimothy Lake

athematica olicy esearch


2/51

Contents

I. Introduction 1

II. Hospital eadmissions in New York State 4

III. Hospital Strateies to educe eadmissions 15

Factors that Contribute to Readmissions 15

Interventions to Reduce Readmissions 16

IV. ayment Incentives to educe eadmissions 19

Direct Costs 19

Indirect Costs 21

Financial Benefts 22

Pay or Perormance 22

Episode-Based Payments 23

V. ayment eorm Simulations 24

Hospital Response to P4P 25

Hospital Response to Episode-Based Payments 27

Eect on the Number o Readmissions 28

Eect on Hospital Payments 29

Direct Payment or Clinical Interventions 30

VI. Summary and Concludin emarks 33

eerences 34

echnical Appendix 36

Reducing Hospital Readmissions in New York State: A Simulation Analysis of Alternative Payment Incentives


3/51


The authors are grateul to the New York State Health Foundation and especially toSenior Vice President David R. Sandman. Dr. Sandman provided valuable suggestions

and support throughout the process, and oered thoughtul comments on early

drats o the research design, fndings, and report. His guidance and insights are

evident in the report.

In addition, we would like to acknowledge Greg Peterson at Mathematica Policy Research,

who helped to develop the literature review, and Donna Dorsey, who produced the drat

and fnal manuscripts.

Support or this work was provided by the New York State Health Foundation

(NYSHealth). The mission o NYSHealth is to expand health insurance coverage,

increase access to high-quality health care services, and improve public and community

health. The views presented here are those o the authors and not necessarily those

o the New York State Health Foundation or its directors, ofcers, and sta.

Acknowledgements


4/51


Hospital readmissions are widely recognized as an important source o avoidablehealth care costs and a potential marker or unacceptable levels o hospital

acquired inections, premature discharge, ailure to reconcile medications,

inadequate communication with patients and community providers responsible or

post-discharge care, or poor transitional care. While not all readmissions result rom problems

with patient care or management, strong evidence exists that some specifc interventions at the

time o discharge can reduce readmissions or certain conditions.

Conronting the urgent need to address health costs, some states have begun to ocus

specifcally on such interventionsincluding adherence to condition-specifc protocols shown

to reduce readmissions, restructuring hospital and post-hospital discharge planning, and use

o standardized discharge orms to improve communication across care settings. Similarly,

some integrated delivery systems and multi-stakeholder collaboratives have begun to investin programs to provide discharged patients with inormation and advice in order to prevent

problems that might lead to readmissions. However, emerging eorts to reduce readmissions

largely ocus on payment incentives.

In this study, we investigate two such incentives: pay-or-perormance (P4P) and episode-based

payments. The P4P strategy we consider is similar to the New York Medicaid programs current

P4P system and also similar to the P4P strategy Medicare will develop as required by the Patient

Protection and Aordable Care Act. The episode-based payment strategy we consider is similar

to a planned Medicare pilot program, bundling payments or hospital and post-acute physician

services to encourage more eective coordination o services and prevent avoidable readmissions.

Rssons n nw YoRk Cost nRY $4 Bon pR YR.In 2008, nearly 15% o all initial (or index) hospital stays in New York resulted in a readmission

within 30 days. These readmissions (nearly 274,000 hospital stays in 2008) cost $3.7 billion,

accounting or 16% o total hospital costs (table s.1). Readmissions or complications or

inections cost $1.3 billion, accounting or nearly 6% o total hospital costs.

Ab S.1. Hospital eadmission ates and the Cost o eadmissions, 2008

ttAl

(tHuSANdS)

AdISSIN

At

pCNtAg

f

AdISSINS

ttAl

pAYNtS

($ bIllINS)

pCNtAg

f pAYNtS f

AdISSINS

pCNtAg

f ttAl HSpItAl

pAYNtS

ll admii 2,087.1 /a /a $23.4 /a 100.0%

ll idexadmii 1,872.6 /a /a $19.9 /a 85.2%

All AdISSINS

Fr ay reaihi 30 day 273.6 14.6% 100.0% $3.7 100.0% 16.0%

Frcmlicai rifeciihi 30 day 72.7 3.9% 26.6% $1.3 34.5% 5.5%

SC: ahemaica plicy Reearch aalyi f ne Yr spRCs hial dicharge daa.

Executive Summary


5/51


Executive Summary(continued)

Patients aged 65 or older accounted or more than hal o all readmissions and readmission

costs in New York State in 2008. The rate and cost o readmissions were highest or Medicareand Medicaid, but readmissions were a source o signifcant cost or private payers as well.

Rsson Rts VRY wY, Vn just FoR Cs .Individual hospitals readmission rates varied substantially in 2008. Nine percent o hospitals

had unadjusted readmission rates that were at least 50% above the statewide average.

Even adjusted or hospital case mix (that is, the prevalence o more severely ill patients),

hospital experience varied. Six percent o hospitals had actual readmission rates that were

at least three percentage points above their expected rates.

pRoVn sCR pRoCsss n post-sCR suppoRtCn RuC ospt Rssons BY on-tR.At least our actors are widely viewed as important to eective discharge planning:

(1) coordination between the hospital-based and primary care physicians, (2) better

communication between the hospital-based physician and the patient, (3) better education

and support or patients to manage their own conditions, and (4) reconciliation o medications

at discharge or immediately aterward. While many o the most promising interventions or

lowering readmission rates address some or all o these actors, ew have been rigorously

evaluated using randomized controlled trials.

Two interventions that have been rigorously evaluated and ound eective are the Care

Transitions Intervention (CTI) and Project Re-Engineered Discharge (RED). Both interventions

engage specially trained nurse advocates who help patients navigate the discharge process,

educating and coaching the patient to manage his or her disease ater discharge. Both also

include a ormal reconciliation o medications ollowing discharge. However, despite strongevidence that both interventions can reduce 30-day all-cause readmissions by 30 to 35%,

relatively ew hospitals have adopted these programs.

wn ConsRn ntRVntons to RuC Rssons,ospts BnC Costs n BnFts.When hospitals are paid ee-or-service (FFS), each admission or day they provide care

represents additional revenue. A hospital that is not paid more when it reduces the probability

o readmissions loses revenue when it readmits ewer patients: each readmission the hospital

avoids represents lost revenue. As a result, it is unsurprising that hospitals might be reluctant

to adopt an intervention to reduce readmissions when they are paid FFS, even when the

intervention is proven to be eective.Increasingly, Federal and state policymakers are looking to payment incentives to re-align

hospital incentives to improve quality on various metrics, including their rates o readmission.

When deciding how to respond to payment incentives, a revenue-maximizing hospital


6/51


would compare the direct and indirect costs with the fnancial beneft o reducing its rate

o readmission. Payment reorm aims to tip the scale by increasing the fnancial beneftto hospitals that reduce their readmission rates; however, little analysis has been done to

determine how eective dierent payment reorms might be.

The two payment reorms considered in this studyP4P and episode-based payments

represent the two ends o a continuum o payment approaches designed to reduce the perverse

incentives o FFS payment that encourage readmissions. The versions o P4P and episode-

based payment we selected are clearly dierent rom one another, but either could be modifed

in ways that would make them more similar (or example, by integrating shared savings

between the payers and hospitals).

The P4P approach that we consider is similar to that recently adopted by New Yorks Medicaid

program, and potentially like that which Medicare will implement soon. P4P continues

to provide FFS payments or each readmission, but adds an incentive or hospitals withmore readmissions than would be expected (given their case mix) to work at lowering their

readmission rates. With P4P, payers can easily recoup savings: not only do they beneft rom

reduced readmissions (resulting in reduced payments), but they pay hospitals with high rates o

readmission less per admission.

In contrast, episode-based payments (which Medicare has piloted but currently does not plan

to adopt more widely) would replace ee-or-service payment entirely. Each hospital would be

paid more or an index stay by an amount equal to its expected cost o readmissions, adjusted to

its case mix. Episode-based payments would provide an incentive or everyhospital to reduce

its readmission rate by, in eect, putting the hospital at fnancial risk or each readmission.

However, because hospitals would retain the savings associated with reduced readmissions,

payers would beneft only as payments are benchmarked to lower readmission rates over time.

pYnt nCntVs Cn RuC Rssons n Costs.Based on a simulation o hospital responses to P4P and episode-based payments, we estimated

that either would result in reduced readmissions and lower total hospital payments. However,

both the magnitude o the response and the expected short-term cost savings would vary,

depending on the payment incentive.

Specifcally, when conronted with P4P incentives, most hospitals would ace no payment

reduction; and among the hospitals that would ace a payment reduction, only some would act

to reduce readmissions. We estimate that 7% o hospitals in New York would respond to P4P

by implementing an intervention known to reduce readmissions (CTI or Project RED), resulting

in 1,200 to 2,000 ewer readmissions per year (a reduction o 0.5 to 1%). In contrast, hospitalswould be uniormly more responsive to episode-based payments. We estimate that at least hal

o hospitals (and as many as 82%) would implement either clinical intervention, resulting in

19,000 to 45,000 ewer readmissions per year (a reduction o 7 to 16%).



7/51


Reducing the number o readmissions generates cost savingsbut the savings to payers

depend on the type o payment incentive. When hospitals are paid FFS, payers always capturethe cost savings rom ewer readmissions. While the P4P incentives we modeled would induce

less change in hospital behavior than episode-based payments, payers would capture most o

the cost savings. As a result, simulated total payments to hospitals (or all admissions) would

all by about $200 million (1%) (table s.2).

With episode-based payments, many more hospitals would respond and, thereore,

readmissions would all more than they have under P4P; however, because the hospitals would

be paid their risk-adjusted expected cost o readmissions (reecting recent past perormance

statewide), the reduction in total payments would lag behind the reduction in readmissions.

Under episode-based payments, simulated total payments would all $188 to $286 million

(0.8 to 1.2%). Payers would save more over time as they rebased episode-based payments to

reect lower rates o readmission, but in the short term, hospitals would retain most o thesavingsand, thereore, have an incentive to urther reduce readmissions.

Ab S.2. Simulated ayment eorm ects on Hospital ayments, 2008

ttAl pAYNtS

f All AdISSINS

($ bIllINS)

SIulAtd CHANg IN ttAl pAYNtS

dllAS

($ IllINS)

pCNtAg

CHANg

cal exeriece $23.4 /a /a

pay fr erfrmace

All HSpItAlS tHAt SpNd IplNt:

Ct $23.2 -$200.3 -0.9%

prjec R $23.2 -$205.3 -0.9%

ide-baed ayme

All HSpItAlS tHAt SpNd IplNt:

Ct $23.2 -$187.5 -0.8%

prjec R $23.1 -$285.5 -1.2%

SC: ahemaica plicy Reearch aalyi f ne Yr hial dicharge daa.

RCt pYnt FoR VnC-Bs sCR pRoCsssn post-sCR suppoRt Cou B oR FFCtV.Under either payment incentives that we modeled, cost savings were less than would have

occurred had more hospitals been induced to adopt CTI or Project RED, or had payers been able

to retrieve all o the savings rom reduced readmissions immediately. Payers might achieve

both greater change and immediate savings simply by paying hospitals directly to implementevidence-based interventions. For example, we estimate that New York Medicaid might have

spent $19.9 million to implement Project RED in all hospitals to achieve a net savings o $116

million. I Medicare had paid directly or evidence-based interventions to reduce readmissions,



8/51



it might have achieved even larger net savings: $427 million by paying or the Project RED in all

New York hospitals. In the aggregate, commercial payers savings would have been smaller, butcomparable to Medicaids and still substantial.

Diverting rom payment incentives to direct payment or reducing hospital readmissions would

be a signifcant step, especially in light o the P4P program that New Yorks Medicaid program

already has implemented. However, the prospect o both greater reduction in readmissions

and greater payer savings rom direct payment to hospitals to adopt evidence-based discharge

procedures raises important questions about whether payers should instead rely on payment

incentives or that purpose. This study demonstrates the need or greater clarity and discussion

among payers and hospitals about how best to achieve the changes that are needed to reduce

readmissions in New York.


9/51

1


Hospital readmissions are widely recognized as an important source o avoidablehealth care costs, as well as a potential marker or problems that reduce the

quality o care.1 High rates o hospital readmissions can indicate unacceptable

levels o hospital-acquired inections, premature discharge, ailure to reconcile

medications, inadequate communication with patients and community providers responsible or

post-discharge care, or poor transitional care. Indeed, appropriate coordination and planning

or ollow-up care that should begin in the hospital appears oten to be lacking: one study

ound that a large percentage o readmitted patients had not seen a physician ater their initial

discharge (Jencks et al. 2009).

Early initiatives to reduce readmissions started with simply educating providers and consumers

about the prevalence o readmissions, and many continue to rely on this method. For example,

Medicare Quality Improvement Organizations use data on readmissions to provide eedback tohospitals about their own perormance. In addition, CMS hosts a Medicare Compare website

to help consumers make more inormed choices when selecting a hospital or inpatient

care. Medicare Compare oers hospital-specifc inormation comparing 30-day Medicare

readmissions or three conditions (heart attacks, heart ailure, and pneumonia).2 At least eight

states (including New York) have data systems comparing hospitals on potentially preventable

readmissions (3-M Health Inormation Systems 2011).3 The Accountable Care Act (ACA) requires

the Department o Health and Human Services also to collect data on readmission rates in

order to calculate and publicly report each hospitals readmission rate.

While not all readmissions result rom problems with patient care or management, there

is strong research evidence that some specifc interventions at the time o discharge can

reduce readmissions or certain conditions (Gwadry-Sridhar et al. 2004, Phillips et al. 2004).Conronting the urgent need to address health costs, some states have begun to ocus

specifcally on such interventionsincluding adherence to condition-specifc protocols shown

to reduce readmissions, restructuring hospital and post-hospital discharge planning, and use

o standardized discharge orms to improve communication across care settings.4

Introduction

1 naially, ad i eleced ae here die f readmii have bee cdced, bh he rae ad c f readmii areigica. Fr examle, early e-fh (19.1%) f edicare aie dicharged frm he hial are readmied ihi 30 day,cig edicare a eimaed $15 $18 billi er year (Cs 2011; jec e al. 2009; edpC 2007). lg rig dy ipeylvaia acr all ayer fd ha early 20% f aie admied fr ay f everal cmm rcedre r diage i 2007ere readmied ihi 30 day f dicharge. aa frm arylad hial fr 2007 fd ha arximaely 10% f aie erereadmied ihi 30 day, cig a eimaed $657 milli er year, r 8% f al iaie charge (sCRC 2011).

2 Fr each cdii, edicare Cmare rer each hial cae-mix-aded readmii rae he aial average.

3 the yem develed by 3 ealh ifrmai syem i m cmmly ed meare eially reveable hialreadmii. decrii f he yem i lace i vari ae i available a

h://.mh.cm/wrh_aerial/peially%20preveable%20Readmii%20(ppR)%20Rer/texa%20ppR%20

ehdlgy%20overvie.df, acceed ril 22, 2011.

4 Fr examle, sae ci vidable Rehializai (stR) i rig ih fr ae (aache, ichiga, ohi, adwahig) hel redce rehializai. stR aem egage ayer, ae ad aial aehlder, aie ad heir

familie, ad caregiver imrve care crdiai befre ad fllig dicharge (ee: h://aar.er.cm/archive/7/2010,acceed ay 22, 2011).


10/51

2


Introduction (continued)

Similarly, some integrated delivery systems or multi-stakeholder collaboratives have begun

to invest in programs to provide discharged patients with inormation and advice to preventproblems that might lead to readmissions. For example, some pay specially trained nurses

or pharmacists to ollow up by telephone to confrm that the patients or caregivers received

discharge instructions, the patient did not receive duplicate or contraindicated prescriptions,

and that patients or caregivers understand what they need to do (such as physician ollow-up

visits) to prevent uture problems or complications (Pittsburgh Regional Health Initiative 2011;

Lake, Stewart, and Ginsburg 2011; Boutwell and Hwu 2009). Two prominent approaches used in

many o these eorts, the Care Transitions Intervention and Project Re-Engineered Discharge

(Coleman et al. 2006; Jack et al. 2009), are discussed in detail in Chapter 3 o this report.

Seeking to expand hospitals eorts to reduce preventable readmissions, both public and

private payers increasingly are turning to the use o fnancial incentivesusing measures o

preventable or all-cause readmissions to select a hospital network, give preerred status ina network, or determine payment levels. For example, in New York, the Medicaid program

reduces payment to hospitals with a potentially preventable readmission rate higher than

a statewide risk-adjusted benchmark or all admissions in the ollowing year.5 In Maryland

(the only state with an all-payer system or establishing hospital payment rates), planning to

incorporate P4P incentives in all-payer hospital rates is underway in an eort to reduce rates

o potentially preventable readmissions (Feeney 2011). Under the Accountable Care Act (ACA),

Medicare also will adjust payment to hospitals with relatively high rates o readmissions or

selected high-volume or high-expenditure conditions, eective October 1, 2012. As set out in

proposed regulations, the readmissions reduction program initially will target acute myocardial

inarction (heart attack), heart ailure, and pneumonia.6

Designing appropriate payment incentives to reduce readmissions raises important questions

related both to the potential eectiveness o payment incentives and to their unintended

consequences. For example, ew hospitals may respond to payment incentives i the magnitude

o incentives is insufcient. Someincluding those that disproportionately serve disadvantaged

populationsmay not have the fnancial or sta resources to respond. In either case, payment

incentives might produce less change that is desired and, urther, might worsen the fnancial

condition o hospitals that serve disadvantaged populations (Bhalla and Kalkut 2010).

This study investigates the potential or two alternative types o payment incentives to

reduce rates o readmission in New York acute-care hospitals. Microsimulation analysis is

used to estimate whether a revenue-maximizing hospital would respond to, respectively, a

conventional P4P payment system or episode-based payments by adopting either o two specifc

5 the sae blic healh la reqire ha rae f ayme fr iaie ervice be redced ch ha e edicaid aymeaeide fall a lea $35 milli fr he erid jly 1, 2010 hrgh arch 31, 2011, ad a lea $47 milli he ex year (ril 1, 2011hrgh arch 31, 2012).

6 the red mehdlgy ad crieria be ed i imlemeig chage he edicare hial iaie recive aymereglai ere ied ril 19, 2010. the deii f alicable hial ad he adme facr by hich ayme ill beredced ill be addreed i he red rle fr FY 2 013 (Cs 2010).


11/51

3

Introduction (continued)


evidence-based interventions to improve discharge procedures and ollow-up with patients

ater discharge. In order to calculate the maximum potential eectiveness o such paymentincentives, we assume that all payersMedicare, Medicaid, and private insurance and employer

planssimultaneously adopt the same system o payment incentives.

The rest o this report is organized as ollows. In Chapter 2, estimates o hospital readmission

rates in 2008 are presented or all payers, measured rom New Yorks hospital discharge

data. Chapter 3 includes a review o the research literature on hospital interventions to

reduce readmissions, explaining the rationale or selecting two specifc evidence-based

interventions or the purpose o this study. In Chapter 4, we present the logic o a revenue-

maximizing hospitals business case or acting to reduce readmissions in response to either

P4P or episode-based payments. We also present estimates o hospital cost or each o the

two evidence-based interventions considered in this study. In Chapter 5, the results o the

simulation analysis are presentedincluding the number and proportion o hospitals thatimplement either intervention in response to payment incentives, the change in the number and

rate o readmissions, and changes in the total cost o inpatient care. Chapter 5 concludes with

an additional analysis investigating the net change in payments or inpatient hospital care that

might occur i payers directly unded interventions to reduce hospital readmission, rather than

relying on payment incentives.


12/51

4


Readmissions are common in New York hospitals, and they are costly. In 2008, nearly

15% o initial (or index) hospital stays in New York resulted in a readmission within

30 days (table .1). The cost o these readmissions totaled $3.7 billion, or about 16%

o total hospital costs that year.7

Ab II.1.Hospital eadmission ates and the Cost o eadmissions, 2008

ttAl

AdISSIN

At

ttAl pAYNtS

($ IllINS)

pCNtAg f ttAl

HSpItAl pAYNtS

ll admii 2,087,087 /a $23,391.1 100.0%

ll idex admii 1,872,564 100.0% $19,925.2 85.2%

INDEx ADMISSIONS FOLLOWED BY A READMISSION WITHIN 30 DAYS

Fr ay rea 273,575 14.6% $3,744.0 16.0%

Fr cmlicai r ifeci 72,656 3.9% $1,290.6 5.5%

INDEx ADMISSIONS FOLLOWED BY A READMISSION WITHIN 14 DAYS

Fr ay rea 175,766 9.4% $2,461.6 10.5%

Fr cmlicai r ifeci 48,079 2.6% $867.4 3.7%


N: Readmii rae are calclaed a he erceage f idex admii flled by a readmii ihi 30 r 14 day. Readmii raead c ere calclaed fr admii ha ccrred beee jaary ad ocber 2008 (rered hrgh ecember 2008) ad he aalized.the c f readmii i eimaed a he charge fr readmii mlilied by he hial 2008 al c--charge rai.

More than one-ourth o readmissions in 2008 (equal to nearly 4% o hospital stays) were

or complications or inections. On average, these readmissions were disproportionately

expensivecosting nearly $1.3 billion or 5.5% o total hospital costs in 2008.

Not all hospital stays are equally likely to be ollowed by a readmission. In 2008, most index

stays in New York were or medical treatment (versus surgery, behavioral health care, or

maternity care). Nearly 18% o medical stays resulted in a readmission, accounting or 67% o

all readmissions and 69% o all readmission costs ($2.6 billion) (table .2).8 Nearly 19% o all

readmissions were or complications or inections ollowing a medical stay; these readmissions

were disproportionately costly, accounting or 24% o all readmission costs ($913 million).

Hospital Readmissionsin New York State

7 oly idex admii are ed deermie he rri f ay ha rel i a readmii. dex admii exclde ayhere he aie died, raferred aher healh care faciliy, lef agai medical advice, r received reame fr a cdiiexeced rel i a beqe readmii, ch a berical care rir l abr ad delivery r reame fr a meaaiccacer. reaer deail decribig h idex admii ere deed i rvided i he echical aedix hi r er.

8 ll admii ha did ivlve rgery, labr ad delivery, r meal healh r bace abe reame ere caegrizeda medical ay.


13/51

5


Hospital Readmissions in New York State (continued)

Ab II.2.Numer and Cost o 30-ay eadmissions y ype o Stay, 2008

Nub f

AdISSINS

AdISSIN

At

pCNtAg

f ttAl

AdISSINS

pAYNtS f

AdISSINS

($ IllINS)

pCNtAg

f ttAl

AdISSIN

CStS

READMISSIONS FOR ANY REASON WITHIN 30 DAYS

nmber f readmii 273,575 14.6% 100.0% $3,744.0 100.0%

tYp f INdx AdISSIN:

edical 182,839 17.9% 66.8% $2,568.6 68.6%

srgical 52,466 11.8% 19.2% $771.8 20.6%

Behaviral healh 33,026 21.5% 12.1% $372.1 9.9%

aeriy 5,243 2.1% 1.9% $31.4 0.8%

READMISSIONS FOR COMPLICATIONS OR INFECTIONS WITHIN 30 DAYS

nmber f readmii 72,656 3.9% 26.6% $1,290.6 34.5%

tYp f INdx AdISSIN:

edical 50,590 4.9% 18.5% $913.4 24.4%

srgical 21,043 4.7% 7.7% $362.9 9.7%

Behaviral healh 930 0.6% 0.3% $13.5 0.4%

aeriy 94 0.0% 0.0% $0.8 0.0%


N: see table .1.

Surgical stays were less likely than medical stays to be ollowed by a readmission. In 2008,

nearly 12% o surgical index stays resulted in a readmission within 30 days, accounting or 19%

o all readmissions and nearly 21% ($772 million) in total payments or readmissions. However,

more than 40% o readmissions ollowing a surgical index stay were due to complications

or inections, a much higher proportion than or any other admission type. Though less than

5% o all readmissions, these readmissions were disproportionately expensive (similar to

readmissions or complications or inections ollowing a medical stay), accounting or nearly

10% o all readmission costs ($363 million) in 2008.

Index stays or behavioral health were the most likely to be ollowed by readmission. More than

21% o patients initially admitted or a behavioral health diagnosis were readmitted within 30 days.

However, because behavioral health stays were less common in the frst place (accounting or lessthan 10% o all stays), readmissions ollowing a behavioral health accounted or a relatively low

share o all readmissions (12%) and a still lower share o readmission costs (10%, or $372 million).

Admissions or labor and delivery were the least likely to be ollowed by a readmission. Just 2%

o index admissions or maternity diagnoses were ollowed by a readmission, accounting or

2% o total readmissions and less than 1% o all readmission costs.


14/51

6


Readmission rates varied by patient age: older patients were much more likely than

younger patients to be readmitted or any reason, and also more likely to be readmitted orcomplications or inections (table .3). Patients aged 65 or older accounted or more than hal

o all readmissions and readmission costs in New York State in 2008. More than one-third o all

readmissions or these patients were linked to complication and inection, compared with 22%

o readmissions or patients aged 45 to 64 and 11% o readmissions or patients aged 18 to 44.

Ab II.3. Numer and Cost o 30-ay eadmissions y atient Ae, 2008

Nub f

AdISSINS

AdISSIN

At

pCNtAg

f ttAl

AdISSINS

pAYNtS f

AdISSINS

($ IllINS)

pCNtAg

f ttAl

AdISSIN

CStS

READMISSIONS

FOR ANY REASON

WITHIN 30 DAYS

273,575 14.6% 100.0% $3,744.0 100.0%

pAtINt Ag:

1824 year ld 8,880 6.9% 3.2% $101.4 2.7%

2544 year ld 43,043 9.4% 15.7% $484.0 12.9%

4564 year ld 78,407 15.1% 28.7% $1,094.8 29.2%

65 r lder 143,245 18.7% 52.4% $2,063.8 55.1%

READMISSIONS

FOR COMPLICATIONS

OR INFECTIONS

WITHIN 30 DAYS

72,656 3.9% 26.6% $1,290.6 34.5%

pAtINt Ag:

1824 year ld 768 0.6% 0.3% $12.8 0.3%

2544 year ld 5,177 1.1% 1.9% $82.2 2.2%4564 year ld 17,550 3.4% 6.4% $323.1 8.6%

65 r lder 49,162 6.4% 18.0% $872.6 23.3%


N: see table .1.

Given the strong association between age and readmission rate, it is unsurprising that most

readmissions occurred ollowing stays or which Medicare paid (table .4). Nearly 20% o

Medicare stays in 2008 resulted in a readmission, accounting or 58% o all readmissions

and 60% o all readmission costs ($2.3 billion). Nearly one-third o readmissions ollowing a

Medicare stay were or complications or inections, accounting or 32% o all readmissions and

24% o all readmission costs ($909 million).

Readmissions were less common ollowing index stays paid by either Medicaid or private

insurance. In 2008, 15% o Medicaid stays were ollowed by a readmission, accounting or 20%

o readmission costs ($757 million). Just 8% o stays paid by commercial insurance resulted

in a readmission, but on average these were more costly than Medicaid readmissions, at least



15/51

7


in part due to higher commercial payment rates or hospital care. Readmissions ollowing a

private-pay stay accounted or 15% o all readmissions costs ($569 million) in 2008.

Ab II.4. Numer and Cost o 30-ay eadmissions y rimary ayer, 2008

Nub f

AdISSINS

AdISSIN

At

pCNtAg

f ttAl

AdISSINS

pAYNtS f

AdISSINS

($ IllINS)

pCNtAg

f ttAl

AdISSIN

CStS

READMISSIONS

FOR ANY REASON

WITHIN 30 DAYS

273,575 14.6% 100.0% $3,744.0 100.0%

xpCtd pAY f INdx AdISSIN:

edicare 157,566 19.6% 57.6% $2,261.6 60.4%

edicaid 57,949 15.0% 21.2% $756.7 20.2%

Cmmercial irace 45,102 8.4% 16.5% $568.9 15.2%

self-ay (ired) 8,288 8.7% 3.0% $101.4 2.7%

ll her 4,669 9.2% 1.7% $55.5 1.5%

READMISSIONS

FOR COMPLICATIONS

OR INFECTIONS

WITHIN 30 DAYS

72,656 3.9%% 26.6% $1,290.6 34.5%

xpCtd pAY f INdx AdISSIN:

edicare 51,050 6.3% 32.4% $909.1 24.3%

edicaid 7,696 2.0% 4.9% $158.9 4.2%

Cmmercial irace 11,478 2.1% 7.3% $183.9 4.9%

self-ay (ired) 1,274 1.3% 0.8% $22.7 0.6%

ll her 1,158 2.3% 0.7% $16.0 0.4%


N: see table .1.

Readmission rates were higher in some regions o the state than in others. In particular,

stays at hospitals in the New York City metropolitan area or Capital District were more likely

to result in a readmission than stays at hospitals in the western or central regions o the

state (table .5). Readmissions associated with hospital stays in the New York Metro areaaccounted or 72% o readmissions statewide and nearly 80% o all readmissions costs, roughly

proportionate to its share o total stays and total hospital costs.

Readmission rates showed little variation by type o hospital. Readmission rates associated

with stays in major teaching hospitals and hospitals serving a disproportionate number o low-



16/51

8



income patients were somewhat higher than or other hospitals, but readmission rates among

these hospital types averaged at most approximately one percentage point higher than thatamong other hospitals. Reecting their large share o all admissions and their generally more

complex case mix (an issue addressed below), major teaching hospitals accounted or 43 o all

readmissions and nearly hal (49%) o all readmission costs. Disproportionate share hospitals

many o them also teaching hospitalsaccounted or a large majority o readmissions (72%)

and readmission costs (75%) in 2008.

Ab II.5.Numer and Cost o 30-ay eadmissions y the ocation

o the Index-Admission Hospital, 2008

Nub f

AdISSINS

AdISSIN

At

pCNtAg

f ttAl

AdISSINS

pAYNtS f

AdISSINS

($ IllINS)

pCNtAg

f ttAl

AdISSIN

CStS

READMISSIONS

FOR ANY REASON

WITHIN 30 DAYS

273,575 14.6% 100.0% $3,744.0 100.0%

lCAtIN f tH INdx AdISSIN HSpItAl:

ne Yr Ciy 127,496 15.2% 46.6% $2,051.1 54.8%

oher ne Yr era 69,515 14.5% 25.4% $941.2 25.1%

weer Regi 36,112 13.6% 13.2% $358.7 9.6%

Ceral 20,446 13.6% 7.5% $208.0 5.6%

Caial iric 20,006 14.5% 7.3% $185.0 4.9%

READMISSIONS

FOR COMPLICATIONS

OR INFECTIONS

WITHIN 30 DAYS

72,656 3.9% 26.6% $1,290.6 34.5%

lCAtIN f tH INdx AdISSIN HSpItAl:

ne Yr Ciy 30,780 3.7% 42.4% $682.3 18.2%

oher ne Yr era 19,501 4.1% 26.8% $338.3 9.0%

weer Regi 10,674 4.0% 14.7% $130.0 3.5%

Ceral 5,957 4.0% 8.2% $74.6 2.0%

Caial iric 5,744 4.2% 7.9% $65.3 1.7%


N: see table .1.

a clde g lad ad ne Rchelle.


17/51

9



Ab II.6.Numer and Cost o 30-ay All-Cause eadmissions y ype o

Index-Admission Hospital, 2008

Nub f

AdISSINS

AdISSIN

At

pCNtAg

f ttAl

AdISSINS

pAYNtS f

AdISSINS

($ IllINS)

pCNtAg

f ttAl

AdISSIN

CStS

READMISSIONS

FOR ANY REASON

WITHIN 30 DAYS

273,575 14.6% 100.0% $3,744.0 100.0%

TYPE OF INDEx ADMISSION HOSPITAL:

wNSHIp:

n fr r 237,373 14.6% 86.8% $3,090.9 82.6%

ll her 36,202 14.9% 13.2% $653.1 17.4%

tACHINg StAtuS:

ajr eachighial

118,642 14.8% 43.4% $1,819.4 48.6%

oher eachighial

96,185 14.4% 35.2% $1,278.9 34.2%

n-eachig 56,380 14.5% 20.6% $645.7 17.2%

dISpptINAt SHA HSpItAl:

Ye 196,450 14.9% 71.8% $2,793.4 74.6%

n 74,664 13.8% 27.3% $903.2 24.1%


NS: see table .1 e. -icme edicare beeciarie receivig lemeal ecriy icme (ss) ad l-icme edicaidbeeciarie acc fr a high rri f iaie day i hial deigaed a dirriae hare hial. l 2008,55% f all ace-care hial i ne Yr sae (129 f 234 i al) ere dirriae hare hial.


18/51

10



nVu ospt VRton n Rsson Rts

Despite relatively little variation between dierent types o hospitals, there was substantialvariation among individual hospitals readmission rates in 2008. While nearly one-third o

hospitals (31%) had readmission rates o 14 to 15%, nearly 13% o hospitals had readmission

rates o 10% or less. Nine percent o hospitals had readmission rates o 20% or moreat least

50% above the statewide average (Figre .1).


1.3%

0.4%

1.3% 1.3%1.7%

0.4%0.9%

2.6%3.0%

5.6%

7.8%8.2%

9.5%

fIg II.1. istriution o Hospitals y 30-ay All-Cause eadmission ates, 2008

10.3%

15.5%15.9%

3.4%

2.2% 2.2%

3.4%3.0%

0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% 15% 16% 17% 18% 19% 20% 21% 22%or

more

18%

10%

12%

14%

16%

4%

2%

8%

6%

0%

perceage f hial

30-ay ll-Cae Readmii Rae


19/51

11



Individual hospitals also perormed very dierently with respect to their rates o readmission or

complications or inections (Figre .2). One-third o all hospitals (33%) had 30-day readmissionrates o 4% associated with complications and inectionsapproximately the statewide average.

But in 9% o hospitals, readmission rates or complications and inections were 6% or more, and

in 2% o hospitals, they were 8% or moreat least twice the statewide average.


4.7%3.0%

11.6%

22.4%

33.2%

16.4%

5.6%

1.3% 1.7%

fIg II.2. istriution o Hospitals y 30-ay eadmission ates orComplications and Inections, 2008

perceage f hial

0% 1% 2% 3% 4% 5% 6% 7% 8%

35%

15%

20%

25%

30%

10%

5%

0%

30-ay Readmii Rae fr Cmlicai ad feci


20/51

12



Some o the variation in readmission rates between hospitals may be because o dierences

in hospital case mix. Across all hospitals, index admissions or some conditionsparticularlyor certain chronic conditionswere much more likely to be ollowed by a readmission.

Approximately 28% o index admissions or heart ailure were ollowed by a readmission within

30 days in 2008 (table .7). Similarly, at least 25% o index admissions or septicemia and

disseminated inections, alcohol or opioid abuse and dependence, or renal ailure were ollowed

by a readmission.

Ab II.7.hirty-day All-Cause eadmission ates y Index-Admission ianosis, 2008

ttAl INdx

AdISSINS SultINg

IN A AdISSIN

AdISSIN

At

pCNtAg

f All

AdISSINS

ALL INDEx ADMISSIONS273,575 14.6% 100.0%

INdx-AdISSIN Ap-dgS wItH tH HIgHSt AdISSIN AtS:a

ear failre 15,059 28.3% 5.5%

Chric brcive lmarydieae (Cop)

9,540 22.9% 3.5%

seicemia & diemiaed ifeci 7,793 26.7% 2.8%

oher emia 7,440 16.9% 2.7%

schizhreia 6,506 23.1% 2.4%

lchl abe ad deedece 5,363 25.1% 2.0%

oiid abe ad deedece 5,082 25.9% 1.9%

Real failre 5,582 25.1% 2.0%

Cardiac arrhyhmia ad cdcidirder 5,456 15.9% 2.0%

percae cardivaclarrcedre

5,317 13.9% 1.9%

Bilar dirder 4,373 20.2% 1.6%

kidey & riary rac ifeci 4,766 17.2% 1.7%

gia ecri ad craryahercleri

4,490 17.5% 1.6%

ajr dereive dirder& her ychi

3,959 17.6% 1.4%

Celllii ad her bacerial iifeci

4,068 12.5% 1.5%

iabee 3,599 17.0% 1.3%

ll her pR-R 175,181 12.6% 64.0%


NS: see table .1 e.

a dex ay ere claied ig ll paie Reed iagi Relaed r (pR-R), hich gr hial ay fr imilardiage ad everiy f ille.


21/51

13



To account or dierences in readmission rates related to dierences in hospital case mix,

we calculated the dierence between each hospitals actual readmission rate and its expectedreadmission rate. A hospitals expected readmission rate is the readmission rate that would

have occurred had it achieved the statewide average readmission rate or its case mix. In 2008,

about one-third o all hospitals (37%) had an actual readmission rate within one percentage

point o their expected rates (Figre .3). All other hospitals either outperormed their expected

rates by more than a percentage point (40%) or underperormed by the same margin (23%).

Six percent o hospitals had actual readmission rates that were at least three percentage points

above their expected rates.

e ha -7 -4 -3 2 3 6 7-6 -5 0 1-2 -1 4 5 8 r mre

-7 -6 -1 0-5 -4 1 2 5 6-3 -2 3 4 7 8

1.7% 1.7% 1.3% 1.7%2.6%

12.1%

19.0% 19.4%

9.9%

7.3%

1.3%

2.6%

0.4% 0.4% 0.4%0.9%

17.2%

fIg II.3. istriution o Hospitals y the ercentae-oint ierence etweenActual and xpected eadmission ates, 2008

perceage f hial20%

12%

14%

16%

18%

10%

6%

4%

2%

0%

8%


NS: negaive vale idicae a hial acal readmii rae a bel (beer ha) i execed rae, give i cae mix f idexadmii. piive vale idicae a hial acal readmii rae a abve (re ha) i execed rae.

perceage pi ifferece Beee he ial' cal ad xeced Readmii Rae


22/51

14



While major teaching hospitals showed, on average, just a slightly higher rate o actual

readmissions in 2008, they were much more likely to have actual readmission rates above theirexpected rates. In 2008, nearly 40% o major teaching hospitals had actual rates o readmission

more than a percentage point higher than their expected rates, compared with 17 to 18%

o other teaching and nonteaching hospitals (table .8). Disproportionate share hospitals

also were much more likely to have readmission rates above their expected rates than other

hospitals (30% versus 14% among other hospitals), although their actual average readmission

rate was only moderately higher.

In the ollowing chapter we describe strategies to reduce rates o hospital readmissions and

the research evidence supporting their eectiveness. We select two o these strategies with

arguably the strongest empirical evidence o eectiveness or the microsimulation analysis that

comprises the balance o this report.

Ab II.8.Actual and xpected eadmission ates y ype o Hospital, 2008

Nub

f

HSpItAlS

ACtuAl AdISSIN At CpAd wItH

tH HSpItAlS xpCtd At

ACtuAl

AdISSIN

At

btt

(AdISSIN At

wAS At lASt N

pCNtAg pINt

blw xpCtd)

Abut quAl

(AdISSIN At

wAS wItHIN +/- 1

pCNtAg pINt

f xpCtd)

wS

(AdISSIN At

wAS At lASt N

pCNtAg pINt

AbV xpCtd)

ALL HOSPITALS 232 14.6% 40.1% 36.6% 23.3%

wNSHIp:

n-fr-r 205 14.6% 39.5% 37.6% 22.9%

oher 27 14.9% 44.4% 29.6% 25.9%tACHINg StAtuS:

ajr teachig 53 14.8% 20.8% 39.6% 39.6%

oher teachig 69 14.4% 44.9% 36.2% 18.8%

n-eachig 108 14.5% 47.2% 35.2% 17.6%

dISpptINAt SHA HSpItAl:

Ye 129 14.9% 34.9% 34.9% 30.2%

n 100 13.8% 47.0% 39.0% 14.0%


NS: see table .1. t f he 234 ace-care hial ideied i hi dy had ay qalifyig a idex admii,ad are iclded i he hial-ecic able ad gre.


23/51

15


Hospital Strategies toReduce Readmissions

The growing body o research investigating actors that contribute to hospital

readmissions has led to broad agreement among clinical experts and other

stakeholders that improving the discharge process and providing support

immediately post-discharge are essential to reducing the number o readmission

(Minott 2008). In turn, a number o clinical interventions have been piloted or populations

most at-risk or readmissions (such as older patients and patients with congestive heart

ailure), and some have been ound to be eective in reducing the likelihood o readmission.

This chapter briey reviews the literature evaluating these interventions.

FCtoRs tt ContRBut to RssonsFragmentation o care across settings is a major contributor to readmissions. Over the past two

decades, patients have become less likely to see their primary care physicians when hospitalized,

and more likely to see a hospitalist (a physician who provides care only to hospitalized patients).

This trend has created a heightened need or care coordination and inormation sharing o among

providers as patients are admitted and discharged rom hospitals, beyond the limits o a typical

discharge process (Bodenheimer 2008). Instead, hospitals discharge processes and timelines

generally are oriented toward documentation (not notifcation) and implicitly assume that patients

primary care physicians were involved with their inpatient care. Fewer than one in fve primary

care physicians report being routinely notifed when their patients are discharged rom a hospital

(Kripalani et al. 2007). Moreover, even when the primary care physician receives a discharge

summary, it oten lacks key inormation about the patients discharge diagnosis, test results,

medications prescribed, or plans or ollow-up care.

When primary care physicians are not notifed about an admission through the discharge

process, patients themselves become the primary source o inormation about their hospital

stay. But patients oten do not understand their condition and treatment plan as well as

hospital-based physicians may assume. One study ound that, while nearly 90% o physicians

believed patients understood key inormation about the side eects o their medications and

when to resume normal activity ollowing discharge, less than 60% o their patients actually

said they understood (Calkins et al. 1997). Such poor communication oten leaves patients

and amily members abruptly expected to manage problems encountered ater discharge

with little preparation (Coleman and Berenson 2004). Unknowledgeable about their condition

and conused about who is responsible or their care immediately ollowing discharge, many

patients may return to the hospital when they experience problems that might have beentreated successully without readmission.

Poor communication between physicians and patients, or between dierent physicians treating

the same patient in dierent settings, can also generate medication errors. One study ound that

nearly 13% o discharged patients suered an injury caused by medication errors; many required


24/51

16


Hospital Strategies to Reduce Readmissions (continued)

re-admission (Foster et al. 2003). Poor communication among the hospitalist, the patient, and the

patients primary care physician was identifed as a contributing actor in most cases.

Medication errors oten occur when the prescribing physician has incomplete inormation about

the patient. In one study, 94% o patients discharged rom an ICU were ound to have medication

errors that the treating physician corrected when presented with current inormation about

the patients other prescriptions and allergies (Provonost et al. 2003). Medication errors also

occur when patients ail to correctly take the medications that were prescribed during their

stay ater leaving the hospital. At least 14% o recently discharged patients may not comply with

the medications specifed in their discharge instructionsoten because the instructions are

illegible or incomplete, contain conicting inormation about what kind o medicine or dosage to

take, or duplicate older prescriptions without inorming the patient that previously-prescribed

drugs must be discontinued (Moore et al. 2003, Coleman et al. 2005).

ntRVntons to RuC RssonsAt least our actors are widely viewed as important to eective discharge planning: (1)

coordination between the hospital-based and primary care physician, (2) better communication

between the hospital-based physician and the patient, (3) better education and support or

patients to manage their own condition, and (4) reconciliation o medications at discharge or

immediately aterward (Kripalani et al. 2007). While many o the most promising interventions

or lowering readmission rates address some or all o these actors, relatively ew have been

rigorously evaluated (Minott 2008).

However, the interventions that have been have been rigorously evaluated, based on one

or more randomized controlled trials (RCTs), are summarized in Table III.1. Most o these

interventions have ocused on enhancing the discharge process or educating patients aboutsel-management o their condition (Boutwell and Hwu 2009). The intervention oten included

ormal review or reconciliation o medications as part o the discharge process or immediately

ater discharge. One meta-analysis o 18 dierent interventions targeting older patients with

congestive heart ailure ound that comprehensive discharge planning plus post-discharge

support reduced the probability o readmission by 25% (Phillips et al. 2004). More intensive post-

discharge services did not appear to be more eective than less intensive services: a single home

visit, multiple home visits, and requent telephone ollow-ups were all eective in reducing the

likelihood o readmissions.

Several o the tested interventions targeted older adults, who typically experience the highest

rates o readmission. One such intervention, the Transitional Care Model, was evaluated or

older patients with congestive heart ailure or respiratory inection, or who underwent cardiac

surgery, orthopedic surgery, or bowel procedures. An advanced practice nurse (APN) visited

each patient regularly during the hospital stay to evaluate patient and caregiver needs, develop

an individualized discharge plan, and educate the patient about sel-care ater discharge. The

APN also provided post-discharge support in the orm o home visits and telephone calls. In three

dierent RCTs, this model was shown to signifcantly reduce readmissions over periods ranging

rom 2 to 52 weeks ater discharge (Naylor et al. 1994; Naylor et al. 1999; Naylor et al. 2004).


25/51


26/51

18


Three interventions targeted patients o all ages. Project Re-Engineered Discharge (RED) uses

a nurse advocate to assist patients in navigating the discharge process, as well as a clinicalpharmacist who telephones the patient 2 to 4 days ater discharge to reconcile medications.

Project RED was shown to reduce 30-day all-cause readmissions by 32% (Jack et al. 2009).

Another intervention that had a pharmacist placing a telephone call to reconcile medications

in the days ater discharge also decreased readmission rates substantially, but the change

was not statistically signifcant (Dudas et al. 2001). An intervention that involved a redesigned

discharge orm, but no ormal reconciliation o medications or additional patient education and

coaching, did not reduce readmissions (Balaban et al. 2008).

Some o the evaluations described above also compared the direct costs o the intervention

to the costs savings due to lower readmissions. In most cases, the amounts that would

have been paid or readmissions were larger than the direct costs o the intervention; even

relatively intense and expensive interventions reduced net medical spending. However,in practice the savings rom lower readmissions accrue to payersMedicare, Medicaid,

commercial health insurance companies, and the patients themselveswhile hospitals

shoulder the direct costs o the intervention. When paid per admission, hospitals would also

ace a revenue loss rom ewer readmissions.

The next chapter explores this problem o incentives or hospitals to invest in reducing

readmissions and reviews changes in payment methods that might cause hospitals to invest

in reducing readmissions. We consider hospital decisions to adopt either o two clinical

interventions described above: CTI or Project RED. Both models have been rigorously evaluated

and tested on a range o patients with dierent diagnoses, and estimates o their resource costs

are public inormation.

Hospital Strategies to Reduce Readmissions (continued)


27/51

19


Policymakers have long been concerned about the lack o incentives or providers to

improve health care quality under a ee-or-service payment system, which provides

ew i any apparent incentives or quality o care and probably contributes to service

overuse and cost (Leatherman et al. 2003). These concerns have motivated some

payers to develop payment reorms that modiy or replace ee-or-service payment.

In this chapter, we consider the decision process o a hospital whose business model is simply

to maximize revenues. Such a hospital would weigh the direct and indirect costs o reducing

readmissions with the likely fnancial benefts, and proceed to invest in reducing readmissions

only i the expected fnancial benefts outweigh the costs. The basic cost components o this

calculus are discussed below with reerence to two alternative clinical interventions that weredescribed in Chapter 3: the Care Transitions Intervention (CTI) and Project Re-Engineered

Discharge (RED). The fnancial benefts are discussed with reerence to two alternative payment

reorms, pay-or-perormance (P4P) and episode-based payments, each structured to encourage

hospitals to reduce readmissions. Because each cost component could dier according to a

hospitals particular circumstances, dierent hospitals could make dierent decisions to invest in

reducing readmissions when conronted with the same payment incentives.

RCt CostsClinical interventions to reduce readmissions entail both labor and capital costs. Labor costs

are incurred to manage transitions rom the hospital to home or other settings. Capital costs,

at least with respect to the clinical interventions that have been rigorously evaluated in theresearch literature, are minimal. Potentially they include only the opportunity cost o unds

allocated to fnance labor.

As described in Chapter 3, CTI targets patients with any o 11 high-risk conditions. CTI entails

use o an advanced practice nurse who serves as an advocate or these patients during the

discharge process, helps them coordinate post-discharge care, and ormally reconciles their

medications during an initial home visit. Estimated or 2008 and calculated regionally or New

York State, these costs per discharge would have varied or hospitals across the state due to

regional dierences in labor costs and by the number o patients per year the advance practice

nurse manages. Hospitals in nonmetropolitan southwest New York State would have paid as

little as $198 per discharge to implement CTI, while hospitals in the New York-White Plains-

Wayne metropolitan area would have paid as much as $348 (table V.1).

In contrast, Project RED targets all patients discharged rom a medical stay. It uses a nurse

advocate to assist patients in navigating the discharge process, as well as a clinical pharmacist

who telephones the patient to reconcile medications. This intervention is less costly per admission

than CTI because it does not entail home visits. In 2008, the cost o implementing Project RED

Payment Incentivesto Reduce Readmissions


28/51

20


Payment Incentives to Reduce Readmissions (continued)

would have ranged rom as little as $111 per discharge or hospitals in the nonmetropolitan

Capital/Northern New York area to $132 per discharge or hospitals in Nassau-Suolk(table V.2). However, because Project RED is not as narrowly targeted (it would apply to all

patients discharged rom the hospital to home, not only those with selected high-risk diagnoses),

the total cost o implementing it can be higher depending on the hospitals case mix.

Ab IV.1.xpected Cost o CI per Admission y eion o New York State, 2007

AVAg ANNuAl CpNSAtIN

f A gIStd NuS

xpCtd AVAg CSt

f A CA tANSItINS INtVNtIN

p AdISSIN

AVAg ANNuAl

wAg

ttAl AVAg

ANNuAl

CpNSAtIN lw StIAt HIgH StIAt

tplItAN AAS

lbay-scheecady-try $56,040 $73,300 $218 $255

Bigham $52,350 $68,474 $204 $238

Bffal-niagara Fall $56,180 $73,483 $219 $255

lmira $48,140 $62,967 $187 $219

le Fall $54,410 $71,168 $212 $247

haca $52,020 $68,042 $203 $236

kig $55,280 $72,306 $215 $251

naa-sffl, erliaivii

$73,440 $96,060 $286 $334

ne Yr-whie plai-waye $78,920 $103,227 $307 $358

pgheeie-nebrgh-iddle $65,810 $86,079 $256 $299

Rcheer $54,860 $71,757 $214 $249

syrace $52,770 $69,023 $205 $240

uica-Rme $52,290 $68,395 $204 $237

NN-tplItAN AA S

Caial/nrher ne Yr $52,170 $68,238 $203 $237

Ceral ne Yr $52,970 $69,285 $206 $241

a-ceral ne Yr $53,720 $70,266 $209 $244

she ne Yr $48,540 $63,490 $189 $220

SCS: ahemaica plicy Reearch. verage aal age are frm eropolia rea Cro-dury imae, ay 2007 (Brea f abrsaiic). the erceage f al cmeai her ha age i he aial average fr hial-baed re, frm mployer Co for

mployee Compeaio, table 14, je 2007 (Brea f abr saiic).NS: the eimae ame ha he raii crdiar i a Rn h i aid he lcal average aal age ad maage 288 aie (higheimae) 336 aie (l eimae) er year, baed a blihed evalai f he Ct demrai (Clema e al. 2006). tal cmeaic iclde he c f all -age cmeai (FC, aid leave, healh irace, reireme bee, ad lemeal ay); agecmeai averaged 30.8 erce f al cmeai fr hial-baed Rn aially i 2007.


29/51

21



Ab IV.2.xpected Cost o roject per Admission y eion o New York State, 2007

ggApHIC

AA


f A pHYSICIAN ASSIStANt


f A pHAACISt xpCtd CSt

f pjCt d

INtVNtIN

p AdISSIN

AVAg

ANNuAl wAg

ttAl

CpNSAtIN

AVAg

ANNuAl wAg

ttAl

CpNSAtIN

tplItAN AAS

lbay-scheecady-try

$80,300 $104,711 $90,900 $118,534 $104

Bigham /a /a $89,000 $116,056 $103

Bffal-niagara Fall $74,900 $97,670 $96,470 $125,797 $101

lmira /a /a $102,010 $133,021 $107

le Fall $84,740 $110,501 $88,100 $114,882 $107

haca /a /a $98,000 $127,792 $106

kig /a /a $98,450 $128,379 $106

naa-sffl,erlia ivii

$87,630 $114,270 $102,690 $133,908 $115

ne Yr-whieplai-waye

$85,880 $111,988 $94,690 $123,476 $110

pgheeie-nebrgh-iddle

$82,270 $107,280 $103,510 $134,977 $110

Rcheer $76,700 $100,017 $108,490 $141,471 $106

syrace $75,810 $98,856 $100,110 $130,543 $103

uica-Rme $84,380 $110,032 $103,470 $134,925 $112

NN-tplItAN AA S

Caial/nrher neYr

$71,630 $93,406 $91,730 $119,616 $96

Ceral ne Yr $85,860 $111,961 $94,230 $122,876 $110

a-ceral ne Yr $86,360 $112,613 $106,500 $138,876 $115

she ne Yr $72,250 $94,214 $99,630 $129,918 $99

SCS: ahemaica plicy Reearch. verage aal age are frm erlia rea Cr-dry imae, ay 2007 (u.s. Brea fabr saiic). the erceage f al cmeai her ha age i he aial average fr hial-baed rfeial, frm mlyerC fr mlyee Cmeai, table 14, je 2007 (u.s. Brea f abr saiic).

NS: the eimaed c er idex admii ame 0.5 hr f harmaci ime ad 1.5 hr f -hyicia rvider ime er aie(jac e al. 2009). tal cmeai iclde he c f all -age cmeai (FC, aid leave, healh irace, reireme bee, adlemeal ay) fr -Rn hial-baed rfeial, hich averaged 30.4% f al cmeai aially i 2007; /a idicae haeimae are available.

nRCt CostsIn considering whether to intervene to reduce readmissions, a revenue-maximizing hospital

presumably would also consider its indirect costs. With ee-or-service payment (even when

hybridized to include P4P incentives), hospitals receive additional revenue when they readmit

a patientand conversely, they lose revenue when they readmit ewer patients. Plainly stated,


30/51

22



a hospital whose business model is to maximize inpatient revenues might not make an eort

to reduce readmissions, all else equalregardless o the level o direct costs the eort wouldrequire. Each readmission it avoids would simply represent revenue oregone.

Indeed, with ee-or-service payment, the fnancial incentives associated with hospitals indirect

costs are strongly perverse, as long as the hospital expects to at least break even on the next

readmission. A hospitals indirect costs would be highest (and, thereore, its ee-or-service

incentives to reduce readmissions would be lowest) when a relatively high share o its revenue

comes rom readmitting patients it had previously discharged.9 Moreover, the more eective

the intervention, the greater this hospitals indirect cost would beand, thereore, the less

likely the hospital would be to implement it.

FnnC BnFts

For a revenue-maximizing hospital to make an eort to reduce readmissions, it would needto anticipate a fnancial reward or reducing readmissions o sufcient magnitude to oset its

cost disincentives. We consider two payment reorms intended to oer such a reward: P4P and

episode-based payments. Each is discussed below.

ay or erormance

P4P is a particular instance o a broader set o payment reorms called value-based purchasing

(Rosenthal et al. 2006, Rosenthal 2009, Miller 2009). P4P initiatives typically use evidence-

based measures o quality, eectiveness, and efciency to classiy or select providers, and to

determine how much they are paid. Commercial P4P systems oten use hybrid approaches,

combining ee-or-service payment with payment bonuses or withholds that reect provider

perormance on specifc measures o quality or patient satisaction (Bernstein et al. 2010).

Evidence o the eects o P4P systems is mixed, in no small part because o the difculty o

discerning their impact on care delivery, costs, or outcomes. However, many may simply oer too

little fnancial incentive or providers to invest in quality improvement (Rosenthal et al. 2005).

A P4P system intended to reduce readmissions, when hybridized with ee-or-service, would

pay hospitals with low case-mix-adjusted readmission rates more than it would pay hospitals

with high case-mix-adjusted readmission rates. To invest in reducing readmissions, a revenue-

maximizing hospital would need to see a fnancial beneft that outweighs its direct and indirect

costs. That is, the hospital would need to succeed in increasing its P4P payment rate enough

to oset both its higher cost per discharge and its revenue loss due to lower patient volume.

Whether any one hospital would see such a net beneft is not immediately obvious. Indeed, the

hospitals response would depend both on its own circumstances and on the particular P4P

strategy used. All else equal, a hospital that expects a larger reduction in its readmission ratewould be less inclined to invest in reducing readmissions, even when paid less per admission.

9 ial al receive revee by readmiig aie dicharged frm her hial. Fr he re f hi aalyi, e ameha a hial decii a heher imleme a iervei redce readmii i made ideedelyha i, eachhial cider ly he acial ceqece i migh rdce he acig ale.


31/51

23



For payers, a potential advantage o P4P is the opportunity to retrieve savings immediately,

even i hospital behavior does not change. Payers are most likely to retrieve P4P savingso the top by paying low-perorming hospitals less, while making little or no adjustments

to payment rates or high-perorming hospitals. This practice limits high-perorming hospitals

incentives to urther reduce readmissions, although it might be efcient i currently low-

perorming hospitals oer the greatest potential or reducing aggregate readmissions.

pisode-based aymentsEpisode-based payments would eliminate payments or some or all readmissions within some

interval ater discharge, removing muchi not allo the perverse incentive associated with

ee-or-service payments. For each index admission, the hospital would receive an increase in

payment equal to the expected cost o readmissions. Each readmission would represent wholly

unreimbursed cost.

With episode-based payment, every hospital would see a fnancial beneft rom reducing

readmissions regardless o how low its readmission rate might already be. However, a revenue-

maximizing hospital would choose to invest in reducing readmissions only i its expected

savings (that is, the expected change in its costs or readmissions) exceeded the direct cost o

the intervention.

Thus, the hospitals decision to invest in reducing readmissions hinges on the same general

actors as those that come into play or P4P, but the hospital considers them dierently.

All else equal, a hospital that expects a greater reduction in its readmission rate would be more

inclined to invest in reducing readmissions.

For payers, the downside o episode-based payments would be their inability to retrieve

immediate savings. I payments are set equal to the expected value o current payments oradmissions plus expected (case-mix-adjusted) readmissions, aggregate payments in the short

term would change very little, i at all. The only signifcant opportunity to retrieve savings would

occur over time, as hospitals succeeded in reducing readmissions and payers re-benchmarked

payments to a lower expected rate o readmissions.

In the next chapter, we estimate hospital responses to P4P and episode-based payment

reorms, using a simulation modeling approach. We estimate whether hospitals in New York

would adopt either CTI or Project RED under either payment reorm and examine the resulting

eects on statewide readmission rates and payments or hospital care.


32/51

24


Payment Reform Simulations

In this chapter, we present the results o our payment reorm simulations and theresulting eects on readmissions and aggregate hospital payments. We estimate

hospitals adoption o either CTI or Project RED in response to two alternative payment

incentivesP4P and episode-based payments. The simulations assume that each

hospital would compare its cost o implementing either intervention to its expected fnancial

beneft rom reducing readmissions. A hospital would implement an intervention to reduce

readmissions only i its expected beneft exceeded its expected cost.

The frst payment reorm we simulated is a P4P approach similar to those recently adopted

by New York Medicaid and soon to be implemented by Medicare. Under this P4P approach,

hospitals with readmissions exceeding an expected rate o readmissions, reerred to as

excess readmissions, would have their overall payments per admission reduced in an amount

roughly equal to payments or those excess readmissions.

The second payment approach is an episode-based payment model, in which hospitals no

longer receive incremental payments or each readmissionas under ee-or-servicebut

instead are paid an additional amount or each initial index stay intended to cover the expected

costs o any readmissions that may occur. Thus, hospitals are at fnancial risk or each

readmission, with no incremental revenue received or actual readmission costs.

These two payment approaches represent the two ends o a continuum o reorms designed to

modiy hospitals incentives under ee-or-service (FFS) to increase the volume o admissions

including readmissions. On the one hand, P4P would continue to pay hospitals or each

readmission, but it would reduce the level o payment to hospitals that have relatively high rates o

readmissions. Since P4P retains FFS payment, ewer readmissions result in lower total payments.

Payers, thereore, can immediately recoup the savings associated with lower rates o readmission.

In contrast, episode-based payments remove all ee-or-service incentives or hospitals to

readmit the patients they discharge. However, in the short-term, hospitals retain the savings

associated with reduced readmissions, so there is less immediate beneft to payers.

Variants o these payment reorms all along a continuum in terms o how strongly they alter

the existing FFS incentives (MedPAC 2007). For example, P4P approaches might incorporate

stronger fnancial incentives that aect a greater number o hospitals, not only those with

above-average rates o readmission. They might also reduce, though not eliminate, FFS

payments or each readmission.

We observe very dierent hospital responses to P4P and episode-based payments. Each would

result in reduced readmissions and total hospital payments, but both the magnitude o theresponse and the expected cost savings in the short term vary. As described in the concluding

section, directly paying hospitals to undertake one o the evidence-based interventions to

reduce readmissions (Project RED) might yield both more change in hospital processes o care

and more savings to payers than instituting payment incentives.


33/51

25


Payment Reform Simulations (continued)

ospt Rspons to p4p

The P4P simulation assumed that each hospital would receive reduced payment in 2008 iits readmissions in 2007 exceeded a benchmark rate. The benchmark rate or each hospital

was calculated as the statewide average readmission rate in 2006, adjusted or the hospitals

case mix. For each payer, the amount o the potential payment reduction was calculated as

the sum o payments or excess readmissions (that is, the number o readmissions above

the benchmark rate) divided by total payments or all admissions to that hospital. Thus, i 1%

o payments to a hospital in 2007 were or excess readmissions, then the payment per

admission or all stays in 2008 would be reduced by 1%.

The simulation conronted each hospital with deciding whether to implement an intervention

to reduce readmissions in 2007 in order to avoid reduced ee-or-service payments the

ollowing year. I the hospitals expected reduction in total payments in 2008 exceeded the sum

o its direct costs or the intervention in 2007 plus its indirect costs (lost revenue rom reducedreadmissions) in 2007, then it would implement the intervention to reduce readmissions.

We assumed that any hospital that chose to implement the intervention in 2007 would continue

the intervention into 2008.

Based on this calculation, we ound that relatively ew hospitals would undertake the cost o

an intervention to reduce readmissions: 7% o hospitals in New York would have implemented

CTI in 2007, and 3% would have implemented Project RED (table V.1).10

Ab V.1.Simulated Adoption o vidence-based Interventions under

Alternative ayment Incentives, 2007

pAYNt

INCNtIV

All

HSpItAlS

HSpItAlSSubjCt

t pAYNt

INCNtIVS

HSpItAlS tHAt wuldIplNt CtI

HSpItAlS tHAt wuldIplNt pjCt d

Nub pCNtAg Nub pCNtAg

pay-fr-erfrmace

238 82 17 7.1% 6 2.5%

ide-baedayme

238 238 194 81.5% 121 50.8%


Under P4P, most hospitals would decline to invest because their 2007 readmission rates were

below their benchmark expected readmission rates. These hospitals would have aced no

payment reduction under P4P and, thereore, no beneft or reducing readmissionsonly directand indirect costs. Among the 82 hospitals that would have aced a payment reduction, only

a small raction would have acted to reduce readmissions.

10 we amed rae f ime referece i hi decii. ad hial diced rer i 2008, relaive he c f iveig i2007, i i ible ha ill feer hial ld have derae he iveme ad imac readmii ld have bee illle ha e imlaed.


34/51

26



Among adopters and non-adopters o the CTI intervention, the average readmission rate

was almost identical (17%), as were the average direct and indirect costs o the intervention(table V.2). However, the cost o excess readmissions among hospitals that would choose

to adopt CTI ($3.2 million) was nearly twice as large as or non-adopters ($1.5 million). As a

result these hospitals would have aced much larger potential payment reductions in 2008.

Ab V.2.Simulated Costs and benets o educin eadmissions or Hospitals facin

ayment eduction under 4, 2007

Nub

fHSpItAlS

AVAg

AdISSINAt IN 2007

AVAg CSt

f xCSS

AdISSINS

IN 2007($ IllINS)

AVAg

ptNtIAl

pAYNt

duCtIN

IN 2008($ IllINS)

AVAg

pAYNt

duCtIN

AVIdd

IN 2008 If 2007

AdISSINS

A duCd($ IllINS)

AVAg

INdICt CStS

f duCINg

AdISSINS

IN 2007($ IllINS)

AVAg

dICt CStS

f duCINg

AdISSINS

IN 2007($ IllINS)

CtI

der 17 17% $3.2 $3.4 $1.8 $1.2 $0.5

n-ader

66 17% $1.5 $1.4 $0.9 $1.0 $0.4

pjCt d

der 6 23% $2.0 $2.0 $1.8 $1.2 $0.3

n-ader

77 17% $1.8 $1.7 $1.7 $3.4 $0.6


Reecting the dierent populations and conditions that each intervention targeted and

the dierence in the costs o each intervention, the hospitals that would choose to adopt CTI

were (with just two exceptions) not the same hospitals that would adopt Project RED (data not

shown). Hospitals that would adopt CTI expected to reduce the incidence o the most expensive

readmissions, bringing down the cost o excess readmissions aster than the number o

readmissions.11 In contrast, Project RED targets all medical admissions (not only conditions

with high-cost or high-likelihood readmissions) and is cost-eective under P4P only or hospitals

with high readmission rates across all types o medical admissions. Hospitals that would adopt

Project RED had higher readmission rates (23% versus 17% among non-adopters) and also aced

much lower direct and indirect costs to achieve the same beneft as non-adopters.

Among hospitals that would decide to invest in reducing readmissions, the business case

or doing so was generally strong. Calculated across all hospitals that would implement CTI,

the estimated rate o return was 41% on an average intervention cost o $464,000 per hospital.

11 Cmared ih prec R, he Ct iervei i mre exeive er dicharge b arge a maller mber f high-c,high-readmii cdii ch a cgeive hear failre. the imlai rel migh chage if readmii erfrmaceder hee p4p refrm ere baed umberf readmii abve he execed mber raher ha c f readmiiabve he execed c.


35/51

27



The estimated rate o return or hospitals that would implement Project RED was still greater:

112% on an average intervention cost o $292,000 per hospital (table V.3).

Ab V.3.Simulated Hospital Costs and benets o Implementin vidence-based Interventions under

Alternative ayment Incentives, 2007

HSpItAlS tHAt wuld IplNt CtI

HSpItAlS tHAt wuld IplNt

pjCt d

pAYNt INCNtIV

Nub f

HSpItAlS

AVAg

dICt CSt

($ tHuSANdS)

AgggAt

At

f tuN

Nub

f

HSpItAlS

AVAg

dICt CSt

($ tHuSANdS)

AgggAt

At

f tuN

pay-fr-erfrmace 17 $463.7 41% 6 $291.9 112%

ide-baed ayme 194 $415.0 111% 121 $711.4 75%


NS: verage dllar am are rered er hial. the average rae f rer der ay-fr-erfrmace i calclaed a he dicedredci i he ayme ealy i 2008, e f l readmii revee ad iervei c, a a erceage f he iervei c. the rae frer der eide-baed ayme i he redci i reimbred readmii a a erceage f iervei c.

ospt Rspons to pso-Bs pYntsWe expect that revenue-maximizing hospitals would be uniormly more responsive to episode-

based payments than to P4P (as we modeled these approaches). Under episode-based

payments, hospitals would receive an enhanced payment or a patients initial admission, but no

payments or any subsequent readmission within 30 days. The enhanced payment is equivalent

to the hospitals current ee-or-service payment or the initial stay, plus an additional payment

intended to cover the average cost and likelihood o readmission or any reason.12 Hospitals that

successully reduce readmissions to below the average statewide rate retain the dierence.

In each subsequent year, episode-based payments would be rebased to the current statewidereadmission rate, reecting hospital responses to episode-based payment incentives.

Because hospitals do not lose revenue rom reducing readmissions (they are no longer paid

ee-or-service or each readmission), they undertake investment i the beneft o avoiding an

unreimbursed readmission costs exceeds the cost o the intervention. In addition, all hospitals

stand to beneft rom reducing readmissions under episode-based payment, not only those

hospitals with relatively high case-mix-adjusted readmission rates.

Consistent with hospitals universal exposure to episode-based payment incentives, our out o

fve hospitals (82%) would implement CTI, and more than hal (51%) would implement Project

RED (table V.1). As was the case under pay-or-perormance, more hospitals making a cost-

beneft calculation would adopt CTI (which targets high-cost, high-readmission conditions),

despite the higher average cost o CTI compared with Project RED.

12 the lielihd f readmii fr ay rea ihi 30 day i baed he aeide average readmii rae fr he R. a rel,hial ih ler-ha-execed readmii rae give heir cae mix ld receive higher ayme ha hey crrely d derhe FFs yem, hile hial ih ler-ha-execed readmii rae ld exeriece ayme redci cmared hecrre ayme yem.


36/51

28



As under P4P, the hospitals that would implement either CTI or Project RED under episode-

based payments generally had a strong business case or doing so. The estimated rate oreturn among hospitals that would implement CTI was 111% on an average intervention cost

o $415,000 (table V.3). Among those that would implement Project RED, the estimated rate o

return was 75% on an average intervention cost o $711,444.

FFCt on t nuBR oF RssonsThe simulations assumed hospitals would succeed in reducing all-cause readmissions at

about the same rate as reported in the research literature when they adopted CTI (35%) or

Project RED (30%) (Kanaan 2009; Jack et al. 2009). Assuming that all payers adopt P4P, New

York hospitals would have readmitted an estimated 1,000 to 2,000 ewer patients (depending on

the intervention)0.5 to 1% ewer readmissions than actually occurred in 2008. The hospitals

that would fnd it cost-eective to implement Project RED were mostly small hospitals with

relatively ew stays; in contrast, three times as many hospitals would fnd it cost-eective to

implement CTI. Consequently, statewide readmissions would have allen less with the adoption

o Project RED (by about 1,000 stays in 2007 and 2008) than with the adoption o CTI (by 2,000

stays), notwithstanding CTIs ocus on a narrower range o diagnoses (table V.4). Assuming that

hospitals that invest in reducing readmissions in 2007 would continue that investment in 2008,

the percentage reductions in readmissions would have been similar in both years.

Ab V.4. Simulated ects o ayment Incentives on eadmission ates, 20072008

Nub f AdISSINS

(IN tHuSANdS)

CHANg IN AdISSINS

Nub (IN tHuSANdS) pCNtAg

ACTUAL ExPERIENCE:

2007 263.1 /a /a

2008 273.6 /a /a

PAY FOR PERFORMANCE:

CA tANSItINS INtVNtIN (CtI)

2007 261.1 -2.0 -0.8%

2008 271.5 -2.0 -0.7%

pjCt d INtVNtIN

2007 261.9 -1.2 -0.5%

2008 272.3 -1.3 -0.5%

EPISODE-BASED PAYMENTS:

CA tANSItINS INtVNtIN (CtI)

2007 244.4 -18.7 -7.1%

2008 254.3 -19.2 -7.0%

pjCt d INtVNtIN

2007 220.1 -43.0 -16.3%

2008 228.9 -44.7 -16.3%


N: ial ha imlemeed he iervei i 2007 are amed cie imlemeai i 2008.


37/51

29



Because many more hospitals would respond to episode-based payments, the reduction

in readmissions in 2007 and 2008 would have been substantially greater. Depending on theintervention, the reduction in readmissions under episode-based payments would have ranged

rom 19,000 ewer readmissions (7%) i hospitals had considered adopting CTI, to 45,000 ewer

readmissions (16%) i they had considered adopting Project RED.

FFCt on ospt pYntsReducing the number o readmissions generates cost savingsbut depending upon the

payment method used, either payers or hospitals may realize those savings in the short

term. When hospitals are paid FFS, payers immediately capture the cost savings rom ewer

readmissions. Unless modifed (or example, by P4P) FFS oers hospitals no fnancial reason

to help patients avoid readmission. In contrast, when hospitals assume ull risk (as would occur

with episode-based payments) they capture nearly all o the short-t

Documents

Readmission report