Health Policy Commission study

8/20/2019 Health Policy Commission study

1/28

COMMONWEALTH OF MASSACHUSETTS

HEALTH POLICY COMMISSION

2015COST TRENDS

REPORTPROVIDER PRICE

VARIATION


2/28


3/28

2015 Cost Trends Report: Provider Price Variation | 1

Executive Summary

Massachusetts has been a national leader in ensuring ac-cess to high quality care and, with the passage of Chapter224 of the Acts of 2012, the Commonwealth took stepsto lead the nation in slowing the growth of healthcarecosts. However, signicant and persistent variation inprovider prices for the same sets of services that is not tiedto value threatens both of these goals of healthcare accessand affordability. While some variation in prices may be warranted to support activities that are benecial to the

Commonwealth (e.g., provision of specialized services orphysician training), work by multiple state agencies overthe last six years has documented signicant variation inprovider prices that is not tied to measurable differencesin quality, complexity, or other common measures ofvalue. Tis unwarranted price variation, combined withthe large share of patient volume at higher-priced pro-viders, results in increased healthcare spending. It alsoperpetuates inequities in the distribution of healthcareresources that threaten the viability of lower-priced, highquality providers.

In this Special Report, the Health Policy Commission(HPC) builds on its past research and work by the Massa-chusetts Attorney General’s Offi ce (AGO) and the Centerfor Health Information and Analysis (CHIA), and demon-strates that the prices that different healthcare providersreceive for the same sets of services vary signicantly,1 pricevariation is not decreasing over time, and the combinationof price variation and the large share of patient volume athigher-priced providers drives higher healthcare spending. We also report on the results of a rigorous analysis of thefactors associated with inpatient hospital prices, nding

that a substantial amount of price variation reects theleverage of certain providers to negotiate higher prices with commercial insurers, rather than value-based factorssuch as higher quality of care.

Why do Provider Prices Vary? How Commercial

Health Care Prices are Set

Commercial prices for healthcare services (including fee-for-

service prices, global budgets, and other units of payment) and

other contract terms are established through negotiations be-

tween payers and providers. The results of these negotiations

are inuenced by the bargaining leverage of the negotiating

parties. Market structure, such as high market share, can create

bargaining leverage that impacts payer-provider contract ne-

gotiations because a payer network that excludes “important”

providers will be less marketable to purchasers (employers and

consumers). If a provider has a substantial market presence

such that there are few or no e ective substitutes for that

provider in its market, the potential cost to a payer of excluding

the provider from that payer’s network will be high. The provider

may use that leverage to command higher, supracompetitive

prices (and other favorable contract terms) from the payer,

and the payer may be motivated to agree to such terms in

order to keep that “important” provider in its network. On the

other hand, providers who have less market leverage may be

motivated to agree to lower prices (and less favorable contract

terms) to stay in the payer network to ensure needed patient

volume. In both cases, the prices may not re

ect the relativequality of the di erent providers, or other indicia of value. This

di erential pricing is generally not transparent to consumers

(e.g., through di erences in premiums or patient cost-sharing).

SUMMARY OF FINDINGS

1. Provider prices vary extensively for the same sets ofservices. Since 2010, multiple state agencies have docu-mented extensive variation in both hospital and physicianprices in Massachusetts for the same sets of services; thehighest-priced hospitals and physician groups have been

found to have prices two to four times those of the low-est-priced hospitals and physician groups among the threelargest commercial payers, with higher variation amongsome smaller payers. Prices vary both among all hospitalsand among cohorts of hospitals with similar characteristics;for example, relative price percentiles vary by more than70 points among community hospitals. Prices also varyacross different payment methods, including both fee-for-service prices and alternatives such as global budgets.Spending for episodes of care also varies extensively, drivenby differences in price.


4/28

2 | 2015 Cost Trends Report: Provider Price Variation

2. Provider price variation has not diminished overtime. Te HPC has found that neither hospital nor phy-sician prices are converging. Both the extent of variationand the distribution of hospital prices have been generallyconsistent since 2010, and the variation in physician priceshas increased somewhat since 2009. Te price positionsof individual hospitals and physician groups relative tothe market tend to be consistent over time, particularlyfor providers at the top and the bottom of the relativeprice distribution.

3. Unwarranted price variation contributes to higherhealthcare spending due both to the prices and to the

large share of volume at higher-priced providers. Pricevariation has a signicant impact on total spending notonly because some providers receive far higher prices thanothers for the same sets of services, but also because the

providers with high prices tend to have high volume. Forthe three major commercial payers, hospitals with thehighest inpatient relative prices had approximately six toeight times as many inpatient stays as hospitals with thelowest relative prices, and approximately 18 to 23 timesas much inpatient revenue, adjusting for differences inthe number of hospitals. Tis share of inpatient volumeand revenue at the highest-priced hospitals increased from2010 to 2014 for two of the three major payers. Volumeand revenue is also concentrated among the highest-pricedhospitals for outpatient services; the highest-priced hospi-tals had two to four times as many outpatient visits andfour to eight times as much outpatient revenue as hospitalsin the lowest-priced group.

4. Higher hospital prices are not generally associat-ed with higher quality or other common measures of value; market leverage continues to be a signicant

driver of higher prices. Past research has found thathigher prices are not generally associated with factors thatare often believed to add measurable value for consumers(e.g., quality or patient acuity). Te HPC used a new,multivariate analysis to further explore the relationship

between inpatient hospital prices and various potentialexplanatory factors. Using this rigorous methodology,the HPC found that, holding all other factors constant,including case mix (i.e., patient acuity):

• Less competition is associated with higher prices

• Membership in certain hospital systems affects prices, with membership in some systems predicting higherprices and membership in other systems predictinglower prices

• Large system size is associated with higher prices

• Provision of higher-intensity services and status asa teaching hospital are associated with higher prices

• Higher prices are not generally associated with mea-sures of higher quality of care or hospital costs

• Higher shares of patients covered by public payers areassociated with lower commercial prices

Additional HPC analysis suggests that where policymakershave dened value-based factors on which provider pricesmay vary, such as in Maryland, some variation still occurs,but the extent of this variation on value-based factors issubstantially less than the variation in Massachusetts.

5. Unwarranted price variation is unlikely to diminishover time absent direct policy action to address the

issue. Massachusetts has undertaken signicant healthcaremarket reforms that have increased the transparency ofprovider price variation and may have prevented furtherincreases in variation over time. However, there has notbeen meaningful progress in reducing unwarranted varia-tion in provider prices over the past six years, and currentreforms do not hold signicant promise for meaningfullyreducing this variation.

In light of these ndings and the lack of evidence that themarket is rectifying this dysfunction on its own throughnew payment and care delivery models or insurance prod-

uct designs, the HPC recommends direct policy actionto address unwarranted provider price variation in theCommonwealth. Following the release of this report, theHPC will promptly convene stakeholders to present anddiscuss specic, data-driven policy options for consider-ation by the legislature, other policy makers, and marketparticipants. Te HPC looks forward to working withthese stakeholders to reduce unwarranted price variationin support of more sustainable and equitable healthcaresystem.


5/28


Findings

Provider prices vary extensively for the

same sets of services

Prices vary extensively across different payment methods,including both fee-for-service prices and alternatives suchas global budgets.

Extensive variation in both hospital and

physician fee-for-service prices has been

documented for six yearsExtensive variation in provider prices for the same sets ofservices has been consistently documented in the Com-monwealth since 2010. For example, the AGO’s seminal2010 Examination of Health Care Cost rends and CostDrivers report found that the highest-priced hospitals in2008 had prices almost two times those of the lowest-pricedhospitals for one of the three major commercial payers inMassachusetts,i and for the other two payers, paymentsto the highest-priced hospitals were three to four timesthose of the lowest-priced hospitals even after adjustmentsfor factors such as volume, product mix, and service mix.Similarly, the highest-priced physician groups had pricesand adjusted payments that were approximately two tothree times those of the lowest-priced groups.2 Te AGOreported similar variation for 2009 in its 2011 report.ii Also in 2011, the Division of Health Care Finance andPolicy (DHCFP) (the predecessor to CHIA) studied vari-ation in payments for 14 selected diagnosis-related groups

i In this report, we focus on variation among commercial payersrather than, for example, among Medicaid Managed Care plans.

Among commercial payers, we often report results for BlueCross Blue Shield of Massachusetts, Harvard Pilgrim Health

Plan, and ufts Health Plan, the three largest commercial payersin Massachusetts. Tese three payers account for 67% of thecommercial market by enrollment. C. H I. A, A R P M- H C S, at 6 (Sept. 2014), availableat http://www.chiamass.gov/assets/chia-annual-report-2014.zip(last visited Jan. 13, 2016).

ii AGO 2011 Report, supra endnote 1, at 15 (nding that all threemajor payers paid their highest-priced hospital more than twoand a half times the prices of their lowest-priced hospital, andtwo payers paid their highest-priced hospital over four times theprices of the lowest-priced hospitals; for physicians, one payerpaid their highest-priced physician group almost two and a halftimes the prices of their lowest-priced group, and two payerspaid their highest-priced group over three times the prices oftheir lowest-priced group).

(DRGs),iii and found a 3- to 29-fold variation in pricespaid to hospitals in 2009.3

o monitor this variation in prices, the Legislature man-dated development of a “relative price” measure to compare

prices paid to different providers within a payer’s network(see Sidebar: Relative Price).4

iii DRGs are “a classication system that groups similar clinicalconditions (diagnoses) and the procedures furnished by thehospital during [an inpatient hospital] stay.” DRG assignmentis determined by “the beneciary’s principal diagnosis and upto 24 secondary diagnoses.” C. M MS., A C H I P PS, at 2 (Apr. 2013), available at https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/AcutePaymtSysfctsht.pdf (last visited

Jan. 12, 2016).

Relative Price

The Legislature directed DHCFP (now CHIA) to develop a methodto measure price variation within payer networks. The relative

price metric shows variation by comparing provider prices to the

average price paid in the network. For example, a hospital with a

relative price of 1.10 is paid 10% more than the network average,

while a hospital with a relative price of .90 is paid 10% less than

the average. Relative price is calculated separately by payer for

di erent types of providers (hospitals, physicians, community

health centers, etc.), and hospital relative price is comprised of

separate inpatient and outpatient relative price metrics. The relative

price calculations are structured to control for quantity and types

of services provided, as well as the di erent types of insurance

products (e.g., HMO, PPO) o ered by the payer. In addition, the

calculation for inpatient relative price incorporates the number ofcase-mix-adjusted discharges from each hospital, which means

that inpatient relative price controls for case mix, the most widely

used hospital-level measure of patient acuity. CHIA calculates

relative price by payer for all Massachusetts hospitals and for the

top 30 physician groups based on share of total payments by

each payer (all other physician groups are reported together in

the aggregate). In 2012, the top 30 groups represented 87.9% to

99.9% of all physician payments by these three major commer-

cial payers. CHIA also calculates relative price percentiles, which

dene each provider’s price ranking within a payer’s network. For

example, a hospital in the 80th percentile of inpatient relative price

has higher inpatient relative price than 80% of hospitals. Relative

price percentiles use the same scale for all payers, so the relative

position of the provider may be compared between payers or

combined across payers into a composite relative price percentile.


6/28


Using the relative price metric, CHIA found extensivevariation in both hospital and physician prices for thesame sets of services in its 2012 and 2013 reports.5 Ex-hibit 1 shows the ratios of the highest-to-lowest relativeprice for hospitalsiv and physicians that CHIA reportedfor the three major commercial payers in Massachusetts,Blue Cross Blue Shield of Massachusetts (BCBS), HarvardPilgrim Health Care (HPHC), and ufts Health Plan(HP).

CHIA continued to nd wide variation in prices for hos-pitals and physician groups in 2012 and 2013.6 While thestatistics cited above only show variation across the threemajor commercial payers in Massachusetts, hospital pric-es varied across all commercial payer networks in Massa-chusetts, as shown in Exhibit 2. In fact, CHIA has foundthat variation tends to be higher for commercial payers with a smaller Massachusetts presence,7 meaning that theresults above and in the remainder of this report that

focus on these three largest commercial payers likely un-derstate the full extent and consequences of price variation.

iv Tis Report uses the term “hospital relative price” or “relativeprice for hospitals” to refer to CHIA’s blended hospital relativeprice metric, which combines the hospital inpatient and hospitaloutpatient relative price metrics.

Te HPC has also found wide variation in prices forhospitals. As described in the next section, we found thatin 2014, the highest-priced hospitals were paid 2.71 to3.36 times the prices of the lowest-priced hospitals for thethree major commercial payers.v

CHIA found that hospital prices varied not only acrossall hospitals but also within hospital cohorts. CHIA denesfour cohorts of general acute care hospitals: academicmedical centers (AMCs), teaching hospitals, communityhospitals, and community-Disproportionate Share Hos-pitals (community-DSH).vi While AMC and teachinghospital median relative price percentiles (see Sidebar:Relative Price) (73rd percentile and 60th percentile, re-spectively) were above those of community and commu-nity-DSH median relative price percentiles (43rdpercentile and 39th percentile, respectively), Exhibit 3

shows wide variation in price within each cohort. Even amongthe six AMCs, relative price percentiles varied by as much as30 points, while for the other, larger cohorts relative pricepercentiles varied by more than 60 to more than 70 points.

v Te HPC calculated these ratios based on data for only thosehospitals reported by each payer in all data years from 2010 to2014. For this reason, our results may vary from calculations basedon those hospitals reported by each payer in data year 2014.

vi AMCs are dened as principal teaching hospitals for their

respective medical schools with case mix intensity greater than5% above the statewide average, extensive research programs,and extensive resources for tertiary and quaternary care. eachinghospitals are non-AMC hospitals that report at least 25 full-timeequivalent medical school residents per 100 inpatient beds.Non-teaching hospitals are broken into two cohorts: communityhospitals (those with a public payer mix of less than 63%) andcommunity-DSH hospitals (those with a public payer mix of63% or more). In addition, all hospitals, not just communityhospitals, with a public payer mix of 63% or more are denedas DSH. C. F H I. A, MH P A: D F Y , at E-6 (Nov. 2015), available at http://www.chiamass.gov/assets/docs/r/hospital-proles/2014/FY14-Pro-les-ech-Appendix-Final.pdf (last visited Jan. 11, 2016).

Hospital PriceVariation

Physician GroupPrice Variation

2010 2011 2009 2010

BCBS 2.73 2.88 2.56 2.64

HPHC 3.05 3.08 2.67 2.83

THP 3.58 3.70 2.85 2.77

Exhibit 1: CHIA 2012 and 2013 Report Findings: Ratios of

Highest to Lowest Relative Price

0.0

0.5

1.0

1.5

2.0

2.5

3.0

BCBS HPHC Tufts Aetna Cigna Fallon

R e l a t i v e P r i c e

Exhibit 2: Distribution of Acute Hospital Inpatient Relative

Prices by Payer (2013)

Source: CHIA Relative Price Databook (2015).8

Academic Medical CentersN=6

(40%)

TeachingN=9

(14%)

CommunityN=20(19%)

Community, DSHN=26(14%)

C o m p o s i t e R P

P e r c e n t i l e ( b l e n d e d )

20th

40th

60th

80th

100th

73

60

43

39

Exhibit 3: Acute Hospital Composite Relative Price

Percentile by Hospital Cohort (2013)



7/28


As shown in Exhibit 4, CHIA found that physician pric-es similarly varied among all commercial payers, withsome of the smaller payers showing even higher levels ofvariation.

Te HPC also found variation in prices for physiciangroups. As described in the next section, we found thatin 2013, the highest-priced physician groups were paid2.62 to 3.32 times the prices of the lowest-priced groupsfor the three major commercial payers.vii

Variation is also extensive in global budgets

and episode spending

While the ndings above focus on variation in fee-for-ser-

vice unit prices, this variation in fee-for-service rates alsotranslates into widely divergent resources available to pro-vider organizations to care for HMO and Point of Servicepatients under risk contracts because global budgets aregenerally based on historic spending, embedding past pricedifferentials. Like fee-for-service prices, resources availableunder risk contracts, including budgets and non-budget-ary incentives, are negotiated. In 2013, the AGO foundsignicant variation in health status adjusted budgetsavailable to providers under risk contracts with each ofthe three major commercial payers to care for patients

of comparable health. Across each payer’s risk contracts,the provider groups with the highest effective budgets (allpayments pursuant to the risk contract, including healthstatus adjusted budget and non-budgetary incentives suchas quality and infrastructure payments) had negotiatedtotal resources 27% to 62% higher than the groups withthe lowest effective budgets to care for comparable pop-

vii Te HPC calculated these ratios based on data for only thosephysician groups reported by each payer in all data years from2009 to 2013. For this reason, our results may vary from calcu-lations based on those physician groups reported by each payerin data year 2013.

ulations.11 In 2015, the AGOreported that this variation hadpersisted, nding that for onemajor commercial payer’s riskcontracts in 2013, the provider

with the highest effective budget

had an effective budget that was37% larger than the provider with the lowest effective budgetto care for comparable patientpopulations; risk contracts forthe other two major payersshowed a similar pattern.12

Te HPC has also found that spending levels for commonepisodes of care vary considerably. Episodes of care includeall services across settings (professional, hospital, post-

acute, etc.) associated with a procedure. For example, theHPC’s 2015 Cost rends Report shows that maternityepisode spending varies from approximately $9,722 to$18,475 for low-risk pregnancies.13 While spending bydifferent providers on an episode of care could vary dueto differences in prices, differences in utilization, or a com-bination of the two, the HPC has found that this variationis driven by variation in the price of the inpatient stay ratherthan variation in prenatal or postnatal utilization patterns.viii

viii See HPC 2014 C R, supra endnote 1, at 23(nding that average spending for hip replacements rangedfrom $26,200 at the least expensive hospital to $41,700 at themost expensive hospital, while knee replacements ranged from$22,300 to $38,000 and PCI episodes ranged from $25,600to $34,800, driven primarily by differences in the price for theprocedures rather than utilization of services before or afterthe procedures). Te HPC also found variation in spendinglevels for outpatient laboratory tests. Te HPC found that forten common tests, prices at hospital outpatient departmentsvaried considerably; prices at the 90th percentile were at leastdouble the prices at the 10th percentile for all tests. See HPC2015 C R, supra endnote 13.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

BCBS HPHC Tufts Fallon HNE Aetna


Exhibit 4: Distribution of Physician Group Relative Prices

by Payer (2012)


PrimaryC-Section

Rate“D” 20% 25% 27% 22% 31% 24% 25% 25% “D” 33% 28% 25% 21% 27%

Volume ofDeliveries

$5K

$10K

$15K

$20K

M a s s G e n e r a l H o s p

B W H

o s p

N e w t o n W e l l e s l e y H o s p

H a l l m a r k H e a l t h S y s t e m

U M a s s M e m o r i a l M C

S o u t h S h o r e H o s p

M i l f o r d R e g i o n a l M C

B I D M C

E m e r s o n H o s p

S o u t h c o a s t C h a r l t o n

W i n c h e s t e r H o s p

N o r t h e a s t H o s p

L o w e l l G e n e r a l H o s p

M t . A u b u r n H o s p

S t . V i n c e n t H o s p

$18.5$17.3

$15.7 $15.6 $15.3$14.7 $14.6 $14.5 $14.5

$12.2 $12.2

$14.0$13.4

$12.8 $12.4

Exhibit 5: Average Payments for Deliveries by Hospital

Source: HPC Cost Trends Report (2015).14

Note: “D” indicates that the hospital declined to voluntarily submit rates.

Alternative payment meth-

ods, such as global bud-

gets, can perpetuate price

disparities because they are

generally based on historic

spending.

Past reports have shown

that provider groups with

the highest global budgets

can receive 20% to 40%

more than other provider

groups to care for compa-

rable populations.


8/28


9/28


horts.xii Te distribution of hospitals around the networkaverage price has also generally persisted over time.xiii

Further, the HPC has found that over time, a given hos-pital tends to receive prices that are at similar levels aboveor below the network average. Tat is, a hospital thatreceived above-average prices in 2010 likely continued toreceive higher prices through 2014, relative to other hos-pitals. In Exhibit 7, we show relative prices for the six AMCsxiv over time. In Exhibit 8, we show relative prices

xii Te AGO found that for the three major commercial payers,

there was no change in variation within the group of AMCs andslight to moderate decreases in variation for teaching hospitals.wo payers showed slight decreases in variation for communitynon-DSH hospitals and one showed a moderate increase, whiletwo payers showed no change in variation for community-DSHhospitals and one showed a slight increase. AGO 2015 R,supra endnote 1, at 20.

xiii Consistent with the relative price ranges used by CHIA in its2015 report, the HPC analyzed the number of hospitals thatreceived prices that were more than 20% below the networkaverage; between the average and 20% below average; betweenthe average and 20% above average; and higher than 20%above average from years 2010 to 2014. Te HPC found thatthere was relatively little compression in price variation overtime across the three major commercial payers. For example,68.8% of hospitals in the BCBS network had inpatient relative

prices within 20% of the network average in 2010 and 65.6%of hospitals in this range in 2014. If variation were decreasing, we would have expected to see the share of hospitals close toaverage price levels signicantly increase over time rather thandecrease as observed here. Te change in the BCBS networkreects an increase in the proportion of hospitals receiving thelowest inpatient prices in the network.

xiv Note that in 2014, BCBS changed the way it reported relativeprice for ufts Medical Center, Massachusetts General Hospitaland Brigham and Women’s Hospital; in previous years, BCBSreported a single relative price for both urban and suburbanhospital campuses, and in 2014 instead reported separate relativeprices for the urban and suburban campuses. Te HPC blendedthe urban and suburban relative prices by computing an averagerelative price weighted by the revenue of each campus.

over time for the community hospitals with the highestand lowest relative prices.xv

Tis consistency in price position, especially at the top andthe bottom of the relative price distribution demonstratesthe persistence of price variation in the Commonwealth.Note that these graphs show changes in relative price, notabsolute price. For example, in the community hospitalgraph we see that in 2010, Fairview Hospital receivedprices 35% above the 2010 network average (relativeprice 1.35) while in 2014, Fairview Hospital receivedprices 32% above the 2014 network average (relative price

1.32). Tis does not mean that Fairview Hospital receivedslightly lower prices in 2014 than in 2010, but rather thattheir prices were slightly closer to the 2014 network averagethan to the 2010 network average.

xv Te HPC’s analysis included only general acute care communityhospitals for which the payer reported data in all ve years toallow consistent comparison. However, the island hospitalsMartha’s Vineyard Hospital and Nantucket Cottage Hospital

were excluded as low volume coupled with unique patient owpatterns resulting from their locations on islands make compar-isons diffi cult between these hospitals and other Massachusettshospitals.

0.5

1.0

1.5

2.0

2.5

2010 2011 2012 2013 2014


max/min2.7

max/min2.9

max/min3.2

max/min2.7

max/min3.4

Exhibit 6: Hospital Relative Price Distribution (BCBS)

Source: CHIA Relative Price Databooks (2012 – 2015).18

Exhibit 7: Relative Prices for Academic Medical Centers

(BCBS)



MassachusettsGeneral

Brigham & Women’s

Beth IsraelDeaconessMedical Center

Tufts MedicalCenter

UMass MemorialMedical Center

Boston MedicalCenter

0.50

0.75

1.00

1.25

1.50

20142013201220112010

Exhibit 8: Relative Prices for Highest- and Lowest-Priced

Community Hospitals (BCBS)



Falmouth

Cape Cod

Fairview

AtholMemorial

Noble

MerrimackValley

0.50

0.75

1.00

1.25

1.50

20142013201220112010

L ow er - pr i c e d h o s pi t a l s

H i gh er - pr i c e d h o s pi t a l s


10/28


11/28


Unwarranted price variation contributes to

higher healthcare spending

Tis substantial variation in provider prices can have signi-cant implications for healthcare spending. Broadly speaking,healthcare spending is comprised of two factors: utilization

(total number of services as well as the mix of services thatpatients receive) and price (each provider’s individual ratesas well as mix of providers that patients utilize for care).Tere is strong evidence, doc-umented by DHCFP/CHIA,the AGO, and the HPC, thathigher prices explain the vastmajority of recent increasesin Massachusetts healthcarespending.23 Past research bythe HPC and others has alsoshown that the higher prices

that some providers receive aregenerally not offset by savingsfrom improved care delivery orreduced utilization.xviii

xviii As discussed in the HPC’s 2014 and 2015 Cost rends Reports,higher spending for joint replacement, percutaneous coronaryinterventions, and maternity care (driven largely by differences in

price) are not associated with better patient outcomes, stronglysuggesting that price differences are also not offset by improvedoutcomes. See HPC 2014 C R, supra endnote1; HPC 2015 C R, supra endnote 13. Someresearchers have also acknowledged that while an effi cient orga-nization can reduce the volume of services provided compared tothe average by perhaps 20 percent, variation in spending drivenby higher prices is far greater than 20 percent in most marketsor, “put more pithily, higher prices eat decreased volume forlunch.” See Robert Berenson, Acknowledging the Elephant: Moving

Market Power and Prices to the Center of Health Policy, H A B, (June 3, 2014), available at http://healthaffairs.org/blog/2014/06/03/acknowledging-the-elephant-moving-market-power-and-prices-to-the-center-of-health-policy/ (lastvisited Jan. 11, 2016).

However, price increases impact spending differently de-pending on a provider’s initial price level and patientvolume. Price variation has a signicant impact on total

spending not only because some providers receive farhigher prices than others for the same set of services,but also because the providers with high prices also tendto have high volume. For the three major commercialpayers, a similar number of hospitals receive inpatientrelative prices that are more than 20% above the networkaverage (the highest-priced group) as receive inpatientrelative prices that are lower than 20% below the networkaverage (the lowest-priced group). However, hospitals inthe highest-priced group had approximately six to eighttimes as many inpatient stays in 2014 as hospitals in thelowest-priced group, and approximately 18 to 23 times

as much inpatient revenue as the lowest-priced group, afteradjusting for the difference in the number of hospitals inthese groups. As shown in the chart below, the highest-priced

hospitals have consistently greater volume and revenue thanhospitals with the lowest prices; even when the proportionof hospitals in the lowest price category increases, theirtotal share of volume and revenue remains a small fractionof the total.xix

xix Te shares of volume and revenue among the highest-pricedhospitals increased somewhat from 2010 to 2014 for BCBS andHP. In 2010, hospitals in BCBS’s highest-priced group had ap-

proximately 5.5 times the inpatient stays and just under 16 timesthe inpatient revenue of hospitals in the lowest-priced group,compared to approximately seven times the stays and 23 timesthe revenue in 2014. In 2010, hospitals in HP’s highest-pricedgroup had just over ve times the inpatient stays and nearly 16times the inpatient revenue of hospitals in the lowest-pricedgroup, compared to approximately 6.5 times the stays and 19times the revenue in 2014. For HPHC, the concentration ofinpatient stays at the highest-priced hospitals remained consistent(approximately 7.5 times the number of inpatient stays at thehighest-priced group compared to the lowest-priced in both2010 and 2014) and revenue concentrated at the highest-pricedproviders slightly decreased (from approximately 20 times therevenue to approximately 18 times the revenue concentratedin the highest-priced hospitals compared to the lowest-priced).

In 2014, hospitals with in-

patient prices more than

20% above the network

averages for the three

major payers had ap-

proximately six to eight

times as many inpatient

stays and approximately

18 to 23 times as much

inpatient revenue as hos-pitals with inpatient pric-

es lower than 20% below

the network average.


12/28


Te HPC found similar patterns for hospital outpatientservices. Hospitals with the highest outpatient relative price

had approximately two to four times as many outpatientvisitsxxi in 2012 as hospitals in the lowest-priced group,

and approximately four to eight times as much outpatientrevenue in 2012 as hospitals in the lowest-priced group,after adjusting for the difference in the number of hospitalsin these groups.xxii,xxiii

xx Tese ndings are consistent with 2015 CHIA ndings thatacross all commercial payers, higher-priced hospitals received86% of total inpatient payments. CHIA 2015 P VR, supra endnote 1, at 3.

xxi Te HPC counted outpatient hospital visits utilizing claimsdata in the All-Payer Claims Database by identifying claimsassociated with inpatient facilities but not associated with DRGsor admissions dates, and combining claim lines with the samepatient identier, service date, and provider identier into asingle outpatient visit. In other words, a single outpatient visitis a set of services provided to the same patient, on the same day,at the same hospital. Due to data constraints, we only analyzedoutpatient visits using 2012 claims data.

xxii We lack comparative data on outpatient visits for multiple years,but in analyzing revenue distribution over time, we nd onlymoderate change in revenue distribution. Hospitals with thehighest relative price for BCBS went from having approximately

9 times the revenue of those with the lowest relative price in2010 to approximately 7 times in 2014. Hospitals with thehighest relative price for HPHC had approximately 3 times therevenue of those with the lowest relative price in both 2010 and2014. Hospitals with the highest relative price for HP wentfrom having approximately 6 times the revenue of those withthe lowest relative price to approximately 5 times.

xxiii We lack data on physician volume, but do nd that for the threemajor commercial payers, physician groups with higher pricesalso receive a high share of revenue, and that this share has in-creased over time. In 2013, physician groups with above-averagerelative price received 66% to 80% of physician group revenue,up from 26% to 78% in 2009, while groups with the highestrelative prices received 21% to 53% of revenue, up from 18%to 26% of revenue in 2009.

< 20% BelowAverage Prices

Average Price

20% BelowAverage toAverage Prices

Average to 20%Above AveragePrices

> 20% AboveAverage Prices

Exhibit 11: Distribution of Inpatient Volume and Revenue at Higher- and Lower-Priced Providers (THP)

Source: CHIA 201024 and 201425 Raw Relative Price Data.xx

201420102014201020142010

Number of Hospitals Inpatient Stays Inpatient Revenue

26.8%

17.9%

32.1%

23.2%

26.8%

47.2% 47.9%

64.6% 64.7%

19.4%

11.8%

2 144.0%

16.0%

15.9%

2 13.5%

25.4%

17.8%

2 1

8.9%

21.2%

23.9%

2 1

7.7%32.1%

21.4%

19.6%

H i g h e r

P r i c e

L o w e r

P r i c e

H i g h e r

P r i c e

L o w e r

P r i c e

H i g h e r

P r i c e

L o w e r P

r i c e

H i gh er

- pr i c e d h o s pi t a l s r e c ei v e d 8 4 .1 %

of

i n p a t i en t

h o s pi t a l r ev en u ef or t h i s p a y er i n

2 0 1 4

In 2012, hospitals with outpatient prices higher than 20%

above the network averages for the three major payers had

approximately two to four times as many outpatient visits

and approximately four to eight times as much outpatient

revenue as hospitals with prices lower than 20% below the

network average.


13/28


Higher hospital prices are not generally

associated with higher value

Past reports have found a relationship

between higher prices and market leverage,

but have not generally found higher prices

to be associated with higher quality, patientacuity, or DSH status

Prior research by the Massachusetts AGO, CHIA, and theHPC has demonstrated that the higher prices that someproviders receive are not explained by better quality, higherpatient acuity, or other factors that provide benet to theCommonwealth. In 2010, for example, the AGO found no

connection between hospital price and the quality of caredelivered or how sick the patients served were; however,it did nd an association between hospital market shareand price, suggesting that hospitals seeing more patients were able to negotiate higher rates with commercial pay-ers.26 Further research presented by the AGO in 2015found almost no correlation between price and qualitymeasures for hospitals or physicians.xxiv A 2011 SpecialCommission on Provider Price Reform similarly foundno statistically signicant relationship between qualityof care and price for any commercial payer, and only a weak correlation between patient acuity and price for onepayer’s inpatient prices, with no signicant correlation forother payers.27 Te Special Commission found that DSHhospitals tended to have lower prices,28 and CHIA alsofound that DSH hospitals had lower prices while AMCs

and teaching hospitals had higher prices.29

The HPC’s rigorous multivariate analysis

shows that a substantial portion of hospital

price variation is associated with market

structure, and is not generally associated

with higher quality

Te HPC used rigorous multivariate analyses, employing16 different model variations, to further explore the re-lationship between inpatient hospital prices and variousfactors, isolating the independent associations betweeneach factor and price. Tat is, the analysis of each factor

xxiv AGO 2015 R, supra endnote 1, at 21. Te report foundno correlation between hospital price and quality performanceas measured by Mass-DAC cardiac procedure outcomes, AHRQquality indicators, patient experience scores (HCAHPS), mor-tality and readmission rates, and a process measure composite;there was a slightly positive correlation with the AHRQ IQI90 Mortality Composite for Select Conditions. Te report alsofound no relationship between physician group relative priceand physician quality performance as measured by 2012 HEDIS

Adult and Pediatric Clinical Quality Measures and 2013 Adultand Pediatric Patient Experience Survey Measures (CG-CAHPS).

holds all other factors constant, so that we can estimatethe effect of, for example, hospital system size, separatelyfrom any other factor. We analyzed the relationship ofprice position both to factors indicative of measurablyhigher value for which we might be willing to pay higherprices (e.g., higher quality of care) and factors that are notgenerally indicative of value (e.g., the level of competitiona hospital faces).xxv See the echnical Note for more detailson methods.

Te HPC found that, consistent with past ndings, 2013inpatient hospital prices in Massachusetts were tied tothe level of competition a hospital faced and the hospitalsystem with which it was affi liated. We also found thatteaching status and provision of more tertiary services alsoplayed a role. We found that measures of quality and localincome levels, both of which might justify higher prices,

were not generally associated with price. In addition, car-ing for more public-payer patients was actually associated with lower prices, suggesting that rather than higher com-mercial prices offsetting lower payment rates from publicpayers as some providers contend, hospitals serving higherproportions of Medicare and Medicaid patients are alsodisadvantaged by generally lower commercial prices. Teresults of our analysis are detailed below.

Less competition is associated with higher prices

Te HPC found that, consistent with past work andnational research, less competition (as measured by the

number of community or teaching hospitals with over-lapping service areas) was associated with higher prices,

xxv Te HPC conducted a multivariate regression analysis, analyzingthe determinants of a hospital’s inpatient relative price percentileusing Ordinary Least Squares regressions, with standard errorsclustered at hospital system level. Te inpatient relative pricemeasure underlying the inpatient relative price percentile iscalculated to hold payer-level case mix constant. As a measureof quality, we used either the 2013 PSI-90 or the 2013 otalPerformance Score. Other variables include: whether a hospitalis a community, teaching, or AMC hospital; mean householdincome by zip code in a hospital’s service area; the share ofa hospital’s inpatient services that are tertiary; the share of ahospital’s discharges paid for by MassHealth, Commonwealth

Care, or Health Safety Net; the share of a hospital’s dischargesthat are paid for by Medicare; the number of community andteaching hospitals with service areas that overlap with a hospital’sservice area; whether a community or teaching hospital has aservice area that overlaps with the service area of an AMC; thesystem to which a hospital belongs; and the number of staffedbeds in the system to which a hospital belongs. Our results

were robust across eight model specications using our baselinemethodology (Ordinary Least Squares). We also ran these eightspecications using a different methodology, a generalized linearmodel with a logit link, to further explore the sensitivity of ourmodel. Tis methodology yielded qualitatively similar resultsfor all variables. However in three model specications, worsePSI-90 scores were associated with lower price. See the echnicalNote for more details on methods.


14/28


and more competition was associated with lower pricesin Massachusetts. For example, a community or teachinghospital whose service area does not overlap with the servicearea of any other community or teaching hospital has apredicted relative price percentile 2.3 to 2.7 points higherthan if it had two such competitors. In addition, wherecommunity or teaching hospital service areas overlap withthose of AMCs, this competition effect was stronger.xxvi A community or teaching hospital that does not share itsPSA with an AMC has a predicted relative price percentile9.2 to 11.1 points higher than a similar hospital with atleast one AMC competitor. Tese ndings indicate thatless competition is associated with higher prices, whilemore competition is associated with lower prices.

Tere is substantial empirical evidence to support the con-clusion that healthcare markets with less competition and

greater market concentration tend to have higher prices forservices.30 A 2006 study that reviewed 13 empirical studiesfound that signicant increases in market concentration(i.e., signicant reductions in competition), particularly inalready-concentrated markets, increase providers’ ability toleverage higher prices and other favorable contract termsfrom commercial payers. Te authors explained that,“[s]tudies that examine consolidation among hospitals thatare geographically close to one another consistently ndthat consolidation [i.e., removal of a competitor from themarket] leads to price increases of 40 percent or more.”31 More recently, a working paper from the National Bureau

of Economic Research examined 2007-2011 commercialclaims data from UnitedHealth Care, Cigna, and Aetna,investigating the factors underlying hospital price variation within and across regions. Te study found that hospitalprices are positively associated with indicators of hospitalmarket power; controlling for a range of other factors,hospital prices in monopoly markets were 15.3% higherthan those in markets with four or more hospitals, whilemarkets with two dominant hospitals had prices 6.4%higher than markets with four or more hospitals.32 As onestudy author explained, “[t]he reason why health insurance

for the privately insured is expensive is because the pricesfrom hospitals with a lot of market power are higher.”33

xxvi Tis suggests that AMCs are acting as effective substitutes tocommunity and teaching hospitals in many markets, perhapsreecting the degree to which patients are increasingly choosingto receive routine care at AMCs as highlighted in other researchby the HPC. See M H P C’, C R: J S, at 25-26 (July 2014),available at http://www.mass.gov/anf/docs/hpc/07012014-cost-trends-report.pdf (last visited Jan. 11, 2016).

Te size of hospital systems, and membership in certainhospital systems, affect prices

When system affi liation was analyzed by size of the system(as measured by staffed beds), the HPC found that addingstaffed beds, at smaller system sizes, was associated withlower prices, suggesting that increased size could initially

create some effi ciency or cost savings. However, with in-creasing size, this effi ciency slowed and, at larger systemsizes, size was associated with higher prices, suggestingthat any effi ciency was offset by gains from market powerallowing larger systems to negotiate higher prices. Forexample, holding all other factors equal, a hospital thatis part of a system the size of Partners HealthCare System(Partners) has a predicted relative price percentile 13.1points higher than the same hospital would have as partof an average-sized system.

Te HPC also examined the effect of membership in spe-cic hospital systems, compared with being unaffi liated with a system. Tis allows us to consider, controlling fora variety of other factors in the regression, the distincteffect of being in a specic system. Tese system effectsincluded system size, but held all other factors constant(e.g., the number of competitors they face, whether theyare a teaching hospital, their share of public-payer patients,and the proportion of their services that are tertiary).Tis allowed us to measure the effect of specic systemaffi liations, including both the size of that system anddiffi cult-to-measure variables such as the impact of that

system’s brand. We found that in most cases, being partof a specic system had measurable and statistically sig-nicant effects on prices. Specically, holding all of thefactors listed above constant, hospitals in the BerkshireHealth System, Cape Cod Healthcare, Partners,xxvii andSouthcoast Health systems had higher prices than otherfactors would otherwise predict. Conversely, hospitals inthe Baystate Health, Beth Israel Deaconess Medical Center,

Circle Health, Heywood Healthcare, and Steward HealthCare systems had lower prices than other factors wouldotherwise predict.

xxvii Tis applies to hospitals that are owned by the Partners system. Weincluded a separate variable for independently-owned hospitalsfor which Partners established contracts in 2013, and foundthat for these hospitals, the affi liation was associated with lowerprices.


15/28


Provision of higher-intensity services and teachingstatus are associated with higher prices

Te HPC also analyzed the proportion of each hospital’sservices that were higher intensity, or “tertiary.”xxviii Eventhough our model held case mix (i.e., patient acuity)constant, we found that a higher proportion of tertiary

services was associated with higher prices across all inpa-tient services.xxix For example, we found that holding all elseequal, a community hospital with a relatively high share oftertiary services (at the 75th percentile among communityhospitals) has a predicted relative price percentile 5 pointshigher than a community hospital with a relatively lowshare of tertiary services (at the 25th percentile).

Consistent with past research, the HPC also found thatteaching status, compared to status as a community hospi-tal, is signicantly associated with higher price. A hospital’spredicted relative price percentile is approximately 10 to11 points higher if it is a teaching hospital rather than acommunity hospital, holding all else, including their shareof tertiary services, equal. Although it is not clear empiri-cally whether training and employing medical residents is anet nancial cost or benet to teaching hospitals,34 payerssuch as Medicare often provide additional payments toteaching hospitals, reecting the social benets of trainingnew physicians. Our analysis suggests that commercialpayers also pay higher rates to teaching hospitals comparedto community hospitals.xxx

Higher prices are not generally associated with mea-sures of higher quality of care or indicia of higherhospital costs

We measured hospital quality using the Centers for Medi-care & Medicaid Services (CMS) otal Performance Score,a nationally recognized and validated composite of mul-tiple quality measures including Clinical Process of Care,Outcomes, Patient Experience, and Effi ciency metrics.35

xxviii While we refer to these as “tertiary” DRGs, we also includehigh-intensity services that some may consider quaternary. Forour purposes, tertiary DRGs were dened as DRGs that are inthe top 10% of DRGs by case weight and are typically performed(at least 50% of discharges in 2011) at hospitals with an average

case mix index of 1 or greater.xxix Notably, AMC status was not associated with higher prices,

likely because after controlling for provision of tertiary services,the fact that a hospital is an AMC does not play a large role indetermining prices.

xxx Commercial payers also pay higher prices to AMC hospitalscompared to community hospitals. Some of the AMC effectis measured in the regression model through the competitionvariable that measures whether a community or teaching hospitalhave an AMC hospital in their PSA. Some of the AMC effect ismeasured through the share of services that are tertiary. Holdingthese and all other factors constant, we found no additionaldifference between designation as a teaching hospital and des-ignation as an AMC.

As a sensitivity analysis, we also examined the PSI-90, acomposite measure of a hospital’s rate of complications. Across all eight models in which we used the more ro-bust quality measure, otal Performance Score, we foundno signicant association between hospital quality andprice. Similarly, in ve out of eight model specications in which we used PSI-90, a hospital’s rate of complications was also not associated with price. In the three modelspecications where we found any statistically signicantassociation between any measure of quality and price, therelationship between a hospital’s complication rate andprice remained small.xxxi

Some hospitals operate in locations that have higher costs,particularly for labor. For this reason, the HPC also stud-ied the relationship between price and the mean incomefor the zip codes comprising each hospital’s service area.

We found that these area income levels were also notsignicantly associated with price, indicating that higherprices are likely not driven by a need to account for higherlocal labor costs.

A higher share of patients covered by public payers isassociated with lower commercial prices

Generally, public payers (e.g., Medicare and MassHealth)reimburse providers at lower rates than commercial pay-ers.xxxii Some providers identify these lower public rates asa valid reason for price variation and as justication forthe higher commercial rates that they receive.

However, in our analyses, we found that the more pub-lic-payer patients a hospital has, the lower its commercialprices tend to be. Both higher shares of Medicare patientsas well as higher shares of patients covered by state pro-grams (MassHealth fee-for-service, MassHealth managed

xxxi In the three model specications that found a relationshipbetween price and PSI-90 complications rates, increasing (wors-ening) a hospital’s PSI-90 by a full standard deviation (20%)above the mean decreased relative price percentile by about 3points. (By comparison, the same reduction in relative pricepercentile is achieved by increasing the number of competitorsthat community and teaching hospitals face from zero to two,

which represents only one third of the standard deviation of thatvariable.) In one of the three specications showing statisticalsignicance, the ndings were also only signicant at the 10%level.

xxxii For example, according to a survey of community hospitals bythe American Hospital Association, in 2013 private insurerspaid, on average, just over 140% of hospital costs per discharge while Medicaid and Medicare each paid just under 90% of costs,factoring in disproportionate share payments. See AH A., A H HS, C .: A H P--C R P P, M, M,1993 – 2013 (Apr. 2015), available at http://www.aha.org/research/reports/tw/chartbook/2015/chart4-6.pdf (last visited

Jan. 13, 2016).


16/28


care, Commonwealth Care, and Health Safety Net) wereassociated with lower prices independently of each other,though we note that many DSH hospitals have both higherMedicare discharges and higher Medicaid and other stateprogram discharges. We found that a hospital has a pre-dicted relative price percentile 3.2 to 4.2 points lower if ithas a share of Medicare discharges comparable to that ofDSH hospitals (53.7%) versus if it has the average shareof Medicare discharges of non-DSH hospitals (45.5%).Similarly, we found that a hospital has a predicted relativeprice percentile 1.3 to 1.8 points lower if it has a share ofdischarges paid by state programs comparable to that ofDSH hospitals (21.6%) versus a share comparable to thatof non-DSH hospitals (17.8%). If a hospital had shares ofMedicaid and Medicare discharges that were much higheror lower than these averages, we would likewise expect theimpact on pricing to be greater. Tis runs counter to the

assertion by many providers that their higher commercialrates make up for lower reimbursement by public payers;rather, hospitals with less need to balance lower publicpayer payments (i.e., hospitals that serve fewer patientscovered by public payers) are more likely to have highercommercial prices.

Some states like Maryland have limited

variation to certain value-based factors;

the extent of this value-based variation

is signicantly less than the variation in

Massachusetts.

Te presence of price variation in multiple markets acrossthe country suggests that the market dynamics that driveextensive variation in provider prices for the same setsof services are not unique to Massachusetts. As detailedabove, some of the wide variation in prices in Massachu-setts is driven by factors, such as those relating to marketstructure, that do not reect value for consumers or theCommonwealth. Again, this observation is not unique toMassachusetts. As discussed above, other New Englandstates experience signicant price variation. Like Massa-chusetts, other New England states experiencing signicant

price variation have also not found that higher prices areassociated with objective measures of value. xxxiii

However, evidence suggests that where policymakers havedened value-based factors on which provider prices mayvary,xxxiv such as in Maryland through its all-payer rate

setting program, some variation still occurs, but the extentof this variation on value-based factors is substantially lessthan the variation in Massachusetts. By design, all pricevariation in Maryland is limited to objective measuresof value, as determined through a regulatory process.Tese include case mix (patient acuity), reasonable hos-pital costs (as measured against peer hospitals), area wagevariations, payer mix, and level of uncompensated careprovided, as well as extra payments for graduate medicaleducation and an incentive program to reward hospitalsfor quality performance.36 While limiting price variation

to these specic value-based factors was a consequence ofMaryland’s rate-setting scheme, such an approach doesnot require rate-setting.

o compare variation in Massachusetts with Maryland,the HPC compared variation in median charges by Mary-land hospitalsxxxv with variation in median payments toMassachusetts hospitals for 14 DRGs, broken out by theseverity level of the inpatient stay.xxxvi Exhibit 12 shows the

xxxiii In New Hampshire, higher inpatient prices were associated withhigher occupancy rates, commercial cost per discharge (notcase-mix-adjusted), and the percent of inpatient charges billedto Medicare, while higher outpatient prices were associated

with higher commercial cost per case-mix-adjusted episode,the percent of outpatient charges billed to Medicare, and thepercent of discharges billed to Medicare. Higher rates of Med-icaid patients were associated with lower outpatient prices. NH P V 2012, supra footnote ix, at 5-6.In Rhode Island, higher-cost hospitals tended to be paid morethan hospitals with lower costs, and researchers found no linkbetween quality and price. R I P V2012, supra footnote ix, at 32-39.

xxxiv Chapter 224 of the Acts of 2012 directs the HPC, through astakeholder process, to identify acceptable and unacceptablefactors of provider price variation, and potentially to recommendmaximum reasonable adjustments from network median rates forservices or sets of services. Because Maryland has implementeda version of this policy, the HPC examines here the effect ofsuch an approach on price variation.

xxxv While median payment data were unavailable for Marylandhospitals, under Maryland’s rate-setting system, hospital chargesand hospital payments are comparable and, according to theMaryland Health Services Cost Review Commission (HSCRC),the variation in charges and payments in Maryland are approx-imately equal. Te HPC is grateful for the assistance of theMaryland HSCRC in providing this data.

xxxvi Data on Massachusetts payments is from the DHCFP 2011report discussed above, which studied payment variation forselect DRGs. DHCFP 2011 R, supra endnote 1, at 9. BothDHCFP and the Maryland HSCRC used APR-DRGs, whichare divided into 4 severity levels. DHCFP reported on variationfor 2 to 4 severity levels for each of 14 DRGs, and we comparedthese with Maryland data, for a total of 44 observations.


17/28


difference between price variation for specic diagnosesof a given complexity in Massachusetts versus Marylandin 2009. Blue bars indicate that Massachusetts variation was greater than Maryland variation (the ratio of Massa-chusetts variation to Maryland variation is over 100%), while orange bars indicate that Maryland variation wasgreater (the ratio of Massachusetts to Maryland variationis less than 100%).

As shown below, we found greater variation among pay-ments to Massachusetts hospitals than among charges byMaryland hospitals for more than three quarters of sever-ity-level DRGs. Further, for more than half of these DRGs,the variation among Massachusetts hospitals was morethan twice the level of that in Maryland. For low-severitypneumonia (DRG 139), the extent of variation in Mas-sachusetts was nearly seven times (700%) that of Maryland.

Variation is greaterin Maryland

Variation is greaterin Massachusetts

0

1

2

34

5

6

7

5 6 0

- V a g

i n a l d

e l i v e r y -

S e v e r i t y

3

5 6 0

- V a g

i n a l d

e l i v e r y -

S e v e r i t y

2

5 6 0

- V a g

i n a l d

e l i v e r y -

S e v e r i t y

1

5 4 0

- C

- S

e c t i o n -

S e v e r i t y

4

5 4 0

- C

- S

e c t i o n -

S e v e r i t y

3

5 4 0

- C

- S

e c t i o n -

S e v e r i t y

2

5 4 0

- C

- S

e c t i o n -

S e v e r i t y

1

5 1 3

- U t e r i n e a n

d a

d n e x a -

S e v e r i t y

3

5 1 3

- U t e r i n e a n

d a

d n e x a -

S e v e r i t y

2

5 1 3

- U t e r i n e a n

d a

d n e x a -

S e v e r i t y

1

4 0 3

- P r o c e

d u r e s

f o r o

b e s

i t y -

S e v e r i t y

3

4 0 3

- P r o c e

d u r e s

f o r o

b e s

i t y -

S e v e r i t y

2

4 0 3

- P r o c e

d u r e s

f o r o

b e s

i t y -

S e v e r i t y

1

3 1 3

- K n e e a n

d l o w e r

l e g p r o c e

d u r e s -

S e v e r i t y

3

3 1 3

- K n e e a n

d l o w e r

l e g p r o c e

d u r e s -

S e v e r i t y

2

3 1 3

- K n e e a n

d l o w e r

l e g p r o c e

d u r e s -

S e v e r i t y

1

3 1 0

- I n t e r v e r t e

b r a

l d i s c e x c

i s i o n -

S e v e r i t y

3

3 1 0


b r a

l d i s c e x c

i s i o n -

S e v e r i t y

2

3 1 0


b r a

l d i s c e x c

i s i o n -

S e v e r i t y

1

3 0 2

- K n e e r e p

l a c

e m e n t -

S e v e r i t y

3

3 0 2

- K n e e r e p

l a c

e m e n t -

S e v e r i t y

2

3 0 2

- K n e e r e p

l a c

e m e n t -

S e v e r i t y

1

3 0 1

- H i p r e p

l a c

e m e n t -

S e v e r i t y

3

3 0 1

- H i p r e p

l a c

e m e n t -

S e v e r i t y

2

3 0 1

- H i p r e p

l a c

e m e n t -

S e v e r i t y

1

2 6 3

- L a p a r o s c o p

i c c

h o

l e c y s t e c t o m y -

S e v e r i t y

3

2 6 3


i c c

h o

l e c y s t e c t o m y -

S e v e r i t y

2

2 6 3


i c c

h o

l e c y s t e

c t o m y -

S e v e r i t y

1

2 2 5

- A p p e n

d e

c t o m y -

S e v e r i t y

2

2 2 5

- A p p e n

d e

c t o m y -

S e v e r i t y

1

1 9 4

- C o n g e s t i v e

h e a r t

f a i l u r e -

S e v e r i t y

4

1 9 4


h e a r t

f a i l u r e -

S e v e r i t y

3

1 9 4


h e a r t

f a i l u r e -

S e v e r i t y

2

1 9 4


h e a r t

f a i l u r e -

S e v e r i t y

1

1 9 0

- A M I

- S e v e r i t y

4

1 9 0

- A M I

- S e v e r i t y

3

1 9 0

- A M I

- S e v e r i t y

2

1 9 0

- A M I

- S e v e r i t y

1

1 4 0 -

C O P D

- S e v e r i t y

3

1 4 0 -

C O P D

- S e v e r i t y

2

1 4 0 -

C O P D

- S e v e r i t y

1

1 3 9

- P n e u

m o n

i a -

S e v e r i t y

3

1 3 9

- P n e u

m o n

i a -

S e v e r i t y

2

1 3 9

- P n e u

m o n

i a -

S e v e r i t y

1

Exhibit 12: Ratio of Massachusetts Variation to Maryland Variation

Sources: DHCFP 2011 Report;37 Maryland Health Services Cost Review Commission.38


18/28


Unwarranted price variation is unlikely to

diminish over time absent direct policy

action to address the issue

As described throughout this Special Report, variation inprovider prices for the same sets of services continues to

be signicant, price variation is not decreasing over time,price variation drives increased healthcare spending, andmuch of variation in prices is not attributable to higherquality or other common measures of value, but rather tomarket leverage. Tese points underscore the necessity ofrectifying this persistent issue.

Te Commonwealth has instituted a multitude of reformsto directly combat rising healthcare costs and prevent worsening market dynamics, such as through Chapter224 of the Acts of 2012 and Chapter 288 of the Actsof 2010. While some of these initiatives have increased

transparency and may have prevented worsening of unwarranted provider price variation, none directly addressedreducing unwarranted price variation, and none currentlyhold signicant promise for meaningfully reducing suchvariation.xxxvii For example, while the state’s healthcare costgrowth benchmark is an important tool to keep the growthin healthcare expenditures in line with growth of the state’seconomy, the benchmark focuses on year-over-year growthrather than the allocation of healthcare dollars within thehealthcare system to different providers.xxxviii Early resultsshow that the benchmark has not changed behavior in amanner that would reduce price disparities, such as by en-couraging payers to reduce rate increases for higher-pricedproviders; even after the benchmark was in place in 2013,payers continued to negotiate higher increases for certainhospitals with already-higher inpatient prices.xxxix Similarly,

xxxvii For example, the requirement under Chapter 288 of the Acts of2010 for DHCFP (now CHIA) to collect data on relative pricehas signicantly enhanced our understanding of price variation.

xxxviii If evaluations of provider spending growth under the benchmark were adjusted to account for baseline spending levels, moreeffi cient providers with lower prices would have more room togrow than less effi cient providers. See AGO 2015 R, supra endnote 1, at 25-27.

xxxix AGO 2015 R, supra endnote 1, at 25-27. Tese priceincreases for higher-priced providers may reduce the availabilityof price increases for lower-priced providers. For example, the

AGO found that where increases in utilization and pharma-ceutical spending are expected, permitting even small increasesfor higher-priced providers could result in little to no priceincreases available for lower-priced providers while staying undera benchmark rate of growth, assuming no changes to the sizeor health status of the population. Specically, the AGO foundthat conservative estimates of 12.5% growth in pharmaceuticalspending from 2014-2015 and 1% growth in utilization, wouldleave 0.8% growth ($142 million) available for price increasesif the state were to meet the benchmark. In this scenario, if thehigher-priced providers received 3% price increases, all otherproviders would have to accept a price cut of 0.3%, actuallyincreasing price disparities over time.

while Chapter 224’s encouragement of the adoption ofalternative payment methods may hold promise for increas-

ing providers’ effi ciency, the construction of global budgetsthus far has been based on providers’ historic spendinglevels, entrenching historically higher fee-for-service pricesin larger global budgets as well. Further, while alternativepayment methods should encourage providers to referpatients to lower-priced providers so as to reduce spendingrelative to their risk budgets, other market forces, includingrelationships between providers, have limited this effect. As a result, extensive variation remains in risk-adjustedglobal budgets for all three major payers.xl

Recognizing that price variation has not diminished todate, and that existing policy initiatives do not appear well suited to addressing the problem, it is unlikely thatunwarranted provider price variation will diminish with-

out additional direct policy action. Tis is particularlylikely given the extent of the variation in the market. oillustrate the extent of price variation in our system, theHPC modeled the time it would take for the lowest-pricedhospitals to reach the pricelevel of the 75th percentilein 2013, with an aggressiveassumption of annual 3.6%price increases.xli At this rateof increase, it would take 16to 19 years for some hospitalsto reach the prices of the 75th

percentile in the three majorpayers’ networks.

xl For one major payer in 2013, some providers in risk contractshad approximately one third more resources (including healthstatus adjusted budgets and non-budgetary payments) availableto them than other providers on a risk adjusted basis to care forpatients. AGO 2015 R, supra endnote 1, at 20. Tis isa similar level of variation from the AGO’s previous ndings;in its 2013 report, the AGO found that in 2011 (BCBS and

HP) and 2010 (HPHC), variation in health status adjustedbudgets between the provider groups with the highest and lowestbudgets ranged from approximately $93 per-member-per-monthto approximately $220 per-member-per-month. AGO 2013R, supra endnote 1, at 21-27.

xli Note that the cost growth benchmark applies to all spendingincreases, not just price increases. Tis means that if providers

were to receive 3.6% price increases in conjunction with anyutilization increases in the Commonwealth (e.g., due to changesin the economy, changing demographics, new pharmaceuticals,etc.), it is likely that the Commonwealth would fail to meet thebenchmark. Tus, it is highly unlikely that many providers canin fact receive 3.6% increases without threatening the Common-

wealth’s ability to meet the benchmark. For more discussion,see AGO 2015 R, supra endnote 1, at 27.

Due to the extent of the

price variation in the mar-

ket for the same sets of

services, it would take 19

years for some hospitals

to reach the prices of the

75th percentile in 2013,

even if they received 3.6%

annual price increases


19/28


Conclusions

Action is required to address unwarranted

price variation and its impact on overall

spending and the sustainability of lower-

priced providers

Given the strong and consistent evidence of persistent unwar-

ranted price variation in the Commonwealth, and evidencethat the market has not made meaningful progress towardrectifying this dysfunction, the HPC recommends direct

policy action to address unwarranted provider price variation.

o inform such action, the HPC will be undertaking ad-ditional research and analyses into different policy optionsand payment structures to fairly compensate providersfor value and will be promptly convening stakeholders tobegin discussing these specic, data-driven policy optionsfor consideration by the legislature, other policy makers,and market participants in support of a more sustainableand equitable healthcare system.

Examples of policy options that should be the subject of

additional analysis and discussion to determine whetherthey have potential to reduce unwarranted price variation without increasing overall healthcare spending include:

1. Policies to enhance healthcare market transparencyand encourage consumers to use high-value providersfor their care

As discussed in the HPC’s 2015 Cost rends Report, payersshould continue to develop and improve value-orientedproducts to create incentives, such as nancial rewards,for members to choose high-value services and providers.Payers should employ strategies such as using transparent,

aligned methods to tier providers; increasing the cost differ-entials between preferred and non-preferred tiers to betterreect value-based differences among providers; improvingeducational and outreach efforts to help employers andemployees better understand the insurance products andtheir benets and tradeoffs; exploring limited networkproducts that are associated with one or more high per-forming accountable care organizations (ACOs); providingcash-back rebates for choosing low-cost providers; andoffering members incentives at the time of primary care

provider (PCP) selection, with the level of incentives tiedto differences in the total cost of care associated with theselected PCP.

Payers should also continue to improve price and qualityinformation available to members. Information, coupled with incentives and choice, is an essential element of a well-functioning market for health care. Massachusettshas already taken steps to greatly increase the amount of

price information available to consumers. However, asrecent reports have demonstrated, the state needs to makemore progress to ensure availability and accuracy of priceestimates from both providers and payers.39 Patient diffi -culty in nding price information and general confusionabout the relationship between healthcare spending andquality also indicates a need for continued discussion ofhow to make prices for services more readily available andaccessible to patients. Te Commonwealth should takesteps to ensure compliance with existing laws requiringprice transparency, and payers should prioritize making

usable cost and quality information available to membersand linking such information with opportunities andincentives to make high-value choices.

Reference pricing (in combination with bundled pay-ments where appropriate)xlii may also be a valuable toolto support enhanced consumer engagement. As describedin HPC’s 2014 Cost rends Report, reference pricing is acost-sharing structure under which “the employer or insurer

pays a predetermined amount for a particular service orprocedure and the consumer is generally responsible forthe remainder of the cost (in addition to any copayments

or coinsurance amounts). Te predetermined amount,or ‘reference price,’ is often based on a pre-identiedlow-cost provider or a median price in a market area.Reference pricing is most applicable in situations whereconsumers seek a well-dened, discrete service that is

xlii For details on the potential opportunities of combining referencepricing and bundled payments, see François de Brantes et al.,R P B P: A M C M, H C I II C P R (Oct. 2013),available at http://www.catalyzepaymentreform.org/images/documents/matchtochangemarkets.pdf (last visited Jan. 11, 2016).


20/28


planned in advance and offered by a number of providersin a region at varying prices.”40 Te HPC encourages payersand purchasers to develop reference pricing plan designsfor appropriate services.

2. Limits on provider charges for emergency out-of-network services and those delivered by out-of-networkproviders located within in-network facilities

As discussed in the HPC’s 2015 Cost rends Report, out-of-network charges for emergency services and servicesprovided by out-of-network providers located in in-network

facilities (“surprise billing”) can create diffi culties for con-sumers as well as impacts on market competition. Manyproviders, and particularly those with signicant emergencyvolume and certain hospital-based providers, have leverageto demand that payers agree to high neg

Documents

Health Policy Commission study