Healthcare Systems and Services Practice
Maximizing value in high-performance networks David Knott, Tom Latkovic, David Nuzum, Jessica Ogden, and Shubham Singhal
1
Maximizing value in high-performance networks Many payors now have experience developing value networks, but they may not yet have optimized their network configuration or approach. Over the long term, payors must be able to maximize the value these networks deliver.
With healthcare costs in America continuing
There are a number of reasons for doing
to rise and an increasing number of value-
this. For example, attitudes toward value
conscious employers and consumers making
networks are changing. Until recently, most
explicit choices about healthcare coverage,
employers and individual consumers have
the time is ripe for payors to focus on the
chosen broad networks as a way to maximize
development of lower-cost, high-perfor-
choice and minimize patient/provider disrup-
mance provider networks. Although these
tions. However, McKinsey’s market research
value networks can take a variety of forms,
(described in the appendix) has found that
they all give members access to only a
when consumers are exposed to a simulated
limited number of quality-credentialed pro-
exchange environment and then asked to
viders, in return for lower premiums, lower
select their own health benefits, up to 80
out-of-pocket costs, or both.
percent of them are willing to trade access for a lower premium. (The percentage varied
Over the past two years, most major payors
in different simulations, depending on the
have begun creating value networks or have
size of the premium reduction and the
expanded their portfolio of such networks.
specific providers involved.) Even many
These efforts were undertaken, in large part,
purchasers in the gold and platinum tiers
to ensure that payors could offer affordable
were willing to make this trade-off.
options to individuals who will be purchasing health insurance on the new state and federal
Furthermore, value networks could play
exchanges, especially those with lower in-
an important role outside the individual
come levels. In many cases, however, devel-
exchange market. These networks could,
opment of the new networks was accelerated
for example, help hold down premiums in
to meet the timelines specified in the Afford-
the growing (but potentially reimbursement-
able Care Act (ACA), resulting in some cases
rate-constrained) Medicare Advantage and
in network designs that address basic require
managed Medicaid markets. They could also
ments but are but not necessarily optimal—
be an attractive option for employers looking
they are, in essence, version 1.0 value networks.
to control benefits costs to their employees.
Now that this deadline has been met, payors
Thus, as payors begin to design their net-
have the opportunity to think more carefully
works for 2015 and beyond, they should ask
about the optimal design of their value net-
themselves how they can maximize value
works and their overall portfolio of networks.
in high-performance networks. To address
David Knott, Tom Latkovic, David Nuzum, Jessica Ogden, and Shubham Singhal
2
McKinsey & Company Healthcare Systems and Services Practice
this issue effectively, payors must carefully
or because the penalties for using out-of-
appraise the dynamics of each local market
network providers are relatively modest
to determine which type (or types) of value
(often, less than a 20-percent difference in
network would work best in each region.
co-payments). Value networks, in contrast,
They must also decide on the approaches
restrict access either by not covering out-of-
they want to use with providers in the near
network services or by imposing significant
term to hold down premiums. In addition,
cost-sharing differentials (often, co-payment
they must contemplate the types of innova-
increases of 20 to 50 percent).
tions they are willing to adopt over the longer term to hold down overall healthcare spend-
The restrictions can be applied to some or
ing and improve care quality.
all categories of providers, including inpatient and outpatient facilities, primary care
We believe that, over the next few years,
physicians, specialists, pharmacies, and
there will be a dramatic increase in the
ancillary service providers, but not emer
prevalence, profile, and importance of
gency care. Typically, value networks limit
value networks. However, careful stake-
a member’s access to at least one major
holder communication and dialogue will
category of provider to between 30 and 70
be critical for payors that want to expand
percent of those in a local market (except
their use of these networks. High-perfor-
for emergency care, where all facilities
mance networks entail a substantial change
continue to be covered). However, the pro
in the way payors interact with providers,
viders included in these narrower networks
and some providers (especially those not
are also quality-credentialed, given the
participating in these networks) may react
networks’ objective of maximizing value,
negatively. Furthermore, a range of other
not just minimizing costs.
stakeholders, including regulators, legis lators, advocacy groups, and the general
Exhibit 1 describes the most common
public, will be following the development
types of value networks: tiered networks,
of these networks closely. Payors that pro
best-of-breed networks, hospital-centric
actively communicate to all stakeholders
networks, physician-centry networks,
the benefits a new value network can deliver
and virtually integrated networks.
to the community will likely increase the chances of successful implementation
Transparent networks are closely related
significantly. The message should be clear:
to value networks, but they do not formally
value networks help give patients access
limit members’ access to providers. Instead,
to more affordable, high-quality care.
they use provider performance report cards
Types of value networks
and/or consumer decision support tools to encourage members to obtain care only from high-performing providers. They often
The majority of payors’ networks today,
also offer incentives to members, providers,
including those considered to be health
or both to achieve this aim. As more and
maintenance organizations, give members
more data about healthcare quality and
broad access, either as an explicit choice
costs become available, value networks
3
Maximizing value in high-performance networks
White Paper - Payor Value Networks — July 2013 Maximizing value in high-performance networks Exhibit 1 of 4
EXHIBIT 1 Common types of value networks
Indemnity
Broad network
Tiered network
Narrow network
Sole provider
Inpatient facility
4
5
Outpatient facility Primary care Specialists
Majority of networks today – including those dubbed “HMO”
1
2
3
5
Pharmacies Ancillary services1
1 Tiered networks use very high cost-sharing differentials to discourage members from going outside an exclusive set of preferred, high-performing providers in all major categories.
2 Best-of-breed networks include only high-performing providers in each category, without regard to existing relationships (e.g., referral patterns) among the providers.
3 Physician-centric networks limit members’ access to physicians to a small panel of primary care providers, specialists, groups, or clinics; these providers are given strong referral/gatekeeper discretion.
4 Hospital-centric networks restrict members to a single hospital system and its affiliated physicians. 5 Virtually integrated networks restrict members to a single hospital system and physician group, which are “virtually integrated” through shared incentives and clinical information systems.
1Includes
radiology, lab, durable medical equipment, chiropractic, etc.
may increasingly use transparency as a way to further encourage members to seek only
Understanding local market dynamics
high-quality, lower-cost care. For a payor, the first step in creating a strong There is no simple formula a payor can use
value network is to understand the dynamics
to determine which one (or ones) of these
of each local market in which it operates,
network types to focus on. The answer
because healthcare delivery remains a local
depends on a range of variables specific to
concern and the healthcare landscape—
each market, as well as the approaches the
including reimbursement pressures—varies
payor thinks are best for reducing the cost
significantly from region to region. The
of care while improving its quality.
primary factors that must be understood on
4
McKinsey & Company Healthcare Systems and Services Practice
White Paper - Payor Value Networks — July 2013 Maximizing value in high-performance networks Exhibit 2 of 4
EXHIBIT 2 Growth in individual market may vary significantly by market US population by coverage type
Increase in share of individual coverage
Millions of members, 2010 and 2019
Percentage points, 2010–2019
Individual1
304
Medicaid
Medicare
Group
328
328
328
24
27
36
+70% – 150%
60
61
64
+30% – 40%
57
57
14 45 44
Uninsured
2010
2–4
5–6
7–8
0–2
4–5
6–7
>8
Scenario 1
57
153
47
_< 0
Scenario 3
156
156
140
32
27
31
1
2
3
–30% – 40%
2019 Scenarios2 1Approximately
75% of future enrollment in the individual market nationally is likely to be through the exchanges (25% off the exchanges). 1: lower employer opt-out, weaker consumer uptake; scenario 2: lower opt-out, stronger uptake; scenario 3: higher opt-out, stronger uptake.
2Scenario
Source: MPACT version 5.0; McKinsey analysis
1
To estimate growth in the individual exchange population in local markets, we often use the McKinsey Predictive Agent-based Coverage Toolkit (MPACT). This analysis enables us to develop a range of scenar ios at the county level, based on varying employer optout and consumer-uptake assumptions. See the appendix for more information about MPACT.
a market-by-market level are the charac
are likely to be. Initially, a significant
teristics of the consumer population, the
portion of the demand for value networks
provider landscape, and the regulatory
will come from consumers in the individual
and legislative environment.
market. However, some regions may see much stronger growth in that market than
Characteristics of the consumer population
others (Exhibit 2).1 Among the factors that
The payor must be able to estimate how
the current number of uninsured people
quickly, in each market, the population
in each region, income distribution levels,
of consumers with control over their
state regulations affecting individual insur-
health plan selection will increase and
ance premiums, and the current industry
how price-sensitive those consumers
and employment mix.
appear to be driving this variability are
5
Maximizing value in high-performance networks
However, the number of individuals who
Value networks can succeed, however,
will be choosing their own healthcare
only if consumers are willing accept them.
coverage is likely to grow significantly in
As discussed earlier, we found that up to 80
other markets as well. For example, popu
percent of participants in simulated exchanges
lation aging makes it highly probable that
appeared willing to accept restrictions in their
the number of people with Medicare Ad
plan design in return for lower premiums.
vantage plans will increase substantially
However, the participants’ willingness to
in many regions. The managed Medicaid
accept restrictions was strongly influenced
market is also poised for significant expan-
by a number of factors, including the size of
sion. Furthermore, many people who retain
the discount offered, the types of insurance
employer-sponsored coverage may find
coverage they currently have, and the provi
themselves selecting their own health in
ders included in each network. For example,
surance products if their companies shift
when we ran repeated simulations in one
to defined-contribution benefit plans and White Paper - Payor Value Networks — July 2013 state, we discovered that the proportion private exchange models. Even some em-
of participants who said they would either
ployers that value opt not make this shift may Maximizing into high-performance networks
change insurance plans or pay extra to go out
come to view value networks as an attrac-
of network if their insurer removed their hospi
tive way3toofhold Exhibit 4 down benefit costs.
tal from the network varied widely (Exhibit 3).
EXHIBIT 3 Consumers place varying levels of importance on whether certain
hospitals are included in their network (disguised state example) Importance that a specific hospital or health system1 is within a plan’s network (%) Unimportant
5
9
7
5
5
46
49
57
61
Important
11
41
71
54
45
44
39
34 18
Provider 1 1May
Provider 2
Provider 3
Provider 4
or may not be the “preferred hospital” to which a participant was affiliated.
Source: McKinsey Consumer Exchange Simulation 2012-2013 (state-level data)
Provider 5
Provider 6
Essential
6
McKinsey & Company Healthcare Systems and Services Practice
Hospitals that have a highly respected brand
oncology services) would be difficult to
within their community may inspire more
exclude from a value network.
consumer loyalty than other facilities. Thus, the extent to which pricing outweighs con-
Hospitals using only a small amount of their
sumer loyalty could differ from health system
available capacity are likely to be eager to
to health system and locality to locality, as
capture additional volume (or defend against
well as by product tier within a given locality.
erosion of existing volume) so that they can spread their fixed costs over more patients.
Provider landscape
These facilities may be willing to offer deep
The payor must also have a deep under-
discounts in exchange for more volume.
standing of the providers within each market.
By contrast, hospitals with more balanced
Obtaining standard market-composition
capacity utilization may see less value in
information (e.g., who the players are, what
trading price for volume.
relationships exist among them, and how much market share each has) is only the first
Each market is unique. No hard rules govern
step. The payor must also develop detailed
the way specific providers will respond.
insights into the level of consolidation among ership of those hospitals (since such owner-
Regulatory and legislative environment
ship influences referral patterns), and each
State regulations on health-system pricing will
hospital’s capacity utilization. Ideally, the
also shape the payor’s pricing strategy, since
payor should also develop detailed insights
they influence both what the payor can do and
into the relative total-cost-of-care perfor-
how providers will respond. Among the issues
mance of each provider (and the level of
that must be considered: Does the state cur-
disparity in that performance within each
rently have balance billing limitations? What are
market). This information will enable the
the usual and customary restrictions on billable
payor to determine whether the market
charges? Based on the above, what level of
composition can support the creation of
reimbursement are providers likely to receive
a value network (and if so, what type) and
for patients who seek care out of their networks?
local hospitals, the extent of physician own-
also to estimate how much of a discount the providers must offer to retain patients
Improving the value delivered
or capture additional volume. To date, most payors have focused primarily For example, a rural health system with
on three approaches to improve the value de-
no major competitors would be difficult
livered through their value networks:
to exclude from a value network, especially
reducing reimbursements to providers,
if smaller, nearby hospitals have capacity
taking advantage of existing variations in
restrictions. Such a health system is also
provider efficiency, and increasing perfor-
not likely to offer a payor significant dis-
mance-based competition for providers.
counts. Similarly, a health system with
The most successful value networks we
unique clinical offerings (e.g., the only facil-
have seen capture value from all three
ity in a region that can provide advanced
sources, wherever possible.
7
Maximizing value in high-performance networks
Reducing reimbursement to providers
after we accounted for each MSA’s capacity
How payors approach contracting with
of volume that could be shifted to the
providers for reduced reimbursement can
highest-efficiency providers) and level
heavily influence the level of discounts
of fragmentation (the less fragmented
achieved. Our experience suggests that
a provider base is, the more difficult it
achieving significant reductions through
is to create a value network). Our calcu-
traditional negotiations is likely to be diffi-
lations also suggested that if capacity
cult. However, alternative approaches that
constraints were relaxed over time, the
reset the context for the negotiations (e.g.,
savings achieved through value networks
by using a request-for-proposal, or RFP,
could be as much as $36 billion annually
approach or by sending an opt-in letter)
in the top 50 MSAs alone.
constraints (which would limit the amount
can help achieve more substantial savings. Sending an RFP to hospitals, for example,
We have worked with several payors that
could allow the payor to include all hospitals
have successfully introduced value-network
in the process (avoiding the need to selec-
products in select MSAs; many of those
tively choose up front which ones to work
payors were able to reduce premiums by
with) and develop a better sense of how
10 to 20 percent in the network’s first year
willing each hospital is to accept lower
simply by taking advantage of existing
reimbursements before negotiations begin.
variations in provider efficiency. Any payor that wants to adopt a similar approach
Taking advantage of existing variations in provider efficiency
must begin by fully analyzing the efficiency
In most markets, payors can capture signi
cost-of-care basis. One provider category
ficant value by analyzing, understanding,
should be analyzed at a time to ensure that
and taking advantage of existing variations
the comparisons among the providers’
in provider total-cost-of-care performance.
performance are fair. The payor should
This source of value is the one we observe
then include in its value network only the
to be most often overlooked by payors,
most efficient providers in each category.
of each provider on a risk-adjusted, total-
perhaps because it is harder to implement than simply reducing reimbursement levels.
Although efficiency on a total-cost-of-care basis can be substantially more difficult to
Considerable evidence suggests that there
measure and analyze than unit costs (or other
are significant variations in the total cost of
common metrics, such unit costs multiplied
care among all provider categories. After
by length of stay), we have found that select-
analyzing hospital costs in the top 50 metro-
ing the most efficient providers is the most
politan statistical areas (MSAs), we estimated
effective way to ensure the delivery of lower
that simply by directing patient volume to
premiums (Exhibit 4). This approach can be
hospitals that are more efficient on a total
the difference between developing a truly
cost of care basis, savings of $18 billion
affordable value network and discovering
to $24 billion could be achieved annually.
that the expected savings erode following
Savings of this magnitude remained even
implementation.
8
McKinsey & Company Healthcare Systems and Services Practice
White Paper - Payor Value Networks — July 2013 Maximizing value in high-performance networks Exhibit 4 of 4
EXHIBIT 4 Inpatient risk-adjusted cost and total capacity by hospital (example) Risk-adjusted cost per discharge ($ thousands) Excluding the three most expensive hospitals in this sample saves ~17% on total medical costs2
Excluding the single most expensive hospital in this sample saves ~3% on total medical costs2 16.0 14.0 13.0
12.5 10.4 8.4 7.0
H1
Hospital 2
Hospital 3
H4
Hospital 5
Hospital 6
H7
Capacity (potential discharges per year1) 1Assumes
average length of stay remains unchanged and a capacity limit of 80% occupancy rate. average plan has utilization patterns that reflect the market overall, and shifted volume is allocated proportionate to excess capacity; only accounts for differences in unit cost, not utilization.
2Assumes
Source: McKinsey analysis; AHD data. Does not include unit-price reduction or performance improvement
Increasing performance-based competition among providers
networks can change provider performance.
Greater use of value networks could foster
system was initially excluded from the new
greater competition among providers. As a
networks that payors were developing in
result, the networks could accelerate growth
Massachusetts. Less than one year later, it
in the number of high-performing providers
announced a major cost-cutting initiative as
(by rewarding them with increased market
a way to improve its overall level of efficiency.
share and thus increasing their ability to
The health system cited concerns about the
expand) and create stronger incentives
number of patients enrolled in tiered net-
for lower-performing providers to improve.
works who were going to other hospitals
Indeed, evidence already exists that value
as one of the reasons for this effort.
For example, a major northeast health
9
Maximizing value in high-performance networks
Maximizing value beyond premium price
accuracy. Having a smaller, more efficient
The approaches just described can help
also give the participating providers direct,
payors reduce premium prices—obviously
positive incentives. (If the increases in the
an important goal, especially now that the
star ratings resulted in increased network
exchanges are about to be launched. Over
enrollment, for example, the providers would
the long term, however, payors may be able
gain more patients.)
network would not only make it easier for the payor to launch such an initiative, but could
to partner in other ways with the providers in their networks to create additional value.
Sharing data to better coordinate care. A
The options include:
smaller network could also make sharing clinical data and coordinating care easier.
Moving to value-based reimbursement. Value
As a result, payors and providers could work
networks can be a good way to introduce
together to identify new ways to improve
new value-based reimbursement models
care quality and make it more cost-efficient.
(e.g., bundled payments, gain sharing, risk sharing) in the commercial market. In turn,
Co-branding products. In some markets,
the value-based reimbursement models
value networks could be constructed around
can reinforce the value network’s goal of
one or two core health systems, which would
providing more affordable options to those
give all stakeholders the opportunity to
who most need or desire them.
co-brand products. Our research has shown that brand familiarity is likely to play a key
Scaling new care delivery models. Because
role in consumer choice on the exchanges.
new care delivery models, such as account-
A co-branded product would enable a
able care organizations and patient-centered
provider to get more market exposure for
medical homes, require time to ramp up to
its name and, simultaneously, offer the payor
their full savings potential, combining them
the positive association of the provider’s
with a value network can help marry near-
brand name.
term and longer-term savings, especially when efficiency is the primary criterion for
Partnering to improve access. In exchange for
network inclusion. A smaller network of more
inclusion in a value network (and the increase
efficient providers can be a more manage-
in volume that such inclusion would presum-
able forum than a broad population of pro
ably bring), some health systems may be
viders when payors are introducing new
willing to offer the network’s members pre
care delivery models as a way to ensure
ferential access or other special services
more affordable products.
(e.g., shuttles, dedicated private rooms, same-day appointments).
Developing additional quality initiatives with providers. A smaller network of more efficient providers may also lend itself to the intro
...
duction of other win-win initiatives, such as
Many payors now have experience develop-
efforts to increase star ratings or coding
ing value networks. However, in the rush to
10
McKinsey & Company Healthcare Systems and Services Practice
meet external timelines, some payors may not have optimized their network configur ation or approach. As they refine their valuenetwork strategies moving forward, these payors would benefit from having a deeper understanding of local market dynamics, as well as stronger analytic tools to calculate each provider’s total cost of care (probably the best way to capture value). In addition, payors must consider how they can work more closely with the providers in their networks to create additional value. For all
David Knott (
[email protected]), a director in McKinsey’s New York office, is co-leader of the Center for US Health Reform. Tom Latkovic (
[email protected]), a director in its Cleveland office, leads its healthcare value work in the Americas. David Nuzum (
[email protected]), is a partner in the Washington, DC, office. Jessica Ogden (jessica_
[email protected]), is an associate principal in the New Jersey office. Shubham Singhal (
[email protected]), a director in the Detroit office, is head of the Health Systems and Services Practice in the Americas.
partners in the networks, the goal should be
Acknowledgments
to improve the efficiency of care delivery and
The authors would like to thank Noam Bauman, Manish Chopra, Jenny Cordina, Amy Fahrenkopf, Jennifer Meyer, Saumya Sutaria, and Frances Wilson for their help in developing the framework for thinking about value networks described in this paper. That team has written a companion article on the implications of value networks for providers (“Winning strategies for participating in narrownetwork exchange offerings”).
enable affordable, high-quality care.
11
Maximizing value in high-performance networks
Appendix: About the research and analysis This article leverages proprietary research
points, network designs, and availability of
and analysis that McKinsey has conducted
dental care or other additional services) on
over the past 18 months. This appendix
consumer buying preferences and choices
describes the major tools and data sources we used.
• See what types of consumers purchased their products, as well as the types that
McKinsey’s Consumer Exchange Simulation With this tool, users (typically, payors) design
preferred competitors’ products • Estimate how their product offerings would
a suite of insurance products that can then
fare in terms of revenue, margin, medical
be sold on a simulated online exchange.
loss ratio, and market share in a real market
Consumers browse the exchange, which high lights information on premiums, deductibles,
• Understand local market dynamics, compe
coverage tiers, and other key product attri-
titive issues, and the effect of subsidies on
butes, before making a selection. As of the
insurance choices
end of 2012, nearly 150,000 consumers across the United States had participated in simu
The “real” consumer feedback gives users
lations. On average, each consumer needs
unique insights into consumer preferences
about 25 minutes to complete the process.
and what their behavior on the exchanges is likely to be, information that is not available
The first round of simulation requires about
through any other source.
five weeks and typically involves some 4,000 local consumers between the ages of 18 and
Several payors have already used the
64, who have incomes above 133 percent of
McKinsey Consumer Exchange Simulation
the federal poverty level. An additional round
to support product design, off-exchange
can be conducted for users that want to test
strategies, and strategies for handling the
detailed product configurations and trade-offs.
transition of existing members from employer-sponsored insurance to individual plans.
The exchange simulation collects a wide cipating consumers, as well as information
McKinsey’s annual Consumer Health Insights (CHI) survey
on their current coverage, health status, and
This unique survey provides information
prior purchase behavior. Thus, the simulation
on the opinions, preferences, and behaviors
allows users to:
of more than 14,500 consumers, as well
range of demographic data about the parti
as the environmental factors that influence • Assess the impact of different product attributes (including brand name, price
their healthcare choices. The survey also enables insights into the current market
12
Payor Value Networks July 2013
environment and can be used to make
tegies, consumer segmentation, consumer
predictions about the choices and trade-
targeting, network configuration design, and
offs consumers are likely to make in the
assessment of new channel opportunities.
post-reform environment. The CHI collects descriptive information
McKinsey Predictive Agent-based Coverage Tool (MPACT)
on all individuals who participate in the
MPACT is a micro-simulation model that uses
survey and their households. It also assess-
a comprehensive set of inputs and a distinc-
es shopping behaviors; attitudes regarding
tive approach to modeling consumer and
health, healthcare, and the purchase and
employer behavior to project how health
use of healthcare services; awareness of
insurance coverage may change post-reform.
health reform; opinions about shopping
MPACT contains 300 million “agents” repre-
for individual health insurance and using
senting all residents of the United States.
an insurance exchange; preferences for
Each agent is characterized by his or her
specific plan designs (including trade-offs
county of residence, type of insurance
among coverage features, such as benefits,
coverage, and eight demographic variables.
network, ancillaries, service options, cost
Over the course of the micro-simulation,
sharing, brand, and price); employee per
agents in each geo-demographic segment
ceptions of the employer’s role in healthcare
make health insurance purchasing decisions
coverage; attitudes about a broad range
depending on their eligibility, prior purchasing
of related supplemental insurance products;
behavior, demographics (including health
opinions, use, and loyalty levels regarding
risk status), subsidy eligibility, penalty impact,
healthcare providers; and attitudes and
and other factors.
behaviors regarding pharmaceuticals and pharmacies.
Provider Reform Impact and Stress-test Model (PRISM)
We supplement the information from the
McKinsey’s PRISM analytic takes hospital
CHI with data from other sources, such
financials, MPACT county-level covered
as information on a consumer’s estimated
lives projections, the Firm’s national hospital
lifetime value to a payor, consumer behavior,
operational benchmarking database, and
and marketplace conditions. This combina-
legislated changes to project hospital perfor-
tion provides a holistic view of healthcare
mance market-by-market. “Add-on” modules
consumers that is not available through
include projections of financial impact and
other sources.
utilization of services at the level of clinical service lines (e.g., cardiology, orthopedics),
We have used CHI data in a range of cus
a bad-debt modeler, and a rapid outside-in
tomized analyses that address both current
module, that takes outside-in data to project
and post-reform healthcare issues. We
the impact of reform on hospital economics.
expect that payors and others will use the
PRISM has built-in flexibility to model a
information primarily in applications that
range of scenarios, based on reform- and
assist with product design, marketing stra
non-reform-related factors.
Contact for distribution: Karen Hardin Phone: +1 (212) 497-5660. Email:
[email protected] Copyright © 2013 McKinsey & Company CONFIDENTIAL AND PROPRIETARY Any use of this material without specific permission of McKinsey & Company is strictly prohibited. www.mckinsey.com/client_service/healthcare_systems_and_services