HMGT 435 WEEK 7 DISC 7

4

Comparative Effectiveness Research

Provide a response to 
TWO of the questions below by Saturday, then provide a response to at least TWO of your peers by Tuesday: 

· Include the two questions that you selected to discuss at the top of your initial posting.

· What kinds of treatments will comparative effectiveness research compare?

· Should comparative effectiveness research include measures of cost?

· What are the concerns if comparative effectiveness findings are used to make coverage decisions?

· Will comparative effectiveness research save money for the health care system?

· What are the pros and cons of using clinical trials in comparative effectiveness research?

· What are the pros and cons of using medical claims data in comparative effectiveness research?

APA Requirements -Include Scholarly Evidence: Include at least TWO APA formatted references with correlating in-text citations. 

CHAPTER

293

18BEHAVIORAL ECONOMICS

Learning Objectives

After reading this chapter, students will be able to

• explain why rational decision making has its limits,
• describe some ways that bounded rationality affects decision making,

and
• identify several ways to use behavioral economics to improve decision

making.

Key Concepts

• Brainpower and time are scarce resources, so decision shortcuts make
sense.

• Some shortcuts result in poor decisions.
• Some decisions appear to reveal inconsistent preferences.
• Status quo bias means that some people tend to avoid even beneficial

changes.
• Overconfidence often leads to poor decisions.
• Problematic shortcuts include availability, anchoring, confirmation, and

hindsight bias.
• Awareness of framing bias is especially important in management.
• Changes in how choices are set up can improve decision making.

18.1 Introduction

Standard economic models start with assumptions that are not really true.
These assumptions include the notions that decision makers are always ratio-
nal, have unlimited willpower, and are concerned only about themselves.
These assumptions were previously viewed as harmless simplifications, but
researchers have demonstrated that being more realistic could be important
in management and policy. For example, cash bonuses may reduce work

Lee.indd 293 1/2/19 3:15 PM

C
o
p
y
r
i
g
h
t

2
0
1
9
.

H
e
a
l
t
h

A
d
m
i
n
i
s
t
r
a
t
i
o
n

P
r
e
s
s
.

A
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d
.

M
a
y

n
o
t

b
e

r
e
p
r
o
d
u
c
e
d

i
n

a
n
y

f
o
r
m

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n

f
r
o
m

t
h
e

p
u
b
l
i
s
h
e
r
,

e
x
c
e
p
t

f
a
i
r

u
s
e
s

p
e
r
m
i
t
t
e
d

u
n
d
e
r

U
.
S
.

o
r

a
p
p
l
i
c
a
b
l
e

c
o
p
y
r
i
g
h
t

l
a
w
.

EBSCO Publishing : eBook Collection (EBSCOhost) – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS
AN: 2144510 ; Robert Lee.; Economics for Healthcare Managers, Fourth Edition
Account: s4264928.main.eds

Economics for Healthcare Managers294

effort (especially if the work is intrinsically interesting or important), but
symbolic payments (e.g., praise) tend to increase work effort (Bareket-
Bojmel, Hochman, and Ariely 2014). For a purely rational worker, that find-
ing would not make sense. Surely praise coupled with cash would be a more
powerful motivator than praise alone. Economics that drops the assumptions
of complete rationality, complete willpower, and complete selfishness is called
behavioral economics.

Behavioral economics addresses the choices that individuals make
when they use shortcuts and rules of thumb in decision making. Our brain-
power and our time are scarce resources, so it makes sense to use rules of
thumb in making decisions. Unfortunately, these shortcuts sometimes result
in poor decisions.

18.2 Inconsistent Preferences

A standard assumption in economics is that consumers make reasonable
forecasts about what they will do in the future and make plans on that basis.
Behavioral economics notes, to the contrary, that many people appear to have
inconsistent preferences. A classic example is the tendency to procrastinate.
For example, we may conclude that the cost of exercising is more than offset
by its benefits, especially if we commit to starting exercising next week. But
when next week arrives, we do not want to work out; we want to put it off for
another week. Last week the costs were in the future; this week they will be
realized right now. Decisions that I make today may conflict with decisions
that I make next week, even though nothing has changed.

This inconsistency appears to involve rather odd patterns of discount-
ing future benefits and costs (Rice 2013). For example, if you regard being
paid $988 today as being just as good as being paid $1,000 in three months,
your personal discount rate is less than 5 percent per year.1 Would you prefer
getting $790 now to getting $1,000 in three months? If so, you are acting as
though your discount rate is more than 150 percent per year. A discount rate
of more than 150 percent per year seems pretty high, but the real anomaly is
that people sometimes use 5 percent and sometimes use 150 percent or more
for seemingly similar transactions.

A standard assumption is that people will use the same discount rate
for short-term financial gains, long-term financial gains, short-term financial
costs, and long-term financial costs because someone could make money
by exploiting discount rate variations. But many people discount the future
heavily and treat short delays much differently from longer delays. For
example, would you be willing to pay an annual rate of more than 300 per-
cent for a $200 three-week loan? More than 18 million taxpayers thought

behavioral
economics
A field of study
that integrates
psychology and
economics.

discounting
Adjusting the value
of future costs and
benefits to reflect
the willingness
of consumers
to trade current
consumption
for future
consumption.
(Usually future
values are
discounted by
1/(1 + r)

n
, with r

being the discount
rate and n being
the number of
periods in the
future when the
cost or benefit will
be realized.)

Lee.indd 294 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 295

this proposition was a good deal in 2011, when they signed up for refund
anticipation checks that allowed them to pay their tax preparation fees out
of their tax refunds (Wu, Fox, and Feltner 2013). The fee for this privilege
was typically $30 or more. Most people who agreed to refund anticipation
checks had very low incomes (so coming up with $200 to pay a tax prepara-
tion fee would be a problem) and probably were not financially sophisticated
(given that a number of ways to have a simple return filled out cost much
less than $200).

Encouraging Employees and Patients
to Be Active

Many struggle to change health-related behaviors. One reason is that
people seeking to lose weight, increase exercise, or stop smoking
act in a time-inconsistent manner. For example, someone joins a gym
but does not go. These inconsistencies not only affect the individual’s
health but increase health insurance costs because of poor health. As
a result, firms, insurers, policymakers, and health professionals are
exploring using financial incentives to change health behaviors (Royer,
Stehr, and Sydnor 2015). Using financial incentives to change behav-
iors has two potential problems. First, participants may just be paid
for doing what they planned to do anyway (i.e., people who go to the
gym three times per week would have done so without the incentive).
Second, participants may revert to their old behaviors when the incen-
tive ends.

Royer, Stehr, and Sydnor (2015) tried two approaches with employ-
ees at a large company. Randomly selected employees were paid $10
per visit to their company’s on-site exercise facility (for up to three
visits per week). After a month, half the group was offered the chance
to fund a commitment contract. This contract allowed participants to
make a pledge that they would continue to use the gym for the next
two months. Employees who kept their pledges got the money back.
For employees who did not keep their pledges, the firm donated the
money to charity. Visits to the gym fell after the incentives ended, but
they fell by less for employees who made pledges.

An alternative strategy is to give a “nudge.” Martin and colleagues
(2015) gave randomly chosen patients wearable activity trackers
that used Bluetooth to connect with their smartphones. The activity

Case 18.1

(continued)

Lee.indd 295 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers296

18.3 Risk Preferences

Why do people smoke or drive without seat belts? That these behaviors are
risky is not exactly news. One could argue that many smokers are addicted,
but that argument just pushes the question back a step. Why do people
start smoking if they know that cigarettes are addictive and that smoking is
dangerous? One possibility is that people who make risky choices like risk.
Another is that they misunderstand the risks they are taking. For example,
many people appear to underestimate health risks, and this underestimation
is a factor in their decision not to buy insurance. Another way to describe
underestimation of risk is to say that people are overconfident (as we discuss
further in section 18.4). Whether we should treat this choice as the result of
overconfidence, bad information about risk, or difficulty in understanding
the meaning of risks does not matter too much. Any of these will lead to
poor decisions.

Some evidence links risk preferences to risky behavior. (Recall from
chapter 4 that risk seekers seek more variable outcomes and risk-averse people

trackers connected to a smart texting system.
Physicians wrote the text-message content, which
mentioned the patient’s physician by name. Com-

bining smart texts with activity tracking increased physical activity
the most. Compared with patients that did not receive texts, nearly
twice as many patients that received texts met the goal of 10,000
steps per day.

Discussion Questions
• Why do people act in a time-inconsistent manner?

• Have you ever acted in a time-inconsistent manner? Why?

• Can you find examples of firms incentivizing workers?

• Can you find examples of insurers incentivizing beneficiaries?

• Can you find examples of providers incentivizing patients?

• How could you avoid paying people to do what they were going to
do anyway?

• How could you reduce backsliding?

• Why did making a pledge increase gym use?

• Why would getting a “nudge” increase exercise?

• Can you find other examples of nudges?

Case 18.1
(continued)

Lee.indd 296 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 297

seek less variable outcomes. Risk seekers seldom buy insurance. Risk-averse
people will buy insurance if the premium is not too much larger than the
expected loss.) For example, Shults and colleagues (2016) found that teens
who did not frequently use seat belts were more likely to be smokers and
drinkers.

Misunderstanding the dangers of risky behavior and the likelihood of
those dangers is a major problem for younger people. Aversion to risk typi-
cally increases with age. Few children are risk averse, a slightly larger share
of adolescents are risk averse, and most adults are risk averse (Romer, Reyna,
and Satterthwaite 2017). Typically, someone who is risk averse tends to dis-
count the future less than someone who is risk seeking, so these two tenden-
cies reinforce each other (Jusot and Khlat 2013).

Not surprisingly, most people who are addicted to cigarettes began
smoking as adolescents (Barlow et al. 2017). Their willingness to accept risk
was high, their concern about the future was low, and their ability to imagine
the consequences of becoming addicted was limited.

18.4 Incorrect Beliefs

Drivers of all ages claim to be more skillful than average (Horswill et al.
2017). But does this overconfidence matter? It does because overconfident
drivers are more likely to use their cell phones while driving, which signifi-
cantly increases the risk of an accident (Engelberg et al. 2015).

More broadly, overconfident decision makers are likely to make bad
choices. They are likely to overestimate their chances of success and apt to
attribute failures to bad luck (hence not learning from them). For example,
the fact that companies often lose money when they buy other companies is
common knowledge. Acquiring a company requires a bid above its current
market valuation, and its current market valuation is as likely to be too high as
it is to be too low. So, it takes a confident management team—one convinced
of their skill and of unrecognized synergies—to buy another company. In
many cases this confidence amounts to overconfidence and the acquisition is
unprofitable (Malmendier and Tate 2015). Overconfident CEOs often make
money-losing acquisitions.

Several cognitive traps feed into overconfidence:

• Availability bias
• Anchoring bias
• Confirmation bias
• Hindsight bias

Lee.indd 297 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers298

We will discuss each of these in turn.
Availability bias can occur because certain outcomes are overly easy

to imagine or overly hard to imagine. For example, if you run a public health
agency, which threat to life should be your top priority, tornadoes or asthma?
If you were asked this question right after reading about a deadly tornado,
you might have said tornadoes. The news reports made them easy to remem-
ber. In fact, the two threats are not even close. Between 2015 and 2017,
tornadoes killed an average of only 30 people each year (National Weather
Service 2018). More than 3,500 people die from asthma each year, and quite
a few of these deaths are preventable (National Center for Health Statistics
2017). If you have never known anyone who died as a result of asthma, you
might have a difficult time imagining asthma as a cause of death and may pay
too little attention to it.

Anchoring bias occurs when some initial estimate, even if it is not
based on evidence or is simply wrong, affects future discussions. In a strategy
discussion about whether to add a long-term care facility to a system, one of
the board members says, “I hope the return on equity is better than the 5
percent that home health care firms earn.” That comment is not really rel-
evant because long-term care and home health care are fairly distinct markets.
True or not, the comment is likely to influence the subsequent discussion.

Irrelevant information can influence decision making. If a job candi-
date starts by mentioning a desired salary of $150,000, the candidate will
probably get a higher offer than if the candidate started by mentioning a
current salary of $85,000. Neither of these numbers may fall within the pay
range for the job in question, but mentioning the $150,000 tends to anchor
the discussion.

Even experienced professionals can be affected by anchoring. For
example, a young girl with a three-week history of weight loss, diffuse
abdominal pain, and fever came to the emergency department (Festa, Park,
and Schwenk 2016). Although the evidence was ambiguous, she was admit-
ted with a preliminary diagnosis of cat scratch fever. Further tests found no
signs of infection, and the patient underwent an MRI (magnetic resonance
imaging) scan and a liver biopsy. Finally, on day 13 of her hospitalization,
hints of intestinal inflammation were found and a colonoscopy confirmed
a diagnosis of Crohn’s disease. In short, the girl received a great deal of
low-value care, largely because the medical team stuck with the diagnosis of
cat scratch fever despite the lack of evidence supporting it (Festa, Park, and
Schwenk 2016).

Confirmation bias occurs when we filter evidence to prove that our
conclusions are right. How did you react when a political candidate that you
support said something stupid? Most of us will offer an example of the oppo-
nent’s failings rather than switch candidates.

availability bias
A cognitive trap
that occurs when
some facts are
overly easy or
overly hard to
recall.

anchoring bias
A cognitive trap
that occurs when
an irrelevant
fact influences a
decision.

confirmation bias
The tendency
to focus on
information that
supports one’s
beliefs.

Lee.indd 298 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 299

A management example of confirmation bias can be found in the
hiring process. Suppose you interview several people, and Ms. Jones seems
to stand out. You are confident that she is the best choice. You call several
references, they say mostly good things about her, and those are the com-
ments that you include in your notes. Ms. Jones turns out to be a disaster.
You used the interview to support your positive impression, not to look for
warning signs. You did not follow up when a reference said, “Well, she wasn’t
here that long,” and another said, “She was only in my unit for about three
months.”

Hindsight bias occurs when you feel that you “knew it all along,”
that is, when you believe that you made a prediction that you did not. This
bias creates two decision traps. First, your overconfidence may grow. Second,
you have no incentive to explore why your predictions were faulty. Neither of
these bodes well for future decisions.

Hindsight bias is widespread, having been documented in diverse situ-
ations including labor disputes, medical diagnoses, managerial decisions, and
public policy (Chelley-Steeley, Kluger, and Steeley 2015). Hindsight bias has
serious consequences because it impairs performance. For example, research-
ers have found that investment bankers who earned the least had the largest
hindsight bias (Merkle 2017). Hindsight bias also makes effective investiga-
tions of accidents and near misses difficult, leaving future patients at risk
because no fundamental changes are made (Zwaan et al. 2017).

18.5 Representativeness and the Law of Small Numbers

To assess a possible merger with another practice, you interview six CEOs
from practices that went through mergers. After you complete the interviews,
you notice that the three CEOs from the successful mergers were accountants
and the three CEOs from the failed mergers were physicians. What should
you infer from that?

You should infer nothing. Your sample is too small and may be biased.
If you looked at a larger, more representative sample, you might find any pat-
tern. For example, you might find that merged practices led by accountants
were more likely to fail. Nonetheless, you may be tempted to think that hav-
ing an accountant as the CEO is important.

Several factors are at work here. First, humans are prone to see pat-
terns even if no pattern exists. We are apt to think that our experience with a
small number of people will be typical of the whole group. This tendency is
called representativeness bias (Saposnik et al. 2016). We are also apt to for-
get that statistics based on small numbers can be misleading. This tendency
is the law of small numbers bias.

hindsight bias
The tendency to
overstate how
predictable an
outcome was
beforehand.

representative-
ness bias
The tendency to
overstate how
typical a small
sample is.

law of small
numbers bias
Generalizations
based on small
samples.

Lee.indd 299 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers300

The problem is greatest when our own experience suggests a conclu-
sion. We easily dismiss colleagues’ stories as being mere anecdotes. Our sto-
ries seem different. They feel meaningful to us. We have no trouble saying,
“The plural of anecdote is not evidence,” unless the anecdote is ours. Our
stories seem compelling.

18.6 Inconsistent Decision Making: Framing

Real-life choices appear to be affected by how they are presented. In fact,
framing appears to be one of the strongest decision-making biases. Framing is
especially relevant to health decisions because the stakes are high and because
older adults (who are more likely to have to make health decisions) appear
more likely to use shortcuts that cause framing bias (Saposnik et al. 2016).

A standard way of illustrating framing bias is via a treatment choice
problem. Treatment 1 is guaranteed to save 200 of 1,000 people with a fatal
disease. Treatment 2 offers a 20 percent chance of saving 1,000 lives and an
80 percent chance of saving no one. Which do you prefer? Most people prefer
treatment 1 because it seems less risky.

Now consider another scenario. If you choose treatment 3, 800 of
1,000 people with a fatal disease will die. With treatment 4, you have an 80
percent chance that everyone will die and a 20 percent chance that no one
will die. Which do you prefer? Most people choose treatment 4.

The only difference between these two scenarios is that the first is
framed in terms of how many people live and the second is framed in terms
of how many die. They are otherwise identical, yet choices typically dif-
fer. By changing the emotional context of a decision, framing can change
choices.

Framing can take several forms. It can describe the attributes of
choices in different ways, describe the outcomes of choices in different ways,
and describe the risks of choices in different ways. For example, people tend
to prefer a choice when its attributes are presented in positive terms (Nanay
2016). Thus, consumers are more apt to choose a hospital that stresses its
high patient satisfaction (a positive attribute frame) rather than its low mor-
tality rates (a negative attribute frame). An example of goal framing would be
to describe the effect of a new strategy as a gain in market share (a positive
goal frame) or as avoiding stagnation (a negative goal frame). Most people
are influenced by loss aversion, meaning that they worry more about avoid-
ing losses than they do about realizing gains. As a result, people may respond
more to negative goal frames. The treatment choice example given earlier in
this section illustrates risk framing. The same problem can be presented in
terms of lives saved (a positive risk frame) or in terms of deaths prevented (a

framing bias
The effect of
presenting the
same data in
different ways.

loss aversion
A focus on
avoiding losses
rather than
maximizing gains.

Lee.indd 300 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 301

negative risk frame). People tend to be more willing to accept risk to avoid
negative outcomes than they are to gain positive outcomes.

The importance of framing appears to vary. Some decision makers
appear to be immune to framing, with decision makers with the best math-
ematics skills the least likely to be affected. In addition, goal framing appears
to have smaller effects than attribute or risk framing (Harrington and Kerr
2017). The challenge for managers is determining when framing will matter
and when it will not.

A number of countries rely on consumer demand to limit medical
costs. An important mechanism is the willingness of consumers to switch
to less expensive health insurance plans, which pressures insurers to offer
low-cost plans and pressures providers to reduce what they charge for care.
However, consumers often find insurance choices daunting, which may result
in reluctance to switch plans. This reluctance dilutes the effects of competi-
tion on costs. For example, in Switzerland (which has an insurance system
similar to the Affordable Care Act) rates of switching between insurers were
low, even though prices varied among comparable plans (Boonen, Laske-
Aldershof, and Schut 2016).

Why were switching rates so low? Behavioral economics offers several
reasons. First, a significant status quo bias is at work. Consumers are often
reluctant to make changes if they do not have to. Second, too many choices
can stop consumers from making any choice. This problem is called decision
overload. Because the average Swiss consumer had 56 plans to choose from,
decision overload appears to have been relevant. Third, consumers tend to
worry about making what turns out to be a bad choice. One way to avoid
regret about a choice is to avoid making a choice. Fourth, consumers appear
to give more weight to avoiding losses than to realizing gains. This loss aver-
sion tends to inhibit making changes.

status quo bias
The tendency not
to change, even
when it would be
advantageous to
do so.

decision overload
Poorer decision
making that occurs
as choices become
more complex.

Children’s Health Insurance

More than two-thirds of the millions of children
without health insurance appear to be eligible for

Medicaid or the Children’s Health Insurance Program (Kenney et al.
2015). For many of these children, health insurance would be free. Not
accepting free health insurance makes sense in standard economics
only if you believe that the hassles of signing up for these programs
outweigh their considerable benefits, but behavioral economics notes

Case 18.2

(continued)

Lee.indd 301 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers302

several reasons for this pattern. First, parents
may focus on the up-front hassles and give much
less emphasis to the future benefits. That is, the

parents may heavily discount the future benefits. Second, the well-
known problem of procrastination means that tomorrow or next week
is always a better time than today to go to the trouble of enrolling a
child. Third, we know that many decision makers have trouble with
probabilities, meaning that the parents of these uninsured children
make poor assessments of the chance that their child will become seri-
ously ill or that better access to medical care will be important.

Between 1984 and 2009, a series of reforms sought to streamline
and simplify enrollment in Medicaid and the Children’s Health Insur-
ance Program. These reforms allowed states to permit continuous
enrollment, to eliminate face-to-face interviews, to simplify verifica-
tion procedures, to grant temporary eligibility, and to use eligibility for
other programs (e.g., the Supplemental Nutrition Assistance Program)
to determine eligibility.

Advances in information technology made these reforms pos-
sible, and the Affordable Care Act financially supported upgrades
to outdated Medicaid eligibility systems, which are integrated with
or connected to health insurance marketplaces in every state. As of
January 2017, 39 states could make Medicaid eligibility determina-
tions within 24 hours, and in 28 states, applicants could apply using
mobile devices (Brooks et al. 2017). Not surprisingly, the increased
convenience of these new systems has boosted enrollment in Med-
icaid and the Children’s Health Insurance Program. For example, Ala-
bama removed asset tests for children, stopped requirements for an
in-person interview, made eligibility last for a full year, and simplified
the application process in other ways. As a result, the share of eligible
children with coverage rose from 91 percent in 2008 to 95 percent in
2015 (Georgetown University Center for Children 2017). Much of this
growth occurred after implementation of the Affordable Care Act, but
not because many children got coverage via marketplace plans. Less
than 1 percent of the eligible children got their coverage this way.

Alabama’s enhancements incorporate ideas from behavioral
economics. They make enrollment easier, rather than emphasizing
traditional outreach strategies or price reductions. Unfortunately,
many children who are eligible for health insurance subsidies remain

Case 18.2
(continued)

(continued)

Lee.indd 302 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 303

18.7 Conclusion

People use shortcuts when they make hard or emotionally charged decisions.
In other words, patients, clinicians, and managers use shortcuts when they
buy insurance, when they make medical decisions, when they make strate-
gic decisions at work, and when they hire and fire employees. Shortcuts are
common.

Sometimes, unfortunately, shortcuts lead to poor choices. Patients
may choose insurance plans that expose them to significant financial risks.
Clinicians may recommend problematic treatment plans. And managers may

uninsured. Rice (2013) suggests that parents’
failure to understand the risks that their children
face, excessive discounting of the future, or lim-

ited grasp of how insurance works might explain this. An experiment
(Flores et al. 2016) suggests that knowledge may be a major issue.
The experiment funded parent mentors (experienced parents with a
child covered by Medicaid or the Children’s Health Insurance Program),
who received two days of training and then helped families apply for
insurance, find providers, and access social services. The result was
that more children got coverage, access to medical and dental care
improved, out-of-pocket costs fell, parental satisfaction increased, and
quality of care improved.

Discussion Questions
• Why are children who are eligible for free coverage uninsured?

• What behavioral economics approaches would further increase
coverage?

• Why are adults who are eligible for low-cost coverage uninsured?

• What behavioral economics approaches would further increase
coverage?

• How does status quo bias affect health insurance decisions?

• How does loss aversion affect health insurance decisions?

• How does decision overload affect health insurance decisions?

• How might insurance decisions be reframed to increase enrollment?

• How could enrollment in health insurance for children be further
simplified?

• How could health insurance be simplified overall?

Case 18.2
(continued)

Lee.indd 303 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers304

make business decisions that harm patients or their organizations. The stakes
can be high.

What can managers do to limit bad decision making due to shortcuts?
Fortunately, a number of strategies are available:

• Look hard for evidence that you are wrong.
• Appoint someone to tear apart your analyses.
• Reward those who express honest disagreement.
• Seek out the opinions of people who disagree with you, and listen

carefully.
• Try to reframe problems to view them from different perspectives.
• Talk about your feelings to see if they are leading you astray.
• Postpone committing to strategies as long as you can.
• Make sure that a review process and an exit strategy are part of

decision making.
• Be aware that your decision making can lead to mistakes.

These steps will not shield you from making errors. They may help you make
fewer mistakes, though.

Exercises

18.1 You will receive a $10,000 insurance payment in two months. If you
are willing to pay for expedited handling, you can be paid in one
month. Would you be willing to pay $50? $100? $200? More?

18.2 You will receive a $20,000 insurance payment in 12 months. If
you are willing to accept a reduced payment, you can be paid in
11 months. Would you be willing to accept $19,800? $19,500?
$19,000? Less?

18.3 What annual interest rate is implied by your answer to exercise
18.1? You calculate this rate by dividing $10,000 by the difference
of $10,000 and the amount you are willing to pay for expedited
handling, then taking the result to the twelfth power and
subtracting 1. For example, if you were willing to pay $100, the
result would be ($10,000/$9,900)12 − 1 = 0.1281781, or 12
percent.

18.4 What annual interest rate is implied by your answer to exercise
18.2? Is it the same as the rate in exercise 18.3? Why would this
comparison matter?

Lee.indd 304 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 305

18.5 How are exercises 18.1 and 18.2 different? How are your answers
to them different?

18.6 How could you use behavioral economics to increase the number of
insured employees in your firm?

18.7 How likely is someone aged 25 to 44 to have an emergency
department visit? What is the probability of having two visits? What
is a typical charge for an emergency department visit? On the basis
of your answers, would someone aged 25 to 44 be willing to buy
coverage for emergency department care (with a $50 copayment) if
it cost $250 per year?

18.8 What was your forecast of emergency department spending in
exercise 18.7? Your forecast should equal the probability of one
emergency department visit times the typical charge plus the
probability of two visits times the typical charge.

18.9 A town has two hospitals. One averages 30 births per day; the other
averages 15. Overall, half of the babies are boys, but some days
more than 60 percent of the babies are boys. Is either hospital likely
to have a greater number of days with a high proportion of boys?

18.10 Eighty percent of the participants at a meeting are physicians. The
rest are nurse practitioners. Your neighbor Amy is there. She is 40,
married, and highly motivated. Colleagues have told you that Amy
is extremely capable and promises to be very successful. What is the
probability that Amy is a physician?

18.11 Will you be in the top half of your class or the bottom? What
proportion of your classmates will forecast that they will be in the
top half? What implications does this scenario have for decision
making?

18.12 You have finished interviewing candidates for an assistant director
position. One of them stands out as the best candidate to you. You
know that this view sets you up for confirmation bias as you check
references. What steps can you take to prevent this bias?

18.13 Your vice president is an accountant and believes that accountants
make the best practice managers. One of the three finalists for a
practice management role has an accounting background. Everyone
on the search team has ranked this candidate lowest of the finalists.
You fear that your vice president will tend to selectively read the
team’s recommendations and lean toward hiring this person. What
can you do to offset this potential confirmation bias?

18.14 Thirty-four percent of the employees in your health system are
obese, and 16 percent of their children are obese as well. Obese
employees are less productive, have higher medical costs, and miss

Lee.indd 305 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers306

more work. Employees with obese children also miss more work,
so persuading employees and their families to lose weight looks
like a good investment for the system. In fact, effective, clearly
cost-effective interventions are available to reduce obesity, and you
offer them to your employees and their families. You have recently
begun to make weight-loss interventions available for free, but
only 1 percent of your employees have signed up for them. What
behavioral economics tools can you use to help your employees lose
weight?

Note

1. If getting $988 is as good as getting $1,000 in three months, your
discount rate is 4.95 percent per year. Dividing $1,000 by $988 gives
a three-month discount factor of 1.012145749. Taking this result to
the fourth power (to convert it to an annual rate) gives 1.04948. It is
customary to subtract 1.00 from this discount factor and express the
results in percentage terms. Doing so gives 4.95 percent.

References

Bareket-Bojmel, L., G. Hochman, and D. Ariely. 2014. “It’s (Not) All About the
Jacksons: Testing Different Types of Short-Term Bonuses in the Field.” Jour-
nal of Management 43 (2): 534–54.

Barlow, P., M. McKee, A. Reeves, G. Galea, and D. Stuckler. 2017. “Time-
Discounting and Tobacco Smoking: A Systematic Review and Network Analy-
sis.” International Journal of Epidemiology 46 (3): 860–69.

Boonen, L. H. H. M., T. Laske-Aldershof, and F. T. Schut. 2016. “Switching Health
Insurers: The Role of Price, Quality and Consumer Information Search.”
European Journal of Health Economics 17 (3): 339–53.

Brooks, T., K. Wagnerman, S. Artiga, E. Cornachione, and P. Ubri. 2017. Medicaid
and CHIP Eligibility, Enrollment, Renewal, and Cost Sharing Policies as of Jan-
uary 2017: Findings from a 50-State Survey. Published January. http://files.
kff.org/attachment/Report-Medicaid-and-CHIP-Eligibility-as-of-Jan-2017.

Chelley-Steeley, P. L., B. D. Kluger, and J. M. Steeley. 2015. “Earnings and Hind-
sight Bias: An Experimental Study.” Economics Letters 134 (September):
130–32.

Engelberg, J. K., L. L. Hill, J. Rybar, and T. Styer. 2015. “Distracted Driving Behav-
iors Related to Cell Phone Use Among Middle-Aged Adults.” Journal of
Transport & Health 2 (3): 434–40.

Lee.indd 306 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 18: Behavioral Economics 307

Festa, N., K. T. Park, and H. Schwenk. 2016. “Anchoring Bias as a Limiting Factor
in High-Value Care: A Case of Fever of Unknown Origin in a Hospitalized
Child.” Hospital Pediatrics 6 (11): 699–701.

Flores, G., H. Lin, C. Walker, M. Lee, J. M. Currie, R. Allgeyer, M. Fierro, M. Henry,
A. Portillo, and K. Massey. 2016. “Parent Mentors and Insuring Uninsured
Children: A Randomized Controlled Trial.” Pediatrics 137 (4): e20153519.

Georgetown University Center for Children. 2017. Alabama: Snapshot of Chil-
dren’s Coverage. Accessed September 24, 2018. https://ccf.georgetown.edu
/wp-content/uploads/2017/02/Alabama-Medicaid-CHIP-new-v2 .

Harrington, N. G., and A. M. Kerr. 2017. “Rethinking Risk: Prospect Theory Appli-
cation in Health Message Framing Research.” Health Communication 32 (2):
131–41.

Horswill, M. S., M. Garth, A. Hill, and M. O. Watson. 2017. “The Effect of Per-
formance Feedback on Drivers’ Hazard Perception Ability and Self-Ratings.”
Accident Analysis & Prevention 101: 135–42.

Jusot, F., and M. Khlat. 2013. “The Role of Time and Risk Preferences in Smok-
ing Inequalities: A Population-Based Study.” Addictive Behaviors 38 (5):
2167–73.

Kenney, G. M., J. M. Haley, N. Anderson, and V. Lynch. 2015. “Children Eligible
for Medicaid or CHIP: Who Remains Uninsured, and Why?” Academic Pedi-
atrics 15 (3, suppl.): S36–S43.

Malmendier, U., and G. Tate. 2015. “Behavioral CEOs: The Role of Managerial
Overconfidence.” Journal of Economic Perspectives 29 (4): 37–60.

Martin, S. S., D. I. Feldman, R. S. Blumenthal, S. R. Jones, W. S. Post, R. A.
McKibben, E. D. Michos, C. E. Ndumele, E. V. Ratchford, J. Coresh, and M. J.
Blaha. 2015. “mActive: A Randomized Clinical Trial of an Automated mHealth
Intervention for Physical Activity Promotion.” Journal of the American Heart
Association 4 (11): e002239.

Merkle, C. 2017. “Financial Overconfidence over Time: Foresight, Hindsight, and
Insight of Investors.” Journal of Banking & Finance 84: 68–87.

Nanay, B. 2016. “The Role of Imagination in Decision-Making.” Mind & Language
31 (1): 127–43.

National Center for Health Statistics. 2017. “Asthma.” Updated March 31. www.
cdc.gov/nchs/fastats/asthma.htm.

National Weather Service. 2018. “Monthly and Annual U.S. Tornado Summaries.”
Modified January 16. www.spc.noaa.gov/climo/online/monthly/newm
.html#2017.

Rice, T. 2013. “The Behavioral Economics of Health and Health Care.” Annual
Review of Public Health 34 (1): 431–47.

Romer, D., V. F. Reyna, and T. D. Satterthwaite. 2017. “Beyond Stereotypes of
Adolescent Risk Taking: Placing the Adolescent Brain in Developmental Con-
text.” Developmental Cognitive Neuroscience 27: 19–34.

Lee.indd 307 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers308

Royer, H., M. Stehr, and J. Sydnor. 2015. “Incentives, Commitments, and Habit
Formation in Exercise: Evidence from a Field Experiment with Workers at a
Fortune-500 Company.” American Economic Journal: Applied Economics 7
(3): 51–84.

Saposnik, G., D. Redelmeier, C. C. Ruff, and P. N. Tobler. 2016. “Cognitive Biases
Associated with Medical Decisions: A Systematic Review.” BMC Medical
Informatics and Decision Making 16 (1): 138.

Shults, R. A., T. M. Haegerich, G. Bhat, and X. Zhang. 2016. “Teens and Seat Belt
Use: What Makes Them Click?” Journal of Safety Research 57: 19–25.

Wu, C. C., J. A. Fox, and T. Feltner. 2013. Something Old, Something New in Tax-
Time Financial Products: Refund Anticipation Checks and the Next Wave
of Quickie Tax Loans. National Consumer Law Center. Published Febru-
ary. https://community-wealth.org/content/something-old-something-new
-tax-time-financial-products-refund-anticipation-checks-and-next.

Zwaan, L., S. Monteiro, J. Sherbino, J. Ilgen, B. Howey, and G. Norman. 2017. “Is
Bias in the Eye of the Beholder? A Vignette Study to Assess Recognition of
Cognitive Biases in Clinical Case Workups.” BMJ Quality & Safety 26 (2):
104.

Lee.indd 308 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

CHAPTER

219

14ECONOMIC ANALYSIS OF CLINICAL AND
MANAGERIAL INTERVENTIONS

Learning Objectives

After reading this chapter, students will be able to

• identify when a cost-minimization analysis is appropriate,
• distinguish between cost–benefit analysis and cost–utility analysis,
• explain why economic evaluation is necessary in healthcare, and
• discuss the importance of comparing the best alternatives.

Key Concepts

• Analyses of interventions are designed to support decisions, not make
them.

• Comparing the most competitive alternatives is vital.
• Four types of analysis are common: cost-minimization analysis, cost-

effectiveness analysis, cost–utility analysis, and cost–benefit analysis.
• The simplest and most productive type of analysis is cost-minimization

analysis.
• Cost–benefit analysis and cost–utility analysis are potentially more

powerful, but their validity is uncertain.
• Modeling costs entails identifying the perspective involved, the

resources used, and the opportunity costs of those resources.
• Focusing on the direct costs of interventions is best.
• Modeling benefits is the most difficult part of economic evaluation of

clinical interventions.

14.1 Introduction

Until recently, economic analyses of clinical interventions were uncommon.
Healthcare decision makers had little or no incentive to assess whether

Lee.indd 219 1/2/19 3:15 PM

C
o
p
y
r
i
g
h
t

2
0
1
9
.

H
e
a
l
t
h

A
d
m
i
n
i
s
t
r
a
t
i
o
n

P
r
e
s
s
.

A
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d
.

M
a
y

n
o
t

b
e

r
e
p
r
o
d
u
c
e
d

i
n

a
n
y

f
o
r
m

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n

f
r
o
m

t
h
e

p
u
b
l
i
s
h
e
r
,

e
x
c
e
p
t

f
a
i
r

u
s
e
s

p
e
r
m
i
t
t
e
d

u
n
d
e
r

U
.
S
.

o
r

a
p
p
l
i
c
a
b
l
e

c
o
p
y
r
i
g
h
t

l
a
w
.

EBSCO Publishing : eBook Collection (EBSCOhost) – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS
AN: 2144510 ; Robert Lee.; Economics for Healthcare Managers, Fourth Edition
Account: s4264928.main.eds

Economics for Healthcare Managers220

procedures were worth their costs, or even whether those procedures could
be done more efficiently. A fee-for-service payment system tells decision
makers what procedures are worth. Practical managers in a fee-for-service
environment will not worry about genuinely balancing value and cost.

The emergence of bundled payment systems and the growth of capi-
tation have made economic analyses of clinical interventions more relevant.
In either system, getting the same outcome at lower cost directly increases
profits. In a capitated system, the options are even greater: Getting the same
outcome more cheaply still increases profits, but strategies such as increasing
prevention, self-care, or adherence to clinically effective protocols can also
have a significant payoff. In short, the value of analyzing clinical interventions
has risen sharply.

Analyses of clinical interventions ask deceptively simple questions,
such as “Are the benefits of this intervention greater than its costs?” and
“Is this intervention better than the alternatives?” Such questions are often
difficult to answer because assessing the benefits of clinical interventions
is difficult. While the second question may sound much like the first, it
is easier to answer because it does not require assigning the benefits an
explicit value.

These questions must be asked because, even in a wealthy society,
resources are limited. When individuals choose to purchase a drug or be
screened for a condition, they cannot use those resources for other purposes.
The same is true for society. If money spent on an electrocardiogram could
be used to greater benefit elsewhere, the resources should be reallocated to
those other uses. Ideally, we would like to use resources to maximum ben-
efit. Practically, we seek to avoid pure waste and interventions in which the
benefits are smaller than the costs.

Why are economic analyses of clinical interventions needed? Public
and private insurers need information on which to base coverage decisions.
Patients seldom are familiar with all the potential outcomes of therapy, their
experience may not be typical, and their perceptions of costs are distorted by
insurance. In addition, providers often need information to make the case for
a new form of treatment. Because the stakes can be high, patients and provid-
ers are reluctant to innovate without evidence.

Analyses of clinical interventions are designed to support decision
making, not to make decisions. By providing a framework for synthesizing
and understanding information, economic analyses can help decision makers
avoid bad decisions.

Four types of analysis are common. Cost-minimization analysis
(CMA), cost-effectiveness analysis (CEA), cost–utility analysis (CUA),
and cost–benefit analysis (CBA) all compare the costs and benefits of

cost-minimization
analysis
An analysis that
measures the cost
of two or more
innovations with
the same patient
outcomes.

cost-effectiveness
analysis
An analysis that
measures the cost
of an innovation per
unit of change in a
single outcome.

cost–utility
analysis
An analysis that
measures the cost
of an innovation
per quality-
adjusted life year.

cost–benefit
analysis
An analysis
that compares
the value of an
innovation with
its costs. (Value
is measured as
willingness to pay
for the innovation
or willingness
to accept
compensation to
not use it.)

Lee.indd 220 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 221

alternative interventions. All four use the same methods to measure costs,
but they use different strategies for assessing benefits.

CMA is the most useful for managers. Although it is more limited
in scope than the others, it is simpler to apply. CMA answers our second
question, “Is this intervention better than the alternatives?” Unfortunately,
it cannot answer it in every case. If the better alternative also costs more or
if the least expensive alternative does not work as well, CMA is not helpful.

CEA extends CMA somewhat. When the better strategy costs more,
CEA answers the question “What is the cost per unit of this gain?” This sim-
ple piece of information is likely to be of genuine value to managers because
it will validate strategies with a small cost per unit and negate those with a
large cost per unit. CEA does not, however, directly compare the costs and
benefits of a strategy as CUA and CBA do.

14.2 Cost Analysis

Before examining these four types of analysis in more detail, we will briefly
review the basics of cost analysis. Measuring costs involves three tasks:

1. identifying the perspective involved,
2. identifying the resources used, and
3. identifying the opportunity costs of those resources.

Costs are often poorly understood (and poorly measured), even though the
issues are seldom very complex.

14.2.1 Identifying a Cost Perspective
Identifying a cost perspective is an essential first step. Confusion about costs
usually arises because the analyst has not been clear about the perspective.
Decision makers usually respond to the costs they see, and different decision
makers typically see different portions of the cost. This notion may seem
abstract, so here is a simple example. An insurance plan (an HMO) wishes
to increase use of a generic drug in place of the brand-name equivalent. The
generic product costs $50, of which $4 is paid by the patient and $46 is paid
by the plan. The branded product costs $100, of which $5 is paid by the
patient and $95 is paid by the plan. From the plan’s perspective, switching
to the generic saves $49. From the consumer’s perspective, switching to the
generic saves $1. From the perspective of society as a whole, switching to the
generic saves $50. These different perspectives are all valid, yet they may lead
to different choices.

Lee.indd 221 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers222

Another example shows how differences in cost perspectives can lead
to different perceptions of the cost of a good or service. Suppose the same
HMO encourages use of an over-the-counter drug because the drug is not
a covered benefit. The over-the-counter product costs $10, of which $0 is
paid by the plan. The prescription product costs $15, of which $5 is paid by
the patient and $10 is paid by the plan. From the consumer’s perspective, the
switch increases costs from $5 to $10. Because consumers share the costs of
covered medications with many other beneficiaries, they will want to switch
to over-the-counter medications only if those medications are more effective
or more convenient than prescription medications. From the insurer’s per-
spective, the switch reduces costs from $10 to $0. The switch makes sense for
the insurer as long as the prescription medication is not “too much better”
than the over-the-counter medication. From the perspective of society, the
switch reduces costs from $15 to $10 and makes sense only if the over-the-
counter medication is “nearly as good” as the prescription medication.

A societal perspective on costs is usually the right perspective for two
reasons. The societal perspective recognizes all costs, no matter to whom they
accrue. Other perspectives typically fail to consider important costs, which
is seldom a good long-run strategy. Those to whom costs have been shifted
try to avoid them and try to avoid contracting with organizations that shift
costs to them.

14.2.2 Identifying Resources and Opportunity Costs
Cost equals the volume of resources used in an activity multiplied by the
opportunity cost of those resources. Keeping these two components of cost
separate is useful because either can vary. A clinical understanding of a pro-
cess helps a manager to identify the resources used in an intervention; a well-
documented clinical pathway is even more helpful.

Most of the time the opportunity cost of a resource simply equals
what you paid for it. The opportunity cost of $100 in supplies is $100. The
opportunity cost of an hour of nursing time is $27 if the total compensation
of a nurse is $27 per hour. Calculating the opportunity cost is more complex
when the cost of a resource has changed since you bought it and you would
not buy it at its current price. In these cases you have to calculate the value
of the resource in its best alternative use.

Economic theory provides a powerful tool for simplifying cost analy-
ses. It says to focus on the resources you add (or do not need) as a result of
an intervention. In other words, focus on incremental costs. This task can be
difficult but is less complex than pondering, for example, exactly what pro-
portion of the chief financial officer’s compensation should be allocated to a
triage process in the emergency room.

societal
perspective
A perspective that
considers all costs
and benefits, no
matter to whom
they accrue.

Lee.indd 222 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 223

14.2.3 Direct and Indirect Costs
Implicit in this advice is a recommendation to focus on the direct costs of
interventions, or those costs that result because an intervention has been
tried. For example, the costs of a drug and its administration are direct costs
of drug therapy. The costs of associated inpatient and outpatient care are
also direct costs. If healthcare costs associated with ineffectiveness or adverse
outcomes are present, those should be counted as well. By the same token,
costs the patient incurs as a result of undertaking the treatment are direct
costs. Added childcare, transportation, and dietary costs that result directly
from therapy should be counted from a societal cost perspective. From the
perspective of the healthcare system, however, these added costs for patients
would not be counted. (Of course, as noted earlier, a cost perspective that
ignores the effects on customers is likely to result in poor decisions.)

Most “indirect” costs represent a confusion of costs with benefits.
Healthier people typically spend more on food, recreation, entertainment,
and other joys of life, but this additional spending is not a part of the costs of
interventions that restored health. (Individuals have independently made the
judgment that this additional spending is worthwhile.) By the same token,
we should not treat a recovered patient’s future spending as a cost of the
intervention that permitted the recovery—unless, as with transplant patients’
immunosuppressive drugs, these costs are an integral part of the intervention.
That a transplant patient feels healthy enough to play tennis certainly signals
that the operation was a success, but if the overenthusiastic athlete suffers an
on-court injury, the cost of knee surgery should not be considered a cost of
the transplant.

14.3 Types of Analysis

We have identified four types of analysis: CBA, CEA, CUA, and CMA. Be
aware that mislabeling is the norm, not the exception. A “cost–benefit analy-
sis” could be anything, and the meaning of “cost-effectiveness analysis” has
changed over the years. Exhibit 14.1 shows when each type of analysis is
needed.

If deciding which strategy is best is difficult, the choice of strategy
should not matter because they all support decision making. If the options
look so similar that choosing the best one is difficult, do not do a detailed
analysis. A coin flip will suffice. Of course, when populations are large, even
small differences in cost or benefit per case can result in significant differences
from society’s perspective. However, for working managers, small differences
are not worthy of attention.

Lee.indd 223 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers224

14.4 Cost-Minimization Analysis

The simplest and most productive type of analysis is CMA, which identi-
fies the intervention with the lowest costs. As long as the intervention has
outcomes at least as good as those of the alternatives, CMA is the analysis of
choice. While CMA avoids most of the problems associated with measuring
benefits, it does not escape them entirely. The most common problem in
CMA is a lack of evidence that the least-cost option has outcomes at least as
good as the other choices.

EXHIBIT 14.1
Using Decision-

Support Tools

Steps in Cost-Minimization Analysis

1. Estimate the expected costs for each option.

2. Show that the least-cost option has outcomes at least as good as
higher-cost alternatives.

An Example of Cost-Minimization Analysis

Treatment guidelines for patients hospitalized with community-acquired
pneumonia recommend antibiotic therapy for eight days. The scientific
basis for eight days of antibiotics is limited, and some researchers
have suggested that briefer treatments may be appropriate. Because
community- acquired pneumonia is a common problem, substantial sav-
ings might be possible with briefer treatments (Scalera and File 2013).

(continued)

Incremental Effectiveness of Intervention

In
cr

em
en

ta
l C

os
t

of

In
te

rv
en

ti
on

More Same Less

More CBA, CEA, CUA CMA

Same Coin Flip

Less CMA CBA, CEA, CUA

Lee.indd 224 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 225

14.5 Cost-Effectiveness Analysis

CEA recognizes that measuring the incremental cost of improving outcomes
may be useful when a more effective intervention costs more. In at least some
cases, the incremental cost will be so high or so low that a decision can be
based on it.

In some cases, CEA is not helpful. If the cost per life year saved is
$35,000 or if the cost per injury prevented is $10,000, the answer will not
seem obvious. In these cases CBA or CUA may be needed.

Steps in Cost-Effectiveness Analysis

1. Estimate the expected costs for each option.

2. Establish how much the higher-cost option improves outcomes.

3. Calculate the cost per unit of improvement in outcome (e.g., the
cost per life year gained or the cost per infection avoided).

An Example of Cost-Effectiveness Analysis

Pregnant women should stop smoking for many reasons, but 17 per-
cent of low-income women smoke during pregnancy (Li et al. 2018).
Trying to increase quit rates, Essex and colleagues (2015) added nico-
tine replacement patches to the standard care for pregnant smokers

(continued)

Uranga and colleagues (2016) conducted a randomized controlled
trial to compare five-day treatment with longer antibiotic therapies. At
day five of treatment, patients with community-acquired pneumonia
who had significantly improved were randomly assigned to discontinue
antibiotics or complete the course of treatment prescribed by their
physician. Patients were then followed for 30 days. Shorter treatments
also led to less antimicrobial resistance, fewer adverse effects, lower
cost, and improved adherence.

(continued)

Lee.indd 225 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers226

14.6 Cost–Benefit Analysis

CBA is also relatively simple, but its validity is unknown. CBA is appropriate
when the option with the best outcomes costs more. CBA begins with a com-
parison of two or more options to find out how their costs differ, followed
by an attempt to estimate the difference in benefits directly. Two different
strategies are used for estimating benefits. One uses statistical techniques to
infer how much consumers are willing to pay to avoid risks. The other uses
surveys of the relevant population to determine whether the added benefits
are worth the cost.

Neither method’s validity has been clearly established. The funda-
mental challenge arises from concerns about consumers’ abilities to make
decisions involving small probabilities of harm. If consumers do not assess
these probabilities accurately, their life choices and their responses to surveys
will not be reliable. In addition, multiple challenges to the validity of sta-
tistical inferences are always present, and statistical estimates of benefits are
imprecise. Surveys may not give us valid measures of willingness to pay or
willingness to accept compensation. First, they ask consumers to make com-
plex assessments of services they have not yet used. Answers to hypothetical,
complex questions are suspect. Second, consumers may misrepresent their
preferences, believing they will have to pay more out of pocket if they answer
willingness-to-pay questions accurately. Therefore, even though CBA can
provide invaluable information to decision makers, its accuracy is not clear.

Two other criticisms are worth noting. Early CBA studies based estimates
of benefits on estimates of increases in labor market earnings. A few minutes of

(which consisted of behavioral support to encourage quitting and
reminder phone calls).

Adding nicotine replacement patches increased costs by $71. It
also increased quit rates by 1.8 percent. Taking into account changes
in cesarean section rates, prenatal hospital admission rates, and neo-
natal unit admission rates, the authors estimated an incremental cost
per quitter of $6,896. This estimate is unsatisfying for two reasons.
First, because treatment costs vary so much, the estimate is highly
variable. (The 95 percent confidence interval ranges from −$159,779 to
$177,446.) Second, the most substantial costs of maternal smoking are
due to the child’s increased probabilities of impairment over a lifetime.

(continued)

Lee.indd 226 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 227

reflection will reveal problems with this approach. Is improved health for retired
persons of no value? Does people’s willingness to pay out of pocket for the care
of their pets (who have no earning power) mean that changes in earnings are a
poor guide to the value of medical interventions? Earnings-based estimates of
benefits have left a legacy of skepticism of CBA among healthcare analysts. A
second complaint is that willingness to pay usually rises with income. This find-
ing is profoundly troubling to analysts who would prefer a healthcare system
that is more egalitarian than the current system in the United States. (While this
complaint is not really a criticism of CBA, it is sometimes presented as such.)

For an illustration of how CBA works, return to the example of the
switch from a branded product to a generic one. Recall that the branded
drug costs $100 and the generic drug costs $50. Uninsured consumers
would buy the branded product only if its benefits were large enough for
them to be willing to pay $100. Few consumers would be willing to pay
this much to get the branded product because branded and generic drugs
seldom differ. Current users of a branded drug, however, face both real and
perceived risks to switching, such as the risk of an allergic reaction to differ-
ent inert ingredients. Remember that from the insured consumer’s perspec-
tive, the cost differential is only $1, from $5 for the branded product to
$4 for the generic. Current users of the drug may be willing to pay $75, in
which case the marginal benefit of the branded drug will appear larger than
its marginal cost. Current users have an incentive to make sure others bear
the financial risk of higher costs. Asking people who are not current users
is also problematic. The opinion of someone who does not have a disease
the drug is intended to treat or who has not used both drugs is not likely
to hold much value.

Steps in Cost–Benefit Analysis

1. Estimate the expected incremental costs of the more expensive
option.

2. Survey consumers to find out if they would be (a) willing to
pay enough to cover the added costs of an option with better
outcomes or attributes or (b) willing to accept payment that would
be less than the cost savings of an option with worse outcomes or
attributes. Alternatively, use market data to estimate how much
consumers are willing to pay to avoid risks or willing to accept to
take on risks.

3. Compare the incremental benefits and costs.

Lee.indd 227 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers228

14.7 Cost–Utility Analysis

CUA rivals CBA as a complete comparison of alternative interventions (note
that a number of analysts do not distinguish between CEA and CUA). CUA
seeks to measure consumer values by eliciting valuations of health states. This
information is then used to “quality adjust” health gains, so that decision
makers can consider the cost per quality-adjusted life year (QALY) saved (we
will explain how QALYs are calculated later).

An Example of Cost–Benefit Analysis

Type 2 diabetes often causes workers to retire. A recent study mea-
sured willingness to pay for work accommodations for people with
type 2 diabetes (Nexo et al. 2017).

Time off with pay was by far the most highly valued option, but
being able to work part time was a close second. Interestingly, people
with type 2 diabetes systematically viewed work accommodations as
less valuable than did a matched sample of other people. People with
type 2 diabetes viewed paid time for medical visits as more valuable
than part-time work, customized work, or additional paid breaks. One
would need to calculate the costs of the options to determine which
offered the highest net benefit (benefit minus cost), but the study
clearly warned against making inferences about value based on the
opinions of individuals who have not experienced the illness.

QALYA QALYB QALYB – QALYA

Discounted

NA UA NA × UA
NB UB NB × UB

0% 3%

Year 1 outcomes 200 0.95 190.00 210 0.96 201.60 11.60 11.26

Year 2 outcomes 195 0.94 183.30 199 0.93 185.07 1.77 1.67

13.37 12.93

Cost per QALY (with a $300,000 cost difference between A and B): $22,438 $23,201

Note: N
A
and N

B
refer to the number of participants. U

A
and U

B
refer to the average utility score of participants.

EXHIBIT 14.2
A Cost–Utility
Analysis

Lee.indd 228 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 229

CUA is complex, and its validity is unknown. It is appropriate when-
ever CBA is, and at a formal level the two are essentially equivalent. At a
practical level, however, the process of calculating benefits is different. CUA
measures how alternative interventions change the health status of patients
and how patients evaluate those changes.

Exhibit 14.2 walks through the calculations for a CUA. Suppose 215
people each get treatments A and B. At the end of one year the number of
survivors differs for the two treatments (NA and NB), as does the average
utility level (UA and UB). We use these data to calculate how many additional
QALYs we get as a result of using treatment B. We then calculate the cost
per QALY if we switch to treatment B.

Four uncertainties are associated with this calculation, aside from the
usual problems of assessing the clinical effectiveness of treatments. First,
should we limit our questions to patients? Family, friends, and strangers are
sometimes willing to help patients afford care. Second, can patients answer
questions about satisfaction adequately and accurately? Third, what discount
rate should we use? While the example uses 3 percent, another rate might
give us different answers, and we do not know what the right rate is. Fourth,
assuming all other calculations are correct, at what cost per QALY should we
draw the line? At the risk of sounding unduly negative, the validity of CUA
hinges on finding satisfactory answers to these questions, which is not likely.

Unlike CMA or CBA, CUA requires that the analyst explicitly dis-
count future QALYs. A technique commonly used in banking and finance,
discounting reflects that benefits we realize far in the future are worth less
than benefits we realize now. Discounting is valid because money can earn
interest. To pay a bill that will come due in the future, one can set aside a
smaller amount today. For example, if we invest $100 at an interest rate of
7 percent, we will have $160.58 at the end of ten years. We can reverse this
calculation to show that the value of a guaranteed payment of $160.58 that
we will get in ten years is $100.

As long as the interest rate is fixed, discounting is easy to figure on a
spreadsheet. A single formula, PV × (1 + r)n = FV, lets us do all the necessary
calculations. In this formula, PV refers to the present value of future costs or
benefits, or the amount we are investing today; r refers to the interest rate;
n refers to the number of time periods involved; and FV refers to the future
value of future costs or benefits, or the amount we will have at the end of the
investment period. We use the same formula to calculate the present value
of future costs and benefits. The formula becomes PV = FV/(1 + r)

n
. If we

knew the size and timing of an intervention’s costs and benefits and the right
discount rate, calculating the present value of the QALYs associated with it
would be a simple matter. In fact, we do not know the right discount rate
and are not sure that the discount rate is constant for a given individual, let

discounting
Adjusting the value
of future costs and
benefits to reflect
the willingness
of consumers
to trade current
consumption
for future
consumption.
(Usually future
values are
discounted by
1/(1 + r)

n
, with r

being the discount
rate and n being
the number of
periods in the
future when the
cost or benefit will
be realized.)

Lee.indd 229 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers230

alone for different individuals. Sensitivity analysis is the best we can do in this
regard. This analysis entails varying the discount rate over a reasonable range
(typically 0% to 10%) and seeing if the answer changes. If not, the result is
insensitive to the value of the discount rate. But if the answer does change,
we have to use our judgment.

In addition, many technical issues remain to be resolved in CUA. In
particular, the validity of the quality adjustment that underlies QALYs is
unknown. Of course, the core idea of CUA—that what happens to an indi-
vidual patient is the only source of value for medical interventions—will not
always be correct.

Steps in Cost–Utility Analysis

1. Estimate the expected costs for each option.

2. Estimate the number of people alive in each year in each cohort.

3. Using a survey of consumers, estimate the average utility score for
each option for each person who is alive in each year.

4. Multiply the utility score (which will range from zero to one) by
the number of people alive in each year for all the cohorts being
compared. The product is the number of quality-adjusted life
years (QALYs) for each cohort.

5. Discount the QALYs using rates of 2 to 5 percent.

6. Add the QALYs for each option, then find the difference.

7. Divide the difference in cost between options by the difference in
QALYs.

8. Decide whether the cost per QALY is too high.

An Example of Cost–Utility Analysis

Are stents cost-effective for patients with stable angina (chest pain
resulting from an inadequate supply of blood to the heart)? Stents
come in two forms: bare metal and drug eluting. A bare-metal stent is
a mesh tube of thin stainless steel or cobalt-chromium alloy wire. A
drug-eluting stent has a coating that slowly releases a medication that
slows the rate of restenosis (the blood vessels narrowing again after
the treatment). Since their introduction in the late 1970s, stents have

(continued)

Lee.indd 230 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 231

been shown to be highly effective in treating angina. During this same
period, however, medical therapy has also improved dramatically.
Hence a team of researchers asked which approach was best given the
options available as of 2011 (Wijeysundera et al. 2013).

The lifetime cost for medical therapy averaged $22,952, the life-
time cost for a bare-metal stent averaged $25,952, and the lifetime
cost for a drug-eluting stent averaged $25,536. Patients who got
medical therapy were forecast to have a quality-adjusted life expec-
tancy of 10.10 years. The forecast was 10.26 years for patients who
got a bare-metal stent and 10.20 years for patients who got a drug-
eluting stent. Because the bare-metal stent cost less and led to a
longer quality-adjusted life expectancy, it dominated the drug-eluting
stent. Compared to medical therapy, a bare-metal stent cost a little
more than $13,000 per QALY. This calculation, which is called the
incremental cost-effectiveness ratio, divides the cost difference by
the QALY difference. On the basis of this calculation, the team con-
cluded that bare-metal stents were cost effective for most patients.
The team also concluded that drug-eluting stents were cost effec-
tive only for certain patients with diabetes, who were at high risk of
restenosis.

This analysis used data from multiple sources. The analysis also
had to rely on a number of assumptions. In recognition of these fac-
tors, the team conducted a wide array of sensitivity analyses, which
entailed redoing their calculations using different data or assumptions.
Not surprisingly, their forecasts of how long patients survived were the
key factors in their conclusions. Modest changes in costs, quality of
life, or survival could change the conclusions. And such changes are
likely, meaning that any conclusion is likely to change as technology
changes.

(continued)

Teledermatology

Most dermatologists reside in metropolitan areas,
so teledermatology should be considered as an

access option for individuals living outside these areas. However, two
questions must be answered. How much does it cost? How valuable is

Case 14.1

(continued)

Lee.indd 231 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers232

it? A team of Veterans Administration researchers
attempted to answer these questions (Datta et al.
2015).

The team used two cost perspectives. One examined costs from
the perspective of the Veterans Administration, estimating how much
it costs to produce teledermatology care and how much it costs to
produce a face-to-face visit. Two challenges emerged from this effort.
First, costs varied considerably. From the Veterans Administration
perspective the average cost of a teledermatology consult was $308,
but the standard deviation was $298. The average cost of a face-to-
face consult was $338, but the standard deviation was $291. Second,
the authors chose not to include the cost of equipment used to take
images of the patient’s skin, arguing that the incremental cost of an
image was negligible.

In looking at costs from a societal perspective, the team added
spending for dermatologic care from providers who did not work for
the Veterans Administration, travel costs, and patient time costs.
From a societal perspective the average cost of a teledermatology
consult was $460, but the standard deviation was $428. The average
cost of a face-to-face consult was $542, but the standard deviation
was $403.

This study was a CUA, so the team measured utility before and
after treatment. They used a time trade-off technique to measure
patients’ quality of life. This technique presents respondents with
directions such as “Imagine that you have ten years left to live. You
can choose to live these ten years in your current health state, or you
can choose to give up some life years to live for a shorter period in
full health. Mark the timeline with the number of years in full health
that you think is of equal value to ten years in your current health
state.”

1 2 3 4 5 6 7 8 9 10

At baseline, average quality of life was 0.90 for both samples. Over
nine months the teledermatology groups’ average increased by 0.03
and the face-to-face visit groups’ average increased by 0.02.

Case 14.1
(continued)

(continued)

Lee.indd 232 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 233

14.8 Conclusion

Except for CMA or possibly CEA, our advice is “Don’t try this at home.”
When you need evidence to make a decision, turn to the literature. If no
guidance is to be found there, do CMA or CEA (or modify existing studies
using your costs). If these tools do not provide a clear direction, use clini-
cal judgment. CBA and CUA are research tools, not management tools.
Still, these techniques can help make your organization more efficient.
Applied judiciously, they will help your organization identify and provide
the most efficient therapies, which will reduce your costs and increase your
options.

The importance of comparing the right options is often lost in the dis-
cussion of these analyses. Failing to compare reasonable alternatives renders
CMA, CEA, CBA, and CUA useless. The best choice will usually be clear if
the most plausible alternatives are compared. And if the best choice is not
clear, either choice may be appropriate.

Discussion Questions
• Would you be willing to use teledermatology?

Why or why not?

• Which perspective on costs seems more valid to you?

• Do you think that the costs of the imaging equipment should have
been included?

• Did the team use the right approach to evaluation? Would a CMA
have been acceptable?

• What is your reaction to the time trade-off technique?

• What is your recommendation for assessing the value of
teledermatology?

• Would you be willing to adopt teledermatology for your health
system?

• Should Medicare use economic evaluation in making coverage
decisions?

• Congress has largely banned considering costs in making coverage
decisions. Do you agree?

• Can you find published examples of CMA? CEA? CBA? CUA?

Case 14.1
(continued)

Lee.indd 233 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers234

Exercises

14.1 Why have economic analyses of clinical and administrative
innovations become more important?

14.2 Why is cost-minimization analysis most likely to be useful for
managers?

14.3 Why would an economist object to including overhead costs in cost-
minimization analysis?

14.4 A clinic finds that it can reduce costs by eliminating appointments.
The clinic is able to eliminate some telephone staff, and physicians
become more productive. Patients wait until the physician is
available, so the physician has virtually no downtime. Does this
analysis adopt a societal view of costs? Why might this analysis result
in a bad managerial decision?

14.5 Treating a patient with lung cancer with modern drugs increases
average life expectancy by 0.25 years. The added cost of therapy is
$24,000. What is the cost per life year? Should modern drugs be
used?

14.6 A test for bladder cancer costs $100. If given to 1,000 individuals,
it will reduce medical costs by $80,000 and increase average life
expectancy from 15.0 to 15.1. What is the cost per life year? Should
you screen this population?

14.7 Compared with a drip system, an infusion pump reduces the cost
of administering chemotherapy from $25 per dose to $20 per dose.
The complication rate of each system is 2 percent. Which should
you choose? What sort of analysis should you do?

14.8 After choosing between the options in exercise 14.7, you discover
that an infusion pump with a dosage monitoring system costs $15
per dose. Its monitoring functions reduce the complication rate to
1 percent. Which of the three options do you prefer? What principle
does this illustrate?

14.9 Switching from one anesthesia drug to another reduces costs by
$100 per patient. What additional information do you need to do a
cost-minimization analysis?

14.10 A vaccine costs $200 per patient. Administration of the vaccine to
1,000 people is expected to increase the number of pain-free days
for this population from 360,000 to 362,000. Calculate the cost per
additional pain-free day due to vaccination. Is vaccination a good
investment?

14.11 An acute care hospital has found that having geriatric nurse
specialists take charge of discharge planning for stroke patients

Lee.indd 234 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 14: Economic Analysis of Cl inical and Managerial Inter ventions 235

reduces the average length of stay from 5.4 days to 5.2 days. On
average, the geriatric nurse specialist (who earns $27 per hour
including benefits) spends 3.3 hours on discharge planning per
patient. Supply and telephone costs are less than $10 per discharge
plan. Your accounting staff tell you that the average cost per
day is $860 and the incremental cost per day is about $340. Is
this innovation financially attractive? Whether it is or not, what
alternatives should the hospital consider?

14.12 The current cost function for a lab that evaluates Pap smears is C
= 200,000 + 25 × Q. Q , the annual volume of tests, is forecast to
be 30,000. The incremental cost is $25 because each evaluation
requires $20 worth of a technician’s time and $5 worth of supplies.
Calculate the average cost of an evaluation.

14.13 You are comparing replacing the current lab, which has a cost
function of C = 200,000 + 25 × Q , with an automated lab that
has a cost function of C = 300,000 + 20 × Q. Doing so would
reduce the error rate from 1.5 percent to 1 percent. Your volume
is expected to be 18,000 tests per year. Should you choose the
automated lab? Briefly explain your logic.

14.14 The expected cost of Betazine therapy is $544. It is effective 57
percent of the time, with a 6 percent chance of an adverse drug
reaction. The table shows data for Alphazine, a new treatment.
Estimate the rate of adverse drug reactions and the expected cost of
treatment. Use Excel to construct a decision tree for this problem.
Should you choose Alphazine or Betazine?

Probability Cost

Effective 63% Adverse drug reaction 5% $700

No adverse drug reaction 95% $500

Ineffective 37% Adverse drug reaction 5% $800

No adverse drug reaction 95% $600

References

Datta, S. K., E. M. Warshaw, K. E. Edison, K. Kapur, L. Thottapurathu, T. E. Moritz,
D. J. Reda, and J. D. Whited. 2015. “Cost and Utility Analysis of a Store-
and-Forward Teledermatology Referral System: A Randomized Clinical Trial.”
JAMA Dermatology 151 (12): 1323–29.

Lee.indd 235 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Economics for Healthcare Managers236

Essex, H. N., S. Parrott, Q. Wu, J. Li, S. Cooper, and T. Coleman. 2015. “Cost-
Effectiveness of Nicotine Patches for Smoking Cessation in Pregnancy: A
Placebo Randomized Controlled Trial (SNAP).” Nicotine & Tobacco Research
17 (6): 636–42.

Li, H., A. R. Hansen, Z. McGalliard, L. Gover, F. Yan, and J. Zhang. 2018. “Trends
in Smoking and Smoking Cessation During Pregnancy from 1985 to 2014,
Racial and Ethnic Disparity Observed from Multiple National Surveys.”
Maternal and Child Health Journal 22 (5): 685–93.

Nexo, M. A., B. Cleal, L. Hagelund, I. Willaing, and K. Olesen. 2017. “Willingness
to Pay for Flexible Working Conditions of People with Type 2 Diabetes: Dis-
crete Choice Experiments.” BMC Public Health 17 (1): 938.

Scalera, N. M., and T. M. File Jr. 2013. “Determining the Duration of Therapy for
Patients with Community-Acquired Pneumonia.” Current Infectious Disease
Reports 15 (2): 191–95.

Uranga, A., P. P. España, A. Bilbao, J. M. Quintana, I. Arriaga, M. Intxausti, J. L.
Lobo, L. Tomás, J. Camino, J. Nuñez, and A. Capelastegui. 2016. “Duration
of Antibiotic Treatment in Community-Acquired Pneumonia: A Multicenter
Randomized Clinical Trial.” JAMA Internal Medicine 176 (9): 1257–65.

Wijeysundera, H. C., G. Tomlinson, D. T. Ko, V. Dzavik, and M. D. Krahn.
2013. “Medical Therapy v. PCI in Stable Coronary Artery Disease: A Cost-
Effectiveness Analysis.” Medical Decision Making 33 (7): 891–905.

Lee.indd 236 1/2/19 3:15 PM

EBSCOhost – printed on 2/6/2023 8:42 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

CHAPTER

355

COMPARATIVE EFFECTIVENESS RESEARCH

A s part of the $787 billion stimulus bill passed in 2010, Congress allocated
$1.1 billion for comparative effectiveness research (CER). The Afford-
able Care Act (ACA) of 2010 included an additional $3 billion for

studies to compare the effectiveness of different treatments for the same illness.
The different treatments include drugs, medical devices, surgery, and other
ways of treating a specific medical condition (Emanuel, Spiro, and Huelskoetter
2016).

CER and the Role of Government

What Is CER?
The scope of CER includes conducting, supporting, and synthesizing research
that compares the clinical outcomes, effectiveness, and appropriateness of ser-
vices and procedures used to prevent, diagnose, and treat diseases and other
health conditions. CER involves three major areas: (1) comparing new treat-
ments for an illness with the best available alternatives for treating that illness,
(2) using the information from CER to improve joint physician and patient
decision making, and (3) basing the data on which these comparative studies
are to be conducted on a sufficiently large population.

Physicians lack information on the effectiveness of alternative treatments
for many diseases. For some illnesses, the relative effectiveness of alternative
treatments has not been studied; for others, the results of effectiveness studies
have not been disseminated to all physicians. The CER’s federal coordinating
council has developed a priority list of diseases and is awarding grants to study
the comparative effectiveness of alternative treatments for diseases highest
on the priority list. CER is a continuing process that is conducted on more
conditions as new treatments become available for illnesses whose alternative
treatments previously were studied (Conway and Clancy 2009).

Why Is the Government Supporting CER?
Public insurance programs, such as Medicare, and private health insurance pay
for medical treatments regardless of how small the benefit or how large the
cost. Under fee-for-service payment, neither insured patients nor their physi-
cians have any incentive not to seek the most advanced medical treatments in
search of a cure. Further, many policy experts acknowledge that insufficient

22

C
o
p
y
r
i
g
h
t

2
0
1
9
.

H
e
a
l
t
h

A
d
m
i
n
i
s
t
r
a
t
i
o
n

P
r
e
s
s
.

A
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d
.

M
a
y

n
o
t

b
e

r
e
p
r
o
d
u
c
e
d

i
n

a
n
y

f
o
r
m

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n

f
r
o
m

t
h
e

p
u
b
l
i
s
h
e
r
,

e
x
c
e
p
t

f
a
i
r

u
s
e
s

p
e
r
m
i
t
t
e
d

u
n
d
e
r

U
.
S
.

o
r

a
p
p
l
i
c
a
b
l
e

c
o
p
y
r
i
g
h
t

l
a
w
.

EBSCO Publishing : eBook Collection (EBSCOhost) – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS
AN: 1907359 ; Paul Feldstein.; Health Policy Issues: An Economic Perspective, Seventh Edition
Account: s4264928.main.eds

Health Pol icy Issues: An Economic Perspect ive356

information exists on which treatments work best for different diseases. New
drugs are typically compared with a placebo rather than with a drug already
on the market.

Given the soaring cost of healthcare and the belief that each year hun-
dreds of billions of dollars are spent on care that is of no value, more accurate
information on which treatments perform better will improve quality of care
and reduce the wide variations in treatment methods, thereby reducing ris-
ing medical expenditures. Few people are opposed to providing consumers,
physicians, and insurers with additional information on treatments that are
more effective.

As shown in exhibit 22.1, several well-known academic medical centers
were compared according to their total reimbursements per decedent, hospital
days per decedent, and reimbursement per day for treating a patient during the
last two years of his or her life. Wide variations existed among these medical
centers in each of these measures. The study authors also showed that wide
variations existed in the underlying resources, such as nurse staffing and physi-
cian hours, used to treat these patients in the institutions. The federal govern-
ment can play a crucial role in aggregating information about the effectiveness
of various medicines and treatments and disseminating that information to
physicians and their patients.

Academic Medical Center

Inpatient
Reimbursements

per Decedent
Hospital Days
per Decedent

Reimbursements
per Day

Johns Hopkins Hospital $93,233 26.5 $3,520

Ronald Reagan UCLA
Medical Center

$79,182 29.1 $2,721

University of Maryland
Medical Center

$78,753 25.3 $3,109

Hahnemann University
Hospital

$63,932 26.7 $2,397

Massachusetts General
Hospital

$53,159 25.6 $2,080

Cleveland Clinic
Foundation

$41,769 21.9 $1,904

Mayo Clinic-St. Mary’s
Hospital

$40,978 16.9 $2,429

Scott & White Memorial
Hospital

$32,707 15.4 $2,118

Source: Data from Dartmouth Institute for Health Policy and Clinical Practice (2017).

EXHIBIT 22.1
Medicare
Spending

per Decedent
During the Last

Two Years of
Life (Deaths
Occurring in

2014), Selected
Academic

Medical Centers

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 357

Advocates of public funding for CER claim that such information has
the characteristics of a public good—that is, everyone benefits from the infor-
mation generated and that information cannot be denied to anyone once it
becomes available. Because the information cannot be restricted to those who
pay for it, the private sector (health plans) will invest too little to collect such
information. Many, therefore, want government to fund CER and assist in the
dissemination of such information.

Concerns over How CER Will Be Used

Using CER for Reimbursement
Funding for CER generated a great deal of controversy when it was enacted
in 2010. Critics were concerned that once the effectiveness of two treatments
or drugs was determined, the relative costs of the two treatments would also
be used to determine which drugs should be used. Fearful of being accused of
promoting “death panels,” Congress prohibited the use of information based
on CER for mandating coverage, reimbursement, or treatment decisions for
public and private payers. However, many remain concerned that under the
fiscal pressures of rising medical costs, use of CER results will eventually move
closer to the way in which European countries use their findings on compara-
tive effectiveness (Nix 2012).

Some opponents of government-funded CER believe that the govern-
ment would ultimately use the findings from such research to establish medical
practice guidelines, limit access to treatments, and refuse to pay for expensive
new drugs. This concern was reinforced by a book by Tom Daschle, who was
nominated by President Obama to become secretary of the US Department
of Health and Human Services. (He subsequently withdrew his name amid a
growing controversy over his failure to accurately report and pay income taxes.)
Daschle had proposed a federal health board that would promote high-value
medical care by recommending coverage of drugs and procedures based on
the board’s research (Daschle, Greenberger, and Lambrew 2008).

Differences in Patient Responses to the Same Treatment
CER is a one-size-fits-all approach to medicine; however, patients’ responses
to different drugs vary widely. Those who do not respond well to the recom-
mended treatment are at a disadvantage. For example, for most patients, a
generic drug is cheaper and works as well as a brand-name drug. However, for
some patients the generic version may cause serious side effects or have little
effect. Thus, although a branded and a generic drug may be equally effective on
average, not paying for the newer, more expensive drug may lead to increased
hospitalization costs and worse health outcomes for those patients who do not
respond well to the cheaper drug.

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Health Pol icy Issues: An Economic Perspect ive358

The problem with a one-size-fits-all approach is illustrated by a CER
analysis of antipsychotic drugs, which found little difference between the effec-
tiveness of older, less costly antipsychotic drugs and newer, more costly drugs.
Using the older drugs could have saved Medicaid $1.2 billion (out of the $5.5
billion spent on these drugs in 2005). However, Basu, Jena, and Philipson
(2011) concluded that the mental health of thousands of patients would have
been worse and societal costs would have been greater than any savings from
using the less costly drugs.

Variations in Medical Practice
While most agree that wide variation occurs in medical practice, huge amounts
of money are wasted on ineffective treatments and testing, and more information
would be beneficial, there is opposition to moving from information genera-
tion and dissemination to basing payment on CER. Zuckerman and colleagues
(2010) attempted to explain why wide variations in medical spending occur.
Using data on Medicare beneficiaries, they found that unadjusted Medicare
spending per beneficiary was 52 percent greater in the highest-spending geo-
graphic region than in the lowest-spending region. The authors then adjusted
the regions based on demographics, baseline health characteristics, and changes
in health status. The difference between the highest and lowest regions then
decreased to 33 percent. Health status was found to explain an important part
of this variation. Although inefficiency in spending per Medicare beneficiary
exists, wide cost differences across geographic areas are not the result of inef-
ficiency alone.

Policies to decrease spending differences per beneficiary between high-
and low-cost areas by reimbursing physicians only for treatments that follow
certain protocols or guidelines should not ignore the legitimate reasons for
some of these variations.

Accuracy and Timeliness of Comparative Effectiveness Studies
One study generally does not provide a definitive answer; several studies likely
will have to be undertaken. For example, bone marrow transplantation for
breast cancer was widely accepted as beneficial, and patients won lawsuits
because some health plans refused to cover it. Subsequently, researchers found
that this treatment was ineffective.

Comparative effectiveness studies may not adequately evaluate alterna-
tive treatments for patients with multiple chronic diseases or rare illnesses.
Similarly, CER often does not include sufficient numbers of women, African
Americans, and Hispanics. Some drugs appear to be more effective in women
than in men, while other medicines are more likely to cause serious complica-
tions in women. CER must include larger numbers of patients in clinical trials
so that gender and minority differences can be considered. As CER studies

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 359

are expanded to account for such differences, the time and money needed to
complete these clinical trials will increase (Chandra, Jena, and Skinner 2011).

Because it takes time to complete CER and for guidelines to be approved
by the government, physicians and their patients may be willing to try untested
therapies, as occurred for AIDS patients. Will they be permitted to do so? Will
providers be reimbursed for these therapies?

CER and

Innovation

An additional concern is that CER might lead to slower adoption of new, more
effective treatments. As new treatments and prescription drugs are developed,
will reimbursement for them be delayed until their comparative effectiveness
has been determined? Physicians might be willing to try new surgical tech-
niques that offer the possibility of improved patient outcomes; will they and
the hospital have to forgo payment because these techniques have not under-
gone CER? Will the healthcare system become more rigid and less innovative
because physicians fear repercussions if their treatments differ from the official
guidelines? Will health plans refuse to reimburse for procedures and treatments
that are not within the federal recommendations?

Medical device and pharmaceutical companies are also likely to face
another layer of government approval that will increase their cost to bring a
product to market, thereby decreasing their incentive for innovation.

Dissemination of CER Findings
CER results must be disseminated. Will dissemination of information, which
often is slow and may go against the financial interest of some physicians, be
sufficient to lead to adoption of the CER? Or will financial incentives and
reporting requirements be necessary? A concern with providing information to
physicians is that their rate of adoption of new practices is very slow. If informa-
tion is to change medical practice, lower costs, and improve quality, physician
practice behavior will have to change more rapidly. However, without appropri-
ate incentives, new information often takes years to change physician behavior.

Exhibit 22.2 provides several examples of the time from dissemination
of information to treatment adoption by physicians. In 1988, the Food and
Drug Administration approved the use of aspirin for treatment of heart attacks,
which resulted in an increase in use from 20 percent to 62 percent. However,
by the mid-1990s, aspirin use had only increased to 75 percent. After studies
were published indicating the potential harmful effects of calcium-channel
blockers, their use declined but remained above 30 percent ten years later.

Possible Stages in Use of CER
Again, the legislation providing government funding for CER states that CER
will not be used for reimbursement or coverage decisions. However, some people

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Health Pol icy Issues: An Economic Perspect ive360

believe that government funding of CER is just the first stage in its evolution.
CER opponents foresee the following stages. First, CER provides information
about the clinical effectiveness of different treatments and drugs for a disease.
Second, their cost-effectiveness is compared. Third, given the rising costs of medi-
cal care and the increasing burden on the federal deficit, the government only
pays for those drugs and other treatments that are cost-effective, even though
the effects may differ among people or population groups. Fourth, instead of
deciding which drug it will pay for on the basis of cost-effectiveness, the govern-
ment decides which drugs and treatments it will pay for by comparing the cost of
the drug with the value of an additional year of life (as occurs in Great Britain).

The following sections examine cost-effectiveness, how CER might be
used for reimbursement, and how CER is used in Great Britain.

Cost-Effectiveness Analysis

Cost-effectiveness analysis compares the additional costs of alternative
approaches to achieve a specific outcome designed to improve health. For
example, an organization interested in decreasing hip fractures would want
to know the different programs that can reduce hip fractures, the cost of

Pharmaceuticals
Year of

Innovation

Pharmaceutical Usea

1973–
1977

1978–
1982

1983–
1987

1988–
1992

1993–
1996

Beta blockers 1962 20.6 41.5 47.5 47.3 49.8

Calcium chan-
nel blockersb

1971 0 0 63.9 59.0 31.0

Angiotensin-
converting
enzyme (ACE)
inhibitors

1979 0 — — — 56.0

Aspirin 1988c 15.0 14.1 20.1 62.0 75.0

aIn hospital or 30-day use.
bCalcium channel blocker use increased rapidly in the early 1980s and then fell, following the publi-
cation of studies documenting potentially harmful effects of their use in acute management.
cIn 1988, the Food and Drug Administration (FDA) proposed the use of aspirin for reducing the risk
of recurrent myocardial infarction (MI), or heart attack, and preventing first MI in patients with
unstable angina. The FDA also approved the use of aspirin for the prevention of recurrent transient
ischemic attacks, or “mini-strokes,” in men and made aspirin standard therapy for previous strokes
in men.

Source: Adapted from Cutler, McClellan, and Newhouse (1999, tables 3 and 5).

EXHIBIT 22.2
Use of Acute

Interventions
(Pharma-

ceuticals) for
Myocardial

Infarction

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 361

expanding each program, and the extent of reduction in hip fractures each
would achieve. Results from a cost-effectiveness analysis are typically presented
in the form of a cost-effectiveness ratio, where the numerator of the ratio is
the additional cost of the intervention and the denominator is some measure
of the outcome of interest.

Alternative approaches for decreasing hip fractures are likely to differ in
costs and effectiveness. Thus, they can be compared according to their cost-
effectiveness ratio, which is the additional cost per averted hip fracture. (For
an example of cost-effectiveness analysis, see exhibit 3.3.)

Calculating the cost-effectiveness ratio for each alternative method of
achieving a given health outcome allows a comparison of the trade-offs from
choosing one alternative over another. Decision makers—whether they are
administrators in government agencies such as Medicaid or health maintenance
organization managers—can make better-informed choices about the relative
costs and effectiveness of alternative interventions by using cost-effectiveness
analysis. When selecting among alternative expenditures to improve health,
alternative interventions can be ranked according to their cost-effectiveness
ratios (e.g., cost per death averted), giving the intervention with the lowest
ratio the highest priority. Choosing interventions on the basis of the lowest
cost-effectiveness ratio maximizes the outcome for a given budget.

Many cost-effectiveness studies have been conducted on the relative
effectiveness of a new drug compared with existing drugs for treating the same
disease. The originators of such studies include health plans seeking to determine
which drugs to include in their formularies and pharmaceutical firms hoping
to use the results to demonstrate to large purchasers the greater effectiveness
of their new drugs compared with those of competitors.

A concern with cost-effectiveness analysis if used for reimbursement by
the government is that, as discussed earlier, patients may differ in their response
to a drug or other treatment. Medical costs could be higher if patients respond
poorly to certain drugs and must be hospitalized. Further, the government’s
cost-effectiveness ratio may be different from the patient’s cost or evaluation
of the treatment’s effectiveness.

Quality-Adjusted Life Years

A specific type of cost-effectiveness analysis uses quality-adjusted life years
(QALYs) as an outcome measure. QALYs indicate the increased utility achieved
as the result of an intervention, such as comparing a new drug to an existing
one. The cost-effectiveness ratios are in terms of the cost per QALYs gained.
The advantage of using QALYs rather than, for example, life expectancy is that
QALY incorporates multiple outcomes—increase in length of life and quality

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Health Pol icy Issues: An Economic Perspect ive362

of life. Using QALYs as an outcome measure also enables comparisons to be
made across different disease conditions.

QALY is calculated as follows: Each additional year of perfect health for
an individual is assigned a value of 1.0, which is the highest value of a complete
QALY. The assigned value decreases as health decreases, with death equal to 0.0.
If the patient has various limitations—such as a disability, physical pain, or receipt
of kidney dialysis—the extra life years are assigned a value between 1.0 and 0.0.
Thus, if new intervention A enables a person to live an additional five years, but
with a quality of life weight of 0.7, then the QALY score for that intervention is 5
× 0.7 = 3.5 QALYs. If intervention B extends life for four years with a quality of life
weight of 0.6, the additional QALYs provided are 4 × 0.6 = 2.4. The net benefit
of intervention A over intervention B is 3.5 QALYs – 2.4 QALYs = 1.1 QALYs.

When comparing alternative interventions according to their additional
cost per QALY, those with a lower cost per QALY are preferred to those with
a higher cost per QALY. A common approach for developing QALYs involves
the use of an activities of daily living (ADL) scale. Patients are asked to rate
their ability to function independently, such as dressing, bathing, and walk-
ing. Patient responses range from unable to perform the function to able to
perform the function without difficulty. These scores are summed over all the
ADL categories to arrive at a patient’s overall functional status.

The calculation of QALYs is the same regardless of a person’s income,
wealth, or age; however, QALYs that occur in later years may be valued less
than QALYs occurring earlier in life.1

QALYs have several drawbacks. For example, QALY does not include
the effects of a patient’s disability on the quality of life of others, such as fam-
ily members. Assigning a quantitative value to a disability may not be accurate
because people differ in their perceptions of the severity of various limitations
on their normal activity. Further, applying these utility measures across a large,
diverse population is unlikely to reflect many individuals’ utility preferences.
Yet using QALYs for a large population is necessary if alternative medical treat-
ments are to be compared.

Applications of QALYs
QALYs have been used in two types of policy analysis. First, they are used as
an outcome measure in cost-effectiveness analysis to compare alternative inter-
ventions in determining which intervention offers the lowest cost per QALY.
Second, and more controversial, cost per QALY has been used to determine
benefit coverage—for example, to decide whether a costly treatment, such as
an expensive new drug, should be provided to a breast cancer patient (Baum-
gardner and Neumann 2017).

Exhibit 22.3 shows the results of several cost-effectiveness studies exam-
ining different drug therapies potentially applicable to the Medicare popula-
tion. Each study describes an intervention compared with the alternative of no

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 363

treatment. The results, in the form of cost per QALY gained, are ranked from
lowest to highest cost per QALY. Therapies whose cost per QALY is relatively
low (e.g., $4,000) are very favorable. Therapies with a relatively high cost per
QALY (e.g., $460,000) are considered unfavorable and less likely to be adopted.

The National Institute for Health and Care Excellence

Great Britain established the National Health Service (NHS) in 1948 as a single-
payer system, administered by the government, funded through taxation, and
provided by public institutions. The British government has a long history of
underfunding the NHS, resulting in long waiting lines and failure to provide
certain types of treatments. To limit expenditures on expensive innovative medical
technology and drugs and to attempt to rationalize its limited budget, in 1999
the government formed the National Institute for Clinical Excellence (NICE)—
now called the National Institute for Health and Care Excellence—a private,

Intervention vs. Base Case in Target Population
Dollars per

QALY Gained

Captopril therapy vs. no captopril in 80-year-old patients surviv-
ing myocardial infarction

$4,000

Treatment with mesalazine vs. no treatment to maintain remis-
sion in Crohn’s disease

$6,000

One-year course of isoniazid (INH) chemoprophylaxis vs. no INH
chemoprophylaxis in 55-year-old white male tuberculin reactors
with no other risk factors

$18,000

Treatment to reduce the incidence of osteoporotic hip frac-
ture vs. no treatment in 62-year-old woman with established
osteoporosis

$34,000

Ticlopidine vs. aspirin in 65-year-old with high risk of stroke $48,000

Chemotherapy vs. no chemotherapy in 75-year-old with breast
cancer

$58,000

Captopril vs. propranolol in persons in the US population aged
35–64 years without the diagnosis of coronary heart disease but
with essential hypertension

$150,000

Antiemetic therapy with ondansetron vs. antiemetic therapy
with metoclopramide in 70-kg patient receiving cisplatin chemo-
therapy who had not been previously exposed to antineoplastic
agents

$460,000

Note: QALY = quality-adjusted life year.

Source: Adapted from Neumann and colleagues (2000, exhibit 3).

EXHIBIT 22.3
Selected Cost-
Effectiveness
Ratios for
Pharma ceu-
ticals, with
a Focus on
the Medicare
Population

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Health Pol icy Issues: An Economic Perspect ive364

independent organization in the Department of Health, to provide guidance on
health technology, clinical medicine, and new prescription drugs. Its decisions
are based on clinical efficacy and cost-effectiveness (NICE 2018; Rawlins 2013).

NICE uses QALYs to determine which treatments to cover in the NHS. Given
that budget constraints exist on the amount the government can spend for medical
services, NICE undertakes cost-effectiveness analysis for new drugs and treatments
in an attempt to provide patients, health professionals, and the public with scientifi-
cally based guidance on current best practices. A NICE committee consisting of
medical and other professionals, such as health economists, statisticians, managers,
patient advocates, and manufacturer representatives, assists in its decision making.

Before NICE was established, the availability of costly treatments varied
greatly throughout the country, as did the level of medical services. NICE has
made the availability of drugs and treatments more uniform throughout the
NHS. Decisions by NICE are transparent to all, and the information on which
it bases its decisions is also publicly available. Further, when NICE believes that
a treatment or drug is cost-beneficial, it attempts to ensure that the treatment
or drug becomes widely available.

The main criticism of NICE is that it bases its recommendations primarily
on cost-effectiveness rather than on clinical effectiveness (Hope 2011; Steinbrook
2008). The criticism that NICE is coldhearted stems from the fact that it uses cost
per QALY to determine cost-effectiveness. One of NICE’s most contentious issues
is how much should be spent per additional year of life that a drug is expected to
provide. NICE’s general threshold is about $66,000 per QALY. If a treatment’s
cost per QALY is higher, NICE will generally deny the treatment. (In 2016, NICE
set the cost per QALY threshold for treatment of rare diseases at $132,000.)

The following example illustrates how NICE uses its cost per QALY
to determine approval for costly treatments. A New York Times article tells
the story of Bruce Hardy, a patient fighting kidney cancer that was spreading
throughout his body (Harris 2008). His physician wanted to prescribe a new
drug from Pfizer called sunitinib malate (Sutent), which delays cancer progres-
sion for six months at a cost of $54,000. NICE, however, decided that the
drug was too costly to be offered free to all those who needed it. According
to NICE, the cost of extending life for six months should be no more than
$22,750; therefore, Hardy could not receive the drug. When NICE rejected
Sutent, some patients mortgaged their homes to pay for the drug on their own.
After much protest, NICE reversed its decision and approved the drug. NICE
has also limited the use of certain breast cancer drugs, such as trastuzumab
(Herceptin), and drugs for osteoporosis and multiple sclerosis.

Great Britain has been explicit in recognizing that resources are scarce
and choices must be made on how to allocate those scarce resources. At some
point, with rising medical costs and a huge government deficit, will the United
States become as explicit or, more likely, make such decisions implicitly by

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 365

limiting healthcare provider reimbursement, thereby limiting the resources
available for new technology and expensive drugs?

Summary

CER should provide additional information to physicians and their patients
regarding the effectiveness of alternative treatments. To the extent that wide varia-
tions in medical practice are the result of lack of information, CER should improve
patient outcomes and reduce medical costs. However, if the CER findings are
used for reimbursement or coverage decisions, some patients may suffer adverse
health consequences, and the medical system could become less innovative.

Some policy experts are concerned that federal funding for CER is but
the first step toward limiting government payment for treatments considered
less effective than others or too expensive with respect to their return in extend-
ing life expectancy. If CER studies demonstrate that a new drug for $1,000
a year usually provides greater benefits than a $50,000 surgical procedure,
but the financial incentive for many surgeons is to continue performing the
more expensive procedure, will insurers and the government continue to pay
for both treatments? Great Britain’s NICE is often cited as an example of a
government agency that determines which medical treatments will be covered
based on cost-effectiveness. Using cost-effectiveness, NICE covers only those
treatments that do not exceed a certain threshold, such as the cost per QALY
not exceeding the value of a life. Although the ACA states that CER shall not
be used as a basis for payment, some are concerned that the United States may
eventually use cost-effectiveness in reimbursement of medical services.

Society cannot spend an infinite amount of money to extend each
person’s life; choices must be made. Economics requires trade-offs because
resources are scarce and can be spent on enhancing life in other ways. The
opportunity cost of spending $100,000 on a new drug that extends the life
of a terminally ill patient by three months is that those same funds could be
spent on prenatal care or to increase the life expectancy of very-low-birthweight
infants. Spending resources on additional medical services to extend one per-
son’s life involves having fewer resources to spend on extending the lives of
others. States and the federal government, faced with higher limits on their
expenditures and increasing demands for costly medical services, will have to
make difficult choices in coming years.

The proposal to create a separate federal health board to make difficult
political decisions regarding which medical services and prescription drugs to
fund insulates legislators from making these difficult choices, such as denying
expensive but potentially life-extending services to a patient whose need for
the treatment has been discussed in the media.

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Health Pol icy Issues: An Economic Perspect ive366

Discussion Questions

1. What are the advantages of CER?
2. What are disadvantages of using CER for federal payment?
3. What are QALYs?
4. How are QALYs used in cost-effectiveness analysis?
5. How does NICE use QALYs in determining whether to approve a new

drug?

Note

1. When interventions produce QALYs over different periods, discounting
may be used to convert them into equivalently valued units at the
present period, similar to discounting future income streams (as is done
in a cost–benefit analysis). To determine the present value of future
QALYs, the number of QALYs in each future year should be multiplied
by (1/1 + rt), where r is the discount rate—such as 0.05—and t
represents the number of years from the future to the present.

Additional Readings

Health Affairs. 2012. “Current Challenges in Comparative Effectiveness Research.”
Published October. www.healthaffairs.org/toc/hlthaff/31/10.

———. 2010. “Comparative Effectiveness Research.” Published October. www.health
affairs.org/toc/hlthaff/29/10.

Neumann, P. J., J. T. Cohen, and M. C. Weinstein. 2014. “Updating Cost-Effectiveness:
The Curious Resilience of the $50,000-per-QALY Threshold.” New England
Journal of Medicine 371 (9): 796–97.

References

Basu, A., A. Jena, and T. Philipson. 2011. “Impact of Comparative Effectiveness Research
on Health and Healthcare Spending.” Journal of Health Economics 30 (4):
695–706.

Baumgardner, J., and P. J. Neumann. 2017. “Balancing the Use of Cost-Effectiveness
Analysis Across All Types of Health Care Innovations.” Health Affairs Blog.
Posted April 14. www.healthaffairs.org/do/10.1377/hblog20170414.059610/
full/.

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

Chapter 22: Comparat ive Ef fect iveness Research 367

Chandra, A., A. Jena, and J. Skinner. 2011. “The Pragmatist’s Guide to Comparative
Effectiveness Research.” Journal of Economic Perspectives 25 (2): 27–46.

Conway, P., and C. Clancy. 2009. “Comparative-Effectiveness Research: Implica-
tions of the Federal Coordinating Council’s Report.” New England Journal of
Medicine 361 (4): 328–30.

Cutler, D., M. McClellan, and J. Newhouse. 1999. “The Costs and Benefits of Intensive
Treatment for Cardiovascular Disease.” In Measuring the Prices of Medical Treat-
ments, edited by J. Triplett, 34–71. Washington, DC: The Brookings Institution.

Dartmouth Institute for Health Policy and Clinical Practice. 2017. “Care of Chronically
Ill Patients During the Last Two Years of Life.” Dartmouth Atlas of Health Care.
Accessed April 2018. www.dartmouthatlas.org/tools/downloads.aspx?tab=40.

Daschle, T., S. Greenberger, and J. Lambrew. 2008. Critical: What We Can Do About
the Health-Care Crisis. New York: Thomas Dunne Books.

Emanuel, Z., T. Spiro, and T. Huelskoetter. 2016. “Re-evaluating the Patient-Centered
Outcomes Research Institute.” Center for American Progress. Posted May 31.
www.americanprogress.org/issues/healthcare/reports/2016/05/31/138242/
re-evaluating-the-patient-centered-outcomes-research-institute/.

Harris, G. 2008. “British Balance Benefit vs. Cost of Latest Drugs.” New York Times.
Published December 2. www.nytimes.com/2008/12/03/health/03nice.html.

Hope, J. 2011. “Breakthrough MS Pill Rejected as Too Expensive by NHS Watchdog
(but You Can Get It in U.S. and Germany).” Mail Online. Published August 5.
www.dailymail.co.uk/health/article-2022767/Miracle-MS-pillrejected-expensive-
NHS-watchdog-available-U-S-Germany.html.

National Institute for Health and Care Excellence (NICE). 2018. “About.” Accessed
April. www.nice.org.uk/aboutnice/.

Neumann, P. J., E. A. Sandberg, C. M. Bell, P. W. Stone, and R. H. Chapman. 2000.
“Are Pharmaceuticals Cost-Effective? A Review of the Evidence.” Health Affairs
19 (2): 92–109.

Nix, K. 2012. “Comparative Effectiveness Research Under Obamacare: A Slippery
Slope to Health Care Rationing.” Heritage Foundation. Backgrounder No.
2679. Published April 12. http://thf_media.s3.amazonaws.com/2012/pdf/
bg2679 .

Rawlins, M. 2013. “NICE—Moving Onward.” New England Journal of Medicine
369 (1): 3–5.

Steinbrook, R. 2008. “Saying No Isn’t NICE—the Travails of Britain’s National Insti-
tute for Health and Clinical Excellence.” New England Journal of Medicine
359 (19): 1977–81.

Zuckerman, S., T. Waldman, R. Berenson, and J. Hadley. 2010. “Clarifying Sources
of Geographic Differences in Medicare Spending.” New England Journal of
Medicine 363 (1): 54–62.

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use

EBSCOhost – printed on 2/6/2023 8:36 AM via UNIVERSITY OF MARYLAND GLOBAL CAMPUS. All use subject to https://www.ebsco.com/terms-of-use