Subido por hguevara

Tutorial on Health Economics and Outcomes Research in Nutrition

PENXXX10.1177/0148607114549770Journal of Parenteral and Enteral Nutrition X(X)Philipson et al
Tutorial on Health Economics and Outcomes
Research in Nutrition
Journal of Parenteral and Enteral
Volume 38 Supplement 2
November 2014 5S­–16S
© 2014 Abbott Nutrition
DOI: 10.1177/0148607114549770
hosted at
Tomas Philipson, PhD1; Mark T. Linthicum, MPP2; and Julia Thornton Snider, PhD2
As healthcare costs climb around the world, public and private payers alike are demanding evidence of a treatment’s value to support approval
and reimbursement decisions. Health economics and outcomes research, or HEOR, offers tools to answer questions about a treatment’s value,
as well as its real-world effects and cost-effectiveness. Given that nutrition interventions have to compete for space in budgets along with
biopharmaceutical products and devices, nutrition is now increasingly coming to be evaluated through HEOR. This tutorial introduces the
discipline of HEOR and motivates its relevance for nutrition. We first define HEOR and explain its role and relevance in relation to randomized
controlled trials. Common HEOR study types—including burden of illness, effectiveness studies, cost-effectiveness analysis, and valuation
studies—are presented, with applications to nutrition. Tips for critically reading HEOR studies are provided, along with suggestions on how
to use HEOR to improve patient care. Directions for future research are discussed. (JPEN J Parenter Enteral Nutr. 2014;38(suppl 2):5S-16S)
nutrition; health economics; outcomes research; burden of illness; efficacy; effectiveness; cost effectiveness; cost utility analysis; value;
oral nutrition supplement; enteral nutrition
Clinical Relevancy Statement
As nutrition interventions are increasingly coming to be evaluated through health economics and outcomes research (HEOR),
clinicians working in the nutrition space can benefit from
developing an understanding of HEOR study types and concepts. Such an understanding will enable clinicians to evaluate
new evidence generated in nutrition using HEOR methods.
Understanding HEOR studies is particularly important given
that some outcomes, such as real-world effectiveness and cost
impacts of interventions, are difficult or even impossible to
study using randomized controlled trials. Understanding a
broader array of relevant study types will give clinicians the
flexibility to gather and evaluate the most appropriate evidence
in their efforts to improve patient care.
Healthcare costs are high and rising, both in the United States
and globally. Given the budgetary pressures this cost growth
imposes, public and private payers alike have become increasingly focused on value in healthcare. For a product or intervention to represent a good value, it should not only be efficacious
but also be worth the scarce resources that were given up to
purchase it. But what does it mean for an intervention to be
worth its cost, and how is this calculation determined?
Answering questions such as these falls within the domain of
health economics and outcomes research, or HEOR.
HEOR has traditionally been used in the study of medical
products (ie, biopharma and devices); however, as cost pressures
have extended to all aspects of healthcare, HEOR has become
increasingly relevant to nutrition. Over the past century, numerous government bodies across the globe have come to require
health technology assessment (HTA) for consideration during
their approval and reimbursement decisions. As nutrition products and interventions must compete for space in healthcare budgets along with pharmaceuticals and devices, nutrition is also
coming to be evaluated through HEOR. In fact, HEOR may
prove to be particularly useful in nutrition, given its low cost
relative to its substantial health benefits, making it compare
favorably with other interventions in comparative effectiveness
From the 1Irving B. Harris School of Public Policy Studies, The University
of Chicago, Chicago, Illinois, and 2Precision Health Economics, Los
Angeles, California.
Financial disclosure: Financial support for the publication of the
supplement in which this article appears was provided by Abbott
Conflicts of interest: T.P. holds the position of partner at Precision Health
Economics. This contribution was prepared by T.P., M.T.L., and J.T.S.
at the request of and within the scope of their employment with Precision
Health Economics, which received support from Abbott Nutrition to
conduct the research appearing in this article, and as such copyright is
owned by Abbott Nutrition.
Received for publication June 24, 2014; accepted for publication August
8, 2014.
Corresponding Author:
Julia Thornton Snider, PhD, Precision Health Economics, 11100 Santa
Monica Blvd, Suite 500, Los Angeles, CA 90026, USA.
Email: [email protected]
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
Table 1. Comparison of Randomized Controlled Trials (RCTs) and Observational Studies.
Observational Study
Prospective—the study is designed and the data are
then collected to measure the study outcomes
Risk of selection bias
Low—controlled by restrictive exclusion criteria,
more complete data collection and observation,
and randomization
Usually small due to cost and practical
Relatively high
Usually finely measured clinical end points such as
change in lean body mass or hand grip strength
Artificially high
Generally short because of cost and difficulty
keeping participants in studies long term
Limited due to narrow study populations and
artificially regulated behaviors
Usually retrospective—the study relies on preexisting
data sources (such as health insurance claims) to study
relationships between treatments and outcomes that
have already occurred
Moderate—heterogeneous population, with
nonrandomized assignment to treatment, but numerous
tools exist to control for bias
Potentially huge
Sample size
Research cost
Outcomes of interest
Follow-up periods
research (CER). As such, using HEOR methods may support
building an evidence base on the exceptional value of nutrition.
HEOR can be based on both trial data and analysis of interventions as they are used by patients in the real world, where
patients are more diverse and usually less adherent than in the
randomized controlled trial (RCT) setting. It can be used to
determine how patients’ everyday decisions reveal the value
they place on their health and nutrition status. And it can be
used to determine how to allocate a fixed budget on nutrition
interventions and other healthcare to maximize patient health
and satisfaction.
Given the growing importance of HEOR for nutrition interventions, this tutorial has 2 basic objectives:
To familiarize the reader with the premise and basic
concepts of HEOR
To demonstrate the value and usefulness of HEOR for
clinicians and researchers working in nutrition
To achieve these objectives, the tutorial will proceed as
follows. First, it will define HEOR and explain how it adds
value in a world where we already have RCTs. Second, it will
provide the reader examples of the types of questions HEOR
can help us answer. Finally, it will offer practical advice for
reading HEOR studies, suggest directions for future research,
and present examples of using nutrition HEOR to advance
patient care.
After reading this tutorial, readers should have a better
understanding of the nutrition HEOR literature and HEOR
concepts they come across in their daily work, and they should
be able to use HEOR to improve their research or practice and,
ultimately, patient outcomes.
Relatively low
Clinical end points, economic outcomes, costs,
readmission risk, etc
Reflects real-world adherence
Potentially long due to nature of data sets (eg, Medicare
data includes all claims for a patient over many years)
More generalizable because study populations and
behavior reflect real-world heterogeneity and decision
What Is HEOR and How Is It Relevant to
HEOR Definition
HEOR is a discipline used to determine the health effects of
interventions, to evaluate those interventions relative to their
full cost, and, based on that evaluation, to allocate scarce
resources to achieve a particular objective. HEOR serves as a
complement to clinical research. While clinical research
focuses on issues such as efficacy and safety, HEOR addresses
the overall benefits and costs of an intervention, including
effectiveness, real-world adherence, complications, and direct
and indirect costs.
Relation to RCTs
While HEOR can be applied to both experimental and observational data, it offers a variety of tools for analyzing observational data that make it particularly well suited for that purpose.
HEOR studies using observational data are useful because
RCTs, while deservedly central to modern medicine, have
important limitations that observational studies can address.
These limitations generally stem from the need to enroll and
retain patients in the RCT and manage the costs of doing so. In
particular, RCTs generally have small, narrowly selected
patient populations; artificially low pricing; artificially high
adherence; and short duration of follow-up (see Table 1).
RCTs are especially useful for providing fine-tuned and
precise estimates of the effect of a given intervention, in a specific setting, with all else being held constant. RCTs are critical
to improving practice because they allow us to see the effect of
Downloaded from by guest on March 5, 2015
Philipson et al
an intervention without the “noise” of the real world. For this
reason, RCTs are generally considered the “gold standard” for
testing the effects of interventions. Once this effect is understood, however, these findings must inform decisions and treatment in the real world. At this point, other important issues
arise, such as the impact on costs in real-world settings, the
effect of adherence on the value of the intervention, or the
long-term cost reductions from, for example, reduced clinical
Because of the need to recruit a relatively homogeneous
sample, RCTs tend to provide us evidence of effect on specific
subgroups of the population, making it difficult to generalize
results to the broader population. This is compounded by the
fact that, due to necessary exclusion criteria, difficulty in
recruiting to trials, and the high cost of intensive data collection and adherence monitoring, trial sample sizes tend to be
small. For example, a recent systematic review of studies
examining the use of oral nutrition supplements (ONS) in
patients with chronic obstructive pulmonary disease (COPD)
included 17 RCTs in its analyses—with a combined total of
632 participants, a sample size much smaller than the thousands or millions of observations often available in observational data.1 In addition, trial researchers follow participants
closely, ensuring adherence to treatment in the arm to which
they are randomized, but there is considerable evidence that
real-world adherence to treatment is far from perfect.2-4
In the field of nutrition, researchers interested in designing
and conducting RCTs face unique challenges.5 Compared with
a pharmaceutical treatment, adherence may be lower with
nutrition interventions, requiring more intensive monitoring by
researchers. In addition, sample sizes may need to be larger
than in pharmaceutical trials because of concerns about adherence and trial retention. The timeframe for measuring nutrition
benefits is also long, raising issues with participant retention
over time and increasing costs. Furthermore, findings are
affected by the general food consumption of participants,
which may increase or decrease the effectiveness of a nutrition
intervention. It may be complicated to hold all participants to a
uniform diet, and maintaining records of all food consumed
may affect study participants’ eating habits, making it difficult
to control for diet. For example, 1 randomized study illustrated
that the act of recording patients’ nutrition intake in the control
arm possibly biased “routine practice” as patients and physicians had a heightened focus of their diets.6 Similarly, the lifestyle and activity level of participants is likely to have a
significant impact on outcomes.
HEOR can complement RCTs by providing tools with
which to study outcomes in observational data, thereby
enabling the study of potentially enormous populations, with
real-world pricing and adherence, and potentially much longer
follow-up than is typical of RCTs. Such studies are generally
retrospective and do not allow researchers the ability to randomize patients into treatment and control groups, thereby
introducing the risk of selection bias. Specifically, in the real
world, whether someone receives a given treatment, as well as
his or her response to that treatment, is likely to be influenced
by underlying health, economic, social, or demographic factors
that may not be observable in the data available to researchers.
Because treatment is not random, the effect of treatment may
be biased by these factors.
HEOR has its roots in economics, a field in which RCT data
have generally been unavailable. (For example, it would be
impractical and unethical to randomly assign half of the country
to experience a recession.) In economics and in the separate but
related field of outcomes research, techniques have been developed to analyze observational data to be able to infer cause and
effect, even in the absence of randomization. Many have argued
that, due to selection bias, observational studies may systematically find larger effects than RCTs, but this criticism has not
been borne out by analyses comparing effect sizes.7,8
Another potential limitation of observational studies is the
fact that they often take a retrospective perspective, in the
sense that they analyze data that already exist at the initiation
of the study and were not set up explicitly for the purpose of
the study (eg, health insurance claims data). Because the data
are preexisting, their collection has not been designed to minimize confounders—hence the need for econometric or statistical techniques to address selection bias. Because both RCTs
and observational studies have limitations, they should be
viewed as complements rather than substitutes, as each offers a
unique and useful vantage point on the advantages and disadvantages of a given treatment.
HEOR methods for identifying treatment effects in observational data are discussed in detail below.
Questions in Nutrition That HEOR Can
Help to Answer
Ultimately, HEOR seeks to inform decision making—from the
decisions of individual providers to those of hospital administrators, policy makers, and patients. HEOR encompasses a
variety of types of studies that can be used to answer questions
of relevance to nutrition. For example, what burden does malnutrition impose? Which interventions have been shown effective to alleviate malnutrition in the real-world setting? Which
interventions are the most cost-effective? And how do patients
reveal their preferences for health and life through their own
everyday decisions? The next section presents common types
of HEOR studies, with applications to nutrition as available.
What Is the Burden on Society From
Nutrition Diseases?
Diseases cost human life, health, and happiness. Disease not
only affects the health of individuals but also has an impact on
society. But how does one measure the collective damage
imposed by a disease, and could it be compared with the “burden” imposed by other diseases or social ills? How do policy
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
makers, researchers, or society in general know which diseases
to prioritize? How much does malnutrition increase the existing burden of other diseases?
These are the types of questions that “burden of illness”
studies help to answer. By better understanding the overall economic, or social, burden of a disease, we may better identify
important areas for research and investment of resources. The
impacts of disease are complex and multifaceted, though,
requiring methods for identifying—and differentiating—these
components without “double counting.”
The greatest costs from illness result from losses in the
quantity and quality of life. Through increased morbidity and
mortality, disease shortens the life expectancies of many of
the ill and decreases the quality of life lived with disease.
Disease also causes direct medical costs from treatment. In
addition, disease has impacts on employment and productivity, burden on caregivers, and other socioeconomic effects.
Clearly, an accurate estimate of all of these societal costs from
disease is ambitious at best, but even a partial estimate is often
Malnutrition may be due to socioeconomic forces as well as
disease. Historically, applications of burden of illness modeling in the nutrition context have focused on undernutrition in
the developing world, especially among children and women
of childbearing age.9 For example, the World Health
Organization (WHO) conducted a study in 2005 to estimate the
burden of malnutrition in children younger than 5 years. The
report outlines the WHO estimation process for calculating the
economic consequences of malnutrition and provides a numerical example using Nepalese children as a sample population.10
In this case, the burden is expressed not in monetary terms but
in terms of disability-adjusted life years (DALYs) lost to malnutrition. As discussed below, DALYs—or, alternatively, quality-adjusted life years (QALYs)—are a common measure of
disease impact that incorporates both morbidity and mortality.
To compare the burden of a disease with the costs of its
treatment, however, it is important to understand how society
values the losses from disease. Losses due to morbidity and
mortality can also be expressed in monetary terms, which
allow for an aggregated estimate of burden that also includes
direct medical costs, lost productivity, and other impacts of
disease. Approaches to understanding how to value losses in
quality or quantity of life are discussed in a later section; examples of the application of such valuation are provided here.
Inotai et al5 developed such an economic model to estimate
the total economic burden of disease-associated malnutrition
(DAM) associated with 10 primary diseases in Europe. To do
so, the authors broke the economic burden of the diseases into
3 components—acute phase, chronic phase, and premature
mortality due to malnutrition—and they included model
parameters such as increased length of stay, medical nutrition
costs, caregiver burden, and health loss, as well as epidemiological factors such as disease incidence and malnutrition prevalence. The authors estimated the burden of DAM in Europe to
be €305 billion annually. In this issue of Journal of Parenteral
and Enteral Nutrition, Snider et al11 present a similar analysis
of the burden of DAM in the community setting in the United
States, using nationally representative data on the prevalence
of diagnosed and undiagnosed disease to estimate disease and
disease-specific malnutrition incidence using real-world data.
The authors find that DAM in the included diseases imposes a
burden of $156.7 billion per year in the United States.
A common criticism of burden of illness studies is that they
invariably find that a given disease imposes a large economic
burden on society but that these analyses do not consider the
cost of reducing this burden or the effectiveness of available
treatments. Both of these dimensions are discussed below. It is
worth noting, however, that strong evidence exists to suggest
that interventions and treatments for malnutrition can be both
low cost and highly effective, making the message of such
studies more compelling and more important for motivating
public concern about the issue.
What Is the Effectiveness of a Nutrition
Intervention, and How Does It Compare
With Its Efficacy?
Knowing that a disease, such as malnutrition, imposes a large
burden on society is only useful if there is something that can be
done about it. One thing that can be done is to conduct research
to discover tools to reduce the burden. Understanding the relative burden from different diseases—and therefore the urgency
of treating them—helps to inform the allocation of resources for
research and treatment development. Once those tools exist,
though, how does one determine whether they successfully treat
the targeted disease? In other words, how does one know
whether the tools are efficacious (ie, achieving the desired treatment effect in a controlled setting) and effective (ie, achieving
the desired treatment effect in a real-world setting)?
Efficacy measures the performance of an intervention under
ideal and controlled circumstances12; it is determined through
RCTs. In RCTs, as discussed above, the decision of whether to
treat a particular patient is made through the randomization of
patients into treatment and control arms. After randomization,
differences in underlying patient health status and other personal characteristics should be small, enabling the identification of the treatment effect. In addition, RCTs eliminate other
factors present in real-world settings that could potentially
affect the perceived effect of a treatment, including patient
access, provider prescribing behaviors, and adherence.12
In contrast, effectiveness measures the performance of an
intervention under “real-world” conditions. These studies may
be more applicable to predicting expected outcomes for realworld patients. For example, patients may be compared with
standard of care rather than a placebo arm, which is a typical
approach in RCTs. Limitations of effectiveness studies stem
from the fact that the real world is not as controlled or homogeneous as the environment of an RCT. The introduction of
Downloaded from by guest on March 5, 2015
Philipson et al
endogenous factors in the real world that are not present in
RCTs could bias the final effectiveness estimate.12
Numerous trials of efficacy in nutrition have been conducted; however, the nature of nutrition poses some unique
challenges for researchers seeking to design and conduct
RCTs. Because nutrition is a fundamental requirement of
everyday life, the ethics of providing a placebo are not straightforward. Moreover, because nutrition—unlike, say, drugs or
devices—is readily available in various forms, adherence can
be harder to measure and maintain. Finally, because of the
nature of the treatment, nutrition trials are often not blinded. In
a recent review of reviews of nutrition efficacy trials, Stratton
and Elia13 discussed these issues and commented on the practical difficulties and ethical issues that limit the ability of nutrition trials to score highly on metrics developed to measure
RCT quality for pharmaceuticals.
Of course, methodological challenges are not limited to
efficacy studies. The potential for endogenous factors to bias
an analysis of effectiveness in a real-world setting is considerable. Fortunately, the various disciplines that form the basis for
HEOR offer a variety of tools for addressing selection bias.
With a careful study design, researchers can often control for
selection bias and measure effectiveness in the real-world setting. We illustrate these tools using an example in nutrition: a
study by Philipson et al14 on the impact of oral nutrition supplements (ONS).
Philipson and coauthors14 sought to identify the effect of
providing ONS to patients in the hospital. To do this, the
authors had to grapple with the issue that, far from being provided at random, ONS was generally given to patients who
were considerably older and less healthy than the average hospital patient. That is, the decision of whom to provide ONS
was subject to strong selection bias. The authors employed 2
techniques to address this bias.
The first technique used by Philipson et al14 was propensity
score matching. Propensity score matching addresses the issue
of unbalanced covariates across treatment and control groups
in observational studies or, in the case of this example, the fact
that the ONS users were older and less healthy on observed
dimensions. It is particularly useful when the number of influential covariates is great enough that matching on individuals
covariates (eg, age and sex) would be impractical.15 If the relevant differences between treated and untreated individuals are
observed, propensity score matching can be employed to
obtain unbiased estimates of the effect of treatment.16,17
Propensity score matching is most commonly performed by
estimating a logistic regression model of the likelihood that a
given individual receives the treatment, as a function of observable characteristics. This regression produces an estimated
probability that each individual in the sample would be given
the treatment. After these estimated probabilities are obtained,
each treated individual is matched to one or more untreated
individuals who had a similar estimated probability of receiving
the treatment. In the Philipson et al14 example, propensity score
matching was used to match each episode in which the patient
received ONS to an episode in which the patient appeared
equally likely to receive ONS (due to similar age, sex, comorbidities, etc) but in fact did not receive ONS. In this way, the
treatment and control groups were made more comparable.
Another example of propensity score matching can be
found in this special issue in the article by Hamdy et al.18 This
study examines the effects of providing glycemia-targeted specialized nutrition (GTSN) to patients with diabetes. By using
propensity score matching, Hamdy et al were able to restrict
the sample of hospitalized patients with diabetes to those who
appeared similarly likely to receive GTSN. This process served
to render the treatment and control groups more comparable.
If propensity score matching is to be used to estimate the
causal effect of a treatment, the researchers must be confident
that all the relevant difference between treated and controlled
individuals can be observed. However, even after matching
treatment and control cases to account for observable differences, crucial unobservable differences may remain, with the
potential to confound the analysis.19 Returning to the Philipson
et al14 ONS example, imagine 2 patients who are identical
along observable dimensions—same age, race, comorbidities,
history of prior hospitalization, and so on. A clinician who
walks into these patients’ rooms to evaluate and plan their
respective courses of treatment is likely to perceive 2 very different patients, despite their similarity in the data. It is reasonable to expect that the clinician’s decision of whether to
provide ONS would be related to the patients’ individual
needs, and therefore, the patients’ clinical outcomes would
result from both their underlying health status and whether
they received ONS. How, then, can we separate these underlying variables from the effect of treatment to determine treatment effectiveness?
HEOR offers several techniques to address bias from unobserved factors. We discuss here a technique called instrumental
variables (IV) analysis, which was employed in the Philipson
et al14 study. Other common techniques include difference-indifference and regression discontinuity design; sources on
these methods are provided in the Further Reading section at
the end of this tutorial.
IV analysis essentially serves as a quasi-randomization
device to sort individuals into treatment and controls groups
with similar observed and unobserved characteristics. It is useful when unobserved characteristics are influential in the treatment decision, and significant heterogeneity in unobserved
characteristics is present even among those who are similar on
observed dimensions. Returning to the Philipson et al14 ONS
example, consider 2 patients who are both 71-year-old women
with comorbid chronic obstructive pulmonary disease (COPD),
one with a body mass index (BMI) of 24 and the other with a
BMI of 16. If we observed only their age, sex, and comorbid
COPD, we would not know the reason why one received ONS
and the other did not (due to unobserved underweight). If we
were to estimate the effects of the intervention without taking
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
Table 2. Types of Cost Analyses.
Study Type
analysis (CEA)
An economic analysis that compares
the costs and outcomes of 2
alternative interventions.
analysis (CBA)
An economic analysis that compares
the costs of an intervention with its
benefits, both expressed in monetary
terms. CBA can also be used to
compare 2 alternative interventions.
Approach to Benefits
Benefits of intervention could
be infections averted, qualityadjusted life years (QALYs)
gained, and life years saved.
Benefits converted to monetary
The result of a nutrition screening
program, in terms of its cost and
life years saved, is compared
with the standard of care.
The result of a nutrition screening
program is quantified in terms
of monetized benefits and costs,
as well as compared with the
monetized benefits and costs of
the standard of care.
this unobserved heterogeneity (the difference in BMI) into
account, we may even come to the erroneous conclusion that
providing ONS caused poorer health outcomes, when in fact
the intervention was given to the woman who faced a worse
prognosis (due to her low BMI) in the first place. Indeed, naive
regression analysis on this sample suggested that ONS was
causing patients to have longer hospital stays, more expensive
episodes, and a higher risk of being readmitted.
To remove confounding in such a situation, an instrument
must be found that predicts treatment but otherwise does not
affect outcomes except through its influence on the treatment. In
this case, the authors used hospitals’ propensity to provide ONS
as an instrument. Because some hospitals were more likely to
offer ONS to any given patient, regardless of the patient’s characteristics, while others were less likely, this variable essentially
simulates a randomization into ONS and non-ONS treatment
groups. To check the validity of the approach, the authors tested
the 2 requirements of a valid instrument: that it predicts the treatment but otherwise does not influence outcomes except through
its influence on the treatment. Specifically, they confirmed that
(1) hospital-level ONS usage rates were a strong predictor of the
individual patient’s ONS use, and (2) hospitals with high ONS
rates did not have (observably) healthier patients or better technology than did hospitals with low ONS rates.
Numerous other researchers have used the IV approach to
measure the effects of interventions. In this issue, Lakdawalla
et al20 employ a similar approach to evaluate the effectiveness
of ONS in pediatric patients. Outside of the nutrition space, IV
has been successfully employed to create natural experiments
in a wide range of contexts, including the estimation of supply
and demand,21 determining the effect of police force size on
crime,22 and calculating the effect of veteran status in the
Vietnam era on mortality.23
How Should Resources Be Directed to
Optimally Reduce Burden of Illness by
Improving Nutrition?
The previous subsection discussed how to gauge whether a
given intervention works, whether in the highly controlled
environment of an RCT (“efficacy”) or in the real world
(“effectiveness”). From a patient’s perspective, this may be all
the information needed to make treatment decisions—if a more
effective treatment exists, that is the treatment a patient would
be expected to want. However, from the broader perspective of
society or health systems, decision makers must determine
how to allocate limited resources to do the most good with a
given budget. To do this, some measure of the “good” an intervention produces must be weighed against its cost.
This is the domain of cost-effectiveness analysis (CEA) and
its closely related cousin, cost-benefit analysis (CBA). Both of
these study types are economic analyses that facilitate an
assessment of the result of an intervention compared with an
alternative intervention. The types differ in terms of how the
results of the intervention are measured, as shown in Table 2.
Both types of studies take into account the costs of an intervention; however, they differ in how they measure the benefits.
Costs should be equal to the value of the opportunity cost had
the resources been used elsewhere.24 For example, when an
individual visits the doctor’s office, the full cost of the treatment includes not only the cost paid for the visit (both by any
insurance and out-of-pocket), but also the cost of transportation from the individual’s home or workplace to the doctor’s
office, as well as the missed wages or home production that
individual did not earn because of the visit to the doctor’s
office. An intervention that improves an individual’s health
may not only reduce doctor’s visits and other healthcare utilization but also improve the individual’s productivity and
reduce caregiver burden. In practice, however, such “indirect”
costs of illness—those apart from the “direct” healthcare
costs—are difficult to measure and are therefore often left out
of CEA studies.
In CEA, benefits are measured in terms of health outcomes,
such as life years, QALYs, or DALYs saved, or cases of disease averted.25 In contrast, in CBA, both benefits and costs are
expressed in monetary terms, which makes it possible to incorporate diverse types of costs and benefits into a single calculation of net benefit.26 Both CBA and CEA can be used to
facilitate policy making,27 whether in the public or the private
sector. For health-related questions, CEA is often preferred to
Downloaded from by guest on March 5, 2015
Philipson et al
CBA because in CEA, the outcome of the intervention (eg,
QALYs saved or cases of disease averted) is transparent and
can be easily compared across studies.25,26,28 Unlike other types
of health interventions, however, nutrition affects every organ
of the body. Therefore, measuring the benefits of nutrition
along a single dimension can be difficult. For this reason, CBA
is often a more appropriate approach. This difficulty is compounded by the fact that, relative to the potential benefits, the
cost of nutrition interventions, especially in the clinical context, tends to be low. As a result, traditional CEA is difficult to
conduct, leading to CEA studies that focus on cost savings,
rather than cost per change in health outcomes. (Typically,
CEA studies report the cost required to attain a given incremental change in a health outcome [eg, $57,600 per life year].
Nutrition interventions often reduce costs while improving
health outcomes. In CEA, such interventions are called cost
saving, and a cost per unit of incremental change in health is
not reported.) Below, we discuss the use of CEA and CBA in
nutrition, highlighting a number of examples from the nutrition
Cost-effectiveness analysis in nutrition. As discussed above,
many of the CEA studies in nutrition focus on cost savings
rather than health outcomes. For example, Doig et al29 present
a cost impact comparison of early (within 24 hours of admission to the intensive care unit [ICU]) administration of enteral
nutrition (EN) with the provision of EN at any point after 24
hours. The authors use a computer simulation, built based on
evidence from the literature, to model the effects of early vs
later EN, using various measures of resource consumption
(ICU length of stay, duration of intubation, and hospital length
of stay) as an estimate of costs. The study found that early EN
resulted in $14,462 in cost savings per patient (95% confidence
interval [CI], US$5464–$23,669).
Modified CEA studies have also focused on oral nutrition
supplementation. One such analysis was performed on a patient
population in the Netherlands undergoing abdominal surgery.
The study demonstrated that ONS resulted in a 7.6% cost savings per patient. The 0.72-day reduction in length of stay was a
driver of the cost savings (8.3% cost savings per patient) and
was sufficient to cancel out the added costs of the ONS therapy.
Therefore, they were able to conclude that ONS is indeed a
cost-effective (indeed, cost-saving) treatment as opposed to
standard care without the nutrition supplement.30 In another
case, a budget impact analysis was conducted on an elderly
patient population in the Netherlands to assess whether adding
the expense of ONS to the nation’s healthcare budget was costeffective. The study affirmed that the use of ONS to treat disease-related malnutrition (DRM) leads to cost-savings, resulting
in €13 million in savings per year (a 4.7% reduction). The additional costs were offset by the reduction in readmission.31
Standardized and consistent reporting across CEA studies is
essential if CEAs are to be used to inform decision making.32
This is a major motivation for the development of cost utility
analysis (CUA), a type of CEA in which the outcome of an
intervention is measured in terms of years of life lived in perfect health (QALYs).33,34 QALYs weight life years by the
health-related quality of life experienced, which can be determined using any of several validated methods of health state
measurement.24,33 This standardized use of QALYs has the
effect of making studies more comparable.24 Studies can also
be more easily compared when they take a common perspective (eg, societal) and a common definition of costs (eg, only
“direct” costs).24 For example, a cost-utility analysis was performed to compare enteral tube feeding (ETF) in patients with
cerebrovascular accident (CVA) in 2 separate locations (home
vs nursing home). Using patient data from insurance charges
and adopting a payer perspective, the study demonstrated that
the cost-effectiveness ratio (cost/QALY) of patients receiving
long-term ETF in their own home (£12,817 per QALY) compares favorably with other interventions.35
Cost-benefit analysis in nutrition. Although CEA and CUA
are often preferred for comparison of health interventions,
CBA is also a valuable tool. CBA assigns a monetary value to
all costs and benefits—a difficult and often contentious process. However, stating all costs and benefits in monetary terms
allows decision makers to compare interventions or policies
that have multiple costs and benefits. Approaches for assigning
value to nonmonetary costs and benefits are discussed in the
next subsection.
CBA is particularly useful in circumstances where the benefits of an intervention are not easily measured along a single
dimension—in a program that not only affects the health of
individuals but also has economic benefits for a neighborhood, for example, or where improved health outcomes themselves lead to further reductions in medical costs or increased
economic productivity. Given the array of potential benefits
from improved nutrition, use of CBA in nutrition is often a
natural fit.
CBA may better convey the full value of nutrition interventions, as well, because it strives to capture all costs and benefits. In fact, the Copenhagen Consensus has repeatedly
considered programs designed to reduce hunger and malnutrition to be among the most promising in terms of the ratio of
benefits to costs.9,36-38
For example, a CBA of the Expanded Food and Nutrition
Education Program in Virginia compared the costs of the program with the benefit of prevented or delayed costs from
chronic diseases. Benefits and costs accrue over time; a delay
of disease onset for 10 years leads to 10 years of benefits, discounted over time. Comparing total benefits with total costs,
the authors found a benefit/cost ratio of $10.64/$1.00.39
Facilitating decision making: the incremental cost effectiveness ratio, or ICER. To decide whether a new intervention is
worth initiating in the place of an old one, CEA can be used to
produce a valuable tool, the incremental cost-effectiveness
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
A new nutrional regimen (regimen “A”) is introduced, which has been shown to improve
survival in a nutrionally at-risk populaon, relave to the standard of care (regimen “B”).
Cost: $10,000
Cost: $20,000
Life expectancy: 1.5 years
Cost: $10,000
Life expectancy: 0.5 years
1.5 years
Life expectancy: 1.0 years
0.5 years
per life year
Is $20,000 per life year “worth it”?
Figure 1. Hypothetical example of an incremental cost-effectiveness ratio (ICER) calculation.
ratio (ICER). An ICER is used to compare the cost to attain a
fixed amount of health benefits under one intervention, relative
to another intervention.24 The formula, illustrated in Figure 1,
is as follows:
Cost A − Cost B
Effect A − Effect B
Policy makers and other decision makers often compare the
performance of interventions in relation to a given ICER threshold. Internationally, use of a threshold of €50,000/QALY is
widely accepted as standard practice in CEA. For example, a
health economic analysis of a population of 114 malnourished
patients showed that a 3-month nutrition intervention with ONS
was cost-effective when its results were compared with the
international ICER benchmark of €50,000/QALY. Two different pricing scenarios for ONS—a minimum and maximum—
were used and the study confirmed that intervention with ONS
increases the quality of life in malnourished patients.40
Limitations of cost-effectiveness analysis. While CEA is a useful tool for healthcare decision making, it does have limitations.25 For example, indirect costs, such as foregone wages
and caregiver burden, can be substantial, but they are difficult
to measure and hence frequently excluded from CEA. Disagreement exists over the proper way to measure health outcomes; for example, there is not yet a consensus on whether to
age-adjust QALYs to account for different utilities experienced
across stages of life.
Moreover, even if indirect costs are obtained and agreement
is reached on how to measure health outcomes, there remains
the question of how much a given health outcome is “worth.”
For example, if a given intervention costs $10,000 to extend
life by 1 QALY, relative to the standard of care, is this a good
value? What if it costs $100,000? Or $1,000,000? Does using
an ICER threshold of €50,000/QALY accurately capture the
priorities or values of society? While different individuals—
and nations—would answer these questions differently, health
economists have developed some techniques for producing
“objective” answers. How the value of health is determined is
the subject of the next section.
Is a Given Intervention “Worth It”? How to
Determine Value
Cost-effectiveness analysis tells us how much it costs to use a
particular intervention to extend life by a given amount, but it
Downloaded from by guest on March 5, 2015
Philipson et al
does not tell us whether it is worth it to pay for the intervention.
While there is no one definitive answer on which technologies
are worth their cost, HEOR does offer methods that stakeholders can use to make these challenging but important decisions.
The value of life extension, or of increasing an individual’s
quality of life, depends on one’s perspective. On the individual
level, the value of a year of life would seem to be priceless—
imagine being told you would lose 1 year of your life unless
you paid for the 1 treatment that would preserve it. Without
limits on budget, you would likely pay any price. The same
holds true for extending life for family, friends, or patients.
In reality, however, we do face budget constraints. In addition, society faces decisions about trade-offs between life extension and cost for many millions of individuals; although we wish
to preserve the lives of all in society, we must make difficult
decisions in allocating resources. For this reason, economists
have developed the concept of the value of a statistical life year,
or the amount society appears to be willing to spend to extend
life by 1 year for an individual whose identity is unknown. After
adjusting for health-related quality of life, this concept then provides us with an estimate of the value of a QALY, which can then
be used in CEA or burden of illness models.
But how should we determine what the value of a QALY
should be? Health economists use several approaches to determine the value an individual places on health and survival.
The simplest approach is known as the human capital
approach. This approach calculates expected lifetime earnings
after adjustments based on one’s health status. While this
approach captures the effects of disease on productivity, it
misses important nuances of valuing one’s remaining life
years.41 For example, this suggests that younger people’s lives
may be more valuable, because they are expected to live longer, but other valuation methods indicate that older people may
value remaining life more.42 Furthermore, many believe that an
individual’s value to society is not adequately represented by
the amount that he or she earns.
Many believe that a better measure of how society values
health and life is seen through the trade-offs people are willing
to make. Studies of revealed preference use observed behavior
to estimate the willingness to assume risk, for example, through
the increased pay required for someone to accept a job with
greater risk of injury or death.
Alternatively, researchers may present subjects with tradeoff decisions that reveal their value of time alive or quality of
life. Such survey-based methods are the domain of contingent
valuation and conjoint analysis. Researchers ask respondents
about their willingness to give up one thing for another (contingent valuation) or rank their preferences for a range of options
(conjoint analysis) and then use regression analysis to learn
about respondents’ valuations of intangible items such as
health and the environment.43 Techniques have been developed
to improve accuracy when eliciting such valuations.44 These
methods are particularly useful in questions related to health.
For example, in studies of time trade-off and standard gambles,
researchers ask the subject to sacrifice time or risk of death,
respectively, in exchange for perfect health relative to his or
her current health state. In addition, this approach is extremely
valuable in determining the relative importance of healthrelated quality of life, which is necessary to accurately estimate
QALYs and perform cost-utility analysis.
How to Use This Tutorial
As we approach the end of this tutorial, we would like to offer
some guidance on how to use the content presented here. We
see 3 potential uses, depending on the aims of the reader. First,
for those interested in understanding the existing nutrition
HEOR evidence base, we offer tips for reading HEOR studies.
Second, for those looking to include HEOR concepts in nutrition research, we identify gaps in the literature and suggest
directions for future research. Third, for those making decisions about nutrition care, we provide examples to illustrate
how clinicians, providers, and payers can use HEOR in nutrition to improve patient outcomes.
Tips for Reading HEOR Studies
When reading an HEOR study, the key questions to ask are the
Are there important differences between treated and
untreated individuals? Or in other words, is the decision to treat some individuals and not others subject to
selection bias?
Given that HEOR usually relies on observational data, the
answer to question 1 will usually be yes. In this case, the reader
should proceed to question 2.
If so, how has the selection bias been addressed?
There are numerous techniques for addressing selection
bias in HEOR studies, as discussed above. Regardless of the
technique, assessing the validity of the study essentially comes
down to carefully considering whether the analyzed treatment
and control groups are really comparable.
For example, if the treatment is providing ONS, one might
ask whether the researchers accounted for the fact that individuals receiving ONS may be older and have more comorbidities than those not receiving ONS. Suppose that the answer
is yes, and those differences were controlled for. In that case,
the reader might ask whether more subtle differences such as
different BMI or laboratory values (indicating malnutrition)
were controlled for. Suppose the answer is no. In this case, the
study approach may still be valid if an appropriate technique
was performed to account for unobserved differences. In that
case, not only should an appropriate technique be used (eg,
instrumental variables, regression discontinuity design, difference-in-difference), but the authors should also make a reasonable case for its validity.
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
In practice, many poorly designed observational studies
suffer from insufficiently comparable treatment and control
groups. Therefore, in asking the above 2 questions, the reader
can go a long way toward gauging the quality of the work.
Additional reading on specific methods, techniques, and
concepts are provided below in a list of suggested further
without needing the approval of a physician. The rule is expected
to reduce healthcare costs and improve patient outcomes.55
In general, HEOR can be used to complement RCTs to
determine the most effective and cost-effective interventions in
the real-world setting. As the available evidence base continues to grow, the tools to improve patient care will continue to
become more powerful.
Directions for Future Research
As this tutorial has shown, HEOR studies in nutrition have
demonstrated the burden of malnutrition, shown the effectiveness and cost-effectiveness of various interventions to improve
nutrition status, and offered tools to gauge whether a given
intervention is “worth it.”
At the same time, gaps exist that future research can help to
fill. For example, although numerous studies have measured
the effects of nutrition interventions in the acute care setting,14,30,35,40,45-53 analyses focused on the effects of interventions in the community setting are less common. While many
studies have shown the health benefits and cost-effectiveness
of nutrition support (whether through ONS, EN, or other
means), questions remain about which type of nutrition and
what timing of nutrition would be best suited to an individual
patient’s circumstances. Research is needed to refine measures
of the benefits from better nutrition, both to individuals and to
society. In addition, an opportunity exists to apply the valuation methods described above—particularly contingent valuation and conjoint analysis—to better understand the impacts of
malnutrition on quality of life and how much patients value
healthy nutrition status. Finally, due to financial and practical
considerations noted above, RCTs in nutrition have often featured small and narrowly selected patient populations, leaving
room for well-designed HEOR studies to investigate how the
effects found in RCTs may generalize to a larger, more diverse,
and likely less adherent real-world patient population.
HEOR can help measure the burden of malnutrition, identify the
most effective interventions for malnourished or nutritionally atrisk patients, facilitate the comparison of the outcomes achieved
through various interventions relative to their costs, and provide
a framework for deciding whether a given intervention is “worth
it.” Such studies may be particularly useful in nutrition, given
that nutrition interventions often come at a modest cost relative
to their health benefits. In a world where private and public payers alike increasingly demand evidence to justify a therapy’s
effectiveness and value, HEOR provides a tool for demonstrating the value of nutrition as a fundamental pillar of health.
Using HEOR to Improve Patient Outcomes
HEOR is not solely a tool for researchers; it can also be used to
allocate resources to achieve the best health outcomes on a
fixed budget, direct patients to the most appropriate treatments
given real-world safety and effectiveness, and improve patient
quality of life.
For example, in the United Kingdom, the National Institute
for Health and Clinical Excellence (NICE) surveyed the available evidence to find ONS valuable for malnourished patients.
Consequently, NICE issued guidelines recommending the use
of ONS by clinicians.54 In the United States, after surveying
the evidence, the Centers for Medicare & Medicaid Services
(CMS) recently issued a new rule that registered dietitian nutritionists be provided more independence in the hospital. Under
this rule, registered dietitian nutritionists will be able to manage patients’ diets and order nutrition-related laboratory tests
The authors thank Nick Summers for research support.
Further Reading
Russell CA. The impact of malnutrition on healthcare costs and economic considerations for the use of oral nutritional supplements. Clin Nutr Suppl.
Burden of Illness
Rice DP. Estimating the cost of illness. Am J Public Health Nations Health.
Zhao F-L, Gao L, Li S-C. Burden of disease studies in the Asia-Pacific region:
are there enough being performed to provide information for evidencebased health policy? Value Health Region Issues. 2013;2(1):152-159.
Efficacy vs Effectiveness
Gartlehner G, Hansen RA, Nissman D, Lohr KN, Carey TS. Criteria for
Distinguishing Effectiveness From Efficacy Trials in Systematic Reviews.
Rockville, MD: Agency for Healthcare Research and Quality; April 2006.
Technical Review 12 (AHRQ Publication No. 06-0046).
Regression Analysis Overview
Wooldridge J.Introductory Econometrics: A Modern Approach. 4th ed. Mason,
OH: South-Western Cengage Learning; 2009.
Zöphel M, Egger C, Riedi H. A Short Critical, Non-Technical, Non-Mathematical
Paper About Regression Analysis. Munich, Germany: GRIN Publishing
GmbH; 2008:
Instrumental Variables
Brookhart MA, Rassen JA, Schneeweiss S. Instrumental variable methods in
comparative safety and effectiveness research. Pharmacoepidemiol Drug
Saf. 2010;19(6):537-554.
Cawley J, Meyerhoefer C. The medical care costs of obesity: an instrumental
variables approach. J Health Econ. 2012;31(1):219-230.
Martens EP, Pestman WR, de Boer A, Belitser SV, Klungel OH. Instru­mental
variables: application and limitations. Epidemiology. 2006;17(3):260-267.
Donald SG, Lang K. Inference with difference-in-differences and other panel
data. Rev Econ Stat. 2007;89(2):221-233.
Downloaded from by guest on March 5, 2015
Philipson et al
Grafova IB, Freedman VA, Lurie N, Kumar R, Rogowski J. The difference-indifference method: assessing the selection bias in the effects of neighborhood environment on health. Econ Hum Biol. 2014;13:20-33.
Gruber J, Poterba J. Tax incentives and the decision to purchase health
insurance: evidence from the self-employed. Q J Econ. 1994;109(3):
Regression Discontinuity Design
Imbens GW, Lemieux T. Regression discontinuity designs: a guide to practice.
J Econometrics. 2008;142(2):615-635.
Linden A, Adams JL, Roberts N. Evaluating disease management programme
effectiveness: an introduction to the regression discontinuity design. J
Eval Clin Pract. 2006;12(2):124-131.
Thistlethwaite DL, Campbell DT. Regression-discontinuity analysis: an alternative to the ex post facto experiment. J Educ Psychol. 1960;51(6):309.
Valuing Life Extensions
Hirth RA, Chernew ME, Miller E, Fendrick AM, Weissert WG. Willingness
to pay for a quality-adjusted life year: in search of a standard. Med Decis
Making. 2000;20(3):332-342.
Rosen S. The value of changes in life expectancy. J Risk Uncertainty.
Ubel PA, Hirth RA, Chernew ME, Fendrick AM. What is the price of life
and why doesn’t it increase at the rate of inflation? Arch Intern Med.
Burden of illness study—designed to quantify the overall burden faced by
patients diagnosed with a particular disease, often aggregated to a societal level and presented in monetary values. Comprehensive estimates
consider both direct costs (eg, cost of treatment, morbidity, mortality)
as well as indirect costs (eg, caregiver burden, productivity effects,
quality of life).
Cost-benefit analysis—economic analysis technique used to compare the
costs and benefits of an intervention, where both values are expressed
in monetary terms.
Cost-effectiveness analysis—economic analysis technique used to compare the relative costs and outcomes of 2 or more alternative
Difference-in-differences—statistical analysis technique used to imitate an
experimental design with observational data by calculating a change over
time for a given outcome within a treatment group and comparing it with
a change over time for the same outcome within a control group.
Discounting—financial analysis technique used to adjust monetary estimates to account for the fact that individuals consider $1 in the present
time to be more valuable than $1 received in the future. Discounting can
also be applied to health outcomes (eg, life years). For example, it
would commonly be considered more valuable to save a year of life
now rather than to save a year of life 100 years in the future.
Effectiveness—measure of how well an intervention performs in a realworld setting.
Efficacy—measure of how well an intervention performs in a controlled
setting (ie, clinical trial or laboratory setting).
Endogenous—economic term used to describe variables included in a statistical model that are correlated with error term. Explanatory variables
may be endogenous if they not only affect but are also affected by the
outcome (ie, if causality runs in both directions). In health studies, treatment may be endogenous due to selection bias. For example, a nutrition
intervention may increase BMI, but it may be given to individuals with
low BMI—meaning that causality runs in both directions. If appropriate
steps are not taken to address the endogeneity, estimating a model with
endogenous explanatory variables will lead to biased estimates of the
treatment effect.
Exogenous—economic term used to describe variables determined outside
of a statistical model. Exogenous variables may influence the outcome of
a statistical model, but they are not influenced by it. For example,
sunshine is exogenous to crop growth. Sunshine influences crop growth,
but crop growth does not influence sunshine.
Instrumental variables analysis—statistical analysis technique used to
imitate an experimental design with observational data by including an
instrumental variable in the model that is correlated with the treatment
but does not otherwise influence the study outcome except through its
influence on the treatment.
Observational study—designed to estimate treatment effects in a real-world
setting using data on observed behaviors (eg, insurance claims, hospital
records, etc) where the assignment of the intervention is not conducted by
the study investigator. There are multiple observational study designs
such as cross-sectional studies, longitudinal studies, and case-control
studies, among others.
Propensity score matching—statistical matching technique used to mitigate the possible bias of real-world treatment selection by identifying a
control sample that has similar observed demographics and health characteristics to the treatment sample.
Randomized controlled trial (RCT)—designed to estimate treatment efficacy in a controlled setting where the assignment of the intervention
and control arms is decided randomly by the study investigator.
Regression analysis—statistical analysis technique used to assess the relationship among variables by predicting the effect that changing a given
variable has on an outcome of interest, all else equal.
Regression discontinuity design—an observational study design that mimics a randomized controlled trial by comparing an outcome of interest
among individuals just above and just below a given arbitrary threshold
that is used in policy. For example, suppose a hospital provides nutrition
support to all individuals with BMI below 18.5. One could study the
effect of nutrition support on outcomes by comparing outcomes among
individuals with BMI just below and just above the cutoff.
1. Ferreira IM, Brooks D, White J, Goldstein R. Nutritional supplementation
for stable chronic obstructive pulmonary disease. Cochrane Database Syst
Rev. 2012;12:CD000998.
2. Dunbar-Jacob J, Mortimer-Stephens MK. Treatment adherence in chronic
disease. J Clin Epidemiol. 2001;54(suppl 1):S57-S60.
3. Kothawala P, Badamgarav E, Ryu S, Miller RM, Halbert RJ. Systematic
review and meta-analysis of real-world adherence to drug therapy for
osteoporosis. Mayo Clin Proc. 2007;82(12):1493-1501.
4. Sajatovic M, Valenstein M, Blow FC, Ganoczy D, Ignacio RV. Treatment
adherence with antipsychotic medications in bipolar disorder. Bipolar
Disord. 2006;8(3):232-241.
5. Inotai A, Nuijten M, Roth E, Hegazi R, Kaló Z. Modelling the burden of
disease associated malnutrition. ESPEN J. 2012;7(5):e196-e204.
6. Ha L, Hauge T, Spenning AB, Iversen PO. Individual, nutritional support prevents undernutrition, increases muscle strength and improves QoL
among elderly at nutritional risk hospitalized for acute stroke: a randomized, controlled trial. Clin Nutr. 2010;29(5):567-573.
7. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med. 2000;342(25):1878-1886.
8. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med.
9. Horton S, Alderman H, Rivera JA. The Challenge of Hunger and
Malnutrition. Copenhagen Consensus Center 2008 Challenge Paper.
2008. Accessed June 18, 2014.
10. Snider J, Linthicum M, Wu Y, et al. Economic burden of communitybased disease-associated malnutrition in the United States. JPEN J
Parenter Enteral Nutr. 2014;38(supp 2):77S-85S.
11. Blössner M, de Onís M. Malnutrition: Quantifying the Health Impact
at National and Local Levels. Geneva, Switzerland: World Health
Organization; 2005.
Downloaded from by guest on March 5, 2015
Journal of Parenteral and Enteral Nutrition 38(Suppl 2)
12. Singal AG, Higgins PD, Waljee AK. A primer on effectiveness and efficacy trials. Clin Transl Gastroenterol. 2014;5:e45.
13. Stratton RJ, Elia M. A review of reviews: a new look at the evidence
for oral nutritional supplements in clinical practice. Clin Nutr Supp.
14. Philipson TJ, Snider JT, Lakdawalla DN, Stryckman B, Goldman DP.
Impact of oral nutritional supplementation on hospital outcomes. Am J
Manag Care. 2013;19(2):121-128.
15. Dehejia RH, Wahba S. Propensity score–matching methods for nonexperimental causal studies. Rev Econ Stat. 2002;84(1):151-161.
16. Rosenbaum PR, Rubin DB. The central role of the propensity score in
observational studies for causal effects. Biometrika. 1983;70(1):41-55.
17. Perkins SM, Tu W, Underhill MG, Zhou XH, Murray MD. The use of propensity scores in pharmacoepidemiologic research. Pharmacoepidemiol
Drug Saf. 2000;9(2):93-101.
18. Hamdy O, Ernst FR, Baumer D, Mustad V, Partridge J, Hegazi R.
Differences in resource utilization between patients with diabetes
receiving glycemia-targeted specialized nutrition vs standard nutrition
formulas in US hospitals. JPEN J Parenter Enteral Nutr. 2014;38(supp
19. Caliendo M, Kopeinig S. Some practical guidance for the implementation
of propensity score matching. J Econ Surv. 2008;22(1):31-72.
20. Lakdawalla DN, Mascarenhas M, Jena AB, et al. Impact of oral nutrition
supplements on hospital outcomes in pediatric patients. JPEN J Parenter
Enteral Nutr. 2014;38(supp 2):42S-49S.
21. Wright PG. The Tariff on Animal and Vegetable Oils (Appendix B). New
York: Macmillan; 1928.
22. Levitt SD. Using electoral cycles in police hiring to estimate the effect of
police on crime. Am Econ Rev. 1997;87(3):270-290.
23. Angrist JD, Imbens GW, Rubin DB. Identification of causal effects using
instrumental variables. J Am Stat Assoc. 1996;91(434):444-455.
24. Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB.
Recommendations of the panel on cost-effectiveness in health and medicine. JAMA. 1996;276(15):1253-1258.
25. Garber AM, Phelps CE. Economic foundations of cost-effectiveness analysis. J Health Econ. 1997;16(1):1-31.
26. Weinstein MC, Stason WB. Foundations of cost-effectiveness analysis for
health and medical practices. N Engl J Med. 1977;296(13):716-721.
27. Hodgson TA, Meiners MR. Cost-of-illness methodology: a guide to
current practices and procedures. Milbank Mem Fund Q Health Soc.
28. Russell LB, Gold MR, Siegel JE, Daniels N, Weinstein MC. The
role of cost-effectiveness analysis in health and medicine. JAMA.
29. Doig GS, Chevrou-Severac H, Simpson F. Early enteral nutrition in critical illness: a full economic analysis using US costs. Clinicoecon Outcomes
Res. 2013;5:429-436.
30. Freijer K, Nuijten MJ. Analysis of the health economic impact of medical
nutrition in the Netherlands. Eur J Clin Nutr. 2010;64(10):1229-1234.
31. Freijer K, Nuijten MJ, Schols JM. The budget impact of oral nutritional
supplements for disease related malnutrition in elderly in the community
setting. Front Pharmacol. 2012;3:78.
32. Siegel JE, Weinstein MC, Russell LB, Gold MR. Recommendations for
reporting cost-effectiveness analyses. JAMA. 1996;276(16):1339-1341.
33. Gold MR, Stevenson D, Fryback DG. HALYS and QALYS and DALYS,
oh my: similarities and differences in summary measures of population
Health. Annu Rev Publ Health. 2002;23:115-134.
34. Dolan P, Gudex C, Kind P, Williams A. Valuing health states: a comparison of methods. J Health Econ. 1996;15(2):209-231.
35. Reddy P, Malone M. Cost and outcome analysis of home parenteral and
enteral nutrition. JPEN J Parenter Enteral Nutr. 1998;22(5):302-310.
36. Behrman JR, Alderman H, Hoddinott J. The Challenge of Hunger and
Malnutrition. Copenhagen Consensus Center 2004 Challenges and
Opportunities. 2004.
files/CP%2B-%2BHunger%2BFINISHED.pdf. Accessed June 18, 2014.
Bhagwati J, Bourguignon F, Kydland FE, et al. Copenhagen Consensus
2008—Results. Copenhagen Consensus Center 2008 Results. 2008. http://
pdf. Accessed June 18, 2014.
Bhagwati J, Fogel R, Frey B, et al. Copenhagen Consensus 2004: The
Results. Copenhagen Consensus Center 2004 Results. 2004. http://www. Accessed June 18, 2014.
Rajgopal R, Cox RH, Lambur M, Lewis EC. Cost-benefit analysis indicates the positive economic benefits of the expanded food and nutrition
education program related to chronic disease prevention. J Nutr Educ
Behav. 2002;34(1):26-37.
Norman K, Pirlich M, Smoliner C, et al. Cost-effectiveness of a 3-month intervention with oral nutritional supplements in disease-related malnutrition: a
randomised controlled pilot study. Eur J Clin Nutr. 2011;65(6):735-742.
Hodgson TA, Meiners MR. Cost-of-illness methodology: a guide to
current practices and procedures. Milbank Mem Fund Q Health Soc.
Philipson TJ, Becker G, Goldman D, Murphy KM. Terminal care and the
value of life near its end. Cambridge, MA: National Bureau of Economic
Research; 2010. NBER Working Paper Series No. 15649.
Stevens TH, Belkner R, Dennis D, Kittredge D, Willis C. Comparison
of contingent valuation and conjoint analysis in ecosystem management.
Ecol Econ. 2000;32(1):63-74.
Arrow KJ, Solow R, Portney PR, Leamer E, Radner R, Schuman H.
Report of the NOAA Panel on Contingent Valuation. Washington, DC:
National Oceanic and Atmospheric Administration; 1993.
de Lucas C, Moreno M, Lopez-Herce J, Ruiz F, Perez-Palencia M, Carrillo
A. Transpyloric enteral nutrition reduces the complication rate and cost in
the critically ill child. J Pediatr Gastr Nutr. 2000;30(2):175-180.
Doig GS, Chevrou-Severac H, Simpson F. Early enteral nutrition in critical illness: a full economic analysis using US costs. Clinicoecon Outcomes
Res. 2013;5:429-436.
Elia M, Stratton RJ. A cost-utility analysis in patients receiving enteral tube
feeding at home and in nursing homes. Clin Nutr. 2008;27(3):416-423.
Amaral TF, Matos LC, Tavares MM, et al. The economic impact
of disease-related malnutrition at hospital admission. Clin Nutr.
Braunschweig C, Gomez S, Sheean PM. Impact of declines in nutritional
status on outcomes in adult patients hospitalized for more than 7 days. J
Am Diet Assoc. 2000;100(11):1316-1322.
Chima CS, Barco K, Dewitt ML, Maeda M, Teran JC, Mullen KD.
Relationship of nutritional status to length of stay, hospital costs, and
discharge status of patients hospitalized in the medicine service. J Am
Diet Assoc. 1997;97(9):975-980.
Correia MI, Waitzberg DL. The impact of malnutrition on morbidity, mortality, length of hospital stay and costs evaluated through a multivariate
model analysis. Clin Nutr. 2003;22(3):235-239.
Efthimiou J, Fleming J, Gomes C, Spiro SG. The effect of supplementary
oral nutrition in poorly nourished patients with chronic obstructive pulmonary disease. Am Rev Respir Dis. 1988;137(5):1075-1082.
Kruizenga HM, Van Tulder MW, Seidell JC, Thijs A, Ader HJ, Van
Bokhorst-de van der Schueren MA. Effectiveness and cost-effectiveness
of early screening and treatment of malnourished patients. Am J Clin Nutr.
National Collaborating Centre for Acute Care. Nutrition Support for Adults:
Oral Nutrition Support, Enteral Tube Feeding and Parenteral Nutrition.
London, UK: National Collaborating Centre for Acute Care; 2006.
Centers for Medicare & Medicaid Services (HHS). Medicare and Medicaid
programs; regulatory provisions to promote program efficiency, transparency, and burden reduction; part II; final rule. Fed Regist. 2014;91:
Downloaded from by guest on March 5, 2015