You are on page 1of 7

CONDUCTING

THE

COST ANALYSIS

An Improved Set of Standards for Finding Cost


for Cost-Effectiveness Analysis
Paul G. Barnett, PhD

Background: Guidelines have helped standardize methods of costeffectiveness analysis, allowing different interventions to be compared and enhancing the generalizability of study findings. There is
agreement that all relevant services be valued from the societal
perspective using a long-term time horizon and that more exact
methods be used to cost services most affected by the study intervention. Guidelines are not specific enough with respect to costing
methods, however.
Method: The literature was reviewed to identify the problems
associated with the 4 principal methods of cost determination.
Findings: Microcosting requires direct measurement and is ordinarily reserved to cost novel interventions. Analysts should include
nonwage labor cost, person-level and institutional overhead, and the
cost of development, set-up activities, supplies, space, and screening. Activity-based cost systems have promise of finding accurate
costs of all services provided, but are not widely adopted. Quality
must be evaluated and the generalizability of cost estimates to other
settings must be considered. Administrative cost estimates, chiefly
cost-adjusted charges, are widely used, but the analyst must consider
items excluded from the available system. Gross costing methods
determine quantity of services used and employ a unit cost. If the
intervention will affect the characteristics of a service, the method
should not assume that the service is homogeneous.
Conclusions: Questions are posed for future reviews of the quality
of costing methods. The analyst must avoid inappropriate assumptions, especially those that bias the analysis by exclusion of costs
that are affected by the intervention under study.
Key Words: cost-benefit analysis, costs and cost analysis/methods,
hospital costs, hospital charges, research design
(Med Care 2009;47: S82S88)

ost-effectiveness analysis (CEA) is an economic method


used to determine if health care interventions yield sufficient improvement in health to justify their cost. CEA has
From the Health Economics Resource Center, US Department of Veterans
Affairs and Department of Health Research and Policy, Stanford University School of Medicine, Stanford, California.
Supported by the US Department of Veterans Affairs Health Services
Research and Development Service (ECN-99-017) and the VA Cooperative Studies Program.
Reprints: Paul G. Barnett, PhD, Health Economics Resource Center, 795
Willow Rd (152), Menlo Park, CA 94025. E-mail: paul.barnett@va.gov.
Copyright 2009 by Lippincott Williams & Wilkins
ISSN: 0025-7079/09/4700-0082

S82

| www.lww-medicalcare.com

the promise of improving health outcomes within the constraints of available funding. An important barrier to greater
use of CEA is decision makers lack of confidence in its
methods. Standardized methods of CEA have been developed
to address this concern.
At least 21 different guidelines for conducting CEA
have been developed by government panels, specialty
organizations, and others.1 Among the most widely cited
works are guidelines from a group from the United Kingdom and Canada2 and a panel organized by the United
States Public Health Service.3
The US panel described a reference case, a standard
method that should be used so that different studies are
comparable. Standardization allows the cost-effectiveness ratios of different interventions to be compared without concern
that differences are methodological artifacts. Standardization
also enhances the generalizability of study findings, allowing
them to be applied to new settings.
These various guidelines largely agree on the principles
to be used to determine costs. They recommend that all
relevant costs be included and that more exact methods be
used to determine the cost of services most affected by the
intervention under study. CEA should examine the incremental change in cost caused by the intervention, known as its
marginal cost.
Resources are to be valued at their opportunity cost,
sometimes called the economic cost. The opportunity cost is
the value of the resources applied to their next best use (ie, the
potential benefit from taking the opportunity to use the
resource in another way). Opportunity cost may differ from
price or reimbursement; health care markets function imperfectly, and these may not reflect the opportunity cost. CEA
guidelines also state that cost should be estimated from the
societal perspective.
Cost and benefits are to be measured over a long-term
time horizon, usually interpreted as the patients remaining
lifetime. There is controversy over whether to include the
cost of treatment of unrelated diseases when an intervention
prolongs survival.
Adopting the long-term time horizon affects the definition of marginal cost. Some costs are regarded as fixed over
the short-term. Examples are building construction, equipment purchases, or intermediate-term commitments such as
provider staffing. In the long-run, these allocations can be
adjusted to the optimal level. Long-run marginal cost includes these costs.
Medical Care Volume 47, Number 7 Suppl 1, July 2009

Medical Care Volume 47, Number 7 Suppl 1, July 2009

Some analysts feel that the existing guidelines are not


specific enough with respect to costing methods.1 Alternative
costing methods, each within CEA guidelines, have been
compared. In some cases, results were sensitive to the choice
of method.4,5
At least 54 different systematic reviews of economic
evaluations of health care have been conducted.6 Some found
modest improvements in the quality of economic analyses
since the promulgation of guidelines. These reviews found a
number of problems with cost-determination methods.
Many studies did not identify the cost of the intervention being evaluated7 or failed to provide disaggregated cost
information.8 Others did not describe the source of cost data9,10
or provided too little information to judge its quality.11 Some
studies failed to include relevant costs.12 Most studies excluded overhead cost.7 Many studies that included overhead
distributed it improperly12 or failed to describe how it was
distributed.13 Many US studies relied on local fee schedules,
without considering how these differed from opportunity
cost.8 Few studies included the start-up cost involved in
implementing a new intervention.7 Most studies had failed to
report cost over the long-term.9,10,14
Most reviews adopted a very low threshold for acceptability of costing methods. They employed only 1 or 2
questions to evaluate cost-determination methods.
This article describes costing methods available to US
researchers. Problems of each method are discussed. New
questions to evaluate cost-determination methods are then
proposed.

Improved Standards for Cost Finding

This article focuses on cost determination for CEA used


in clinical trials and decision analytic models. Its intended
audience is the economist who creates a CEA and the decision maker who wants to use one. Budget Impact Analysis is
a method of finding the effect of an intervention on the budget
of a payer or provider in the short-term. Although budget
impact analysis is an important, perhaps essential, adjunct to
CEA,15 it is beyond the scope of this article.

OVERVIEW OF METHODS USED TO ESTIMATE


HEALTH CARE COSTS
The choice between cost-determination methods represents a trade-off between precision and expense.16 18 Table 1
lists the principal methods, starting with the most accurate but
labor intensive method followed by those that compromise
accuracy for ease of use. These methods include microcosting, activity-based cost allocation systems, administrative
data (cost-adjusted charges or total reimbursement), and
gross costing.
Each method has its limits and its appropriate use.
Multiple methods are needed within a single study. Accurate
methods are needed to assess services that the intervention is
likely to affect. Simpler methods may be employed to avoid
spending scarce resources on precise measurement of unimportant services. Each method involves assumptions. The
analyst must review whether these assumptions are appropriate. An important additional concern for CEA studies is the
wider applicability of study findings to other providers, health
plans, or countries.

TABLE 1. Overview of Costing Methods Available to US Researchers


Method

Advantage

Disadvantage

Issues for Concern

Microcosting

Enumerate staff time, supplies,


and items used to provide a
specific service and estimate
their cost.

Accurate, often needed to find a


cost of a service intervention.

Method is labor intensive


and not useful for
finding overhead cost. It
cannot be used to find
total health care cost.

Activity-based cost
allocation system.

Multistep cost allocation system.


Assign cost of staff time,
supplies, and equipment to
production departments.
Distribute overhead. Use
relative values to find cost of
specific products and assign
cost to specific stays or
encounters.
Billed charges are adjusted by
the ratio of cost-to-charges in
a hospital cost report.

The best available estimate of


economic costs of health
services.

Used by relatively few


hospitals, data may not
be available to the
researcher. Costing
system unlikely to
capture costs of a novel
intervention tested in a
research study.

Need to include all costs:


nonwage labor cost,
person-level and
institutional overhead,
cost of development,
set-up, screening,
supplies, and space.
Quality must be
evaluated. Cost
estimates may not be
generalizable.

Charges routinely created for


most of US health care.
Hospital cost reports are
available from Medicare.

Requires strong
assumptions that charge
is proportional to
economic cost. Charges
difficult to obtain for
care received at other
sites. Difficult to use to
cost ambulatory care.
Strong assumptions about
homogeneity of
services.

Cost-adjusted charges or
total reimbursement.

Gross costing

Description of Method

Quantities of different services


are determined, and cost
estimated using service
specific unit cost.

2009 Lippincott Williams & Wilkins

Relative ease of
implementation.

Use of unadjusted charge.


Exclusion of cost of
physician services to
inpatients. Exclusion of
patient copayments and
deductibles from
reimbursement amount.
Data on characteristics of
service may be
inadequate. Use of
appropriate unit costs.

www.lww-medicalcare.com |

S83

Medical Care Volume 47, Number 7 Suppl 1, July 2009

Barnett

MICROCOSTING
Microcosting, the direct measurement of cost by observation and survey, is needed when administrative data are
insufficient. It is needed to find the cost of a novel intervention. It may be required if administrative data are not sensitive to the effect of an intervention. Because microcosting is
too labor intensive to use for all health care, its use must be
limited to activities most likely to be affected by the intervention
under study. The specific considerations in designing a microcosting method have been described previously19 21 and in
the current supplement.22 Systematic reviews of economic
studies have criticized microcost estimates for failing to
include all relevant costs, such as direct costs, overhead,
labor, and development cost. When microcost estimates are
based on the labor costs and overhead of a particular facility
or payer, the analyst will need to justify that they are typical
of the health care system as a whole.

Inclusion of All Direct Costs


CEA guidelines recommend inclusion of all costs affected by the intervention under study. All interventionrelated activities must be identified. Pretreatment screening
should be included as a cost if it would be needed to replicate
the intervention in clinical practice, for example, if patients
were screened to exclude those who would be harmed by the
intervention. The cost of screening that was exclusively for
research should not be included.

Institutional Overhead
In the short-run, a small expansion in patient care
services does not increase institutional overhead (eg, the cost
of nonpatient care hospital departments such as human resources, finance, administration, and environmental services).
Over the long-run (the perspective of CEA), the size of
overhead departments can be adjusted; adoption of an intervention ultimately increases institutional overhead. For hospitals and other large institutions, it is not feasible to use
microcosting to determine overhead. One approach is to
apply the ratio of overhead to direct expense for a similar
department in the hospital cost report.

Include All Time Spent in Providing


the Intervention
Labor cost is often estimated by determining the number of minutes each worker spends in direct activities to
provide the service. This effort is measured by direct observation, staff activity logs, supervisor report, or other methods.
One important issue is how to allocate activities that have a
joint purpose.
When a new provider is hired exclusively to provide an
intervention, it is clear that the cost of the intervention
includes the full cost of this new employee. Yet no provider
spends their full time in direct contact with patients. An
activity analysis would find the provider engaged in nonpatient care activities: training to maintain credentials, answering the phone, meeting with colleagues, taking vacation, and
going on sick leave. For ease of exposition, these nonpatient
care activities are referred to as person-level overhead.

S84

| www.lww-medicalcare.com

An intervention might also be delivered by a provider


who produces other products. For this provider, the personlevel overhead activities are a joint cost of producing the
intervention and other products. Some of this joint cost is
attributable to the intervention. Failure to include the cost of
person-level overhead implicitly assigns it to other products,
violating the general principal of CEA guidelines to include
all cost of the intervention. One method of distributing
person-level overhead is in proportion to the time spent in
direct delivery of each product.

Labor Costs Should Not Be Limited to Wages


The cost of labor should not be limited to wage costs.
It should also include the employers share of taxes and
benefits. Some studies have excluded these costs, which often
exceed 25% of wages.

Development Cost
Adoption of the societal perspective and a long-term
horizon requires inclusion of the development cost of all
interventions, even if this cost is not repaid to the developer.
Payer reimbursement for a pharmaceutical exceeds the shortterm marginal cost of manufacture and distribution. In the
long-run, the societal cost of a drug includes the cost of manufacture, distribution, innovation, and returns to investment.
Some analysts believe that development costs that are
sunk (ie, they have been incurred and cannot be recovered)
should be excluded from a cost-effectiveness analysis, but
this is a controversial point of view. The more common view
is that an evaluation that ignores development cost will be
biased in favor of the intervention.
The cost of developing new behavioral or organizational interventions may be small relative to the number of
potential beneficiaries. If the number of potential beneficiaries is large, the analyst may successfully argue that per
patient development cost may be small enough to be ignored.

ACTIVITY-BASED COSTING
Activity-based costing (ABC) was developed in the
manufacturing sector and more recently applied to health
services. ABC is similar to microcosting, but uses automated
data collection, handles overhead expenses, and includes all
costs of the enterprise.23
An ABC system is more complex than a traditional
hospital cost report, identifying costs, services, and products
at a much finer level of detail.24,25 Although evidence is slim,
ABC cost estimates are regarded as more accurate than
cost-adjusted charges (described later).
Departments are defined by aggregating similar activities. The cost of staff time, supplies, and equipment is determined for each department. The cost of nonpatient care departments is distributed to patient care departments using a
statistical basis (sometimes called a cost-driver) that is appropriate to that service. For example, the percentage of total
space may be the basis for distributing janitorial cost. The
quantity of health care products is extracted from electronic
databases, known as feeder systems. A schedule of relative
values is used to give proportionate weights to reflect the
different resources used to make different products. Depart 2009 Lippincott Williams & Wilkins

Medical Care Volume 47, Number 7 Suppl 1, July 2009

ment costs are divided by the sum of relative value weights of


all products made by the department. This cost per unit of
relative value is used to find the unit cost of each product.
Unit cost is multiplied by product quantities and product costs
summed to find the cost of each stay and encounter.
An example of an ABC system is a proprietary cost
allocation system called Decision Support System (also
known by the former name of the vendor, Transition Systems
International). This system has been implemented in the
United States. Department of Veterans Affairs,26 the health
care system that serves 5 million veterans of the US military
services. Another example is the Ontario case costing system
adopted by 12 hospitals and now being adopted more widely
in Canada.27 ABC systems have also been adopted by the
Kaiser health care system and a number of US university
hospitals.
ABC cost estimates have been used in a number of
studies of the cost of individual hospitals and in some
multicenter studies.23 An important limitation is that ABC
systems are used by no more than 5% of US hospitals.28 ABC
estimates can be difficult to obtain, regarded as confidential
information needed to negotiate reimbursement rates and
contracts.
Several potential sources of errors in ABC estimates
have been identified.23 If important intermediate products are
under counted or excluded altogether, inappropriately high
cost will be assigned to the products that are being tracked.
Distribution of overhead may be flawed. The relative value of
different products may be based on external data that have not
been validated.
ABC cost estimates may be inappropriately influenced
by short-term fluctuations in workload.29 Implementation of
ABC systems is often idiosyncratic, and this surely affects the
comparability of ABC data from different sites.
Given these concerns, the analyst must confirm that
assigned cost is plausible, determining if they are consistent
with alternate estimates. ABC systems can provide a source
of unit cost that is superior to a fee schedule or reimbursement rate, but estimates that are based on a single facility may
not be representative (a problem whenever unit costs are
based on a small sample, as discussed later). The analyst will
need to justify why ABC cost estimates are representative of
the larger health care system.

Improved Standards for Cost Finding

Commentators widely agree that the raw charge should


not be used as an estimate of the cost of care.31 Hospital
charges can be twice the actual cost of care.5 Use of costadjusted charges makes the strong assumption that the charge
for a specific service is proportionate to economic cost. This
assumption may not always be warranted. Hospitals may set
their charges without knowing the relative cost of different
services. There are strategic reasons to overcharge for some
services and undercharge for others.
Some analysts have found costing to be more accurate
if cost adjustment is done at the department level.5,32 A ratio
of cost-to-charges is found for each department in the hospital
and applied to charges incurred in that department. Other
analysts feel that this extra effort does not increase accuracy.
Departments can be defined differently in claims and cost
report data; use of department-level charge adjustment may
not result in any greater accuracy.30

Physician Services to Inpatients


Most US physicians who care for inpatients are not
employed by the hospital. The charge for their services is not in
the hospital bill and their cost is not in the hospital cost report.
This cost is an additional 20% above the hospital cost.
It is difficult to identify physician bills associated with
a particular stay. The US Medicare program reimburses hospitals using diagnosis-related groups (DRG), categories based on
admitting diagnosis, and procedures performed during the stay.
The mean Medicare payment for physician services associated with stays in each DRG has been estimated.33,34 These
values can be used as the cost of physician services to
inpatients under the assumption they are constant within each
DRG. The recent revision of DRGs will render these studies
obsolete.

Out-of-System Services
Concerns about patient privacy and the restrictions
imposed by Federal law make it difficult to obtain charge data
on care received outside of the system where the patient was
enrolled. A signed release is usually required. The alternative
is to base the estimate on a less precise method of gross
costing, described later.

Ambulatory Care

Cost-Adjusted Charges

Although charges for ambulatory care may be available, physicians do not file cost reports, and there is no source
of a ratio of cost-to-charges. Reimbursement may be the only
available way to estimate ambulatory cost. It should include
any copayment made by the patient. Cost-adjusted charges is
the cost incurred by the provider, while reimbursement is the
cost incurred by the payer. It is not clear which best represents the societal perspective, or whether they differ by much
in the long-run. In the absence of access to an administrative
data system, it is rarely feasible to gather ambulatory care
charges from providers and gross costing must be used.

Charges are cost adjusted by multiplying by a ratio of


cost-to-charges.30 This ratio may be determined from data in
publicly available cost reports that hospitals submit to Medicare. Over an entire hospital, the sum of cost-adjusted charges
is equal to the sum of costs in the cost report.

Gross costing is used when administrative data are not


available. The quantity of each service is multiplied by an
estimate of its unit cost. Service utilization is often obtained

COST ESTIMATES BASED ON


ADMINISTRATIVE DATA
The United States is one of the few countries that
assigns charges and reimbursements to characterize care.
Although they may not be available for services provided by
managed care organizations, they are widely used as a proxy
for health care cost.

2009 Lippincott Williams & Wilkins

GROSS COSTING

www.lww-medicalcare.com |

S85

Barnett

Medical Care Volume 47, Number 7 Suppl 1, July 2009

from the study participant. Gathering these data involves a


trade-off between accuracy and expense. Accuracy can be
improved by surveying patients more frequently and by
employing logs and other memory aids.35

It may be justifiable to characterize ambulatory care as


a hospital clinic visit, ambulatory surgery, emergency room
care, or a visit to a specialist or office-based primary care
physician, and assign a unit cost for each of these types of care.

Inpatient Cost Estimates

Outpatient Pharmacy Cost

Among the unit costs used to estimate the cost of


hospital stays are the average daily rate, the specialty-specific
daily rate, and the diagnosis-weighted rate.23 Use of an
average daily rate makes the assumption that all days of
hospitalization have the same cost. The specialty-specific
daily rate relaxes this assumption by creating a different rate
for different units of the hospital. The diagnosis-weighted
rate reflects the effect of DRG on the daily rate. The Medicare
DRG relative value weight is an index of cost relative to that
of an average stay. In one study, this weight explained 40%
of the variation in cost-adjusted charges, while length of stay
explained less than 5% of the variation.36
Cost may be estimated as the hypothetical Medicare
reimbursement. Medicare payment is not only based solely
on DRG, but also includes payments for high cost outliers,
capital expenses, graduate medical education, and disproportionate service to unsponsored patients.
Patients can provide information on the length of stay,
but are unlikely to provide sufficient information to assign a
DRG. Patients may be able to provide information on
whether the stay involved surgery or was in an intensive care
unit. These additional measures of acuity provide information
that can be used to determine costs in ways not accounted for
by length of stay alone.

The cost of prescription medications is rapidly growing, and now accounts for more than 10% of the US health
care budget. Few researchers have access to prescription
claims data. Estimates of pharmacy utilization are often based
on patient self-report. A common source of unit cost is the
pharmacy Red Book, an annual publication that provides the
average wholesale price of drugs dispensed in the United
States.39 Sponsors receive substantial discounts from this
price, at least 15% for brand name drugs.40 Payments also
reflect a dispensing fee.41 The full average wholesale price
should not be used as the unit cost for pharmacy. An alternative source is the Federal Supply Schedule, but the payment rates may be too low for findings to be generalized
beyond Federal providers.

Standard Cost Estimates


A set of standard unit cost estimates have been offered as
part of the guidelines used in the Netherlands and Australia.37
These unit costs have helped standardize estimates of health
services costs in CEA studies of new pharmaceuticals. Such
standard estimates have limitations. For example, a standard
estimate for an ambulatory visit may include the cost of
associated laboratory tests. This would ignore the incremental
effect of an intervention that generates additional laboratory
orders.

Cost of Ambulatory Care


Outpatient costs can be estimated by multiplying a
count of visits by a unit cost. Cost varies by the types of care
provided.
The analyst can also calculate a hypothetical reimbursement.38 Since Medicare is the largest sponsor of US care, its
reimbursement rules are often used. This requires care to be
characterized by Current Procedure and Terminology codes,
the billing codes used by physicians to identify their services.
The complexity of the coding system and reimbursement
rules must be respected. Multiple Current Procedure and
Terminology codes are often needed to characterize care, but
some codes are mutually exclusive. Different payment rates
apply when care is provided in a facility (such as a hospital
outpatient department or ambulatory surgery facility). When
multiple procedures are provided during a single surgery,
reimbursement is discounted.

S86

| www.lww-medicalcare.com

Limitations of Gross Costing Approach


CEA findings based on a fee schedule or unit cost of a
single provider may not be suitable for general application. If
hospital fee schedules are used, economic findings may
depend on the choice of hospitals and the number of hospital
schedules used.4
An important limitation of gross costing is the need for
measures that reflect the effect of the intervention. Even an
accurate method of assigning cost to hospital stays, such as a
DRG based-costing, will be inappropriate if the intervention
affects resource use without affecting the DRG assignment.
The analyst must honestly consider whether all effects of the
intervention on resource use are captured by the units of
measure.

CONCLUSIONS
Standard questions have been used to evaluate CEA
studies. These questions have identified quality problems, but
they are not sufficiently specific with respect to cost determination.
Table 2 proposes an enhanced checklist for evaluating
the quality of cost-determination methods. It represents implementation of general principles in guidelines: to include all
cost, adopt the societal perspective, use a long-term horizon,
and value resources at the opportunity cost. It considers the
weaknesses of each method, as identified in review of the
literature.
The questions are designed to help the reviewer identify
the analytic assumptions that have been made and whether
they are appropriate. The most important consideration is to
avoid bias in assessing the cost of services most directly
affected by the intervention under study.
The first question addresses choice among methods.
Most studies will employ both microcosting and a less resource intensive method. Since no guideline can anticipate
the complete range of possible studies, the question simply
asks if the analyst has sufficiently justified the choice of
methods. The trade-off between precision and research ex 2009 Lippincott Williams & Wilkins

Medical Care Volume 47, Number 7 Suppl 1, July 2009

Improved Standards for Cost Finding

TABLE 2. Questions for Enhanced Assessment of the Quality of Cost Determination Methods
Question

Microcosting

Activity-Based
Cost Allocation

Cost-Adjusted Charges or
Total Reimbursement

Gross
Costing

x
x
x

x
x
x

x
x
x

x
x
x

x
x

x
x

x
x

Was the choice of costing method(s) justified?


Were methods, data sources, and assumptions described?
If cost estimates were based on a specific payer or provider,
was evidence presented that they were representative of the
health system generally?
Did the cost of the intervention include person-level overhead,
institutional overhead, nonwage labor cost, set-up cost,
space, and supplies?
If a new intervention was being evaluated, was development
cost included?
If the intervention requires pretreatment screening, was the cost
of this screening included?
Did hospital cost reflect differences in diagnosis, surgery,
intensive care, length of stay, and setting? (eg, intensive,
psychiatric, or long-term care)
Were charges adjusted for the ratio of cost-to-charges?
Was the cost of physician services to inpatients included?
Did outpatient cost reflect heterogeneity of services?
Was nonphysician ambulatory care cost included?
Was outpatient pharmacy cost included?
Did pharmacy cost reflect discounts from wholesale price and
dispensing cost?

pense should be considered. Simplifying assumptions are


acceptable if they do not result in bias.
Questions regarding microcosting address the general
principles of including all relevant costs from a long-term
societal perspective. Screening, person-level overhead, facility overhead, start-up, and development costs should not be
neglected.
Few studies use data from ABC costing systems yet,
but those that do need to consider the quality of data and its
generalizability. US economic researchers have a unique
resource in billing data, but it must be appropriately used. The
analyst must take care not to ignore significant costs not
recorded in administrative data.
Gross costing is an important method appropriate for
many studies, but often involves strong assumptions. The analyst should eliminate any analytic assumptions that render the
costing method insensitive to the effect of the intervention.
Existing guidelines have identified important first principles that can be used to identify the cost of care for CEA.
These guidelines may be so broad that they may encompass
a range from success to failure. Systematic evaluations of the
quality of CEA studies have not considered the full range of
problems that can occur in cost determination. A more rigorous set of standards for the quality of cost-determination
methods will give health care decision makers greater confidence in CEA findings.
REFERENCES
1. Adam T, Koopmanschap MA, Evans DB. Cost-effectiveness analysis:
can we reduce variability in costing methods? Int J Technol Assess
Health Care. 2003;19:407 420.
2. Drummond MF, Sculpher MJ, Torrance GW, et al. Methods for the
Economic Evaluation of Health Care Programmes. Oxford: Oxford
University Press; 2005.

2009 Lippincott Williams & Wilkins

x
x

x
x
x
x
x

3. Gold MR, Siegel JE, Russell LB, et al. Cost-Effectiveness in Health and
Medicine. New York, NY: Oxford University Press; 1996.
4. Goeree R, Gafni A, Hannah M, et al. Hospital selection for unit cost
estimates in multicentre economic evaluations. Does the choice of
hospitals make a difference? Pharmacoeconomics. 1999;15:561572.
5. Taira DA, Seto TB, Siegrist R, et al. Comparison of analytic approaches
for the economic evaluation of new technologies alongside multicenter
clinical trials. Am Heart J. 2003;145:452 458.
6. Jefferson T, Demicheli V, Vale L. Quality of systematic reviews of
economic evaluations in health care. JAMA. 2002;287:2809 2812.
7. Balas EA, Kretschmer RA, Gnann W, et al. Interpreting cost analyses of
clinical interventions. JAMA. 1998;279:54 57.
8. Petrou S, Henderson J, Roberts T, et al. Recent economic evaluations of
antenatal screening: a systematic review and critique. J Med Screen.
2000;7:59 73.
9. Kruper L, Kurichi JE, Sonnad SS. Methodologic quality of cost-effectiveness analyses of surgical procedures. Ann Surg. 2007;245:147151.
10. Smith WJ, Blackmore CC. Economic analyses in obstetrics and gynecology: a methodologic evaluation of the literature. Obstet Gynecol.
1998;91:472 478.
11. Manuel MR, Chen LM, Caughey AB, et al. Cost-effectiveness analyses
in gynecologic oncology: methodological quality and trends. Gynecol
Oncol. 2004;93:1 8.
12. Ungar WJ, Santos MT. Quality appraisal of pediatric health economic
evaluations. Int J Technol Assess Health Care. 2005;21:203210.
13. Graves N, Walker D, Raine R, et al. Cost data for individual patients
included in clinical studies: no amount of statistical analysis can compensate for inadequate costing methods. Health Econ. 2002;11:735739.
14. Blackmore CC, Magid DJ. Methodologic evaluation of the radiology
cost-effectiveness literature. Radiology. 1997;203:8791.
15. Mauskopf JA, Sullivan SD, Annemans L, et al. Principles of good
practice for budget impact analysis: report of the ISPOR Task Force on
good research practices budget impact analysis. Value Health. 2007;10:
336 347.
16. Ramsey S, Willke R, Briggs A, et al. Good research practices for
cost-effectiveness analysis alongside clinical trials: the ISPOR RCTCEA Task Force report. Value Health. 2005;8:521533.
17. Reed SD, Friedman JY, Gnanasakthy A, et al. Comparison of hospital
costing methods in an economic evaluation of a multinational clinical
trial. Int J Technol Assess Health Care. 2003;19:396 406.
www.lww-medicalcare.com |

S87

Barnett

18. Luce B, Manning W, Siegel J. Estimating costs in cost-effectiveness


analysis. In: Gold MR, Siegel JE, Russell LB, et al, eds. Cost-Effectiveness in Health and Medicine. New York, NY: Oxford University Press;
1996.
19. Smith MW, Barnett PG. Direct measurement of health care costs. Med
Care Res Rev. 2003;60(suppl 3):74S91S.
20. Findorff MJ, Wyman JF, Croghan CF, et al. Use of time studies for
determining intervention costs. Nurs Res. 2005;54:280 284.
21. Johnston K, Buxton MJ, Jones DR, et al. Assessing the costs of
healthcare technologies in clinical trials. Health Technol Assess. 1999;
3:176.
22. Frick KD. Frontiers of micro-costing. Med Care. In press.
23. Azoulay A, Doris NM, Filion KB, et al. The use of the transition cost
accounting system in health services research. Cost Eff Resour Alloc.
2007;5:11.
24. Ross TK. Analyzing health care operations using ABC. J Health Care
Finance. 2004;30:120.
25. Udpa S. Activity-based costing for hospitals. Health Care Manage Rev.
1996;21:8396.
26. Barnett PG, Rodgers JH. Use of the decision support system for VA
cost-effectiveness research. Med Care. 1999;37(4 Suppl Va):AS63AS70.
27. Ontario Ministry of Health and Long-Term Care. Ontario Guide to Case
Costing. Ontario Case Costing Initiative Toronto. Ontario, Canada:
Ontario Ministry of Health and Long-Term Care; 2006.
28. Emmett D, Forget R. The utilization of activity-based cost accounting in
hospitals. J Hosp Mark Public Relations. 2005;15:79 89.
29. Cooper R, Kaplan RS. The promiseand peril of integrated cost systems. Harv Bus Rev. 1998;76:109 119.

S88

| www.lww-medicalcare.com

Medical Care Volume 47, Number 7 Suppl 1, July 2009

30. Riley GF. Administrative and claims records as sources of health care
cost data. Med Care. 2009;47(suppl 7):S51S55.
31. Finkler SA. The distinction between cost and charges. Ann Intern Med.
1982;96:102109.
32. Shwartz M, Young DW, Siegrist R. The ratio of costs to charges: how
good a basis for estimating costs? Inquiry. 1995;32:476 481.
33. Miller ME, Welch WP. Analysis of Hospital Medical Staff Volume
Performance Standards: Technical Report. Washington, DC: The Urban
Institute; 1993.
34. Mitchell J, McCall N, Burge F, et al. Per Case Prospective Payment for
Episodes of Hospital Care. Boston, MA: Health Economics Research,
Inc; 1995.
35. Bhandari A, Wagner T. Self-reported utilization of health care services:
improving measurement and accuracy. Med Care Res Rev. 2006;63:
217235.
36. Barnett PG. Research without billing data. Econometric estimation of
patient-specific costs. Med Care. 1997;35:553563.
37. Oostenbrink JB, Koopmanschap MA, Rutten FF. Standardization of
costs: the Dutch manual for costing in economic evaluations. Pharmacoeconomics. 2002;20:443 454.
38. Phibbs CS, Bhandari A, Yu W, et al. Estimating the costs of VA
ambulatory care. Med Care Res Rev. 2003;60(suppl 3):54S73S.
39. Fleming T. Red Book. Montvale, NJ: Thomson Healthcare, Inc; 2006.
40. Gencarelli DM. Average wholesale price for prescription drugs: is
there a more appropriate pricing mechanism? NHPF Issue Brief.
2002:119.
41. Gencarelli DM. One pill, many prices: variation in prescription drug
prices in selected government programs. NHPF Issue Brief. 2005:
120.

2009 Lippincott Williams & Wilkins