N
B64 Principles af Pharmac cecannnis
y criteria were Ass
Anportant, The fll
jawin
Nisha checklists and Pu,
J yeenmmendations for the conduct, oat
vend ana af plvarmnacaeconomic evaluntions%*¥ Toga arte ion,
not provided to make readers hypereri and unaece ‘ an
pharmacoceonomic evaluations; few, if ‘any, reports wou) of
criteria in total, However, as a guide, they can be Watt
rit is worth the reader's time to continue u
determining whether -
ifeo, whether results and conclusions are deemed valiq baseg
the methodologies used. Some journals stipulate section ya."
ings, length, and content requirements. This may need to be ‘ad.
nidered when critiquing @ published report. However, the acy ne
of online journals and eles tronic communication makes it ah
Me for authors to readily provide detailed information about,
pharmacoeconomic evaluation to the reader in the form of
pendices or technical reports. Users are encouraged to shee,
these additional data (if possible) when assessing any pharma.
coeconomic evaluation,
CRITERIA FOR EVALUATING PHARMACOECONOMIC STUDIES
1, Introduction
A. Are the problem statement and resultant study questions
clearly and precisely stated?
B, Is the significance of the inquiry discussed with resultant
justification of the research?
C. Whose perspective is considered (society, healthcare pay-
er, patient, healthcare provider)? Was the study perspective
clearly stated?
. Methodology
A. Type of Analysis/Study Design
1, What type of analysis is being performed (cost-minimization,
cost-effectiveness, cost-benefit, cost-utility, cost-consequence)?
2. What study design was employed (eg, prospective vs ret-
rospective, randomized clinical trial, decision analysis, su
vey)? Was an experimental or quasi-experimental desig
uscd? Is the design internally valid? Is the description com
plete enough to allow replication?
3, What is the time horizon? Is the time horizon appropriat®
for the given disease state? Is the time horizon appropriat®
for the study perspective?Chapter 16: Assessing
\g Pharmacoecoy
nomic Studi
Sm 365
’scribed?
bins Considered?
Patient, Population
1. icable, are sources of sampling
4 if orn salient characteristics ofthe
3. Adin sulicent detail?
descr patient population appropriate for
the P:
4. gow will the population limit the ge
we esl?
re the diagnostic criteria adequate} ‘
os approprat® for the study disease sta WY described? Are
wernatives
- “What alternatives are compared?
: ‘Are the options described in sufficient detail?
3, Are the appropriate alternatives compared? Does the
analysis compare the new entity against a current standard?
[sa 4qo-nothing” alternative included? Is it appropriate? Ire
all relevant alternatives considered? F
4, Was the efficacy/effectiveness of each alternative estab-
ished? What are the data used to establish efficacy/effec-
tiveness?
D. Costs and Consequences
1, Are all relevant costs and consequences (including ad-
verse events) identified for each alternative? Is their inclu-
sion supported by available clinical data? If an important
variable is excluded, is justification given?
9, Are the treatment alternatives described in sufficient
detail?
3. Are the costs and consequences included consistent with
the perspective-and time horizon that were selected?
4. How are the costs and consequences measured or counted?
5. How are the costs and consequences valued?
E. Assumptions and Limitations
1, Are assumptions and limitations cle
F. Discounting
1. Are future costs and consequen'
Year discounted?
2. What is the rate of discounting?
" Sensitivity Analysis .
Isa sensitivity analysis conducted for va"
Not be measured with certainty?
the study ques.
neralizability of
arly stated?
ces occurring beyond one
ables that may_
coecon 0™icS
re varied?
366 * Principles of Pharma .
je when factors
2. Do the ronults change whe
IIL, Results
A, Costs and Consequences
ean
. » results of the a sd form?
ee it yrosented i disaggregated fo' a am
the results P sults F »gent the total costs an | consesuences
: 9 rest " af
2. Do Pe gernativ, ‘as well as the tn ed al results?
oe th imvalta of the sensitivity ane lyses presented?
8 inh = sufficient evidence to answer the study ques.
. Is th
alysis clearly presented? Arg
tions?
IV, Discussion and
A. Are results
problem? .
B, Are the limitations of
an ‘ate?
C. Are generalizations approprial (
D. ‘re the results compared with published evaluations
that are similar or focus on the same disease state?
E. Are the practical and ethical implications of implement-
ing the results of the analysis discussed?
Given the various methodologies that exist for conducting
pharmacoeconomic evaluations, it is not possible to provide in-
depth coverage of all the relevant issues for each. This discus-
sion is intended to introduce issues that should be considered
when evaluating a pharmacoeconomic evaluation.
All research reports should clearly state the question that is
being considered. When reading a typical study, its purpose is
presented in the introduction section of the article, usually as
the last sentence. This practice also holds for pharmacoeconom-
ic evaluations. The study question is very important because it
hts cau for ee oni ernie and gives the reader a
rence from which to begin evaluating the aj i
‘ppropri-
ateness and usefulness of the study. Ideally, the study question
should clearly state the t i
‘ e type of pharmacoecono: luati
being conducted, th A mic evaluation
i cted, the alternatives being considered, the outco
or disease state that is being studied, sti ofthe
study. The followi ied, and the perspective of the
The p a are two hypothetical research questions:
urpose of this st
namie of asprin study was to evaluate the pharmacoeco-
The purpose of this study
Perspective, the cost-effe
Conclusions
Gemassed within the context of the origina]
the evaluation discussed?
tive 2s f0 evaluate, from a societal
eness of aspirin versus no treat-_ emparison of acquisition costs, whereas ot
Chapter 16:, Assessing Pharma
ACC
ntion of re mm Ne
waite prove Current acute my ‘a
. mci i
first example, & poorly Written Tegg, vs
; . , en Yes
pond topic that has 'Mplications i 8tion +
= Fgnitions and perspectives, ff does NerO88 a yy eto.
gonse of the focus of the eonomie Bive yy Mber of
ples @ well-written Tesearch ant
the type of evaluation (je, COSt-effagtn Petition ‘ieee
(ie, society), the treatment, al ne jl
rin vs no therapy), and the Outcome eee ee
5 yention of recurrent AMD). A Wellsinitis
i also allows the reader to determine wh N esearch Ques,
if nomic evaluation will be of inte ther the pharm,
xe ith a well-wri ite a
ong Wil : written Tesearch questi
sould adequately justify the need for ‘orion? the authors
‘ nomic evaluation by Presenting the clinical ne Bhar.
pla available and demonstrating gap in the an nee
pation. There is great variation in the clinical aa ce a
ct of therapeutic decisions and, in Seneral, th mia in.
galuation should be commensurate with the i Bia 7
impact. Thus, there are some questions that Ge aie
le
thers demand an ex-
tensive pharmacoeconomic evaluation.*
The pharmacoeconomic evaluation should use a patient popu-
lation that is appropriate and relevant to the disease state that
isbeing considered. Since the ultimate goal of the evaluation is
toaid in decision-making, the salient characteristics of the pa-
tient population should resemble those for whom these results
are targeted. In order to evalute this, the salient characteristies
of the patient population that was included inthe eenorie
analysis should be clearly stated. These ae bestia
formation (eg, age, sex) as well a dagnstc ie
on comorbidities or other factors that ave ety af sucessfl
ence the outeome of the evaluation (2 Pr remnta
treatment, risk of adverse events) SHOU T. inctusion andes
8, .
a ective S' clear descrip:
For prospective and ro sels ment am
clusion criteria should be eel d iat onside
tion of each cohort that 16! databases:
: ther impor” sing Cl!
should be provided. Aner studies
Pp trospective § a>
oa enol
> |nacooconornics —_—___
health care (ef,
stieally af
pility for health in
that are independent of any
ap» Principles of Pha
ent’s access wo
ean drame
jon of lig
ferences
the pati
surance)
Uneven «
can
ornntive: :
ato eventos
inimization, cost-effecti 3, cont-utility, et pe
perspective (og, patient, payer, society) o.
earch question. Although not preferred 7
arly stated. Once these pa thiy
e as it is cle
is acceptable as long ! ror i
ters are established, the evaluation should remain at
with them. This is particularly important for the perspec;
‘The cost-effectiveness of'a medication can, and often does, ieee
a different profile depending on the perspective that is take
For example, if'a third-party payer perspective is selected ty
appropriate to include the ‘aimbursed hospitalization cogr’
physician visits, ‘and medication costs, but not productivity sts,
(eg, lost wages) or caregiver time. Lack of consistency wahae
perspective threatens the validity of the results. the
‘The study design used also has important implications on
validity and generalizability of the results. The most fami bd
design is the randomized clinical trial (RCT). RCTs use ae
cols that limit differences among comparison groups a
treatment alternatives, increasing confidence that the arate
are due to the treatment alternatives and not extrani Tals
ables, However, these strict protocols may not be rej eaten
of rotine clin paste and limit the generlinaility ofthe
results, | e
Lane such cases, efficacy is established but effectiveness is
Conversely, retrospecti ‘ .
ability by Aa ‘ eospactve aesiene try to increase generaliz-
claims databases, medical ré aa routine clinical practice (eg,
ther randomized nor contr sled a), Sines these designs are 13
variables affecting the pa ed, the possibility of confounding
vali Theme ce arereseed, wich Tite
study design or statistical vera can be controlled via the
However, full accountit nalyses (ie, multivariate analyses).
Models (ie, decisis ing of differences is often not possible.
integrate dat: ion analysis, Markov mod used to
cee lata from various sour els) are also used
‘odels can be useful and ces to perform an evaluation.
are necessary when comparativ®
win
jistribull
Jond to itl
Aeron,
jroups cm
ment a
Some
{ie, cost-m!
fit) and the study
rately from the rest
ns Hint the type of angry,
hig,Chapter 16: Assessing Pha i
a 9 macoeconomic Studios " 369
et treatment alternatives are not available,
3g provide justification for selecting the study ¢ on
wf ot arzon of the strdy shouldbe lean ne esi,
Mee the disease state, For example, er cc i
_a tt ternatives for blood pressure that consid a ee
i ¥ er
Eun yal as consequences would appropria Lely use 8 stroke
we »rizon. An evaluation of treatment alternative,
an youd U5e a shorter time horizon (eg,
et zon should also be consistent with
ne ce, the time horizon for the payer pe
ist is typically 12 to 24 months du
a Sueured patients.
oe tment See frould be described in sufficient de-
wait THIS includes i‘ e mes a dose, frequency of adminis-
ration, route of a ministration, and duration of therapy. The
resument alternatives should be dosed comparably to avoid
bias: Ideally, comparisons should be made against the current,
andard(s) of practice to be relevant to decision-making. Many
dinical trials use a placebo comparison (“do-nothing’ alterna-
tive) when evaluating efficacy, which may or may not be appro-
priate for pharmacoeconomic evaluations. If the medication is a
rewentity that is being used to treat a condition for which no
alternative exists, then including a do-nothing alternative in the
sharmacoeconomic evaluation is appropriate. However, if other
treatment alternatives do exist, as is often the case, including a
do-nothing alternative is inappropriate and the results will not
berelevant to clinical practice. In cases where head-to-head
comparison trials are not available, it will be necessary to com-
pare the interventions using a model.
Another critically important component of all pharmacoeco-
nomic evaluations is to establish the effectiveness of the treat-
ment alternatives. This is often an issue of concern for pharma-
‘economic evaluations. Good pharmacoeconomic methodology
‘annot resolve issues concerning questionable effectiveness
82. The authors should clearly state how each probability (eg,
Aflctiveness, adverse events, adherence) was derived. If RCTs
fre the data source, the authors should take into account the ef-
ann tt Ptotocol-driven outcomes (eg, increased adherence, evalu-
of outcomes) in their effectiveness estimates. Effectiveness
“edsures taken from retrospective studies should be adjusted for
alifetime
8 for otitis
3 to 6 months), The
the Perspective, For
Tspective in the Unit.
e to the high turnovervarmacooconomics eee
inciples of Ph
370 = Princip nat could bias the estimate, a
riables th
potential confounding vail ee vs wit
a pho rts included in the mode},
fectiveness meas' Seat
. the pati , : in them
tae er ques ded 8 ATMA gg,
The costs and ¢ ia
evaluatio yo he pers ' :
st be consistent with en, the ou
must be consi ic evaluations depend on a fe ey ib ;
pharmacoeconom! ral ae rela) ie
isition cost,
1g acquis ind consequences asgogj,
isportant tha al relevant oe included in the evaluate’
with the treatment a fom published clinical trials, observation,
‘These can be Lames expert opinion, or personal know].
ere tive or condition. For instance, obvious
edge of a treatment alternal (from a third-party pa
ts and consequences te
aaa nde drug acquisition, disease complications phyei
ee aa ueethetan tia Olier im portent Sea conse,
oonees that shonld not be overlooked include adverse dru
a laboratory or clinical monitoring, administration costs (og
intravenous infusions, supplies), and follow-up care. ,
It is not necessary to include every cost that is associated
with each treatment alternative. In some cases, both alterna.
tives will incur the same costs and will not impact the incre-
mental analysis. In other cases, costs may be insignificant and
will not influence the results. If costs are excluded from the
analysis, the authors should provide adequate justification for
their exclusion. Simple disclosure of the excluded costs does not
constitute adequate justification.
Measuring/counting the costs and outcomes in an economic
evaluation can take on several different forms including case re-
Ports, reviewing medical records, analyzing claims databases, or
making assumptions regarding the amount of resource utiliza-
tion based on published literature or expert opinion when no
data are available, Consistency and validity of measurement
across all the treatment alternatives is important to avoid bias-
ing the evaluation. When the treatment alternatives are similar
(eg, drugs within the same class), this is less likely to be a prob-
lem. However, in cases where the medications or the nature of
the consequences (eg, adverse events) are different, the authors
need to ensure that the events are being measured evenly for all
treatment alternatives, Finally, the authors must state clearlyChapter 16:.
hapter 16: Assessing Pharmacoeconomi
ic Studies
and consequence wi eu
1 ae seation in disagrees ne Pe
4 ' 0 i s
y es jn the Ds ea in disaggregated form ay the val.
s ags goneralizabili 8
ee ean 38 ability. Pharmacoceonors that the
ee Since oo (the actual amount paid ne oat
tt ala Gl t paid), : .
ee yt ws ile, Comon dt wn vee charges
fae atees include Rovertment andl private relmbnee ts
wrt proitary ee lists, andl claims data mem
“a Seen ne
sles Pyatal of the necessary information fora ph
A pharma:
iste valuation i t my
I some evaluation 1s available. Therefore. some
+ some as
eg always necessary. When data are not available fae
“ ¢, the au-
peated explicitly state all of th i :
hors one evaluation and discuss wie the nm me that are
utcome of the analysis. When assumptions aheah affect
the 1a be reasonable and conservative. The cana tion cer
be evaluated in sensitivity analyses to determine the hould
se the results ofthe evaluation ein
with certain types of data, it is possible to calculate confi-
gence intervals around the cost-effectiveness ratio to test the ro-
ess of the results. More commonly, sensitivity analyses are
to evaluate uncertainty in pharmacoeconomic evaluations
. varying the value of one or more variables at a time and see-
ing how the results change. Extensive sensitivity analyses
should be performed over @ plausible range of expected values to
determine how the results change. Some experts recommend
performing sensitivity analyses over extreme ranges. In any
case, the authors should list the variables that were included in
sensitivity analyses along with the ranges selected for sensitivi-
ty analyses and a justification for selecting that particular
range. Conducting a threshhold analysis for the variables that
have the most impact on the results can also be helpful in evalu-
ating the robustness of the results.
Costs and consequences occurring beyond a one-year period
thould be discounted to account for the time preference of mon
ey and outcomes. The authors should explicitly state whether
costs and consequences were discounted and, if they were, the
tate that was used. The United States Panel on Cost-Bifect™
a recommends a3 percent rate as & starting re
7 S appropriate for a range of discount rates 0
ate the impact of discounting on the analysis.TTT
the results of the economig
rated form along with the incrementa}
M ind consequences should be bra,
hh treatment alternative 80 that the reader can
‘rany difference between them. Resulty
ia undiscounted and discounted forry
vinalyses should be clearly presen.
meters that had the most infly,
ence on the results. Most importantly, the owe should Pro.
vide results that specifically answer the researc! question,
The authors should summarize the findings of the results in the
1. The results of the sensitivity analyses and the impact
7 odel should be discussed. In general,
in the m
Fthe findings should be addressed. Limita.
372. & Principles of pharmacoeconomnics
aarly state
‘The authors should clearly ata
evaluation in disagarek “i
ig, Specifically, the co
ani
ken down for eae
evaluate the source ¢
should also be presented !
The results of the sensitivity
ed with specific data on the paran
discussior
of any assumptions i
the overall validity of 7
tions of the study (eg, design, data sources, assumptions, or gener.
liability) should also be discussed. The authors Hiould Sorasets
from previously published evaluations
their findings with those sly p
similar disease states. Finally,
that have examined the same or ses
pharmacoeconomic evaluations are intended to aid in the decision-
making process regarding the use of pharmaceutical products,
Therefore, the authors should address the practical and ethical is-
sues of implementing the results in actual practice.
Economic Evaluations in the Literature
Several issues have been raised concerning the misapplica-
tion of terminology and the methodologic conduct of published
pharmacoeconomic evaluations. For example, simple terms,
such as cost-effectiveness, have been used in various contexts to
mean cost savings, clinically effective, or being worth the addi-
tional cost.” Several studies have documented that methodolog-
« problems are common in published economic evaluations.”
* sen have included poorly defined research questions, not
vous ae ve ae study perspective, improper handling of
counting as citly stating which costs were included, not dis-
ind consequences, inadequate itivity
analyses, and not adequately addressit aera
cal implications of the results. Th ae _ practical and ae
nificant improvements in the is ae studies did not detect sig-
nomic evaluations over time, Oe a Heahed phereChapter 16: Asessing Pharmacoeconoie
Studies
373
rns have been raised about
| pharma
cone
cor itte i
git as gubmitted to national governments for rei
sai og Hil cb al” found that almost h ir re mburae.
oh pt atuations submitted to the Australi of pharma.
ait ie ta ; Lralian govey
fw tjonable comparative efficacy data and ¢ eremant
‘on tion ean Rhee Ver Onesfj
rae ns with their models including Unsubstantiatg
iated as.
pro joblems with the estimation
. and i i
my hide neo
“i other technical issues. poration of
an
207
several roalheees exist for conducting pharma-
omic evaluations, there are widely accepted principles,
cot pose discussed in this chapter that apply to all evalua
ott sing these principles to evaluate pharmacoeconomic lit-
* will help the user determine whether the conclusions of
a studies are valid and relevant. Adherence to these princi-
does not guarantee that an evaluation is useful. Conversely,
Pe nacoeconomic evaluations that do not strictly adhere to all
ibe principles presented in this chapter are not automatically
valid. Every pharmacoeconomic evaluation should be assessed
fused on its own merit. Empirical evidence has demonstrated
that methodologic problems are common in these studies. This
fact underscores the need for users to familiarize themselves
with the principles of pharmacoeconomies in order to critically
eluate data and use the information wisely.
References
| Drumnend ME, Richardson WS, O'Brien BJ, Levine M, Heyland D. Users
guides to the medical literature. XIII. How to use an article on economic
analysis of clinical practice. A. Are the results of the study valid? JAMA
1997;277:1552-1,
2 Drummond ME O'Brien B, Stoddart GL, Torrance GW. Methods the
cxnomis eelvation of heath eare programmes. 2nd ed. New Yor Ox
intPress, 197, etn
. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-effectiveness i
, balhand meicine New York: Oxford Press 1990, agp users
- OBrien BY, Heyland D, Richardson WS, Levine M, Dram com
tides to the medical literature. XIII. How to use ap artic