Cost of Illness in Colorectal Cancer: An International Review

SYSTEMATIC REVIEW

Cost of Illness in Colorectal Cancer: An International Review

Christine Kriza • Martin Emmert • Philip Wahlster •

Charlotte Niederlander • Peter Kolominsky-Rabas

� Springer International Publishing Switzerland 2013

Abstract

Objectives Given the current—and increasing—pressure

to limit expenditure on health care provision in many

countries, a better understanding of the cost burden of

colorectal cancer is needed. Cost-of-illness studies and

reviews thereof can be a useful tool for analysing and

critically evaluating the cost-related development of colo-

rectal cancer, and they highlight important cost drivers.

Methods A systematic review was conducted from 2002

to 2012 to identify cost-of-illness studies related to colo-

rectal cancer, searching the Medline, PubMed, Science

Direct, Cochrane Library and the York CRD databases.

Results Among the 10 studies (from France, the US,

Ireland and Taiwan) included in the review, 6 studies

reported prevalence-based estimates and 4 studies focussed

on incidence-based data. In the studies included in the

review, long-term costs for colorectal cancer of up to

$50,175 per patient (2008 values) were estimated. Most of

the studies in the review showed that the initial and ter-

minal phases of colorectal cancer care are the most

expensive, with continuing treatment being the least costly

phase. One study also highlighted that stage I CRC disease

was the least costly and stage III the most costly of all 4

stages, due to the high cost impact of biological agents.

Conclusions This review has highlighted a trend for ris-

ing costs associated with CRC, which is linked to the

increasing use of targeted biological therapies. COI studies

in colorectal cancer can identify specific components and

areas of care that are especially costly, thereby focussing

attention on more cost-effective approaches, which is

especially relevant to the increased use of biological agents

in the field of personalised medicine. COI studies are an

important tool for further health economic evaluations of

personalised medicine.

Key Points for Decision Makers

• Colorectal cancer is linked to increased economic

expenditure. Data from COI studies can be used as the

basis for modelling and simulation, which in turn are

useful tools for highlighting the drivers of cost increases.

• The use of targeted therapies in CRC is changing COI

trends—it is resulting in increasing costs.

• Future COI studies would greatly benefit from a com-

mon methodology.

1 Introduction

1.1 The Burden of Colorectal Cancer

Colorectal cancer (CRC) constitutes a major burden on

global health, causing significant morbidity and mortality.

CRC is the third most common cancer and the fourth most

common cause of cancer deaths worldwide [1], but there

are significant worldwide variations in the rates of CRC

[2]. The impact of CRC will increase in the coming years

C. Kriza (&) � P. Wahlster � C. Niederlander �P. Kolominsky-Rabas

Interdisciplinary Centre for Health Technology Assessment and

Public Health, University of Erlangen-Nuremberg, National

BMBF-Cluster of Excellence ‘‘Medical Technologies-Medical

Valley EMN’’, Schwabachanlage 6, 91054 Erlangen, Germany

e-mail: [email protected]

URL: http://www.public-health.de; http://www.prohta.de

M. Emmert

School of Business and Economics, Institute of Management,

University of Erlangen-Nuremberg, Lange Gasse 20, 90403

Nuremberg, Germany

e-mail: [email protected]

PharmacoEconomics

DOI 10.1007/s40273-013-0055-4

due to an ageing population. Forecasts for global increases

in annual incidence point to rises of up to 80 % from the

current 1.2 million estimated cases to 2.2 million cases

over the next two decades, with the largest share of the

increase occurring in less developed regions [1]. Public

health efforts to control CRC are therefore increasingly

important, and thus will necessitate the provision of addi-

tional funds [1].

Given the current—and increasing—pressure to limit

expenditure for health care provision in many countries, a

better understanding of the cost burden of CRC as well as

specific determinants and drivers of expenditure is needed.

Cost estimates for individual patients vary widely accord-

ing to the data source considered [3, 4]. In this context,

‘‘cost-of-illness’’ studies are a useful tool and an important

source of information when analysing and critically eval-

uating cost aspects of CRC.

1.2 Background on Cost-of-Illness Studies

Cost-of-illness (COI) studies evaluate the economic bur-

dens exerted by specific health problems or diseases on the

general population [5]. COI studies are a useful public

health and advocacy tool as they can raise awareness about

the importance of certain health issues through descriptive

analysis of their impact on health care resources [5].

Results from COI studies are an important tool that can

inform the decision making and planning of health care

services [5, 6].

Criticism of COI studies has focussed on inconsistencies

across studies, such as varying methodologies and cost

components, a lack of transparency when reporting, and

thus concerns about validity [5, 6]. Critically evaluating

COI with regards to the methodologies used is therefore

crucial. The most current methods used in COI studies have

been reviewed in detail by Akobundu et al. [6].

The purpose of this paper is to offer a review of the

literature on analyses of the costs of CRC, and to highlight

especially costly aspects of care. Challenges to the com-

parability of costs reported in different studies and also for

different countries are highlighted, and recommendations

for the design of future cost-of-illness studies for CRC are

made. The review aims to describe the development of

costs relating to CRC, and to highlight important cost

drivers.

1.3 Methodological Framework of COI Studies

Traditional cost-of-illness studies analyse ‘‘direct costs’’

relating to healthcare and other resource use linked to the

diagnosis, treatment and management of a health condition

or disease, while ‘‘indirect costs’’ are connected to pro-

ductivity losses resulting from morbidity and mortality [5].

These different cost aspects are then expressed in monetary

values and analysed in the COI study [5]. Additionally,

‘‘intangible costs’’ are also considered in some COI studies.

These relate to loss of quality of life and length of life due

to other factors such as pain or suffering [5, 7], but are

more difficult to translate into monetary significance.

COI studies can be conducted from various perspectives,

and are related to the cost bearer. In general, economists

favour the societal perspective of a COI study, as potential

biases resulting from narrow viewpoints are minimised and

these types of studies include all costs for a specific illness

[5]. Other studies with a narrower focus (i.e. from the

perspective of a government, healthcare system, industry or

employer or patient group with the health condition) offer

useful accounts of expenditure in specific sectors, but must

be differentiated from holistic COI studies that include all

of the resources expended or foregone as a result of a

health problem [5].

Most COI studies follow either a prevalence-, or an

incidence-based approach, according to the time frame in

the study. The prevalence-based approach measures the

costs of a disease relating to a fixed time period in the past,

usually one year in length, and includes medical care and

morbidity costs [5, 7]. These costs are analysed indepen-

dent of the first occurrence of the disease, and the mea-

surement of the impact of existing cases usually includes

costs at various stages of the disease [5, 7].

Incidence-based approaches focus on lifetime costs

attributed to a disease [5]. Costs are calculated from the

onset to the conclusion of a disease (i.e. cure, or in the case

of a chronic disease, the remainder of the life) [5]. Incidence

costs measure discounted, lifetime medical, morbidity and

mortality costs for the specific incident cohort [8].

The majority of COI studies use a prevalence-based

approach as it requires less data; however, this approach is

generally less reliable for measuring the potential savings

as a result of preventive interventions [5]. Incidence-based

studies can illustrate how costs differ according to disease

duration, and are therefore more useful for assessing and

planning the use of interventions targeted at specific stages

of disease.

2 Methods

2.1 Search Methodology

The search was performed for journal articles between

2002 to 2012, with no language restrictions applied to the

search. The following databases were searched: Medline,

PubMed, Science Direct, Cochrane Library and the York

Centre for Reviews and Dissemination databases. A com-

prehensive search strategy was developed, using

C. Kriza et al.

Table 1 Cost-of-illness studies in colorectal cancer: summary of main study characteristics (incidence-based studies are shown in italics)

Study Country Perspective Study method Basis of analyses Main data sources

Bouvier

et al.

[13]

France Third-party

payer

(health

insurance)

Retrospective cohort study. Study data

related to 1997–1998

Cancer registry. Health

insurance fund

claims database

Digestive cancer registry Calvados.

Health insurance fund claims

database Caisse Nationale

d’Assurance Maladie des

Travailleurs Salaries

Selke

et al.

[14]

France Societal Retrospective cohort study. Study data

related to 1999

Government health

data (hospital

discharges).

Company registry of

CRC patients

French national hospital healthcare

information system, French

permanent medical prescription

survey, CRC cohort Societe Gaz et

Electricite de France

Chang

et al.

[15]

USA Third-party

payer

(health

insurance)

Retrospective matched-cohort control

analysis. Study data related to

1998–2000

Claims data from

National Survey

Information. Cancer

registry and claims

database

National Medical Expenditure Survey,

National Hospital Discharge Survey,

Surveillance Epidemiology and End

Results (SEER)—Medicare-linked

files

Paramore

et al.

[16]

USA Third-party

payer

(health

insurance)


analysis. Study data related to

1998–2005

Claims database Integrated medical and pharmacy

claims data from the PharMetrics

database

Clerc

et al.

[17]

France Societal Retrospective cohort study. Study data

related to 2004

Cancer registry.

National Health

System Database

CRC registry Burgundy. National

Health System claims database

Song

et al.

[18]

USA Third-party

payer

(health

insurance)


analysis. Direct estimation of mCRC

costs based on treatment and

procedure claims. Study data related

to 2004–2008

Claims databases Thomson Reuters MarketScan

Commercial Claims and Encounter

Database and Medicare

Supplemental and Coordination of

Benefits Database

Ramseyet al.[19]

USA Third-partypayer

Retrospective matched-cohort controlstudy. Modelling techniques used foranalysis of cost data to estimatelifetime CRC attributable health carecosts (Weibull model). SEER datafrom 1986–1994. Cost modellingtime horizon: 10 years

Cancer registry andclaims database

Surveillance Epidemiology and EndResults (SEER)—Medicare linkedfiles and survival data for the generalUS population from US life-tables

Langet al.[20]

USA Third-partypayer

Retrospective matched-cohort controlstudy. Modelling techniques used foranalysis of cost data to estimatelifetime CRC attributable health carecosts (Weibull model). SEER datafrom 1996–2002, followed-up to2005. Cost modelling time horizon:25 years

Cancer registry andclaims database.Survival data fromUS life-tables

Surveillance Epidemiology and EndResults (SEER)—Medicare linkedfiles, SEER Stat database andsurvival data for the general USpopulation from US life-tables

Langet al.[21]

Taiwan Third-partypayer

Retrospective matched-cohort controlstudy. Study data related to1999–2002. Cost modelling timehorizon: 10 years

Cancer registry andclaims database

Taipei Veterans General HospitalCancer Registry and claim data fromthe National Health Insurance for thecontrol group. Cancer registrystatistics from the Taiwan Bureau ofHealth Promotion were used forprobabilities related to survival,cancer fatality and general mortality

Tilsonet al.[22]

Ireland Health carepayer

Static decision tree model of CRCtreatment pathways, estimatingrelated costs. Study data from cancerregistry related to 2004–2005,hospital data 2007. Cost modellingtime horizon: 5 years

Cancer registry.Hospital databases

National Cancer Registry Ireland, datafrom three local university teachinghospitals supplemented by aliterature review and expert clinicalopinion


Table 2 Disease and patient-level characteristics (incidence-based studies are shown in italics)

Study Disease specification Patient characteristics Inclusion/exclusion criteria

Bouvier

et al.

[13]

Colorectal cancer (no further elaboration

on definition)

142 patients, 75 men, 67 women. Mean

age was 67.4 years (range 28–88).

Private sector employees and their

families

Medical expenses not claimed by the

members of the health insurance system,

nonreimbursable medical care, self-

medication and costs incurred in

retirement homes were excluded from

the analysis

Selke

et al.

[14]

Colorectal cancer defined as including

malignant tumor of the colon (ICD-10

code C18), malignant tumor of the

rectosigmoid junction (ICD-10 code

C19), and malignant tumor of the

rectum (ICD C20)

Data from the French national hospital

healthcare information system identified

69,046 hospital stays with a principal

diagnosis of CRC. Data from the

permanent medical prescription survey

identified 172,000 general practitioner

and 78,000 specialist consultations

related to CRC. Company CRC cohort

with 47 incident CRC patients (40 men,

7 women)

Data on paramedical care and the

prescription of complementary tests

were excluded from the study. Patients

had to have no other cancer diagnoses or

treatments in the year before the initial

cancer diagnosis

Chang

et al.

[15]

Colorectal cancer ICD-9 153.xx, 154.0x,

154.1x

2,860 newly diagnosed CRC patients;

study included other cancer types for a

total of 38,127 patients in the control

group and 12,709 in the cancer group.

Median age for the control group was

69, and 67 for the cancer group. Further

disaggregation according to sex, follow-

up and Charlson comorbidity index

given for overall cancer group and

control, but not in a disaggregated way

for the CRC group

Patients had to have at least one

subsequent cancer claim (diagnosis or

treatment) in the 3 months after initial

diagnosis

Paramore

et al.

[16]

Metastatic colorectal cancer. ICD-9-CM

153x or 154.x

Sample of 699 patients with newly

diagnosed mCRC. Control group

matched to cases on a 4:1 basis (2,795

controls). Mean age for both cases and

controls was 58.6 years, both groups

were 43.5 % female. Mean comorbidity

index score was 1.5 for cases and 1 for

controls. Mean duration of follow-up

was 12.8 months for both groups

The criteria that related to the highest

number of patients excluded were

related to patients that did not have two

or more encounters with same mCRC

diagnosis code within 6–180 days of

each other and patients that did not have

at least 1 year of continuous health plan

enrolment before the first mCRC

encounter. Additional exclusion criteria

relate to patients that had disqualifying

chemotherapy or more than 1 medical

encounter with a diagnosis of primary

cancer in the 1-year preindex period

Clerc

et al.

[17]

Colorectal cancer according to ICD-10.

Colon cancers included cancers from

C18.0–C18.9 and rectal cancers C19.0

(rectosigmoid junction) to C20.9 (rectal

ampulla)

384 patients (53.1 % men). Categorisation

of patients into three age groups: under

65: 27.6 %; 65–74: 35.9 %; 75 and

over: 36.5 %

All cases from social health insurance

funds that could not be traced in the

relevant database, resulting in the

inclusion of 76 % of incident cases

Song

et al.

[18]

Metastatic colorectal cancer: patients with

at least 1 inpatient or 2 outpatient CRC

diagnoses (ICD-9: 153.0–153.4,

153.6–153.9, 154.0, 154.1, 154.8) and at

least 1 inpatient or 2 outpatient

diagnoses of metastasis (ICD-9: 196.0,

196.1, 1963.3, 196.5, 197.0–197.4,

197.6–197.8, 198.x, 199.0)

Sample of 6,675 patients with newly

diagnosed mCRC with matching 6,675

controls without cancer. Mean age was

64.1 for cases and 62.6 for controls.

55.5 % men for both groups. Both

groups were matched on the Charlson

comorbidity index, 73.3 % of both

samples had no other indications of

chronic disease in 6 months prior to

index. Medicare beneficiaries 42.8 %

for cases and 41.2 % for controls. On

average, cases were observed for 16.3

months following mCRC diagnosis

Patients with a diagnosis of ICD-9 153.5

(Appendix), 154.2 (anal canal, anal

sphincter), or 154.3 (anus, unspecified)

prior to CRC diagnosis; patients that did

not have 6 months of claims data prior

to the index date, were under 18, or had

other primary cancer diagnoses or

metastatic disease claims in the 6-month

pre-period

C. Kriza et al.

appropriate disease-related MeSH terms and economic

filters (see the Appendix). Search terms included the fol-

lowing: colorectal cancer, colon cancer in combination

with the following: cost, cost of illness and economics. The

keywords were combined and adapted to search the

abovementioned databases. Additional articles were found

in the references and citations of the retrieved articles

(citation snowballing). The search results were indepen-

dently reviewed and screened by two researchers who then

independently extracted the relevant data.

The search methodology employed was in line with

PRISMA guidelines [9] except for the use of the PICOS

review system (population, intervention, comparators,

outcomes, study design). With all intervention terms

specified, relevant cost-of-illness studies might have been

missed [10]. The selection criteria were adapted from the

CHEC-list (Consensus on Health Economic Criteria) [11]

and the BMJ guidelines for authors and peer reviewers of

economic submissions [12]. In addition, the inclusion and

exclusion criteria were designed in accordance with the

cost-of illness evaluation checklist developed by Larg et al.

in the guide to the critical evaluation of cost-of-illness

studies (2011) (see inclusion and exclusion criteria in the

Appendix).

Table 2 continued

Study Disease specification Patient characteristics Inclusion/exclusion criteria

Ramseyet al.[19]

Colorectal cancer (no further elaborationon definition)

Annual costs refer to long-term CRCsurvivors and age-matched controls whosurvived an entire year, for years 6–10after diagnosis. Total sample sizes notindicated but subtotals aredisaggregated for age groups and sex(estimate from paper: 40,000). Costinformation relates to CRC stages I toIV and is aggregated for age groups65-69; 70–74; 75–79; 80–84; 85?

Selected patients were diagnosed at age65 or older with no prior diagnosis ofcancer. Patients had to be entitled toboth parts A and B of Medicare.Patients who were enrolled in prepaidhealth plans were excluded

Langet al.[20]

Colorectal cancer according to ICD-O-3histology codes: 8140, 8210-11,8220-21, 8260-63, 8470, 8480-81, and8490

The study gave details for newlydiagnosed CRC patients, with furtherdisaggregation for colon cancer andrectal cancer. 56,838 CRC patients(41,256 colon cancer, 15,582 rectalcancer), stages I to IV, ages 66–74;75–84; 85?. Mean age of combinedCRC cohort and comparison cohort was77 years, 55.2 % female. Ethnicity forCRC cohort 85.8 % white (86.2 % forcomparison cohort), 7.77 % AfricanAmerican (6.7 % for comparisoncohort), 1.3 % Hispanic (2.1 % forcomparison cohort), 5.2 % other (5 %for comparison cohort). Mean Charlsonscore 1.8 for CRC cohort, 1.7 forcomparison cohort

Patients were excluded if, in the period 12months before, or anytime after theindex date, they were enrolled in aMedicare Health MaintenanceOrganization, not eligible for bothMedicare parts A and B benefits, orwere eligible for other benefits in therenal disease program. Patients wereexcluded who had claims in the12-months pre-index period indicatingany other cancer, were diagnosed withCRC at the time of death or autopsy, orcould not be matched to an appropriatecomparator

Langet al.[21]

Colorectal cancer (no further elaborationon definition), among other cancergroups

2,992 patients with CRC, equal number incontrol group. Mean age for CRC groupwas 65 years, 62 years in control group.34 % female in CRC group, 38 %female in control group. DistinguishedCRC patients in terminal phase, initialphase, continuing phase

For matching control and cancer groups,only patients who were initiallydiagnosed during the study period wereincluded

Tilsonet al.[22]

Colorectal cancer (ICDO2 C18–C20) The study gave details for CRC patients,with further disaggregation for coloncancer and rectal cancer, stages I to IV.Hospital data for 297 CRC patients and46 rectal cancer patients. Figuressupplemented and verified by expertclinical opinion. Data for 4,268 patientsfrom cancer registry. No age or sexdisaggregation of patient data

Modelling assumptions included overviewof data parameters used for model input


3 Results

The search identified 7,171 citations which were poten-

tially relevant. After excluding 3,725 duplicate papers,

3,446 papers were screened by abstract. Of those, 3,391

papers did not fulfil the abovementioned selection criteria.

Fifty-five papers were retrieved for detailed inspection. A

further 45 papers did not meet the selection criteria either.

One additional study was identified in the literature of the

reviewed papers. Thus, in total, 10 papers have been

included in this review.

Collectively, the studies reviewed showed cost-of-ill-

ness results for the following 5 countries: United States (5

studies), France (3 studies), Taiwan (1 study) and Ireland

(1 study).

Studies that used prevalence-based approaches were

grouped separately from studies reporting incidence-based

estimates. Of the total of 10 studies included in this review,

six reported prevalence-based estimates [13–18], and four

focussed on incidence-based data [19–22].

Cost estimates are cited in both the original currency of

the publications and in US dollars. The costs in US dollars

were calculated by using currency exchange rates as well

as purchasing power parities (PPP). According to previous

research, this approach can be used when applying health

care costs from one country to another [23]. The illustra-

tions of the costs in the different groupings follow a

structure used by Payne et al. [23] in a review of the long-

term cost of illness in stroke.

3.1 Study Characteristics

Of the 6 studies focussing on prevalence-based results, data

from registries were used in three and data from adminis-

trative healthcare databases in the other three. These

studies originated from France and the US. The majority of

the studies employed a third-party payer perspective;

mostly a health insurance perspective. The remaining

studies used a societal perspective, including both direct

and indirect cost aspects. Two studies focussed on meta-

static colorectal cancer, with the remaining studies having

a broader focus on CRC. The most important aspects of

each of these studies are summarised in Table 1 [13–18].

Among the 4 studies that focussed on an incidence-

based approach and regarded long-term costs of CRC,

study characteristics varied widely (Table 1, incidence-

based studies shown in italics).

While study authors referred to estimates over a full

episode of care, the time horizons used varied from 5 to 25

years. Three of the studies were designed from a third-party

payer perspective, and one on the health care payer per-

spective. Most studies used a retrospective matched-cohort

control analysis. The basis for most of the analyses was a

combination of cancer registries, claims databases and

population estimates, coupled with modelling techniques

for the analysis of lifetime health-related costs regarding

CRC.

Table 2 shows the details for the disease specifications

used in the studies, along with patient characteristics and

inclusion/exclusion criteria used. While most studies

defined their focus as being on CRC, some studies did not

include specific definitions. The patient characteristics in

the studies varied from the use of patient data from 142

cases up to cohorts of 56,838 cases. The study by Lang

et al. [20] included the most details on the baseline and

demographic characteristics of patients and controls (and

was the only study providing information on the race/eth-

nicity of patients). The disaggregation of cost data

according to gender and age was presented in only two

studies: Lang et al. [20] and Ramsey et al. [19].

3.2 Cost Components

Grouping the prevalence-based studies reveals large vari-

ations in the cost components included in the cost-of-ill-

ness analysis (Table 3). Among the aspects of direct costs,

only the hospital inpatient care and ambulatory outpatient

care components were considered in all of the studies. Only

two of the studies included considerations of indirect costs,

notably lost productivity, disability allowances and out-of-

pocket costs to caregivers [14, 15]. In terms of direct costs,

the US-focussed studies of Song [18] and Chang [15] were

the most wide-ranging in terms of their consideration of

cost aspects. The latter study provided the most encom-

passing and detailed presentation of total costs, broken

down by type and categories within the cost types. The

majority of the studies employed differentiated among

different stages of CRC based on either the TNM (tumor,

nodes, metastasis) staging system most commonly used to

describe the extent of cancer spread [24] or the different

phases of CRC (initial, continuing and terminal phases).

The cost components for incidence-based studies are

summarised in Table 3 (shown in italics). The same

problem with variations in the cost components that was

noted for the prevalence-based studies also applies to the

incidence-based studies. All studies considered hospital

inpatient care in their analyses. It is notable that none of the

incidence-based studies considered indirect costs for CRC.

The study by Tilson et al. [22] included the most details

on cost components. In the same study, unit cost data were

included in the model used to estimate lifetime health-

related costs. That report includes a table describing the

test/treatment/procedure along with unit costs in the most

detailed manner of all the studies. In addition, among all

the research groups, Tilson et al. considered the most up-to

date regimens of chemo- and biological therapy for CRC,

C. Kriza et al.

and their study was the only incidence-based one to report

the costs of such targeted therapies.

3.3 Prevalence-Based Studies: Cost Estimates

An overview of the economic results reported in the

prevalence-based studies is presented in Table 4.

The highest costs are presented in the study by Song

et al. [18] that focuses on treatment costs for metastatic

colorectal cancer, which are substantial. This is the most

up-to-date study, and it examines cost aspects in great

detail. In this new era of CRC management, when com-

binations of biologics and chemotherapies have become the

standard of care, a new trend for rising costs due to the

increasing cost share of biologics is apparent [18].

Advanced methods for the management of CRC are linked

to increasing costs [25]. The study of Song et al. is the only

one to examine the trend in the cost share of biologics, and

they noted that this cost share increased from 4.8 % for

patients diagnosed in 2004 to 9.4 % for those diagnosed in

2008 [18]. All of the studies also identified that hospital

costs comprised a large share of the direct costs (44–98 %).

3.4 Incidence-Based Studies: Cost Estimates

The main economic results for the incidence-based studies

are presented in Table 5.

As all of the incidence-based studies were conducted in

varying healthcare settings and at different time periods, it is

problematic to present valid direct cost comparisons, even

after the conversion to a common monetary unit. However,

certain common aspects and trends become apparent.

Previous studies have shown that the initial and terminal

phases of CRC care are the most expensive, with contin-

uing treatment representing the least costly phase [7, 26].

This trend has been echoed by the studies of Lang et al.

[20, 21]. In addition, Lang et al. [20] have shown that the

initial treatment phase is associated with larger costs than

the terminal phase of CRC care.

Lang et al. [20] have shown that the largest costs can be

attributed to stage IV of CRC. In the study by Tilson et al.

[22], stage I CRC disease was the least costly, with stage

III the most costly of all 4 stages. In this study, the high

cost impact of biological agents on CRC was shown. Costs

associated with diagnosis, treatment and follow-up repre-

sented approximately 4, 91, and 5 % of the total cost.

Ramsey et al. [19] showed that younger age groups and

those with early-stage CRC were associated with the

highest costs. As a result, CRC-attributable costs can be

substantial for long-term survivors. Lang et al. [20] also

presented data demonstrating that CRC patients among the

younger age groups of 66–74 resulted in higher costs,

whereas the oldest age groups of 85 and older represented

the lowest cost attributions. In these circumstances, where

Table 3 Summary of cost components (a checkmark indicates that the component is included; incidence-based studies are shown in italics)

Cost parameters Bouvier

et al.

[13]

Selke

et al.

[14]

Chang

et al.

[15]

Paramore

et al.

[16]

Clerc

et al.

[17]

Song

et al.

[18]

Ramseyet al.[19]

Langet al.[20]

Langet al.[21]

Tilsonet al.[22]

Components

Direct costs

Hospital inpatient care x x x x x x x x x x

Surgery x x x x x

Radiation treatment x x x x x

Chemotherapy only x x x

Chemo- and biological therapy treatment x x x x

Hospital outpatient care x x x

Medical purchases x x x x x x

Ambulatory outpatient care x x x x x x x

Indirect costs

Lost productivity x x

Disability allowances x x

Out of pocket costs to caregivers x

Transportation x x

Differentiation into cancer stages or phases of

treatment

x x x x x x x x


improved patient survival is linked to higher long-term

costs, cost-effectiveness studies are more useful than COI

studies for analysing relationships between costs and sur-

vival, or (specifically) the impact of the use of targeted

therapies.

4 Discussion

This review has primarily highlighted the rising cost bur-

den of CRC, so it is most useful for policy makers, who

face increasing costs associated with CRC. However, this

review may also provide a valuable resource for future

cost-of-illness studies, as a number of issues relating to

improving the methodology of such studies have been

highlighted. A lack of description of detailed cost aspects

was a common problem among both the prevalence-based

and the incidence-based study groups. For costs relating to

incidence-based studies, it is important to note that survival

projections clearly influence cost outcomes, as resource use

will improve patient survival.

Cost estimates can vary widely depending on the data

source utilised [4]. In addition, there was widespread

heterogeneity in the methodologies used and the the

patient-level characteristics applied in the studies. Due to

these challenges, no effort has been made to present

pooled results from the studies in order to provide aver-

age costs for colorectal cancer care. Another major lim-

itation of this review is the limited number of studies

included due to the very strict quality criteria applied (see

Appendix). However, even given the specific limitations

of the studies analysed, certain commonalities and trends

could be extracted from the analysis of the different

studies.

In the studies included in the review, long-term costs for

colorectal cancer of up to $50,175 per patient (2008 values)

[22] have been estimated. Estimates for annual prevalence-

based costs for colorectal cancer patients range from $8415

(1999 values) [14] to $175,020 (2008/2009 values) [18].

Results from incidence-based studies were more useful for

comparing costs incurred at different stages and phases of

disease.

It became apparent that inpatient hospital costs represent

a large part of the overall direct costs. In most studies

included in the review, the initial and palliative treatment

phases of CRC treatment were still the largest cost aspects

of colorectal cancer care, in comparison to continuing

treatment. Tilson et al. [22] pointed to an emerging trend

Table 4 Summary of prevalence-based cost-of-illness parameters and estimated costs of CRC

Study Year of

costs

Prevalence Average cost of treatment in one year Per patient % of hospital

costs of all direct

costs

Indirect

costsLocal

currency

$US (ER) $US (PPP)

Bouvier et al. [13] 1997/1998 21,918 EUR 29,173

USD

27,131 USD 70

Selke et al. [14] 1999 69,046 470m EUR 625m USD 581m USD 8,415

USD

98 x

Chang et al. [15] 1999/2000 44,904 USD 60 x

Paramore et al.

[16]

2005 96,112 USD 44

Clerc et al. [17] 2004 24,966 EUR 33,228

USD

30,902 USD 55.2

Song et al. [18] 2008/2009 175,020 USD 51

Conversion based on 2004 PPP versus $US for gross domestic product as calculated by the OECD, and on currency exchange rates in April 2012

Table 5 Summary of estimated lifetime costs of colorectal cancer patients

Study Year of costs (% discount rate) Cost per patient local $US (ER) $US (PPP)

Ramsey et al. [19] 2000 (3 %) 14,461 $ 14,461 $ 14,461 $

Lang et al. [20] 2006 (3 %) 28,626 $ 28,626 $ 28,626 $

Lang et al. [21] 2002 (3 %) 546,348 NTD 18,872 $ 18,683 $

Tilson et al. [22] 2008 (4 %) 39,607 EUR 52,816 $ 50,175 $

Conversion based on 2004 PPP versus $US for gross domestic product as calculated by the OECD, Taiwan PPP from the IMF, and on currency

exchange rates in April 2012

C. Kriza et al.

for stage III to be the most costly of all 4 stages, due to the

high cost impact of biological agents.

4.1 Targeted Therapies

New treatments for advanced CRC in the era of persona-

lised medicine are increasingly costly, with the most

effective treatments for metastatic CRC incurring the

highest costs [27]. Among the prevalence-based studies

included in this review, only the studies reported post 2005

[16–18] include the costs of targeted therapies, and only

the latest incidence-based study did so [22]. These studies

indicate much higher costs relating to CRC than those

reported in older studies. Therefore, this review has high-

lighted a trend for increasing costs due to biological ther-

apy (and an increasing cost share of biological therapy). It

is important, however, to analyse cost-effectiveness sepa-

rately from cost of illness. Achieving more targeted inter-

ventions is crucial given the economic burden of CRC. In

the short term, from the perspective of a third-party payer,

like a health insurance provider, overall costs relating to

CRC care are higher when more tests are performed to

diagnose a disease or to establish recurrence, as well as

when using customised treatment [28]. However, in the

long term, the application of personalised medicine shows

promise in terms of demonstrating cost-effectiveness, as

genetic and molecular information about CRC patients has

the potential to indicate responsiveness to different pre-

vention and treatment approaches [28]. It is important to

note that the high costs of targeted therapies do not auto-

matically equate to added clinical value or cost-effective-

ness. Indeed, some of the expensive targeted anticancer

therapies may have limited life-expectancy gains, but may

still show clinical utility for a limited group of specific

patients, as well as in combination with other drugs [25].

Attempts to measure value in this regard as an aid to policy

makers are becoming increasingly challenging [25]. In

order to save patients from toxic side-effects of inefficient

drugs and costs relating to ineffective treatment, targeting

novel treatments and prescreening with biomarkers is a

recommended approach [25]. A Japanese study found that

KRAS1 testing (the KRAS test is a CRC-related biomarker

test) of CRC patients before cetuximab therapy would lead

to savings of more than $50 m per year for the health care

system, as opposed to if KRAS-based pre-selection for

therapy was not used [25, 29]. The clinical utility of testing

for other biomarkers such as UGT1A12 in colon cancer is

unclear, so economic savings are unlikely [30].

The studies in the review that indicated costs for tar-

geted treatments did not give information on pre-selection

based on biomarkers, which may be an important addition

to future cost-of-illness studies.

4.2 Indirect Costs

In the review it also emerged that indirect cost aspects have

largely been neglected. Although aspects of lost produc-

tivity are more commonly considered, intangible costs such

as pain and suffering are difficult to value and estimate.

Yabroff et al. [31] have estimated patient time costs

associated with CRC care as an important cost component,

and found the highest net patient time costs for the initial

phase of treatment. Van Houtven et al. [32] have examined

the economic burden for informal caregivers of CRC

patients and found substantial cost attributions. A popula-

tion-based study by Syse et al. [33] examined the

increasing impact of cancer on spouses’ labor earnings, and

noted gender-specific consequences. Cotrim et al. [34]

analysed the impact of CRC on patients and their families,

and found that the lives of patients and their families were

deeply affected when faced with the diagnosis and the

resulting physical and psychological changes. The quality

of life, psychological morbidity, sexuality and body of the

patient altered significantly after the CRC diagnosis, and

the patient’s caregivers suffered from high levels of anxiety

and depression [34]. In future COI studies, a consideration

of indirect costs would lead to a better understanding of all

of the costs associated with CRC.

5 Conclusion

This review has highlighted the trend for rising costs

associated with CRC, which are linked to the increasing

use of targeted biological therapies. Such rising costs may

have important implications for policy makers; one option

is a renewed CRC screening effort. The review also shows

widespread methodological heterogeneity among the dif-

ferent studies included in the review, which is also linked

to limits on the transparency of the studies. Future cost-of-

illness studies would greatly benefit from a common

approach to methodology, especially concerning study

design, description and cost-component data. The provision

of high-quality, transparent and methodologically sound

cost-of-illness studies would be a stepping stone towards

more detailed cost-effectiveness studies, which would

facilitate more effective prioritisation and interventions. In

addition, cost-of-illness studies can identify specific com-

ponents and areas of care that are especially costly.

Role of the Funding Source This research is supported by the

German Federal Ministry of Education and Research (BMBF), project

grant no. 01EX1013B, as part of the Centre of Excellence for Medical

Technology.

1 KRAS: V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog.2 UGT1A: uridine diphosphate glucuronyltransferase.


Appendix

The following criteria were used for study inclusion.

Study methods are described in adequate detail:

• The study population is clearly described

• There is a well-defined research question, posed in an

answerable form

• The disease differentiation is clearly laid out for

colorectal cancer

• The viewpoints of the analysis are clearly stated

There is enough economic detail in the study:

• All costs are measured appropriately, i.e. monetary

values for disease-specific attributed costs

• The chosen time horizon is appropriate for the inclusion

of relevant costs and consequences

• Estimates of indirect and direct aspects of costs for

colorectal cancer are given

• The differentiation of cost aspects is made appropri-

ately and in physical units

• The different cost components are described in suffi-

cient detail

• For long-term costs of colorectal cancer, the estimates

are based on a period of a minimum of 5 years of care

following the onset of disease

The following exclusion criteria were used:

• Cost-of-illness studies referring to colon cancer alone

(excluding rectal cancer)

• Studies focusing on co-morbidities like diarrhoea-

related diseases as a result of chemotherapy

• Studies referring to costs of colorectal screening

interventions only

• Studies comparing cost-effectiveness among different

treatments for colorectal cancer

• Cost-of-illness studies focusing on just one disease

stage of colorectal cancer

• Reviews of existing economic studies relating to

colorectal cancer

• Studies not reported in English or German

C. Kriza et al.

References

1. Karsa LV, Lignini TA, Patnick J, Lambert R, Sauvaget C. The

dimensions of the CRC problem. Best Pract Res Clin Gastroen-

terol. 2010;24(4):381–96.

2. Center MM, Jemal A, Smith RA, Ward E. Worldwide variations

in colorectal cancer. CA Cancer J Clin. 2009;59(6):366–78.

3. Gellad ZF, Provenzale D. Colorectal cancer: national and inter-

national perspective on the burden of disease and public health

impact. Gastroenterology. 2010;138(6):2177–90.

4. Yabroff KR, Warren JL, Banthin J, Schrag D, Mariotto A,

Lawrence W, et al. Comparison of approaches for estimating

prevalence costs of care for cancer patients: what is the impact of

data source? Med Care. 2009;47(7 Suppl 1):S64–9.

5. Larg A, Moss JR. Cost-of-illness studies: a guide to critical

evaluation. PharmacoEconomics. 2011;29(8):653–71.

6. Akobundu E, Ju J, Blatt L, Mullins CD. Cost-of-illness studies: a

review of current methods. PharmacoEconomics. 2006;24(9):

869–90.

7. Damm O, Hodek JM, Greiner W. Methodological standards for

cost-of-illness studies using breast cancer, prostate cancer and

colon cancer as an example. Zeitschrift fur Evidenz, Fortbildung

und Qualitat im Gesundheitswesen. 2009;103(6):305–16. Me-

thodische Standards von Krankheitskostenstudien am Beispiel

von Brust-, Prostata- und Darmkrebs.

8. Segel J. Cost-of-illness studies—a primer. Research Triangle Park:

RTI International; 2006. http://www.rti.org/pubs/coi_primer.

pdf. Updated 24/03/2012.

9. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC,

Ioannidis JPA, et al. The PRISMA statement for reporting sys-

tematic reviews and meta-analyses of studies that evaluate health

care interventions: explanation and elaboration. PLoS Med.

2009;6(7):e1000100.

10. Tappenden P, Chilcott J, Brennan A, Pilgrim H. Systematic review

of economic evidence for the detection, diagnosis, treatment, and

follow-up of colorectal cancer in the United Kingdom. Int J Technol

Assess Health Care. 2009;25(4):470–8.

11. Evers S, Goossens M, de Vet H, van Tulder M, Ament A. Criteria

list for assessment of methodological quality of economic eval-

uations: consensus on health economic criteria. Int J Technol

Assess Health Care. 2005;21(2):240–5.

12. Drummond MF, Jefferson TO. Guidelines for authors and peer

reviewers of economic submissions to the BMJ. BMJ.

1996;313(7052):275–83.

13. Bouvier V, Reaud JM, Gignoux M, Launoy G. Cost of diagnostic

and therapeutic management of colorectal cancer according to

stage at diagnosis in the Calvados Department, France. Eur J

Health Econ. 2003;4(2):102–6.

14. Selke B, Durand I, Marissal JP, Chevalier D, Lebrun T. Cost of

colorectal cancer in France in 1999. Gastroen Clin Biol.

2003;27(1):22–7.

15. Chang S, Long SR, Kutikova L, Bowman L, Finley D, Crown

WH, et al. Estimating the cost of cancer: results on the basis of

claims data analyses for cancer patients diagnosed with seven

types of cancer during 1999 to 2000. J Clin Oncol. 2004;22(17):

3524–30.

Search term Database

Medline PubMed CRD York

(incl. NHS

EED)

Cochrane Science Direct

‘‘Cost’’ AND

‘‘colorectal cancer’’

1,521 results on 12/03/

2012

None relevant, or they

were duplicates

1,637 results on 23/02/2012

None relevant, or they were

duplicates

337 results on

12/03/2012

None relevant

164 results

on 08/03/

2012

None

relevant

407 results on 17/03/2012

(abstract, title,

keywords)


were duplicates

‘‘Cost’’ AND ‘‘colon

cancer’’

326 results on 12/03/2012


were duplicates

358 results on 23/02/2012.

Of these, 2 papers was

deemed relevant on 08/03/

2012

77 results on

12/03/2012

None relevant

3 results on

08/03/2012

117 results on 17/03/2012

(abstract, title,

keywords)


were duplicates

‘‘Cost of illness’’

AND ‘‘colorectal

cancer’’

57 results on 12/03/2012

PubMed result duplicates

57 results on 23/02/2012

Of these, 23 papers were


2012

8 results on

12/03/2012

None relevant

2 results on

08/03/2012

None

relevant

150 results on 12/03/2012

3 relevant

‘‘Cost of illness’’

AND ‘‘colon

cancer’’

18 results on 12/03/2012

PubMed result duplicates

18 results on 23/02/2012

None relevant, or they were

duplicates

3 results on

12/03/2012

None relevant

1 result on

08/03/2012

Not relevant

83 results on 12/03/2012

None relevant

‘‘Economics’’ AND

‘‘colorectal cancer’’

40 results on 12/03/2012

Of these, 1 paper was

deemed relevant on

12.03.2012

1,141 results on 28/02/2012



2012

234 results on

12/03/2012

None relevant

21 results on

08/03/2012

None

relevant

15 results on 17/03/2012

(abstract, title,

keywords)

None relevant

‘‘Economics’’ AND

‘‘colon cancer’’

8 results on 12/03/2012


were duplicates

235 results 28/02/2012



2012

55 results on

12/03/2012

None relevant

1 result on

08/03/2012

Not relevant

5 results on 17/03/2012

(abstract, title,

keywords)

None relevant


http://www.rti.org/pubs/coi_primer.pdf

http://www.rti.org/pubs/coi_primer.pdf

16. Paramore LC, Thomas SK, Knopf KB, Cragin LS, Fraeman KH.

Estimating costs of care for patients with newly diagnosed met-

astatic colorectal cancer. Clin Colorectal Cancer. 2006;6(1):52–8.

17. Clerc L, Jooste V, Lejeune C, Schmitt B, Arveux P, Quantin C,

et al. Cost of care of colorectal cancers according to health care

patterns and stage at diagnosis in France. Eur J Health Econ.

2008;9(4):361–7.

18. Song X, Zhao Z, Barber B, Gregory C, Cao Z, Gao S. Cost of

illness in patients with metastatic colorectal cancer. J Med Econ.

2011;14(1):1–9.

19. Ramsey SD, Berry K, Etzioni R. Lifetime cancer-attributable cost

of care for long term survivors of colorectal cancer. Am J Gas-

troenterol. 2002;97(2):440–5.

20. Lang K, Lines LM, Lee DW, Korn JR, Earle CC, Menzin J.

Lifetime and treatment-phase costs associated with colorectal

cancer: evidence from SEER-Medicare data. Clin Gastroenterol

Hepatol. 2009;7(2):198–204.

21. Lang HC, Wu SL. Lifetime costs of the top five cancers in Tai-

wan. Eur J Health Econ. 2012;13:347–53.

22. Tilson L, Sharp L, Usher C, Walsh C, O’Ceilleachair A, et al.

Cost of care for colorectal cancer in Ireland: a health care payer

perspective. Eur J Health Econ. 2012;13:511–24.

23. Payne KA, Huybrechts KF, Caro J, Craig Green TJ, Klittich WS.

Long term cost-of-illness in stroke: an international review.

PharmacoEconomics. 2002;20(12):813–25.

24. Kwon SJ. Evaluation of the 7th UICC TNM staging system of

gastric cancer. J Gastric Cancer. 2011;11(2):78–85.

25. Sullivan R, Peppercorn J, Sikora K, Zalcberg J, Meropol NJ,

Amir E, et al. Delivering affordable cancer care in high-income

countries. Lancet Oncol. 2011;12(10):933–80.

26. Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. Pro-

jections of the cost of cancer care in the United States:

2010–2020. J Natl Cancer Inst. 2011;103(2):117–28.

27. Wong YN, Meropol NJ, Speier W, Sargent D, Goldberg RM,

Beck JR. Cost implications of new treatments for advanced

colorectal cancer. Cancer. 2009;115(10):2081–91.

28. Verma M. Personalized medicine and cancer. J Pers Med.

2012;2(1):1–14.

29. Shiroiwa T, Motoo Y, Tsutani K. Cost-effectiveness analysis of

KRAS testing and cetuximab as last-line therapy for colorectal

cancer. Mol Diagn Therapy. 2010;14(6):375–84.

30. Wong WB, Carlson JJ, Thariani R, Veenstra DL. Cost effec-

tiveness of pharmacogenomics: a critical and systematic review.

PharmacoEconomics. 2010;28(11):1001–13.

31. Yabroff KR, Warren JL, Knopf K, Davis WW, Brown ML.

Estimating patient time costs associated with colorectal cancer

care. Med Care. 2005;43(7):640–8.

32. Van Houtven CH, Ramsey SD, Hornbrook MC, Atienza AA, van

Ryn M. Economic burden for informal caregivers of lung and

colorectal cancer patients. Oncologist. 2010;15(8):883–93.

33. Syse A, Tretli S, Kravdal O. The impact of cancer on spouses’

labor earnings: a population-based study. Cancer. 2009;115(18

Suppl):4350–61.

34. Cotrim H, Pereira G. Impact of colorectal cancer on patient and

family: implications for care. Eur J Oncol Nurs. 2008;12(3):

217–26.

C. Kriza et al.

Cost of Illness in Colorectal Cancer: An International Review

Documents

Transcript of Cost of Illness in Colorectal Cancer: An International Review