Original Article

Costs Associated With Complications

Are Lower With Capecitabine
Than With 5-Fluorouracil in Patients
With Colorectal Cancer
Edward Chu, MD1, Kathy L. Schulman, MA2, Susan Zelt, DrPH, MBA3, and Xue Song, PhD2

BACKGROUND: Capecitabine, an oral alternative to 5-fluorouracil (5-FU) in patients with colorectal cancer
(CRC), has equal clinical efficacy and a favorable safety profile; however, its use may be limited because of
unit cost concerns. In this study, the authors measured the cost of chemotherapy-related complications
during treatment with capecitabine- and 5-FUbased regimens. METHODS: Patients with CRC who
received at least 1 administration of capecitabine or 5-FU during 2004 and 2005 were identified from the
Thomson MarketScan research databases. Monthly frequency and cost for 23 complications were recorded.
Logistic regression was used to predict complication probability. General linear models were used to pre-
dict monthly complication cost and total monthly expenditure. RESULTS: In total, 4973 patients with CRC
met the inclusion criteria for this analysis. Although the most frequently observed complications were the
same between capecitabine and 5-FU (nausea and vomiting, infection, anemia, neutropenia, diarrhea),
each was observed with greater frequency in 5-FUbased regimens. The mean predicted monthly compli-
cation cost was significantly higher (by 136%) with 5-FU monotherapy than with capecitabine monotherapy
(difference, $601; 95% confidence interval [95% CI], $469-$737). In addition, the mean predicted monthly
complication cost for 5-FU oxaliplatin was higher than the cost with capecitabine plus oxaliplatin (differ-
ence, $1165; 95% CI, $892-$1595). When acquisition, administration, and complication costs were taken into
consideration, there were no significant differences in the total cost between capecitabine regimens and
5-FU regimens. CONCLUSIONS: Capecitabine compared well with 5-FUbased therapy in patients with
CRC and was associated with lower complication rates and associated costs. Cancer 2009;115:141223.
V
C 2009 American Cancer Society.

KEY WORDS: capecitabine, 5-fluorouracil, oxaliplatin, combined capecitabine and oxaliplatin, combined
5-fluorouracil, leucovorin, and oxaliplatin, colorectal cancer, complications, pharmacoeconomics.

Fluoropyrimidine -based therapy has improved median survival significantly in patients with met-
astatic colorectal cancer (CRC), from 6 months with best supportive care to about 20 months with 5-fluo-
rouracil (5-FU) combined with either oxaliplatin or irinotecan (with or without bevacizumab).1,2

Corresponding author: Edward Chu, MD, Yale Cancer Center, Yale University School of Medicine, 333 Cedar Street, WWW-221, New Haven,
CT 06520; Fax: (203) 737-5698; chueyale@yahoo.com
1
Section of Medical Oncology, Yale Cancer Center, Yale University School of Medicine, New Haven, Connecticut; 2Thomson Healthcare, Thomson
Reuters, Cambridge, Massachusetts; 3Medical Data Management and Analytics, Roche Laboratories, Inc., Nutley, New Jersey
We thank Lynda Wiseman, PhD, and Steven Melvin, who provided medical editing and styling assistance, and Nicole Morandi, MA, of Thomson
Medstat, who provided statistical assistance in the preparation of this article.
Received: May 15, 2008; Revised: September 29, 2008; Accepted: October 3, 2008
Published online: February 4, 2009, V
C 2009 American Cancer Society

DOI: 10.1002/cncr.24131, www.interscience.wiley.com

1412 Cancer April 1, 2009


Currently, intravenous 5-FU combined with leucovorin
(LV) is the most widely used fluoropyrimidine backbone
in combination regimens used for this disease.
The oral fluoropyrimidine capecitabine has demon-
strated similar clinical efficacy and a favorable safety pro-
file compared with intravenous 5-FU. Therefore, oral
capecitabine offers an attractive alternative for patients
with CRC.3-7 Recent studies that compared combined
capecitabine and oxaliplatin (XELOX) with combined 5-
FU and oxaliplatin demonstrated that both regimens were
tolerated well and had a manageable safety profile.8,9 The
incidences of grade 3 and 4 diarrhea and grade 1 and 2
stomatitis were much less common with capecitabine,
whereas grade 1 and 2 hyperbilirubinemia and grade 1
and 2 hand-foot syndrome were more common with the
capecitabine-based regimen.8,9
Oral administration of capecitabine allows patients
the flexibility to accommodate their treatment regimen
within their daily lives. In addition, oral capecitabine
removes the requirement for indwelling intravenous cath-
eters and infusion pumps, which are associated with sub-
stantial morbidity and complications. The advantages of
home drug administration and fewer hospital visits are
reflected in the results from a recent randomized, cross-
over, patient-preference trial in which patients reportedly
preferred oral capecitabine monotherapy over intravenous
5-FU/LV (Mayo Clinic regimen) and, to a lesser extent,
over infusional 5-FU/LV as administered by the biweekly
De Gramont regimen.10
With the development of active cytotoxic chemo-
therapy agents and the integration of novel biologic
agents, the treatment of CRC is associated with substan-
tial economic implications. In the United States, the esti-
mated annual expenditures for CRC are between US $5.3
billion (2000 values) and US $6.5 billion (1994 values),
including both direct and indirect costs,11 and the lifetime
cost for the treatment and care of a patients with CRC
recently was estimated as approximately US $100,000.11
Consequently, the comparative costs and cost effective-
ness of various treatment regimens represent important
elements that must be factored into the treatment deci-
sion-making process.
Although the acquisition cost of capecitabine is
higher than the cost of 5-FU, this expense may be offset
by other factors, including the requirement for office vis-
its, the cost of drug administration, complication rates
Cancer April 1, 2009
Capecitabine and Cost of Complications/Chu et al
and associated costs, and other costs, such as lost produc-
tivity. The objective of this study was to quantify the
impact of chemotherapy-related complications on ex-
penditure during treatment with a capecitabine- or 5-
FUbased regimen in a real-world setting.
MATERIALS AND METHODS
Study Design and Data Source
For this retrospective analysis, we used claims data from
the Thomson Healthcare MarketScan Commercial
Claims and Encounters and Medicare Supplemental and
Coordination of Benefits databases to estimate the overall
healthcare costs associated with CRC in patients who
were with capecitabine- or 5-FUbased regimens. The
data sources contain the inpatient and outpatient medical
claims and outpatient prescription drug claims for
approximately 25 million individuals who are covered
under a variety of fee-for-service and capitated provider
reimbursement schemes. In compliance with the Health
Insurance Portability and Accountability Act, patient data
that were included in this analysis were deidentified and,
thus, were exempt from institutional review board
approval.
Patient Selection
Patients were identified with CRC if they had a claim
with a diagnosis (primary or secondary) of colon or rectal
cancer (International Classification of Diseases, Ninth
Revision [ICD-9] diagnosis codes 153 and 154) on a non-
diagnostic claim (for services other than venipuncture,
laboratory, and radiology) on 3 different days between
January 2000 and December 2005. Patients who met
these criteria subsequently were selected into the study if
they had at least 1 claim for chemotherapy during the
study window from January 1, 2004 to December 31,
2005. The study index date was the first observed chemo-
therapy episode in the study window. Patients had to be
enrolled continuously in MarketScan for at least 6 months
before and at least 30 days after the index date. Patients
who were in the middle of a chemotherapy treatment epi-
sode at the beginning of 2004 were not included in the
analysis. Patients were followed from the study index date
until death, disenrollment, or study end.
1413
Original Article
Data Analyses
Patient characteristics
Demographic variables included age, geographic
region, area of residence (urban vs rural), payer type (com-
mercial vs Medicare), and length of follow-up (months).
Demographic variables were measured as of the study
index date. Clinical variables that were evaluated during
the 6-month preindex period included total direct health-
care costs, the Charlson Comorbidity Index (CCI), the
Chronic Disease Score, select comorbidities, and preindex
treatments (radiation therapy, surgery). The presence of
metastases was measured using all data available before
the study index date.
Treatment episodes
Each chemotherapy episode was categorized accord-
ing to treatment regimen and treatment setting. The treat-
ment regimens of interest included capecitabine
monotherapy, XELOX, 5-FU monotherapy, and com-
bined 5-FU and oxaliplatin. Because patients could
receive more than 1 type of chemotherapy during the
study window, a single patient may have been represented
multiple times in the data file.
The initial chemotherapy drug regimen was estab-
lished after a review of medical and prescription pharmacy
claims in the 30 days after the study index date and con-
tinued until an eligible episode-ending event was
recorded. An eligible episode-ending event was defined as
a gap
45 days in therapy, a change in an existing regimen
(eg, the addition of a new agent or the subtraction of an
existing agent), or the complete replacement of 1 regimen
with a subsequent regimen (switched therapies). The
length of each episode was calculated based on the dura-
tion between the start date of a treatment episode and the
end date of the same episode. Neoadjuvant treatment regi-
mens ended on the date of surgery.
Complications of chemotherapy
Complications of chemotherapy included anemia,
alopecia, asthenia, constipation, cough, dehydration, der-
matitis, diarrhea, esophagitis, fever, gastritis, headache,
infection, insomnia, mucositis, nausea and vomiting, neu-
tropenia, night sweats, weight loss, and complications of
vascular access devices (central line infection, central line
thrombosis, pneumothorax, and secondary thrombocyto-
1414
penia). These complications were identified if the patient
had a diagnosis code specific to the condition during the
treatment episode or if the patient received a treatment
specific to these events (for example, the administration of
erythropoietin-stimulating agents) (Table 1). It is note-
worthy that it was not possible to distinguish whether
treatment was given as prophylaxis (eg, antiemetics, sleep
aids) or in response to a treatment-related adverse event.
Moreover, the ICD-9 code that was used to identify
hand-foot syndrome encompasses all dermatitis because
of substances taken internally and, thus, is not specific
only to hand-foot syndrome. Accordingly, the data
reported in the current study include all drug-related
adverse events that had an impact on the skin.
Healthcare utilization and expenditure
Direct healthcare expenditures were accumulated
for all services during each chemotherapy episode and
standardized to cost per month in 2005 US dollars. These
costs were categorized based on site of service (inpatient
and outpatient medical, outpatient pharmacy), and type
of service (office visits, hospitalizations, emergency room
visits, oncology-related prescription drugs). Outpatient
pharmacy costs included drug acquisition costs and the
costs of antianemic, antianxiety, antidepressant, antie-
metic, anti-infective, antidiarrheal, laxative, hormone,
growth factor, and other medications.
The cost of complications was defined as the total
dollar amount associated with a claim in which a diagnosis
or treatment related to a complication was recorded. Cost
estimates were based on paid amounts of adjudicated
claims, including insurer and health plan payments,
copayments, and deductibles. Because of the variation in
episode lengths, costs were reported in monthly incre-
ments based on observed expenditures within each treat-
ment episode.
Statistical Analysis
Descriptive analyses were used to compare unadjusted
expenditures. Categorical variables were summarized in
frequency tables, and continuous and other numeric vari-
ables were summarized by presenting the number of
observations, the mean value, and the standard deviation.
Multivariate analyses were performed to adjust for differ-
ences in patient demographic and clinical factors, which
Cancer April 1, 2009
 Capecitabine and Cost of Complications/Chu et al

Table 1. International Classification of Disease, Ninth Revision Diagnosis Codes Used to Identify Complications

Complication Diagnosis Code Description

Anemia 281, 283, 284, 285 Diagnosis of iron or other deficiency anemia or unspecified anemia; patients receiving
antianemic medications in the absence of a diagnosis also were flagged
Alopecia 704.0x Unspecified alopecia, alopecia areata, telogen effluvium, and other forms of alopecia,
such as hypotrichosis; it did not include madarosis or syphilitic alopecia
Asthenia 780.7x (or HCPCS codes ICD-9 780.7x included general malaise and fatigue and unspecified asthenia;
G9029 through G9032) the HCPCS codes capture patient reports of fatigue during chemotherapy
Constipation 546.0x Slow transit, outlet dysfunction, other or unspecified constipation, or evidence of any
of the following: laxatives (bulk form, emollient, saline, stimulant) enemas, stool
softeners, or other cathartics
Cough 786.2x, 786.3x, 786.4x Cough with or without hemorrhage or abnormal sputum or evidence of a prescription
for an antitussive
Dehydration 276.50 or 276.51 Dehydration or volume depletion
Dermatitis 693.0x, 693.8x, 693.9x Dermatitis caused by substances taken internally
Diarrhea 007, 009.x, 787.91 Diarrhea associated with specific infections or nonspecific or evidence of the use
of antipropulsive
Esophagitis 530.1x Acute and reflux esophagitis, and abscess of the esophagus
Fever 780.6x Fever or pyrexia with and without chills either nonspecific or of unknown origin
Gastritis 535.xx Duodenitis and gastritis
Headache 784.0x Pain in head or face of unspecified origin; this excludes migraine and
tension headache
Infection 0001.xx-018.xx, 030.xx-041.xx, Use of any antibiotic or antifungal medication or any of the ICD-9 codes listed that
050.xx-057.xx, 110.xx-118.xx, covered intestinal infectious diseases, tuberculosis, other bacterial diseases,
070.xx-079.xx, 130.xx-136.xx, viral diseases accompanied by exanthem, mycoses, other diseases caused by
400.xx-486.xx, viruses and chlamydia, other infectious and parasitic diseases, pneumonia,
045.xx - 049.xx038.0x, 038.19, poliomyelitis and other nonanthropod-borne viral disease of the central nervous
038.80, 038.9x 995.91, 995.92 system, sepsis
Insomnia 780.51, 780.52 Insomnia with or without apnea or prescriptions for sleep aids or benzodiazepines
Mucositis 528.0x, 528.1x, 528.2x, 528.3x, Stomatitis, cancrum oris or oral aphthae, cellulitis or abscess of the mouth, oral hairy
528.6, 529.0x, 54.2x leukoplakia, glossitis, herpetic gingivostomatitis
Nausea and vomiting 787.0x, or CPT codes G9022, Diagnosis of nausea or vomiting or a claim for chemotherapy assessment for nausea
G9023, G9024 and vomiting, a prescription for an anti-emetic or an Food and Drug
Administrationapproved nerve stimulator for treatment of nausea and vomiting
Neutropenia 288.0x All forms of agranulocytosis, including neutropenia and Kostmann syndrome; patients
who received growth factors also were classified as having neutropenia
Night sweats 780.8x Generalized or secondary hyperhidrosis or diaphoresis
Weight loss 783.2x or 783.0x Abnormal weight loss or nonpsychogenic anorexia
Complications of vascular 999.3x, 996.62, 996.74, 512.0x, Central line infection, central line thrombosis, pneumothorax, and secondary
access devices 512.1x, 512.8x, 287.4x thrombocytopenia

HCPCS indicates Healthcare Common Procedure Coding System; ICD-9, International Classification of Disease, Ninth Revision; CPT, Common Procedural Code.

confounded the mean expenditure and utilization values
from the descriptive analysis.
Evaluated outcomes included total expenditure
per patient per month on treatment, complication cost
per patient per month on treatment, and cost by site of
service (inpatient, outpatient, pharmacy) per patient
per month on treatment. Expenditure outcomes were
estimated by using generalized linear models (GLM).
Wald tests were performed to test for the presence of
structural differences between each of our comparison
groups: 1) capecitabine monotherapy versus 5-FU
monotherapy and 2) XELOX versus 5-FU plus oxali-
platin. In both cases, we observed that the estimated
coefficients were significantly different between capeci-
tabine and 5-FU regimens, suggesting that there is a
different impact on expenditure between the treatment
regimens of the same covariates. Accordingly, we esti-
mated a GLM separately for each comparison. The
mean costs were calculated on the basis of the different
models and bias corrected, and accelerated bootstrap
confidence intervals (CIs) were calculated for the mean
costs. The bootstrap method also was applied to obtain
the CIs for the difference in the mean cost between
each pair of comparisons.

Cancer April 1, 2009 1415

Original Article
The resulting model predicted the cost of a capecita-
bine treatment episode assuming the mean characteristics
of patients in a 5-FU treatment episode. The difference
between costs based on mean characteristics would be the
cost difference if capecitabine users and 5-FU users had
shared the same characteristics, ie, all other things being
equal. The ratio of the difference in complication costs to
the difference in total costs is the percentage of total cost
difference between capecitabine and 5-FU that could be
explained by the difference in complication cost. The
probability of any complication was estimated using logis-
tic regression analysis.
RESULTS
Demographic and Clinical Characteristics
The characteristics of the 4973 patients who met the
inclusion criteria for this analysis and who had at least 1
drug treatment episode with capecitabine monotherapy,
5-FU monotherapy, XELOX, or 5-FU plus oxaliplatin
are summarized in Table 2. In total, 4666 of 4973
patients (94%) were administered 1 of these 4 regimens as
their index treatment (first observed chemotherapy during
the study window). The remaining patients received these
regimens subsequent to other chemotherapy regimens.
Patients in the sample had a total of 6716 treatment
episodes during the analysis period: 42.1% (n 2830) of
the treatment episodes were on 5-FU monotherapy,
33.3% (n 2234) were on 5-FU plus oxaliplatin, 19.5%
(n 1309) were on capecitabine monotherapy, and 5.1%
(n 343) were on XELOX.
Patients who received a capecitabine regimen, on av-
erage, were 2 years older than their counterparts who
received a 5-FU regimen and, accordingly, were more
likely to have Medicare as their primary payer. Patients
who received capecitabine monotherapy had more comor-
bid illnesses (CCI score, 5.1 vs 4.4) and also were more
likely to have both regional and distant metastases (41%
vs 33%) compared with patients who received 5-FU
monotherapy. Accordingly, they were more likely to be
treated in the first- or second-line settings, whereas
patients who received 5-FU were more likely to be treated
in an adjuvant setting. The clinical profile generally was
similar for the patients who received either of the combi-
nation regimens (Table 2).
1416
Rate of Complications
After adjusting for differences in patient demographics,
clinical history, and treatment setting, the probability of
developing any of the 23 complications that were studied
during an episode on capecitabine monotherapy was
lower than during treatment with 5-FU monotherapy
(P .71 vs P .90). Differences in the estimated proba-
bilities were most pronounced for the most frequent com-
plications: nausea and vomiting, infection, anemia,
diarrhea, and neutropenia. Detailed probabilities for the
monotherapy comparison are provided in Figure 1a.
The predicted complication rates for both oxalipla-
tin combination regimens (XELOX and 5-FU plus oxali-
platin) was 100%. Differences in the estimated
probabilities were most pronounced for fever, constipa-
tion, infection, and anemia. Detailed probabilities for the
combination therapy comparison are shown in Figure 1b.
Because the ICD-9 code for hand-foot syndrome,
which is a common complication of capecitabine therapy,
is not specific for hand-foot syndrome but also includes
dermatitis because of substances taken internally, we con-
ducted a sensitivity analysis that included all ICD-9 codes
indicative of an inflammatory dermatologic reaction
(codes 690-698). However, the rates remained low
(0.3%) in patients who received capecitabine therapy.
Cost of Complications
The unadjusted monthly complication cost was lower
during capecitabine-based monotherapy regimens than
during 5-FUbased monotherapy regimens ($568 vs
$1177; P < .001). Similarly, the monthly complication
cost for XELOX was lower than the cost for 5-FU plus
oxaliplatin ($1465 vs $2515; P .001). The monthly
complication cost was highest for anemia, infection, neu-
tropenia, and nausea/vomiting. Cost differences between
the regimens were driven both by differences in the fre-
quency of events and by differences in the resource inten-
sity of events. Table 3 details the cost of complication by
regimen and complication type.
After multivariate adjustment, the predicted cost of
complications per month on 5-FU monotherapy was sig-
nificantly higher, by 136% ($601; 95% CI, $469-$737),
than the cost of complications per month on capecitabine
monotherapy. Differences in complication cost between
Cancer April 1, 2009
 Capecitabine and Cost of Complications/Chu et al

Table 2. Demographic and Clinical Characteristics of the Study Population

Characteristic Capecitabine 5-FU P Capecitabine 1 5-FU 1 P All

Monotherapy Monotherapy Oxaliplatin Oxaliplatin Patients
No. of patients* 1112 2491 310 2063 4973

Demographicsy
No. of women (%) 503 (45.2) 1083 (43.5) .326 116 (37.4) 909 (44.1) .028 2167 (43.6)
Age: Mean 6 SD, y 65.2  12.4 62.9  11.7 <.001 61  11.9 59.2  11.2 .009 62.3  11.9
No. covered by Medicare (%) 590 (53.1) 1136 (45.6) <.001 115 (37.1) 712 (34.5) .373 2184 (43.9)
No. residing in urban area (%) 874 (78.6) 1852 (74.3) .006 239 (77.1) 1544 (74.8) .392 3746 (75.3)

Clinical characteristics
No. of episodes 1309 2830 NA 343 2234 6716
Treatment setting:
No. of patients (%)
Neoadjuvant setting 20 (1.8) 46 (1.8) .921 4 (1.3) 10 (0.5) .084 79 (1.6)
Adjuvant setting 341 (30.7) 1231 (49.4) <.001 79 (25.5) 974 (47.2) <.001 2580 (51.9)
First-line setting 451 (40.6) 869 (34.9) .001 121 (39) 667 (32.3) .02 2108 (42.4)
Second/third-line settings 426 (38.3) 638 (25.6) <.001 134 (43.2) 555 (26.9) <.001 1502 (30.2)
Charlson Comorbidity Index: 5.1  3.3 4.4  3.3 <.001 5.2  3.4 5.7  3.3 .013 5  3.3
Mean  SD
Chronic disease score: 3.7  3.4 3.2  3.2 <.001 3.1  3.3 3.2  3.2 .609 3.3  3.3
Mean  SD

Comorbid conditions: No. of patients (%)

Anemia 244 (21.9) 526 (21.1) .576 55 (17.7) 504 (24.4) .01 1128 (22.7)
Anxiety/depression 32 (2.9) 65 (2.6) .646 11 (3.5) 65 (3.2) .711 134 (2.7)
Cardiac arrhythmia 131 (11.8) 286 (11.5) .795 31 (10) 189 (9.2) .635 550 (11.1)
Cerebrovascular disease 61 (5.5) 95 (3.8) .023 9 (2.9) 54 (2.6) .77 196 (3.9)
Congestive heart failure 63 (5.7) 103 (4.1) .043 8 (2.6) 73 (3.5) .386 216 (4.3)
Coronary artery disease 188 (16.9) 385 (15.5) .271 36 (11.6) 273 (13.2) .429 753 (15.1)
Chronic obstructive 81 (7.3) 136 (5.5) .033 15 (4.8) 109 (5.3) .743 294 (5.9)
pulmonary disorder
Diabetes 188 (16.9) 399 (16) .505 44 (14.2) 329 (15.9) .429 819 (16.5)
Hypertension 280 (25.2) 702 (28.2) .062 58 (18.7) 558 (27) .002 1376 (27.7)
Hypercalcemia 0 (0) 3 (0.1) .247 0 (0) 2 (0.1) .583 5 (0.1)
Renal complications 39 (3.5) 76 (3.1) .472 16 (5.2) 65 (3.2) .069 160 (3.2)
Metastatic disease|| 546 (49.1) 907 (36.4) <.001 169 (54.5) 1162 (56.3) .549 2271 (45.7)
No metastases 652 (58.6) 1659 (66.6) <.001 166 (53.5) 980 (47.5) .1303 2915 (58.6)
Regional metastases 107 (9.6) 316 (12.7) .005 29 (9.4) 328 (15.9) .0107 653 (13.1)
Distant metastases 353 (31.8) 516 (20.7) <.001 115 (37.1) 755 (36.6) .7368 1405 (28.3)

Previous treatment: No. of patients (%)

Vascular access device 15 (1.3) 31 (1.2) .796 5 (1.6) 40 (1.9) .695 74 (1.5)
Surgery 557 (50.1) 1555 (62.4) <.001 128 (41.3) 1258 (61) <.001 2959 (59.5)
Radiation therapy 197 (17.7) 682 (27.4) <.001 32 (10.3) 215 (10.4) .957 903 (18.2)
Hormone therapy 10 (0.9) 10 (0.4) .063 0 (0) 11 (0.5) .198 27 (0.5)
Baseline expenditure: 3956  5103 4035  4782 .654 3920  4693 4534  4481 .025 4226  4834
Mean 6 SD, $

5-FU indicates 5-fluorouracil; SD, standard deviation; NA, not applicable.

* One patient may receive multiple regimens; therefore, the total number of study-eligible patients will be less than the sum of each regimen.
y Demographics were measured on the date the first study-eligible chemotherapy was administered.
z One patient may receive treatment in multiple settings; therefore, the total number of study-eligible patients will be less than the sum in each setting.
Measured in the 6 months before the first study-eligible chemotherapy unless otherwise specified.
| Measured using all available data before index.
Mean per month, over 6 months preindex.

Cancer April 1, 2009 1417

Original Article

FIGURE 1. Estimated probability of an event (adjusted) by event type. The top 10 most likely events are shown. (a) Estimated
probability of a complication event in capecitabine or 5-fluorouracil monotherapy regimens. (b) Estimated probability of a com-
plication event in capecitabine or 5-fluorouracil combination regimens.

the oxaliplatin combination regimens were not statisti-
cally significant ($1165; 95% CI, $892-$1595). Compli-
cation cost represented between 6% and 17% of the total
mean monthly expenditure incurred. Figure 2 depicts
these proportions graphically by regimen.
Direct Medical Costs: Pharmacy and
Medical Expenditures
Total mean monthly expenditures (unadjusted), includ-
ing the cost of complications, were similar for capecita-
bine and 5-FU monotherapies ($8176 and $8066 per
month, respectively) and for each of the combination regi-
mens with oxaliplatin (XELOX and 5-FU plus oxalipla-
tin; $14,360 and $14,940 per month, respectively).
Monthly per patient expenditure also was evaluated
by site of service. There was an increase in pharmacy ex-
penditure during episodes on capecitabine monotherapy
compared with 5-FU monotherapy ($3808 vs $2571; P <
.001); however, capecitabine was associated with lower
inpatient expenditure ($1270 vs $1888; P .015). The
cost of direct administration of each regimen (drug acqui-
sition, office visit, vascular access, and infusion therapy)
was $1957 for each month of capecitabine monotherapy

1418 Cancer April 1, 2009

Cancer
April 1, 2009
Table 3. Unadjusted Monthly Complication Cost During Treatment By Treatment Regimen: The Top 10 Most Expensive Complications*

Capecitabine 5-FU Monotherapy, Capecitabine 5-FU 1 Oxaliplatin,

Monotherapy, N 5 2840 1 Oxaliplatin, N 5 2250
N 5 1317 N 5 343
Complication Mean SD Mean SD P Mean SD Mean SD P
Chemotherapy
Anemia $214 $860 $432 $1754 <.001z $531 $1942 $564 $1046 .638
Infection $55 $454 $281 $4042 .043z $99 $865 $329 $3426 .216
Neutropenia $41 $608 $173 $1076 <.001z $316 $1897 $930 $2044 <.001z
Nausea and vomiting $168 $1366 $143 $877 .484 $405 $1843 $388 $1272 .822
Diarrhea $34 $472 $40 $352 .653 $27 $222 $47 $526 .478
Dehydration $8 $177 $23 $661 .429 $30 $375 $34 $751 .909
Central line infection $3 $94 $23 $443 .104 $0 $0 $106 $2963 .509
Fever $9 $197 $11 $139 .69 $6 $67 $20 $195 .184
Asthenia $2 $22 $11 $101 .002z $5 $21 $22 $267 .217
Central line thrombosis $0 $2 $11 $174 .028z $20 $335 $29 $441 .737
Total complication expenditure $568 $2039 $1177 $4912 <.001z $1465 $3860 $2515 $5874 .001z

5-FU indicates 5-fluorouracil; SD, standard deviation.

* The mean monthly expenditure during the treatment episode for all claims carrying codes indicative of each complication specified.
y Includes capecitabine in combination with oxaliplatin; may include additional chemotherapeutic agents other than 5-FU and irinotecan.
z Statistically significant difference.

1419
Capecitabine and Cost of Complications/Chu et al
Original Article

compared with $617 for each month of 5-FU
monotherapy.
Pharmacy expenditure was similar between compa-
rators during combination regimens with oxaliplatin
($10,049 vs $10,026; P .954), whereas all other medical
costs were similar. The cost of direct administration of
each regimen (drug acquisition, office visit, vascular
access, and infusion therapy) was $7062 for each month
of XELOX compared with $6484 for each month of 5-
FU and oxaliplatin.
Total expenditure per month on the drug was
adjusted for differences in demographics, baseline comor-
bidity, stage of disease, and treatment setting. The result-
ing expenditures associated with capecitabine and 5-FU
monotherapies were $7952 and $7441, respectively.
Adjusted pharmacy cost (all pharmacy expenditures) was
higher for capecitabine monotherapy than for 5-FU
monotherapy ($3423 vs $2222); however, capecitabine
monotherapy was associated with lower inpatient and out-
patient costs (in aggregate, differences in total monthly
expenditures did not reach statistical significance; $511;
95% CI, from $1160 to $221).
Differences in mean monthly expenditure for all
healthcare services were lower during episodes on XELOX
compared with episodes on 5-FU and oxaliplatin ($1212;
95% CI, from $29 to $2538), although the difference
was not statistically significant. The mean, adjusted
monthly expenditures were $13,398 during episodes on
XELOX and $14,610 during episodes on 5-FU plus oxali-
platin. Monthly expenditure in all sites of service (inpa-
tient, outpatient, and pharmacy) appeared lower with
capecitabine-based combination regimens as opposed to
5-FU plus oxaliplatin. Table 4 summarizes the mean
monthly difference adjusted by treatment regimen and
site of service.

DISCUSSION
The introduction of novel cancer treatments often is iden-
tified as a key driver in the escalation of healthcare cost.
This focus is largely because of the high unit cost associ-
ated with the newer cytotoxic or biologic therapies, yet
few studies have evaluated their cost impact in a real-
world setting. To address this gap, in the current study,
we used a database that reflected the healthcare expendi-
ture of 25 million individuals of all ages who were covered
by employer-paid commercial or Medicare supplemental
insurance, integrating inpatient, outpatient, and pharma-
ceutical drug claims. We characterized the patient popula-
tion treated with either the traditional fluoropyrimidine
FIGURE 2. Adjusted expenditure per month on treatment, by
treatment regimen.
5-FU or the oral fluoropyrimidine capecitabine and then
quantified the cost per patient during treatment with

Table 4. Differences in Mean Monthly Expenditure During Treatment With 5- Fluorouracil Versus Capecitabine*

5-FU Monotherapy vs Capecitabine 5-FU 1 Oxaliplatin vs Capecitabine 1 Oxaliplatin

Unadjusted Adjusted Unadjusted Adjusted
Expenditure Mean Mean CI Mean Mean CI

Inpatient $618 $621 $316, $1018* ($290) $358 $91,$1406

Outpatient $509 $216 $96, $609 $893 $1035 $781,$1360*
Pharmacy ($1237) ($1202) $1462,$891 ($23) $245 $536, $1127*
Total ($110) ($511) $1160, $221 $580 $1212 $29, $2538*

5-FU indicates 5-fluorouracil.

* Values greater than zero reflect higher monthly expenditure during 5-FU regimens (negative values are in parentheses).

1420 Cancer April 1, 2009


these regimens as well as the cost of complications fre-
quently reported with these treatments.
Capecitabine regimens were associated with a lower
incidence of adverse events compared with 5-FU therapy,
which translated into a lower monthly cost for treatment
of complications. Both unadjusted and adjusted monthly
complication costs per treatment episode were signifi-
cantly lower for capecitabine regimens. This reduction in
costs was most evident when comparing the adjusted costs
for treating complications associated with capecitabine
and 5-FU monotherapies, in which the cost associated
with 5-FU was >2-fold higher than the cost associated
with capecitabine.
Pharmacy costs were higher for capecitabine mono-
therapy compared with 5-FU monotherapy, which was
primarily because of the higher acquisition cost of capeci-
tabine. However, total monthly healthcare expenditure
adjusted from multivariate analyses did not differ signifi-
cantly for capecitabine monotherapy compared with 5-
FU (total adjusted monthly costs, $7952 vs $7441,
respectively). In addition, the XELOX regimen also com-
pared well with 5-FU plus oxaliplatin (estimated monthly
cost saving, $1212). The difference in cost of treating
complications was a contributing factor in the lower total
cost of the capecitabine regimen.
The findings in our analysis are similar to pharma-
coeconomic data published previously indicating that
capecitabine is not more costly than other 5-FU regi-
mens.10-13 However, these findings are in sharp contrast
to what was presented recently in a commentary reporting
that, based on drug acquisition cost, infusion room
charge, and pump cost, the estimated cost of XELOX is
almost double the cost of the 5-FU plus oxaliplatin com-
parator regimens (approximately $50,000 vs $25,000-
$27,000 for 12 weeks of treatment).14 It is noteworthy
that the drug acquisition cost assumptions for XELOX in
that analysis were more than double the actual acquisition
cost observed during the current study. Moreover, the
analysis was flawed, because it did not factor in the costs
associated with insertion and maintenance of central ve-
nous catheters, the costs associated with the use of infu-
sion pumps, and the costs associated with chemotherapy-
and catheter-related complications.
In a recent prospective comparison of medical
resource use (examined from the US third-party payer
and societal [time and travel costs] perspective), Garrison
Cancer April 1, 2009
Capecitabine and Cost of Complications/Chu et al
and colleagues demonstrated that patients who received
combined oxaliplatin, LV, and 5-FU (FOLFOX-4)
required approximately 15 more office visits for drug
administration than patients who received XELOX.15 In
addition, those authors observed that patients who
received intravenous FOLFOX-4 spent at least 60 hours
more time for office and clinic visits compared with
patients who received XELOX. In a cost comparison of
medical resource use, the total direct medical cost esti-
mates were comparable between the XELOX regimen
($44,500) and the FOLFOX-4 regimen ($45,800),
whereas indirect patient time cost estimates were lower
with XELOX (reduced by $1500 to $1700 per patient)
compared with FOLFOX-4.15
A cost-minimization analysis conducted in France
that compared XELOX with FOLFOX-6 indicated that
the capecitabine regimen decreased hospital resource con-
sumption compared with FOLFOX-6.16 In that analysis,
the average cost of chemotherapy per cycle per patient was
significantly lower with XELOX (608  446 vs
1043  787; P < .001). Thus, the findings of these
prospective pharmacoeconomic and cost analyses are con-
sistent with the findings in our retrospective analysis indi-
cating that capecitabine-based therapy has a favorable cost
comparison with 5-FUbased therapy in patients with
CRC.
In interpreting the findings of our retrospective
database analysis, several factors should be considered.
First, although the current study was based on a large,
national sample of patients with CRC, it was not a ran-
dom population sample, and the sources of the data are
worth reviewing. The MarketScan databases comprise
employer-sponsored health insurance for active employ-
ees, retirees, and their dependents and, thus, may not gen-
eralize to the uninsured or Medicaid-insured populations.
Second, diagnostic and procedural information was
recorded to the extent required for reimbursement; but,
otherwise, the clinical information available to researchers
is limited. Third, metastatic status was concluded using
the Medstat disease staging algorithm, which relies on
accurate ICD-9 assignment but cannot be linked to spe-
cific American Joint Committee on Cancer stages. Thus,
the algorithm may understate the frequency of metastatic
disease. Fourth, certain adverse events that are especially
common in patients undergoing chemotherapy may not
be recorded on a claim unless they are deemed clinically
1421
Original Article
relevant. For example, the rate of hand-foot syndrome
was low in our analysis. Hand-foot syndrome is recorded
using the ICD-9 code 693, which includes dermatitis
because of substances taken internally. This code was
recorded in 0.3% of patients who received capecitabine
monotherapy and in no patients who received XELOX.
To ensure that hand-foot syndrome was not being mis-
coded, we conducted a sensitivity analysis using all ICD-9
codes that were indicative of an inflammatory dermato-
logic reaction (codes 690-698), but the rates still remained
low. It is possible that the episodes of hand-foot syndrome
that were observed were not clinically significant enough
to warrant coding on the part of clinicians.
In addition, clinical and laboratory measures were
not available to confirm specific clinical events (eg, ane-
mia, neutropenia, and venous thromboembolism). There-
fore, the frequency of complications may be understated.
To offset this phenomenon, medications that were
designed specifically to treat these events also were identi-
fied and used to flag the presence of complications. To the
extent that these treatments were initiated as prophylaxis,
the frequency of complication may be overstated. Finally,
a key problem that often plagues observational studies is
the lack of randomization in assigning individuals to ei-
ther treatment or control groups. Given this concern, the
estimation of the effects of treatment may be biased by the
existence of confounding factors. For the current study,
we used multivariate models to adjust for these pretreat-
ment differences. Given the strength of the selection bias
observed in the current study, future studies may want to
incorporate a propensity model approach in combination
with multivariate adjustment.
To our knowledge, this is the first US study to
include a detailed cost comparison between oral and intra-
venous fluoropyrimidine chemotherapy in patients with
CRC. Our findings suggest that, despite the higher drug
acquisition costs associated with capecitabine, monthly
per patient expenditures are comparable to those observed
during treatment with 5-FU as a result of the lower com-
plication costs associated with capecitabine. This retro-
spective analysis highlights the overall pharmacoeconomic
advantage of capecitabine-based chemotherapy compared
with intravenous 5-FU-based therapy and suggests the
continued use of oral capecitabine as a way to reduce the
ever-increasing healthcare costs associated with CRC
treatment.
1422
Conflict of Interest Disclosures
Supported by Roche.
References
1. Grothey A, Sargent D, Goldberg RM, Schmoll HJ. Survival
of patients with advanced colorectal cancer improves with
the availability of fluorouracil-leucovorin, irinotecan, and
oxaliplatin in the course of treatment. J Clin Oncol.
2004;22:1209-1214.
2. Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizu-
mab plus irinotecan, fluorouracil, and leucovorin for meta-
static colorectal cancer. N Engl J Med. 2004;350:2335-
2342.
3. Twelves C, Wong A, Nowacki MP, et al. Capecitabine as
adjuvant treatment for stage III colon cancer. N Engl J
Med. 2005;352:2696-2704.
4. Van Cutsem E, Twelves C, Cassidy J, et al. Oral capecita-
bine compared with intravenous fluorouracil plus leucovo-
rin in patients with metastatic colorectal cancer: results of a
large phase III study. J Clin Oncol. 2001;19:4097-4106.
5. Hoff PM, Ansari R, Batist G, et al. Comparison of oral
capecitabine versus intravenous fluorouracil plus leucovorin
as first-line treatment in 605 patients with metastatic color-
ectal cancer: results of a randomized phase III study. J Clin
Oncol. 2001;19:2282-2292.
6. Scheithauer W, McKendrick J, Begbie S, et al. Oral capeci-
tabine as an alternative to intravenous 5-fluorouracil-based
adjuvant therapy for colon cancer: safety results of a
randomized, phase III trial. Ann Oncol. 2003;14:1735-
1743.
7. Cassidy J, Twelves C, Van CE, et al. First-line oral capeci-
tabine therapy in metastatic colorectal cancer: a favorable
safety profile compared with intravenous 5-fluorouracil/leu-
covorin. Ann Oncol. 2002;13:566-575.
8. Diaz-Rubio E, Tabernero J, Gomez-Espana A, et al. Phase
III study of capecitabine plus oxaliplatin compared with
continuous-infusion fluorouracil plus oxaliplatin as first-line
therapy in metastatic colorectal cancer: final report of the
Spanish Cooperative Group for the Treatment of Digestive
Tumors Trial. J Clin Oncol. 2007;25:4224-4230.
9. Porschen R, Arkenau HT, Kubicka S, et al. Phase III study
of capecitabine plus oxaliplatin compared with fluorouracil
and leucovorin plus oxaliplatin in metastatic colorectal can-
cer: a final report of the AIO Colorectal Study Group.
J Clin Oncol. 2007;25:4217-4223.
10. Twelves C, Gollins S, Grieve R, Samuel L. A randomised
cross-over trial comparing patient preference for oral capeci-
tabine and 5-fluorouracil/leucovorin regimens in patients
with advanced colorectal cancer. Ann Oncol. 2006;17:239-
245.
11. Jansman FG, Postma MJ, Brouwers JR. Cost considerations
in the treatment of colorectal cancer. Pharmacoeconomics.
2007;25:537-562.
Cancer April 1, 2009

12. Cassidy J, Douillard JY, Twelves C, et al. Pharmacoeco-
nomic analysis of adjuvant oral capecitabine vs. intravenous
5-FU/LV in Dukes C colon cancer: the X-ACT trial. Br J
Cancer. 2006;94:1122-1129.
13. Ward SE, Kaltenthaler E, Cowan J, Marples M, Orr B,
Seymour MT. The clinical and economic benefits of cape-
citabine and tegafur with uracil in metastatic colorectal can-
cer. Br J Cancer. 2006;95:27-34.
14. Mayer RJ. Should Capecitabine replace infusional fluorour-
acil and leucovorin when combined with oxaliplatin in met-
astatic colorectal cancer? J Clin Oncol. 2007;25:4165-4167.
Cancer April 1, 2009
Capecitabine and Cost of Complications/Chu et al
15. Garrison L, Best J, Cassidy J, et al. Cost comparison of
XELOX vs. FOLFOX-4  bevacizumab in metastatic col-
orectal cancer. Presented at the 43rd Annual Meeting of
the American Society of Clinical Oncology, June 1-5,
2007, Chicago, Illinois. Poster 4074.
16. Perrocheau G, Bennouna J, Ducreux M, et al. Cost-mini-
mization analysis of a phase III study of capecitabine
oxaliplatin (XELOX) vs. infusional 5-FU/LV oxaliplatin
(FOLFOX-6) as first-line treatment for metastatic colorectal
cancer (MCRC) in the French setting [abstract]. J Clin
Oncol. 2007;25:184s. Abstract 4083.
1423