Jco 1997 15 3 974

Reliability and Validity of the Functional Assessment of
Cancer Therapy-Breast Quality-of-Life Instrument

By Marianne J. Brady, David F. Cella, Fei Mo, Amy E. Bonomi, David S. Tulsky, Steven R. Lloyd, Shirley Deasy,
Melody Cobleigh, and Gail Shiomoto
Purpose: This is the first published report on the vali- Well-Being (FWB) subscale, and the BCS. Sensitivity to
dation of the Functional Assessment of Cancer Therapy- change in QL as measured by the Functional Living Index-
Breast (FACT-B), a 44-item self-report instrument de- Cancer (FLIC) was documented in the FACT-B total score,
signed to measure multidimensional quality of life (QL) PWB, FWB, and Emotional Well-Being (EWB). Additional
in patients with breast cancer. The FACT-B consists of the validity and reliability data were obtained from a larger
FACT-General (FACT-G) plus the Breast Cancer Subscale sample (n = 295). The alpha coefficient (internal consis-
(BCS), which complements the general scale with items tency) for the FACT-B total score was high (a = .90), with
specific to QL in breast cancer. The FACT-B was developed subscale alpha coefficients ranging from .63 to .86. Evi-
with an emphasis on patients' values and brevity and is dence supported test-retest reliability, as well as conver-
available in nine languages. gent, divergent, and known groups validity.
Methods and Results: Two validation samples were Conclusion: The FACT-B is appropriate for use in on-
used for this report. The first (n = 47) was tested twice cology clinical trials, as well as in clinical practice. It dem-
over a 2-month period to assess sensitivity to change. onstrates ease of administration, brevity, reliability, va-
Significant sensitivity to change in performance status lidity, and sensitivity to change.
rating (PSR) was demonstrated for the FACT-B total score, J Clin Oncol 15:974-986. 1997 by American So-
the Physical Well-Being (PWB) subscale, the Functional ciety of Clinical Oncology.
AMONG North American and European women, curative. In these circumstances, the quality of a pa-
breast cancer is the most frequently diagnosed tient's survival may be as important as the length of
malignancy.' It is estimated that in the United States survival. 12
alone, more than 184,000 women will be diagnosed QL issues for breast cancer patients include such fac-
with this disease in 1996.2 This large number is reflec- tors as pain, fear of recurrence, and fatigue, 5' concerns
tive of an increased incidence of breast cancer, with, common to a wide range of cancers. However, breast
for example, incidence increasing 32% between 1980 cancer patients also identify more unique concerns, for
and 1987. 3 This sharp increase has been attributed to example, an altered sense of femininity, feelings of de-
the more widespread use of mammography, with a con- creased attractiveness,' ° and problems associated with
sequent earlier detection of disease.3 Earlier detection, treatment-related arm swelling.' 3 Assessing these health-
and therefore earlier treatment, is expected to result in related QL issues in cancer patients is now widely advo-
improved survival rates,3 with some indications that cated,' 4 "'5 with empirical QL data collected as end points
this improved survival may already be apparent. 4 How- in many clinical trials and in clinical practice organiza-
ever, the quality of that survival can be affected sig- tions. QL is also recognized as important to informed
nificantly by both the disease and its treatment, 5 '6 with decision-making.
even the most conservative current treatments dis-
rupting quality of life (QL).7 -" Furthermore, for pa- MEASURING QL IN BREAST CANCER PATIENTS
tients with metastatic disease, current therapy is not
There is currently no gold standard for measuring QL
in breast cancer patients. Scales specific to breast can-
cer,' 6 '1 as well as more general cancer-related QL
From the Rush CancerInstitute, Chicago; Cook County Hospital,
Chicago, IL; University of Washington, Seattle, WA; The Psycholog- scales,' 1 20 have been used in breast cancer research. How-
ical Corporation, San Antonio, TX; and Emory University, Atlanta, ever, each type of scale has its drawbacks. The more
GA. specific scales can provide adequate coverage of breast
Submitted January 2, 1996; accepted September 23, 1996. cancer issues, but this specificity can prevent comparisons
Supported by grants no. PBR87 (American Cancer Society, At-
across different cancer sites and treatments. While the
lanta, GA) and R29CA51926 (National Cancer Institute, Bethesda,
MD). general scales allow for such comparisons, they can fail
Presented in part at the Fourth International Congress of Behav- to capture important breast cancer-specific QL issues
ioral Medicine, Washington, DC, March 13-16, 1996. (eg, feelings of femininity, self-consciousness about
Address reprint requests to David F. Cella, PhD, Division of
clothes) that can differ across current treatments. It is our
Psychosocial Oncology, Rush Cancer Institute, 1725 W. Harrison
St, Suite 863, Chicago, IL 60612; Email [email protected]. position that a scale incorporating both generalizability
© 1997 by American Society of Clinical Oncology. and specificity is the ideal compromise between these two
0732-183X/97/1503-0050$3.00/0 competing considerations.
974 Journal of Clinical Oncology, Vol 15, No 3 (March), 1997: pp 974-986
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102
Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
CANCER THERAPY-BREAST QUALITY-OF-LIFE INSTRUMENT 975
THE FUNCTIONAL ASSESSMENT OF CANCER Although the FACT-B is currently being used rather
THERAPY-BREAST extensively in national and even international clinical tri-
Since 1987, our group has been involved in the devel- als and clinical practice, this article is the first published
opment of multidimensional health-related QL instru- report on the development of the BCS and the validation
ments that are simple and brief, yet sensitive to meaning- of the FACT-B. This article also reports on the reliability
ful change over time. It was decided that there would be and validity of a 23-item outcome measure, the Trial
a set of general questions that would be applicable to Outcome Index-Physical/Functional/Breast (TOI-PFB),
patients with all types of cancer, and perhaps other which can be used in analyses as a summary measure of
chronic illnesses. Additional symptoms or concerns spe- PWB and FWB in breast cancer patients.
cific to a condition (such as breast cancer), or to a treat- METHODS
ment (such as bone marrow transplant), could then be Development of the FACT-B (Version 3)
added as a subscale to the general questionnaire. We As the science of QL measurement develops, it is important that
would therefore have adequate sensitivity to specific is- there be continual refinement of instrumentation. Our group has been
sues, while retaining the ability to compare general QL committed to refining the FACT scales over time as experience with
issues across cancer sites. Patient input in the generation the scales becomes more extensive. Therefore, the FACT scales,
of questions has been emphasized because relevance to including the FACT-B, are now in their third version. The original
development of version 1 of the FACT and modifications that led
patient values is central to the concept of QL. Along these to versions 2 and 3 are described in detail elsewhere.2 ' 23 The spe-
lines, we have also sought to provide the patient with the cifics of the FACT development as they apply to the FACT-B are
opportunity to weight the various QL dimensions. This as follows. In-depth interviews were conducted with 15 breast cancer
is necessary for a true estimation of treatment costs and patients and five experts, defined as credentialed MD or RN oncology
benefits specific to the individual. professionals with at least 3 years experience treating more than 100
patients with breast cancer. The purpose of the interviews was to
The result of this effort has been the Functional Assess- develop a large pool of questions, or items, relevant to QL in breast
ment of Cancer Therapy (FACT) Measurement System, cancer patients.
which includes the FACT-General (FACT-G) Scale, as Patients were eligible for this study if they met the following
well as many disease-specific, symptom-specific, and criteria: (1) presented with advanced (stage III or IV) breast cancer;
treatment-specific subscales.2 '24 The FACT-G is multidi- (2) were currently receiving chemotherapy and/or radiation therapy;
(3) could read and speak English; and (4) were not impaired cogni-
mensional, consisting of subscales assessing Physical tively with CNS metastasis, overt psychosis, major depression, or
Well-Being (PWB), Emotional Well-Being (EWB), So- delirium. All 15 breast cancer patients who were approached ("to
cial Well-Being (SWB), Functional Well-Being (FWB), help develop a measure of QL for people with your illness") agreed
and Relationship with Doctor (RWD). The FACT-G to participate. The open-ended interviews with the 20 initial partici-
pants (15 patients; five experts) resulted in the production of 137
yields a total score, as well as individual subscale scores, candidate questions. In the next step, 30 FACT-naive breast cancer
with higher scores reflecting better QL. patients used a four-point scale to rate the relative importance to
We have developed a Breast Cancer Subscale (BCS) QL of each question. The purpose of this step was to ensure that
for the FACT. The BCS is comprised of nine items spe- content considered by patients to be relevant to QL would be retained
cific to QL in breast cancer but not already included in in the questionnaire. The results were reviewed by an expert panel.
Comparable data from patients with lung and colon cancer were also
the FACT-G. When the BCS is added to the FACT-G, obtained and reviewed. Questions rated as highly important (median
the 44-item instrument is known as the FACT-Breast 2 3.0 on a scale of one to four) were retained. Questions that
(FACT-B). overlapped with the responses from patients with other types of
We are aware of only one other cancer-related QL cancer became part of the FACT-G. The nine responses rated as
instrument using a modular approach and offering a breast highly important by the breast cancer patients, but evidencing little
overlap with the FACT-G, were retained as the BCS. The FACT-G
cancer-specific module or subscale as an addition to the plus the BCS make up the FACT-B. After extensive testing of the
general scale. Like the FACT, the European Organization FACT-B, one item from the BCS version 1, "I want to encourage
for Research and Treatment of Cancer Quality of Life other women to examine their breasts," was deleted because of poor
Questionnaire (EORTC QLQ) 2 527 - is multidimensional, statistical and conceptual fit with the rest of the subscale. It was
brief, and has been extensively translated. Differences replaced by the item "I am able to feel like a woman." The FACT-
B has to date been translated in the following languages: Spanish,
between the FACT and EORTC QLQ are that the FACT Flemish (Dutch), French (appropriate for use in Canada, France, and
allows for a total QL score, broader coverage of the SWB Belgium), German (appropriate for use in Germany, Austria, and
domain, and the opportunity for patients to provide indi- Switzerland), Italian, Norwegian, Swedish, and Japanese. The trans-
vidualized weighting for the various QL domains. Re- lation procedures are described elsewhere. 24
search comparing the FACT and EORTC QLQ is cur- Reliability and Validity of the FACT-B
rently underway, and objective data relative to the Subjects. The data used in this report were derived from two
advantages of each scale will be forthcoming. samples. The first sample was tested twice over a 2-month period
976 BRADY ET AL
and was therefore used for the sensitivity to change analysis. Because Table 2. Demographic and Clinical Characteristics:
this sample also completed a QL instrument, the Functional Living Reliability/Concurrent Validity Sample
Index-Cancer (FLIC), 2 ° data from this sample were used to assess No. of Patients
concurrent validity as well. This first sample consisted of 47 breast (n = 295) %
cancer patients who completed version 1 of the FACT-B as part of
Ethnicity
a funded psychosocial QL intervention project targeting patients
White non-Hispanic 136 46
with advanced (stage III or IV) breast cancer. All patients were
Black non-Hispanic 151 51
being treated at Rush-Presbyterian-St. Luke's Medical Center in
Hispanic 6 2
Chicago, IL. Patients completed the FACT-B once at baseline, and
Other 1 .3
again 2 months later. Demographic and clinical data for these pa-
tients are listed in Table I. Missing 1 .3
Age (years)
The second sample consisted of 295 breast cancer patients who
Median 55
completed version 3 of the FACT-B as part of a large 3-year valida-
Range 28-86
tion study of the FACT Measurement System. Data from this admin-
istration of the FACT-B were used for reliability analyses, as well Education level (years)
Median 12
as to further assess validity. These patients were recruited from the
Range 3-28
following medical centers: Rush-Presbyterian-St. Luke's Medical
Marital status
Center, Chicago, IL; Grady Memorial Hospital (Emory University),
Single 44 15
Atlanta, GA; and Cook County Hospital, Chicago, IL. To participate
Married 127 43
in this study, patients had to meet the following criteria: (1) be
Separated 26 9
diagnosed with breast cancer (any stage of disease); (2) be older
Divorced 48 16
than 18 years; (3) have no metastases to the brain or CNS; and (4)
Widowed 46 16
not be using psychotropic medication. Demographic and clinical
Missing 4 1
Living status
Table 1. Demographic and Clinical Characteristics: Alone 56 19
Sensitivity to Change Sample With other adults (no children) 149 51
With other adults and children 72 24
No. of
Patients With children only 16 5
In 47) % In an institution, retirement home, or shelter 2 1
Extent of disease
Ethnicity
No evidence of Disease 125 42
White non-Hispanic 33 70
Local disease 67 23
Black non-Hispanic 12 26
Regional disease 43 15
Asian 1 2
Distant metastases 58 20
Hispanic 1 2
Missing 2 1
Age (years)
ECOG PSR
Median 56
Normal activity 149 51
Range 32-80
Have symptoms/no extra time in bed 89 30
Education level (years)
< 50% daytime in bed 42 14
Median 14
> 50% daytime in bed 12 4
Range 4-18
Unable to get out of bed 0 0
Marital status
Missing 3 1
Single 2 4
Married 35 75
Separated 0 0
data for these patients are listed in Table 2. A subset (n = 32) of
Divorced 8 17
these patients completed the FACT-B on a second occasion, 3 to 7
Widowed 2 4
days after the first administration, to assess test-retest reliability
Living status
(stability of the instrument over time). These patients were a consec-
Alone 5 11
utive subsample, drawn from Cook County Hospital and Rush-Pres-
With other adults (no children) 25 53
byterian-St. Luke's Medical Center.
With other adults and children 15 32
Materials and procedure. Patients completed multiple instru-
With children only 2 4
ments in each of the two studies from which data in this report are
Institution or retirement home 0 0
derived. Two of the most relevant instruments are the FACT-B
Disease stage
and the Eastern Cooperative Oncology Group (ECOG) performance
III 11 23
status rating (PSR). 2 8 Validity studies for QL instruments frequently
IV 36 77
use physician-rated performance status as indicators of health status
ECOG PSR
and clinical change.2 0 25
' '26 29 The PSR has been shown to be respon-
Normal activity 21 45
sive to changes in symptom status and clinical course, 2930 as well to
Have symptoms/no extra time in bed 13 28
changes in physical and role functioning. 2 6 However, it is generally
< 50% daytime in bed 10 21
acknowledged that there is discrepancy between physician and pa-
> 50% daytime in bed 3 6
tient-rated indices of performance status and symptomatology, with
Unable to get out of bed 0 0
the patient rating viewed as more important because it contains
32
subjective information only the patient can provide.1, We therefore ment correlation (r) was used to assess the degree of association
2 ' 23
include patient-rated versions in the FACT validity studies. " Both between the measures. It was expected that there would be high
physician and patient-rated performance status ratings have been correlations between the FLIC and the FACT-B because both purport
33
shown to be valid prognostic indicators. to measure QL. It was also expected that there would be high correla-
All patients completed the FACT-B in one sitting, with staff per- tions between the EWB subscale of the FACT-B and the POMS-
sonnel available to answer any questions or clarify any item meaning. SF, a measure of emotional distress. It was further hypothesized that
Patients were also shown the ECOG PSR descriptors and asked to the more physical FACT-B subscales would relate most highly with
assign themselves to one of the five subgroups (0, fully ambulatory the more physical POMS-SF subscales, POMS-Vigor and POMS-
without symptoms; 1, fully ambulatory with symptoms; 2, requires Fatigue. No significant relationship was expected between the
bed rest < half of the waking day; 3, requires bed rest > half of FACT-B and the M-CSDS, a measure of social desirability, since
the waking day; 4, bedridden). Independently, a physician-rated PSR QL and social desirability are different concepts.
was obtained from each patient's physician. Known-groups (discriminant) validity was assessed by determin-
Other relevant instruments administered included the FLIC QL ing if the FACT-B could differentiate among groups known to differ
instrument20 (administered to the first sample only), the Profile of in PSR or extent of disease. Because these two groupings are physi-
Mood States Short Form (POMS-SF),3 4 and the 10-item short form cally based, it was expected that the more physical FACT-B sub-
3
of the Marlowe-Crown Social Desirability Scale (M-CSDS). 5 These scales would be best able to differentiate these groups. As in the
instruments were administered to obtain data relevant to validity. sensitivity to change analyses, the first step in these analyses was a
Plan of analysis. All data were analyzed using the System for multivariate test, including the set of six FACT-B subscales. One-
Statistics (SYSTAT) or the Statistical Package for the Social Sci- way univariate ANOVAs were then used to specify the source of
ences (SPSS). To explore the value of combining subscales most the effect.
relevant to physical and functional well-being into a summary score,
an aggregate of the PWB, FWB, and the BCS was created. This 23- RESULTS
item TOI-PFB was subjected to the same analyses as the FACT
Descriptive Statistics
subscales.
A critical aspect of a QL instrument's validity is its sensitivity to Tables 3 (first sample, n = 47) and 4 (second sample,
change over time. Such sensitivity allows for the documentation of n = 295) list the means and SDs for each of the FACT-
change in a clinical trial, or in patients being followed-up in clinical B subscales as well as for the FACT-G total score, the
practice. Because PSR is a well-known and commonly-used indica-
tor of patient functional status, we established the ECOG PSR as a
FACT-B total score (FACT-G plus BCS), and the 23-
criterion of change. Patients were followed for change in PSR and item TOI-PFB.
categorized as improved, worsened, or unchanged. Analysis of vari-
ance (ANOVA) was used to determine if the FACT-B could distin- Reliability
guish among these groups. The same analyses were conducted with Internal consistency. Internal consistency data are
the FLIC, to assess the responsiveness of the FACT-B to changes
in a measure more multidimensional than the PSR.
listed in Table 4. Chronbach's alpha for the nine-item
Because of the fact that multiple FACT-B subscales were evalu- BCS was .63. Alpha coefficients for the other subscales
ated in these analyses, a conservative multivariate analysis of vari- were in the 0.69 to 0.86 range. FACT-G and FACT-B
ance was first conducted using the set of the six FACT-B subscales. total score alpha coefficients were both 0.90. The alpha
To further clarify where the effect was taking place, significant coefficient for the TOI-PFB was 0.88 (Table 4).
multivariate analyses were followed-up with a series of univariate
Test-retest reliability. Test-retest correlation coeffi-
ANOVAs. The aggregate and total FACT scores were evaluated
only by univariate analyses because their collinearity prevented the
cients for subscales and aggregate scores never previously
use of multivariate analysis. Significant sensitivity to change in PSR published were .88 for the BCS, .89 for the TOI-PFB,
(a scale highly dependent on symptoms and physical functioning) and .85 for the FACT-B total score, indicating a high
was expected for the subscales and aggregate scores with a number
of items relating to the more physical aspects of QL (PWB, FWB,
BCS, TOI-PFB, FACT-G total score, FACT-B total score). Sensitiv- Table 3. FACT-B Descriptive Statistics: Sensitivity to Change Sample'
ity to change in the FLIC, a multidimensional QL instrument, was Scale Meant SD
expected for PWB, FWB, EWB, and the aggregate scores. Sensitivity
to change in the FLIC was not expected for SWB or BCS subscales PWB (7-item) 21.0 5.8
because the FLIC does not provide much coverage of SWB, nor SWB (7-item) 23.3 4.2
does it address breast cancer-specific issues (such as arm swelling RWD (2-item) 7.1 0.9
and feelings of femininity). EWB (5-item) 15.6 2.9
For the sensitivity to change analyses, data from version 1 were FWB (7-item) 20.3 5.1
FACT-G total score (28-item) 87.2 13.7
used. Since one item from the version 1 BCS has been replaced,
BCS (9-item) 24.6 4.6
that item was treated as missing in the scoring. The scores for eight
FACT-B total score (37-item)t 111.8 16.3
items were therefore prorated to be consistent with the current nine-
FACT-B TOI-PFB (23-item)§ 65.9 12.5
item BCS.
Construct validity (does the FACT-B measure QL?) was assessed *n = 47.
by examining a pattern of relationships between the FACT-B and tHigher scores reflect better QL.
instruments that measure similar concepts (convergent validity) and tFACT-G plus BCS.
dissimilar concepts (divergent validity). The Pearson product-mo- §PWB plus FWB plus BCS.
978 BRADY ET AL
Table 4. FACT-B Descriptive Statistics: was significant (F(df = 12, 78) = 2.59; P = .006). Univari-
Reliability/Concurrent Validity Sample'
ate analyses revealed that, as expected, sensitivity to
Scale Meant SD Coefficient alpha change in PSR was obtained with FACT-B total score, the
PWB (7-item) 22.1 5.3 0.81 FACT-G total score, the PWB subscale, the FWB subscale,
SWB (7-item) 22.7 5.2 0.69 the BCS, and the TOI-PFB (Table 5).
RWD (2-item) 7.0 1.6 0.78
Mean change scores associated with change in the
EWB (5-item) 16.3 3.5 0.69
FWB (7-item) 20.6 6.4 0.86 FLIC are listed in Table 6. As in the previous analysis,
FACT-G total score (28-item) 88.8 16.3 0.90 a multivariate ANOVA including the set of six FACT-B
BCS (9-item) 24.1 6.5 0.63 subscales was significant (F(12, 78) = 4.25; P < .001).
FACT-B total score (37-item)t 112.8 20.9 0.90
Univariate analyses revealed that the expected results
FACT-B TOI-PFB (23-item)§ 66.9 15.3 0.88
were obtained. Sensitivity to change in the FLIC was
*N 295.
demonstrated for the PWB subscale, the EWB subscale,
tHigher scores reflect better quality of life.
tFACT-G plus BCS.
the FWB subscale, the TOI-PFB, the FACT-B total score,
§PWB plus FWB plus BCS. and the FACT-G total score (Table 6).
Construct validity. The FACT-B does evidence con-
degree of stability over time periods (3 to 7 days) during vergent validity in that it correlates significantly with an-
which no change would be expected. other measure of QL, the FLIC. Correlations between the
FLIC and the FACT-B total score (r = .87; P < .001),
Validity the FACT-G total score (r = .86; P < .001), the TOI-
Sensitivity to change. Forty-seven patients were tested PFB (r = .86; P < .001), and the BCS (r = .53; P <
at 2-month intervals to evaluate sensitivity to change. Sen- .001) (n = 47) reflect the expected overlap between these
sitivity to change in both PSR and the FLIC was evaluated. measures. Convergent validity was also demonstrated by
Mean FACT-B change scores associated with change in the pattern of correlation coefficients using the larger
PSR are listed in Table 5. The overall multivariate analysis sample (n = 295) and listed in Table 7. As was expected,
Table 5. Sensitivity of FACT-B to 2-Month Changes in PSR
Scale Group No. of Patients (N = 47) Mean Change Score (S.D) t (1,44) P
PWB (7-item) Declined PSR 8 -3.3 (1.6)

No change 29 1.2 (3.3) -4.37 < .001
Improved PSR 10 3.1 (4.2)
SWB (7-item) Declined PSR 8 0.1 (4.0)
No change 29 0.2 (2.8) 0.70 .491
Improved PSR 10 -1.0 (4.4)
RWD (2-item) Declined PSR 8 0.5 (0.5)
No change 29 -0.2 (1.0) 1.21 .250
EWB (5-item) Declined PSR 8 0.0 (1.5)
No change 29 0.7 (2.8) -0.68 .499
FWB (7-item) Declined PSR 8 -1.9 (4.5)
No change 29 -0.1 (3.1) 2.22 .031
FACT-G total score (28-item) Declined PSR 8 -4.5 (4.1)
No change 29 1.8 (8.2) -2.68 .010
BCS (9-item) Declined PSR 8 -3.0 (4.6)
No change 29 0.1 (3.8) -2.61 .012
Total FACT-B (37-item)* Declined PSR 8 7.5 (5.9)
No change 29 1.9 (8.8) -3.53 .001
TOI-PFB (23-item)t (PWB + FWB + BCS) Declined PSR 8 -8.1 (5.0)
No change 29 1.2 (6.5) -4.92 < .001
*FACT-G plus BCS.

tPWB plus FWB plus BCS.
Table 6. Sensitivity of FACT-B to 2-Month Changes in FLIC
Scale Group No. of Patients

(N = 47) Mean change (SD) t P
PWB (7-item) Declined FLIC 16 -1.5 (3.1)
No change 18 0.1 (3.2) -4.237 < .001
Improved FLIC 13 3.6 (3.4)
SWB (7-item) Declined FLIC 16 0.9 (3.3)
No change 18 -0.6 (3.4) 0.926 .359
Improved FLIC 13 -0.3 (3.4)
RWD (2-item) Declined FLIC 16 -0.2 (1.4)
No change 18 -0.3 (0.9) -1.347 .185
EWB (5-item) Declined FLIC 16 -0.5 (1.9)
No change 18 0.3 (2.6) -3.216 .003
FWB (7-item) Declined FLIC 16 -2.3 (3.4)
No change 18 -0.6 (3.1) -4.092 < .001
FACT-G total score (28-item) Declined FLIC 16 -3.7 (7.8)
No change 18 -1.1 (5.6) -4.966 <.001
BCS (9-item) Declined FLIC 16 -0.6 (4.6)
No change 18 -1.3 (3.6) -1.662 .104
Total FACT-B (37-item)' Declined FLIC 16 -4.3 (8.2)
No change 18 2.4 (6.7) -5.506 < .001
TOI-PFB (23-item)t (PWB + FWB + BCS) Declined FLIC 16 -4.5 (6.7)
No change 18 -1.8 (5.8) -5.716 < .001
'FACT-G plus BCS.
tPWB plus FWB plus BCS.
Table 7. Construct Validity: Pearson Product-Moment Correlation Coefficients'
ECOG-Phys POMS PWB SWB RWD EWB FWB FCS FACT-G FACT-B TOI-PFB M-CSDS
ECOG-Pt 0.34t 0.34t -0.59t -0.13 0.02 -0.25t -0.50t -0.40t -0.49*" -0.51t -0.58t -0.04
ECOG-Phys 0.12 -0.29t -0.05 -0.03 -0.14t -0.29t -0.14t -0.26- -0.25t -0.28+ -0.01
POMS -0.49t -0.30t -0.18t -0.60t -0.49t -0.55+ -0.66" -0.70t -0.61+ -0.14
PWB 0.24 0.19t 0.42t 0.63+ 0.50t 0.55** 0.59t 0.64t 0.02
SWB 0.34+ 0.36t 0.41 0.33t 0.43" 0.44+ 0.40+ -0.03
RWD 0.24+ 0.26+ 0.14 0.33** 0.30t 0.23t -0.04
EWB 0.48t 0.55t 0.54** 0.60t 0.58t 0.12
FWB 0.53t 0.68" 0.69t 0.66t 0.06
BCS 0.61- 0.61t 0.57+ 0.12
FACT-G 0.61 0.57+ 0.04
FACT-B 0.57t 0.07
TOI-PFB 0.08
POMS-T -0.35t -0.22t -0.13 -0.57t -0.36t -0.45t -0.49t -0.54t -0.46t
POMS-D -0.37t -0.35t -0.18t -0.53t -0.38t -0.50t -0.58+ -0.64t -0.50t
POMS-A -0.28t -0.26t -0.18t -0.45t -0.22+ -0.36t -0.41 -0.45t -0.34+
POMS-V 0.46t 0.21t 0.13 0.35t 0.57+ 0.44t 0.53t 0.55t 0.58+
POMS-F -0.54+ -0.22t -0.13 -0.46+ -0.44+ -0.47t -0.60t -0.64+ -0.57+
POMS-C -0.27+ -0.20+ -0.07 -0.34t -0.28t -0.31 -0.37 -0.40t -0.34+
NOTE. Because of occasional missing data, the actual sample size of each coefficient varies from 272 to 293. Coefficients of FACT subscales and TOI
with FACT total scores (FACT-G, FACT-B) have been corrected for overlapping items.
Abbreviations: ECOG-Phys, physician-Completed ECOG; ECOG-Pt, patient-completed ECOG;
'N = 295.
tP < .01.
tP < .001.
980 BRADY ET AL
the POMS-Depression (POMS-D), the POMS-Tension the SWB subscale, in that there was little difference be-
(POMS-T), and the POMS-Anger (POMS-A) subscales tween those participants with a PSR of 0 and those whose
correlated more highly with the EWB subscale than with score was 1, and for the EWB subscale there was no
other FACT subscales. Also as expected, the POMS- significant difference between participants with a PSR of
Vigor (POMS-V) subscale correlated highly with the 0 those with a PSR of 3. Table 8 presents the detailed
PWB and FWB subscales, and the POMS-Fatigue analysis, with specifics relative to the PSR group differ-
(POMS-F) subscale correlated highly with the PWB sub- ences.
scale. Validity is also supported by the pattern of correla- The previous analysis was also performed using physi-
tions between the POMS-SF total score (an overall mea- cian-rated PSR. The overall multivariate test was signifi-
sure of distress) and the FACT subscales. As would be cant (F(18, 784) = 2.84; P < .001). Univariate analyses
expected, the POMS-SF total correlated most highly with revealed that the PWB, FWB, and BCS subscales, as well
the EWB subscale, and least highly with the RWD sub- as the TOI-PFB, FACT-G and FACT-B total scores differ-
scale. Divergent validity was evidenced in that the corre- entiated groups by physician-rated PSR category (Table
lations between the M-CSDS and the FACT subscales 9). Scheffe' post-hoc comparisons demonstrated that, in
were uniformly small, indicating the expected lack of general, participants rated as 0 on the physician-rated PSR
relationship between these differing concepts. The small had higher FACT scores on these dimensions than did
correlations between the M-CSDS and the FACT-B sub- participants with a PSR of 1. Patients with a PSR of 0
scales also indicate that FACT-B scores are not signifi- also demonstrated higher scores on the PWB and FWB
cantly influenced by a response style characterized by an subscales than did patients with a PSR of 2, 3, or 4.
effort to answer questions in a socially desirable manner. Additional evidence supporting validity was obtained
Known-groups validity. Further evidence supporting by comparing the FACT-B and its subscales across
validity was obtained by comparing the FACT-B and its groups differing in extent of disease. The multivariate
subscales across groups differing in ECOG PSR (patient- test that included the six subscales was significant (F(18,
rated). The overall multivariate test was significant (F(18, 787) = 3.9; P < .001). Univariate analyses demonstrated
784) = 10.27; P < .001). Results of the univariate ANO- that all aggregate scores as well as the PWB, FWB, and
VAs are listed in Table 8. With the exception of the two- EWB subscales were able to differentiate according to
item RWD subscale, all subscales, as well as the TOI- extent of disease (Table 10). Scheffe' post-hoc compari-
PFB, FACT-G, and FACT-B total scores, differentiated sons demonstrated that, in general, patients with no evi-
groups by PSR category. Scheffe' post-hoc comparisons dence of disease had higher FACT scores than patients
demonstrated that, in general, participants rated as 0 on evidencing disease. However, except for the PWB scale,
the PSR (ambulatory without symptoms) had higher the differences in FACT scores between patients with no
FACT scores (better QL) in all dimensions than did parti- evidence of disease and those with distant metastases,
cipants with a PSR of either 1, 2, or 3. Exceptions were while in the expected direction, were nonsignificant.
Table 8. Oneway ANOVAs: FACT-B by Patient PSRs*
Subscale PWB SWB RWD EWB FWB BCS FACT-G FACT-BTotal TOI-PFB
PSR
0 (n - 149)
Mean 24.7 23.2 7.1 17.1 23.2 26.3 95.2 121.5 74.2
SD 3.8 5.2 1.6 3.2 5.1 5.3 12.6 15.5 10.6
1 (n = 88)
Mean 21.3 23.1 6.9 15.9 20.0 23.4 87.3 110.6 64.8
SD 4.3 4.7 1.7 3.7 5.9 6.3 15.0 19.3 13.6
2 (n = 42)
Mean 16.5 20.2 7.2 14.6 14.3 18.4 72.6 90.9 49.2
SD 5.4 6.3 1.5 3.7 6.3 6.4 17.0 21.8 14.0
3 (n 12)
Mean 15.0 22.7 6.7 15.0 14.5 21.2 73.9 95.1 50.7
SD 5.7 4.3 1.9 3.4 6.5 7.7 15.8 21.2 15.4
P < .0001 < .01 NS < .0001 < .0001 < .0001 < .0001 < .0001 < .0001
Post-hoc comparisons 0 >1 > 2
(Scheff6) 0 > 1 > 2, 3 0, 1 > 2 0>2 0 > 1 > 2, 3 0>3 0 > 1 > 2, 3 0 > 1 > 2, 3 0 > 1 > 2, 3
Abbreviation: NS, not significant.

*N = 295.
Table 9. Oneway ANOVAs: FACT-B by Physician PSR*
Subscale PWB SWB RWD EWB FWB BCS FACT-G FACT-B
Total TOI-PFB
PSR
0 (n = 207)
Mean 23.0 23.1 7.1 16.6 21.7 24.8 91.6 116.3 69.5
SD 5.0 5.0 1.5 3.3 5.8 6.0 15.0 19.2 13.9
1 (n = 67)
Mean 20.5 21.5 6.7 15.6 18.3 22.0 82.3 104.2 60.8
SD 5.6 6.3 2.1 4.1 7.5 7.6 18.9 24.3 17.3
2 n = 11)
Mean 17.9 23.4 7.3 14.6 15.8 24.5 78.7 103.3 58.2
SD 4.1 3.3 0.8 3.7 4.6 6.3 7.7 12.5 11.5
3, 4 (n = 6)
Mean 16.5 23.6 7.2 15.5 14.7 21.8 77.4 99.2 53.0
SD 5.3 2.7 1.3 2.6 6.0 6.1 12.0 16.3 14.6
P < .0001 NS NS NS < .0001 < .05 < .0001 < .0001 < .0001
Post-hoc comparisons
(Scheff6) 0>1, 2, 3 0>1,2, 3 0>1 0>1 0>1 0>1,3
*N = 295.
DISCUSSION has been described elsewhere. 21 This report details the

The FACT-B QL instrument, a 44-item inventory of formation and testing of the nine-item BCS, which com-
multidimensional health-related QL, is now in its third plements the FACT-G by addressing additional concerns
version and is available in nine languages. This is the of breast cancer patients not included in the general scale.
first report on the reliability and validity of the English The FACT-B is comprised of the FACT-G plus the BCS.
language translation. The English version was the one Review of Tables 3 and 4 reveal average FACT sub-
initially developed, using standardized test construction scale scores that are slightly higher than those reported
methods. A large pool of candidate items, generated em- in general (mixed diagnosis) cancer patients2 and lung
pirically from patients and experts, was then narrowed cancer patients." Subscale variances and alpha coeffi-
down through an item review process that combined pa- cients are comparable to those previously reported.
tient prioritization with conceptual judgment and expert Reliability and validity data were drawn from two sam-
input. The formation of the primary (FACT-G) subscales ples: one (n = 47) to determine concurrent validity with
Table 10. Oneway ANOVAs: FACT-B by Extent of Disease'

Subscale PWB SWB RWD EWB FWB BCS FACT-G FACT-B
Total TOI-PFB
Extent of disease
1. No evidence of disease (n = 125)
Mean 23.8 22.9 7.4 17.0 22.4 25.0 93.4 118.4 71.2
SD 4.9 5.0 1.1 3.4 5.4 6.2 14.4 18.7 13.5
2. Local disease (n = 66)
Mean 21.7 21.0 6.3 15.5 18.3 22.7 82.7 105.4 62.6
SD 4.9 6.3 2.1 3.8 7.6 7.3 18.8 23.9 16.7
3. Regional disease n = 40)
Mean 19.9 23.9 7.0 15.9 18.5 23.8 85.2 109.0 62.2
SD 6.5 5.1 1.7 3.9 7.2 6.5 18.5 23.7 17.6
4. Distant metastases (n = 56)
Mean 20.4 23.6 7.2 16.1 20.7 23.6 88.0 111.6 64.6
SD 4.9 4.2 1.5 3.1 5.3 5.7 12.7 16.5 13.0
P < .0001 NS < .0001 < .05 < .0001 NS < .0001 < .0005 < .0001
Post-hoc comparisons
(ScheffCl 1 > 3, 4 1, 4 > 2 1 >2 1 > 2,3 1 > 2,3 1 >2 1 > 2,3
*N = 295.
982 BRADY ET AL
the FLIC and sensitivity to change in both the FLIC and arate physical symptoms from psychosocial distress. We
PSR over time; and one (n = 295) to assess internal consis- have purposely eliminated such questions from our psy-
tency, test-retest reliability, and aspects of validity other than chosocial scales. For example, in the SWB subscale, the
that obtained from the smaller sample. Test-retest reliability questions pertain to feelings of closeness or distance,
coefficients reflected the stability of the questionnaire across rather than social activity, because such activity is laden
administrations. Internal consistency (alpha) coefficients with the physical parameters of strength or energy.
ranged from acceptable (.63 for BCS) to strong (> .80 for FACT-B validity was further supported by the data
PWB; FWB; TOI-PFB), indicating that, for the most part, relevant to sensitivity to change in the FLIC (Table 6).
the subscales as constructed perform well as homogeneous The FLIC is more multidimensional than the PSR, so it
(unidimensional) indicators. Despite its somewhat low was expected that, along with the more physical sub-
alpha, the nine-item BCS has sufficient unidimensionality scales, the EWB subscale of the FACT-B would also be
to show expected correlations with other measures, and is sensitive to change. This proved to be the case. Also as
sensitive to change in performance status over time. It is not anticipated, the SWB and the BCS subscales were not
surprising that the BCS has a relatively low alpha coefficient related to change in the FLIC, possibly reflecting the lack
because the BCS items share only their specificity to breast of coverage of these areas on the FLIC. For example,
cancer, and no attempt was made to keep the item content while the FLIC includes questions pertaining to spending
homogeneous. The goal was to include items identified by time with friends and family, it does not specifically ad-
breast cancer patients as important, but not covered in the dress feelings of emotional closeness or issues of sexual-
general scale. When one examines the BCS items, it is ity. These were identified as important issues by patients
apparent that they would likely be placed in different FACT- in our FACT development sample and are therefore in-
G subscales. It should be remembered that the BCS is meant cluded in the SWB subscale. We are currently conducting
only as a complement to the FACT-G, and is never used studies with measures that would more appropriately as-
alone. When combined with other subscales as recom- sess the validity of SWB.
mended (eg, FACT-B total score and TOI-PFB), the alpha Table 7 data on concurrent validity of the subscales in
coefficients are high. Also, as Helmstadter3 6 states, when a the larger (n = 295) sample provide further supporting
scale is heterogeneous, the internal consistency coefficient evidence for convergent and divergent validity. Squaring
is not an appropriate measure of reliability, as it is only a the correlation coefficients in Table 7 reveals the amount
measure of homogeneity. With more complex scales (such of variance shared between measures. None of the sub-
as the BCS), other measures of reliability (eg, test-retest scales or aggregated scores shared even 2% of the vari-
reliability) are more appropriate indicators of error. Thus, ance with the M-CSDS, a concept presumed to be unre-
even though it is not meant to be used in isolation, the lated to QL. On the other hand, POMS-SF subscale scores
BCS does demonstrate good reliability, in that test-retest share considerable variance with FACT-B subscale scores
reliability was high. in a logical pattern. For example, POMS-SF Tension and
The data for sensitivity to change in PSR (Table 5) are FACT-B EWB subscale shared 35% variance, whereas
very encouraging in that they show changes in scores the amount of shared variance between POMS-SF Ten-
on subscales that conceptually are linked to performance sion and other FACT-B subscales was lower. On the
status. All three subscales of the TOI-PFB (PWB; FWB; other hand, POMS-SF Fatigue was most closely related
BCS) were responsive to changes in performance status. to FACT-B PWB subscale, as would be expected.
On the contrary, and as expected, the SWB, EWB, and As expected, the relationship between FACT-B scores
RWD subscales were not responsive to the same change and PSRs (as rated by both patients and physicians) was
in PSR. This supports not only sensitivity to change, strongest in areas of PWB, FWB, and the BCS. This is
important for any QL instrument used in longitudinal exemplified somewhat crudely by the correlation coeffi-
studies or clinical practice, but also convergent and diver- cients in Table 7, and detailed in Tables 8 (FACT-B to
gent validity of the FACT-B subscales. This differential patient PSR) and 9 (FACT-B to physician PSR). As was
sensitivity to physical health parameters on the part of the case with the smaller sample reported in Table 5, this
physical and psychosocial measures is a common find- differential sensitivity of the more physical subscales of the
ing.21 25 37 38 It is desirable in that it allows for adequate FACT-B to PSR provides further supportive evidence of
discrimination between physical and psychosocial impair- construct validity. Overall, the aggregated FACT-G and
ment. A major problem encountered in using many earlier FACT-B scores are able to separate groups of patients ac-
measures of psychologic distress in the medical popula- cording to differences that are known to exist between them
tion was that these measures were too laden with physical (known-groups validity). The separation of groups is more
symptoms (eg, appetite, sleep, fatigue) to adequately sep- dramatic when FACT-B scores are compared across patient
PSR compared with provider PSR. To some extent, this scales possess unique value, they were not created to be
probably reflects the fact that the FACT-B/provider compar- administered in isolation. This is particularly relevant to
ison entails two different methods of data collection (patient the BCS, because it is by design a miscellaneous subscale
self-report/observer rating), which commonly lowers the co- containing items of importance to women with breast
efficient of association. Indeed, Table 7 reflects only 12% cancer, but which are not already included in the FACT-
shared variance between patient- and provider-rated PSR. G. The BCS therefore will work best when aggregated
Similar lack of agreement has been reported elsewhere.3 3' '""
39
with other subscales, such as in the computation of the
This speaks to the necessity of obtaining patient self-report TOI-PFB or the FACT-B total score.
regarding QL, rather than relying on provider estimation. Given the high internal consistency, precision, and sen-
The FACT-B was also able to separate groups of pa- sitivity of the of the 23-item TOI-PFB (which includes
tients differing in extent of disease. As hypothesized, the the BCS as a component), we recommend using the TOI-
effects were largest for the PWB and the FWB subscales, PFB as an efficient and precise summary measure of PWB
replicating previous findings with the FACT-G, 2 ' as well and FWB in clinical trials. We stress that we recommend
as with the EORTC QLQ-Core 30.26 The BCS scores for this as an analytic strategy for purposes of enhancing
patients with no evidence of disease were higher than measurement sensitivity, reducing data points, and sum-
those for patients with evidence of disease, although not marizing physical/functional outcomes. It is always rec-
significantly so. This may be because of the fact that BCS ommended that the entire FACT-B be administered and
items (eg, feeling like a woman, self-consciousness about evaluated because the TOI-PFB will inadequately capture
clothes) may be less related to extent of disease than to the psychosocial dimensions of QL, dimensions that con-
other variables, such as type of treatment. tribute significantly to patients' QL estimations.
Previous studies further document the validity of the In conclusion, the FACT-B is a self-report instrument
EWB and SWB subscales. For example, the EWB sub- designed to measure multidimensional QL in patients
scale was able to document change effected by a psy- with breast cancer. It is reliable, relates to similar mea-
chosocial intervention.4 The EWB subscale was sensitive sures in an expected pattern, and performs as predicted
to this change, whereas the POMS-SF total score and the in relation to change in clinical status over time. The
Impact of Event Scale 4 ' were not. With respect to the FACT-B was created with an emphasis on patients' val-
SWB subscale, it was able to predict willingness to accept ues and brevity. It is written at the sixth-grade reading
aggressive cancer treatment, 42 highlighting a role for QL level, takes approximately 10 minutes to complete, and is
data in clinical decision-making. available in nine languages. Its psychometric properties,
The FACT-B is intended to be administered in its en- brevity, and relevance to patients' values make its suitable
tirety as a 44-item questionnaire. Although different sub- for use in both research and clinical settings.
APPENDIX. The FACT-B QL Instrument (Version 3)
FACT-B (Version 3)
Below is a list of statements that other people with your illness have said are important. By circling one number per line,
please indicate how true each statement has been for you during the past 7 days.
not a little some- quite very
PHYSICAL WELL-BEING at all bit what a bit much
1. I have a lack of energy ................................................................................. 0 1 2 3 4

2. I have nausea ................................................................................................ 0 1 2 3 4
3. Because of my physical condition, I have trouble meeting the needs of
m y family ...................................................................................................... 0 1 2 3 4
4. I have pain .................................................................................................... 0 1 2 3 4
5. I am bothered by side effects of treatment .................................................. 0 1 2 3 4
6. I feel ill ......................................................................................................... 0 1 2 3 4
7. I am forced to spend tim e in bed ................................................................ 0 1 2 3 4
8. Looking at the above 7 questions, how much would you say your
PHYSICAL WELL-BEING affects your quality of life? (circle one number)
0 1 2 3 4 5 6 7 8 9 10
Not at all Very much so
984 BRADY ET AL

SOCIAL/FAMILY WELL-BEING at all bit what a bit much
9. 1 feel distant from my friends ...................................................................... 0 I 2 3 4
10. I get emotional support from my family ..................................................... 0 1 2 3 4
11. I get support from my friends and neighbors .............................................. 0 1 2 3 4
12. My family has accepted my illness ............................................................. 0 2 3 4
13. Family communication about my illness is poor ........................................ 0 2 3 4
14. I feel close to my partner (or the person who is my main support) .......... 0 I 2 3 4
15. Have you been sexually active during the past year?
No _ Yes _ If yes: I am satisfied with my sex life ...................... 0 I 2 3 4
SOCIAL/FAMILY WELL-BEING affects your quality of life? (circle one number)
0 1 2 3 4 5 6 7 8 9 10

RELATIONSHIP WITH DOCTOR at all bit what a bit much
17. 1 have confidence in my doctor(s) ............................................................... 0 2 3 4
18. My doctor is available to answer my questions .......................................... 0 I 2 3 4
RELATIONSHIP WITH THE DOCTOR affects your quality of life?....... (circle one number)
0 1 2 3 4 5 6 7 8 9 10

EMOTIONAL WELL-BEING at all bit what a bit much
20. I feel sad ....................................................................................................... 0 1 2 3 4
21. 1 am proud of how I'm coping with my illness .......................................... 0 1 2 3 4
22. I am losing hope in the fight against my illness ......................................... 0 1 2 3 4
23. I feel nervous ................................................................................................ 0 2 3 4
24. I worry about dying ...................................................................................... 0 I 2 3 4
25. I worry that my condition will get worse .................................................... 0 I 2 3 4
EMOTIONAL WELL-BEING affects your quality of life'? ....................... (circle one number)
0 1 2 3 4 5 6 7 8 9 10

FUNCTIONAL WELL-BEING at all bit what a bit much
27. I am able to work (include work in home) ................................................. 0 1 2 3 4
28. My work (include work in home) is fulfilling ............................................ 0 1 2 3 4
29. 1 am able to enjoy life .................................................................................. 0 1 2 3 4
30. I have accepted my illness ........................................................................... 0 1 2 3 4
31. I am sleeping well ........................................................................................ 0 1 2 3 4
32. I am enjoying the things I usually do for fun ............................................. 0 l 2 3 4
33. 1 am content with the quality of my life right now .................................... 0 2 3 4
FUNCTIONAL WELL-BEING affects your quality of life? ..................... (circle one number)
0 1 2 3 4 5 6 7 8 9 10
ADDITIONAL CONCERNS at all bit what a bit much
35. I have been short of breath .......................................................................... 0 1 2 3 4
36. I am self-conscious about the way I dress .................................................. 0 1 2 3 4
37. One or both of my arms are swollen or tender ........................................... 0 1 2 3 4
38. I feel sexually attractive ............................................................................... 0 1 2 3 4
39. I am bothered by hair loss ........................................................................... 0 1 2 3 4
40. I worry about the risk of cancer in other family members ........................ 0 1 2 3 4
41. I worry about the effect of stress on my illness ......................................... 0 1 2 3 4
42. I am bothered by a change in weight .......................................................... 0 1 2 3 4
43. I am able to feel like a woman .................................................................... 0 1 2 3 4
44. Looking at the above 9 questions, how much would you say these
ADDITIONAL CONCERNS affect your quality of life? ........................... (circle one number)
0 1 2 3 4 5 6 7 8 9 10
REFERENCES
1. Parkin DM, Liiri E, Muir CS: Estimates of the worldwide 17. Levine MN, Guyatt GH, Gent M, et al: Quality of life in
frequency of sixteen major cancers in 1980. Int J Cancer 41:184- stage II breast cancer: An instrument for clinical trials. J Clin Oncol
197, 1988 6:1798-1810, 1988
2. The American Cancer Society: Cancer Facts and Figures-1996. 18. Ganz PA, Schag CA, Lee JJ, et al: The CARES: A generic
New York, NY, American Cancer Society, 1996 measure of health-related quality of life for patients with cancer.
3. Garfinkel L, Boring CC, Heath CW Jr: Changing trends: An Qual Life Res 1:19-29, 1992
overview of breast cancer incidence and mortality. Cancer 74:222- 19. Schag CA, Ganz PA, Heinrich RL: Cancer Rehabilitation
227, 1994 Evaluation System-Short Form (CARES-SF): A cancer specific re-
4. Smigel, K: Breast cancer death rates decline for white women. habilitation and quality of life instrument. Cancer 68:1406-1413,
J Natl Cancer Inst 87:173, 1995 1991
5. Meyerowitz BE: Psychological correlates of breast cancer and 20. Schipper H, Clinch J, McMurray A, et al: Measuring the
its treatment. Psych Bull 87:108-131, 1980 quality of life of cancer patients: The Functional Living Index-Can-
6. Glanz K, Lerman C: Psychosocial impact of breast cancer: A cer: Development and Validation. J Clin Oncol 2:472-483, 1984
critical review. Ann Behav Med 14:204-212, 1992 21. Cella DF, Tulsky DS, Gray G, et al: The Functional Assess-
7. Levy SM, Haynes LT, Herberman RB, et al: Mastectomy ver- ment of Cancer Therapy (FACT) Scale: Development and validation
sus breast conservation surgery: Mental health effects at long-term of the general measure. J Clin Oncol 11:570-579, 1993
follow-up. Health Psych 11:349-354, 1992 22. Cella DF: F.A.C.T. Manual: Manual for the Functional As-
8. Fallowfield L: Quality of life in breast cancer-Results from sessment of Cancer Therapy (FACT) scales and the Functional As-
3 Cancer Research Campaign studies. Acta Clin Belg 48:19-23, 1993 sessment of HIV Infection (FAHI) scale. Chicago, IL, Rush-Presby-
9. Sneeuw KCA, Aaronson NK, Yarold JR, et al: Cosmetic and terian-St. Luke's Medical Center, 1994
functional outcomes of breast conserving treatment for early stage 23. Cella DF, Bonomi AE, Lloyd SR, et al: Reliability and valid-
breast cancer. 2. Relationship with psychosocial functioning. Ra- ity of the Functional Assessment of Cancer Therapy-Lung (FACT-
diother Oncol 25:160-166, 1992 L) quality of life instrument. Lung Cancer 12:199-220, 1995
10. Kemeny MM, Wellisch DK, Schain WS: Psychosocial out- 24. Bonomi AE, Cella DF, Hahn E, et al: Multilingual translation
come in a randomized surgical trial for treatment of primary breast of the Functional Assessment of Cancer Therapy Measurement Sys-
cancer. Cancer 62:1231-1237, 1988 tem. Qual Life Res 5:309-320, 1996
11. Ganz PA, Schag AC, Lee JJ, et al: Breast conservation versus 25. Aaronson NK, Ahmedzai S, Bergman B, et al: The European
mastectomy: Is there a difference in psychological adjustment or Organization for Research and Treatment of Cancer QLQ-C30: A
quality of life in the year after surgery? Cancer 69:1729-1738, 1992 quality-of-life instrument for use in international clinical trials in
12. Coates A, Gebski V, Stat M, et al: Improving the quality of oncology. J Natl Cancer Inst 85:365-376, 1993
life during chemotherapy for advanced breast cancer. N Engl J Med 26. Osoba D, Zee B, Pater J, et al: Psychometric properties and
317:1490-1495, 1987 responsiveness of the EORTC Quality of Life Questionnaire (QLQ-
13. Knobt MT: Symptoms and rehabilitation needs of patients C30) in patients with breast, ovarian and lung cancer. Qual Life Res
with early-stage breast cancer during primary therapy. Cancer 3:353-364, 1994
66:1392-1401, 1990 27. Sprangers MAG, Groenvold M, Arraras JI, et al: The Euro-
14. Johnson JR, Temple R: Food and drug administration require- pean Organization for Research and Treatment of Cancer Breast
ments for approval of new anticancer drugs. Cancer Treat Rep Cancer-Specific Quality-of-Life Questionnaire Module: First results
69:1155-1157, 1985 from a three-country field study. J Clin Oncol 14:2756-2768,
15. Nayfield SG, Ganz PA, Moinpour CM, et al: Report from a 1996
National Cancer Institute (USA) workshop on quality of life assess- 28. Zubrod CG, Scneiderman M, Frie E, et al: Appraisal of meth-
ment in cancer clinical trials. Qual Life Res 1:203-210, 1992 ods for the study of chemotherapy of cancer in man: Comparative
16. Priestman TJ, Baum M: Evaluation of quality of life in pa- therapeutic trial of nitrogen mustard and triethylene thiophosphora-
tients receiving treatment for advanced breast cancer. Lancet 24:899- mide. J Chron Dis 11:7-33, 1960
901, 1976 29. Bergman B, Aaronson NK, Ahmedzai S, et al: The EORTC
986 BRADY ET AL
QLQ-LC13: A modular supplement to the EORTC core quality of 36. Helmstadter GC: Principles of Psychological Measurement.
life questionnaire (QLQ-C30) for use in lung cancer clinical trials. New York, NY, Appleton-Century-Crofts, 1964
Eur J Cancer 30:635-642, 1994 37. Bergman B, Sullivan M, Sorenson S: Quality of life during
30. Ganz P, Haskell CM, Figlin RA, et al: Estimating the quality chemotherapy for small cell lung cancer II. A longitudinal study of
of life in a clinical trial of patients with metastatic lung cancer using the EORTC core quality of life questionnaire and comparison with
the Karnofsky performance status and the Functional Living Index- the Sickness Impact Profile. Acta Oncol 31:19-28, 1992
Cancer. Cancer 61:849-856, 1988 38. McHorney CA, Ware JE, Raczek AE: The MOS 36-item
31. Slevin ML, Plant H, Lynch D, et al: Who should measure short-form health survey (SF-36): II. Psychometric and clinical test
quality of life, the doctor or the patient? Br J Cancer 57:109-112, of validity in measuring physical and mental health constructs. Med
1988 Care 31:247-263, 1993
32. Moinpour CM, Savage M, Hayden KA, et al: Quality of life 39. Ganz PA, Schag AC, Cheng HL: Assessing the quality of
assessment in cancer clinical trials, in Dimsdale JE, Baum A (eds): life-A study in newly-diagnosed breast cancer patients. J Clin
Quality of Life in Behavioral Medicine Research. Hillsdale, NJ, Law- Epidemiol 43:75-86, 1990
rence Erlbaum, 1995, pp 79-95 40. McCain NL, Zeller JM, Cella DF, et al: The influence of
33. Loprinzi CL, Laurie JA, Wieand S: Prospective evaluation of stress management training in HIV disease. Nursing Res 45:246-
prognostic variables from patient-completed questionnaires. J Clin 253, 1996
Oncol 12:601-607, 1994 41. Horowitz M, Wilner N, Alvarez W: Impact of events
34. Cella DF, Jacobsen PB, Orav EJ, et al: A brief POMS measure scale: A measure of subjective stress. Psychosom Med 41:209-
of distress for cancer patients. J Chron Dis 40:939-942, 1987 218, 1979
35. Strahan R, Gerbasi KC: Short homogeneous versions of the 42. Yellen SB, Cella DF: Someone to live for: Social well-being,
Marlowe-Crowne Social Desirability Scale. J Clin Psychol 28:191- parenthood status, and decision-making in oncology. J Clin Oncol
193, 1972 13:1255-1264, 1995

Jco 1997 15 3 974

Uploaded by

Copyright:

Available Formats

Jco 1997 15 3 974

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jco 1997 15 3 974

Uploaded by

Copyright:

Available Formats

Reliability and Validity of the Functional Assessment of

Cancer Therapy-Breast Quality-of-Life Instrument

974 Journal of Clinical Oncology, Vol 15, No 3 (March), 1997: pp 974-986

Table 5. Sensitivity of FACT-B to 2-Month Changes in PSR

PWB (7-item) Declined PSR 8 -3.3 (1.6)

*FACT-G plus BCS.

Scale Group No. of Patients

Table 7. Construct Validity: Pearson Product-Moment Correlation Coefficients'

Table 8. Oneway ANOVAs: FACT-B by Patient PSRs*

Abbreviation: NS, not significant.

DISCUSSION has been described elsewhere. 21 This report details the

Table 10. Oneway ANOVAs: FACT-B by Extent of Disease'

APPENDIX. The FACT-B QL Instrument (Version 3)

1. I have a lack of energy ................................................................................. 0 1 2 3 4

not a little some- quite very

not a little some- quite very

not a little some- quite very

not a little some- quite very

You might also like