Support Care Cancer (2013) 21:2267–2277

DOI 10.1007/s00520-013-1789-4

ORIGINAL ARTICLE

The impact of clinical and sociodemographic features

on quality of life in patients with early stage cancers
using the Functional Assessment of Cancer
Therapy-General assessment tool
Marko Popovic & Nicholas Lao & Liang Zeng &
Liying Zhang & David Cella & Jennifer L. Beaumont &
Ronald Chow & Nicholas Chiu & Leonard Chiu &
Henry Lam & Michael Poon & Edward Chow

Received: 28 September 2012 / Accepted: 7 March 2013 / Published online: 22 March 2013
# Springer-Verlag Berlin Heidelberg 2013

Abstract age and education. Currently, the literature conflicts in its

Purpose This study aims to compare the Functional
Assessment of Cancer Therapy-General (FACT-G) quality
of life (QOL) scores of patient groups with varying clinical
and sociodemographic features in the early stage cancer
population.
Methods A literature search was conducted on both the
Embase and OvidSP platforms. Weighted analysis of vari-
ance (ANOVA) was performed for binary predictors and
weighted linear regression analysis was conducted for con-
tinuous predictors.
Results Six binary features predicted at least one domain of
QOL: primary cancer site (homogeneous versus heteroge-
neous), total per capita healthcare expenditures, mean age,
previous chemotherapy, radiotherapy, and previous hormon-
al therapy. Two continuous factors had predictive value with
respect to QOL: completion of postsecondary education and
marital status.
Conclusion Although there are limitations of the current
study, similar correlations to our own have been previously
described between QOL and healthcare expenditures, mean
M. Popovic : N. Lao : L. Zeng : L. Zhang : R. Chow : N. Chiu :
L. Chiu : H. Lam : M. Poon : E. Chow (*)
Rapid Response Radiotherapy Program, Department of Radiation
Oncology, Odette Cancer Centre, Sunnybrook Health Sciences
Centre, University of Toronto, Toronto, ON, Canada
e-mail: Edward.Chow@sunnybrook.ca
D. Cella : J. L. Beaumont
Department of Medical Social Sciences, Northwestern University
Feinberg School of Medicine, Chicago, IL, USA
analysis of previous radiotherapy and chemotherapy as pre-
dictors of QOL. No published evidence exists describing the
presently found relationships in primary cancer site, marital
status and hormonal therapy. Future work may focus on
determining cause and effect relationships between these
predictors and QOL.
Keywords Early stage cancer . Quality of life . Functional
Assessment of Cancer Therapy-General . Treatment
Introduction
In patients with early stages of cancer, overall prognosis is
favorable. Treatments for these patients tend to have a curative
intent with a focus on improving survival. Although enhanc-
ing survival remains the primary endpoint in this patient
population, improvement of quality of life (QOL) also merits
attention [1–3]. In addition to serving as a secondary treatment
endpoint, QOL must be taken into account in order to evaluate
efficacy of treatments, such as radiotherapy, chemotherapy,
surgery, and hormonal therapy, on the well-being of patients
[4].
QOL is a self- or proxy-administered subjective construct
that evaluates health and overall well-being of a patient. To
quantitatively measure the QOL of cancer patients, two
validated questionnaires are most commonly used: the
European Organization for Research and Treatment of
Cancer questionnaire C30 (EORTC QLQ-30) and the
Functional Assessment of Cancer Therapy-General (FACT-
2268
G). The FACT-G has been shown to be a reliable and robust
tool for evaluating QOL in cancer patients [5–8]. An
earlier version of the FACT-G was a 28-item question-
naire with five subscales: physical well-being (PWB),
functional well-being (FWB), social/family well-being
(SWB), emotional well-being (EWB), and relationship
with doctor (RWD). The RWD subscale was removed
from the general version to be expanded into a larger
treatment satisfaction subscale, reducing the FACT-G to
27 items (27 rather than 26 because one question was
added to the EWB subscale).
The literature notes that sociodemographic and clinical
characteristics can affect QOL in cancer patients [9–13].
Pretreatment characteristics have a tendency to affect per-
ceptions of well-being after diagnosis by influencing patient
reactions to their condition and to treatment [9]. In the
research setting, pretreatment variables may confound out-
comes and must, thus, be accounted for in QOL trials [13].
In the clinical setting, these factors may play a role in
enticing both patients and healthcare professionals towards
particular treatment modalities and away from others, and
thus must be appropriately considered in devising and eval-
uating an appropriate course of treatment.
In the early stage cancer population, no analysis of clin-
ical and sociodemographic factors and QOL exists that
considers data from a wide variety of clinical trials. If
available, such an analysis could validate current relation-
ships hypothesized by individual clinical trials and could
also potentially introduce novel associations. In light of this
information, the purpose of this review is to identify clinical
and sociodemographic predictors of QOL in the early stage
cancer population by using the FACT-G.
Methods
We performed a literature review on both the Embase plat-
form (1994 to 2012) as well as on OvidSP in MEDLINE
(1994 to 2012). The search term “FACT-G” was combined
with the following terms: “quality of life,” “palliative,”
“cancer,” “curative” or “curative treatment,” and “FACT-
G.mp.” Reference lists of articles included in the review
were cross-referenced for additional pertinent articles. We
excluded all non-English studies from review as well as all
repeat data and abstracts. The literature search was indepen-
dently sorted through by three authors.
We only included studies that assessed QOL in
nonadvanced cancer patients at baseline (i.e., before treat-
ment for the latest diagnosis) using the FACT-G assessment
tool or any tumor-specific FACT questionnaire that reported
subscale scores of the FACT-G. Included articles must have
disclosed at least two of the following: PWB, EWB, FWB,
SWB, RWD, or total FACT-G score without RWD. Only
Support Care Cancer (2013) 21:2267–2277
study arms that had greater than 50 % of their cohort with
histologically confirmed nonadvanced carcinoma (either
stage I or II cancer using the Roman Numeral Staging
system) were included in the review.
We extracted the following information from each in-
cluded study: authorship; year of publication; country of
origin; primary cancer site; admittance status; mean sur-
vival time after treatment; gender; Karnofsky Performance
Status (KPS) or Eastern Cooperative Oncology Group
Performance Status (ECOG PS); mean/median age; prev-
alence of previous chemotherapy, radiotherapy, surgery or
any other treatment modalities; marital status; education;
disease progression; and all relevant FACT-G total and
subscale scores. FACT-G scores were stratified by
sociodemographic/clinical parameters.
Statistical analysis
To compare FACT-G scores in early stage cancer patients
with different clinical and sociodemographic characteristics,
weighted analysis of variance (ANOVA) was performed for
binary predictors and weighted linear regression analysis
was conducted for continuous predictors. General linear
model procedure (PROC GLM) was performed for the
unbalanced data. The total number of patients from each
study arm was considered a weighting variable. The weight-
ed arithmetic means and the weighted standard deviations
(SD) of FACT-G total and subscale scores were also calcu-
lated in this patient population. P
wi x i
The weighted mean was defined as xw ¼ Pi w , and the
i
P i
weighted variance was defined as s2w ¼ d1 i wi ðxi  xw Þ2 ,
where wi is the weight for the ith study arm, xi is the ith
variable value, and the variance divisor d is n −1. The
weighted variance is a measure of variability and is com-
puted as the sum of the squared weighted distance from the
data value to the mean divided by the variance divisor.
To normalize the distribution of FACT-G scores, natural
log transformation was applied to each score coming from a
different study arm. A p value of less than 0.05 constituted
statistical significance. All analyses were conducted by
Statistical Analysis Software (SAS version 9.2 for Windows).
Cutoffs for total per capita healthcare expenditures
Study arms were grouped by total per capita healthcare expen-
ditures of their originating countries. One thousand US dollars
(USD) was established as an arbitrary cutoff to ensure compa-
rable sample sizes between the two groups. China was the only
country with per capita expenditures below 1,000 USD, while
the United States, Sweden, Japan, Australia, and South Korea
all had above 1,000 USD in per capita expenditures.
Support Care Cancer (2013) 21:2267–2277
Results
The search produced 953 different articles of interest. When
all results were sorted for potential inclusion, a total of 17
applicable articles were found that collectively included 23
study arms (Table 1) [4, 9, 14–28]. FACT-G total and subscale
scores of each included study can be found in Table 1.
Study objectives
Three different study objectives were noted (Table 2). Nine
studies (52.9 %) aimed to identify determinants of QOL [9,
16, 18, 20, 21, 24–26, 28]. Seven studies (41.2 %) sought to
assess QOL in their respective cohorts [4, 14, 15, 17, 19, 22,
27]. One study (5.9 %) evaluated the effectiveness of a
particular treatment regimen using QOL as an endpoint [23].
Overall sociodemographic/clinical characteristics
An overview of overall sociodemographic and clinical fea-
tures included in the review can be found in Table 2. Briefly,
20 of the 23 included study arms (87.0 %) analyzed patients
of only one primary cancer site, of which eight focused on
solely breast cancer patients (40.0 %) and another five
considered patients with gynecologic cancers (25.0 %). A
great number of study arms came from the United States
(15/23; 65.2 %). Many cohorts (18/23; 78.2 %) were <50 %
male. Eight cohorts (36.4 %) had a mean age range of 50–
59, while a further eight had a mean range of 60–69. The
majority of enrolled patients were married and had not
completed postsecondary education programs. Similarly, a
majority of patients had undergone chemotherapy prior to
being administered the FACT-G at baseline.
Sociodemographic/clinical features in relation to FACT-G
scores
Binary predictors
Six binary predictors produced statistically significant re-
sults after weighted ANOVA was performed (Table 3). One
further binary factor, frequency of previous surgery, was
analyzed for its correlation to QOL scores, although it was
statistically insignificant for all FACT-G total and subscale
scores.
Study arms were stratified based on whether they analyzed
cohorts that included patients with a variety of primary cancer
sites (i.e., heterogeneous cohorts) or whether analysis was
limited to one primary cancer site (homogeneous cohorts).
When compared to heterogeneous samples, cohorts with ho-
mogeneous populations had better relationships with their
doctors (weighted mean: 6.75 versus 4.80; p=0.0088) as well
2269
as higher global QOL (mean FACT-G score with RWD: 93.33
versus 73.40; p=0.0020).
When arms were grouped based on total per capita national
healthcare expenditures, cohorts from countries with >1,000
USD in healthcare expenditures had significantly better FWB
(mean: 18.45 versus 13.70; p<0.0001), EWB (mean: 17.47
versus 13.70; p<0.0001), SWB (mean: 21.61 versus 19.30;
p=0.0002), global QOL without RWD (mean FACT-G score:
78.79 versus 68.50; p=0.0003), RWD (mean: 6.75 versus 4.
80; p=0.0088), and global QOL with RWD (mean: 93.33
versus 73.40; p=0.0020) when compared to China (<1,000
USD in expenditures). The total per capita health expenditures
predictor was, therefore, statistically significant for all FACT-
G total and subscale scores except PWB.
Study arms were grouped based on average cohort age,
with cohorts with an average age of <60 years sorted into
one group and cohorts with an average age ≥60 years sorted
into a second group. Statistical significance was reached in
the domains of PWB (mean: 24.76 versus 20.96; p=0.0022),
FWB (mean: 21.60 versus 16.15; p=0.0074), global QOL
without RWD (mean: 85.45 versus 73.77; p = 0.0072),
RWD (mean: 6.75 versus 4.80; p=0.0088), and global
QOL with RWD (mean: 93.33 versus 73.40; p=0.0020).
Cohorts with older patients, therefore, had significantly
better PWB, FWB, global QOL without RWD, RWD, and
global QOL with RWD when compared to cohorts with
younger patients.
The reported frequency of previous chemotherapy was
analyzed for its value as a predictor of QOL. Cohorts were
grouped based on whether they had <50 % or ≥50 % of
patients undergoing chemotherapy before baseline assess-
ment. Cohorts with <50 % of patients going on chemotherapy
had significantly better PWB (mean: 23.74 versus 18.93; p=0.
0003), FWB (mean: 21.21 versus 15.25; p=0.0002), EWB
(mean: 18.04 versus 15.98; p=0.0183), and SWB (mean: 22.
45 versus 20.44; p=0.0202). Additionally, global QOL ex-
cluding RWD was superior in cohorts with <50 % of patients
going on chemotherapy (mean: 85.37 versus 70.62; p<0.
0001). The sample sizes for the comparisons of the RWD
subscale and the FACT-G with RWD scale were too small to
conduct a meaningful binary analysis.
When cohorts were arranged into two groups based on
previous radiotherapy (group one: <50 % of patients under-
taking previous radiotherapy versus group two: ≥50 % of
patients going on radiotherapy regimens), it was found that
cohorts with fewer patients undergoing radiotherapy before
baseline had significantly better PWB (mean: 22.94 versus
18.71; p=0.0048), FWB (mean: 19.93 versus 15.31; p=0.
0082), and global QOL excluding RWD (mean: 82.22 ver-
sus 70.51; p=0.0063).
In addition to previous chemotherapy and radiotherapy,
the frequency of previous hormonal therapy was analyzed as
a binary predictor (group one: <50 % of patients undergoing
 2270

Table 1 Summary of identified study arms reporting FACT-G scores in early stage cancer patients

Author (year) Primary cancer Gender Mean Previous Previous Previous PWB FWB EWB SWB Total FACT-G RWD Total FACT-G
(% male) age chemotherapy radiotherapy surgery (%) score (without score (with
(years) (%) (%) RWD) RWD)

McQuellon et al. (2006) [4] Gynecologic cancer 0.0 % 46 100.0 % 92.0 % n/a 17.70 13.50 14.20 20.80 66.00 6.50 93.6
McQuellon et al., second Gynecologic cancer 0.0 % 48 100.0 % 91.0 % n/a 17.40 14.20 14.70 20.50 66.80 4.80 73.4
study arm (2006) [4]
Gil et al. (2007) [9] Gynecologic cancer 0.0 % 58.8 n/a n/a 100.0 % 23.30 20.20 17.50 22.30 83.30 n/a n/a
Esper et al. (1997) [14] Prostate cancer 100.0 % 66 n/a n/a n/a 26.20 21.60 15.50 23.50 86.80 n/a n/a
Yu et al. (2000) [15] Various 60.0 % 55.8 n/a n/a n/a 21.80 13.70 13.70 19.30 68.50 n/a n/a
Rao et al. (2008) [16] Various 32.6 % 56.2 n/a n/a n/a 21.70 19.60 18.20 23.00 82.50 n/a n/a
Rao et al., second study arm Various 36.1 % 53.3 n/a n/a n/a 20.20 18.10 18.80 20.90 78.00 n/a n/a
(2008) [16]
Crippa et al. (2008) [17] Pancreatic cancer 50.0 % 65 n/a n/a 0.0 % n/a n/a n/a n/a 76.30 n/a n/a
Paull et al. (2006) [18] Lung cancer 95.0 % 63 27.0 % 8.0 % 100.0 % 23.80 19.00 18.40 20.80 82.00 7.30 90.8
Monga et al. (2005) [19] Prostate cancer 100.0 % 67.8 n/a 100.0 % 0.0 % 23.90 20.90 18.00 20.70 83.50 n/a n/a
von Gruenigen et al. Gynecologic cancer 0.0 % 60.4 n/a 44.7 % 100.0 % 24.37 21.89 17.26 22.69 86.21 n/a n/a
(2005) [20]
von Gruenigen et al., second Gynecologic cancer 0.0 % 57.6 n/a n/a 100.0 % 23.05 20.42 17.54 22.29 83.50 n/a n/a
study arm (2005) [20]
Koinberg et al. (2006) [21] Breast cancer 0.0 % 60 10.0 % 28.0 % 2.0 % 21.60 18.90 18.60 23.70 82.80 n/a n/a
Koinberg et al., second study Breast cancer 0.0 % 62 13.0 % 30.4 % 6.5 % 23.10 19.90 18.80 24.40 85.70 n/a n/a
arm (2006) [21]
Granda-Cameron et al. Various 36.4 % 44.5 100.0 % n/a 90.9 % 17.64 14.65 15.80 20.40 68.49 n/a n/a
(2011) [22]
Crew et al. (2007) [23] Breast cancer 0.0 % n/a 30.0 % n/a 15.0 % 19.90 20.50 19.60 20.10 80.10 n/a n/a
Taira et al. (2011) [24] Breast cancer 0.0 % 53.3 57.1 % 42.9 % 100.0 % 21.00 17.50 16.50 20.80 76.00 n/a n/a
Head et al. (2011) [25] Head and neck cancer 87.3 % 60 72.0 % 99.0 % 37.0 % 19.00 16.10 17.90 21.80 74.80 n/a n/a
Milne et al. (2008) [26] Breast cancer 0.0 % 55.2 69.0 % 51.7 % 41.4 % 17.50 16.40 16.50 20.10 70.60 n/a n/a
Milne et al., second study Breast cancer 0.0 % 55.1 72.4 % 69.0 % 62.1 % 18.60 13.60 17.70 19.80 69.60 n/a n/a
arm (2008) [26]
Ramsey et al. (2000) [27] Colorectal cancer 25.7 % 70.4 5.8 % 1.2 % 2.3 % 25.00 22.80 17.40 22.20 87.40 6.70 94.1
Yoo et al. (2005) [28] Breast cancer 0.0 % 42.9 100.0 % n/a 100.0 % 19.17 14.72 18.40 17.03 69.32 n/a n/a
Yoo et al., second study arm Breast cancer 0.0 % 44 100.0 % n/a 100.0 % 19.38 12.16 16.94 17.30 65.80 n/a n/a
(2005) [28]

n/a not applicable—information not given in the study arm, PWB physical well-being, FWB functional well-being, EWB emotional well-being, SWB social/family well-being, RWD relationship with
doctor, FACT-G Functional Assessment of Cancer Therapy-General
Support Care Cancer (2013) 21:2267–2277
Support Care Cancer (2013) 21:2267–2277 2271

Table 2 Sociodemographic and clinical features of included patients Table 2 (continued)

at baseline
Feature No. (%)
Feature No. (%)
Previous radiotherapy (n=12)
Number of studies 17 ≤50 % 6 (50.0 %)
Number of study arms 23 >50 % 6 (50.0 %)
Purpose of study (n=17) Previous surgery (n=17)
Identify determinants of quality of life 9 (52.9 %) ≤50 % 8 (28.6 %)
Assess quality of life 7 (41.2 %) >50 % 9 (14.3 %)
Evaluate effectiveness of a treatment 1 (5.9 %)
Primary cancer (n=23)
Heterogeneous 3 (13.0 %) hormonal therapy versus group two: ≥50 % of patients
Homogeneous 20 (87.0 %) going on hormonal therapy). The two subscales of EWB
Primary cancer site (n=20) (mean: 18.70 versus 16.64; p=0.0182) and SWB (mean: 24.
Breast 8 (40.0 %) 04 versus 20.61; p=0.0060) reached statistical significance;
Gynecologic 5 (25.0 %) cohorts with a smaller percentage of patients undergoing
Prostate 2 (10.0 %) hormonal therapy had better EWB and SWB.
Colorectal 1 (5.0 %)
Head and neck 1 (5.0 %)
Continuous predictors
Lung 1 (5.0 %)
Pancreatic 1 (5.0 %)
Sarcoma 1 (5.0 %)
Four continuous predictors were examined using weighted
Country of origin (n=23)
linear regression analysis (Table 4). Two features, percent-
United States 15 (65.2 %) age of patients being married and percentage of patients
Sweden 2 (8.7 %) completing college/university, were statistically significant
Australia 2 (8.7 %) at predicting linear trends in at least one of the FACT-G
South Korea 2 (8.7 %) subscales. Two other characteristics, percentage of patients
Japan 1 (4.3 %) being male and percentage of patients completing high
China 1 (4.3 %) school, were not significant in the analysis.
Gender (n=23) When weighted linear regression was conducted for the
0–25 % male 13 (56.5 %) completion percentage of college/university, statistical sig-
26–50 % male 5 (21.7 %) nificance was reached for the domains of FWB (weighted
51–75 % male 1 (4.3 %) coefficient: 0.8877; p=0.0029), SWB (coefficient: 0.3345;
76–100 % male 4 (17.4 %) p=0.0078), and total FACT-G score excluding the RWD
Mean age (years) (n=22) subscale (coefficient: 0.4417; p = 0.0032). The positive
40–49 5 (22.7 %) weighted coefficients indicated that a positive correlation
50–59 8 (36.4 %) between QOL domains (FWB, SWB, and global QOL) and
60–69 8 (36.4 %) percentage of patients completing college or university ex-
70–79 1 (4.5 %) ists. For every extra 10 % of patients completing
Marital status (n=15) college/university, FWB was, on average, 8.9 points higher,
<70 % married 6 (40.0 %) SWB was 3.3 points higher, and the total FACT-G score
≥70 % married 9 (60.0 %) excluding RWD was 4.4 points higher.
High school education (n=8) When percentage of patients being married was analyzed
≤50 % completed high school 6 (75.0 %) as a continuous predictor, only the EWB subscale was
>50 % completed high school 2 (25.0 %) statistically significant, suggesting a negative correlation
College/university education (n=11) between marital status and EWB (coefficient: −0.5858; p=
≤50 % completed college/university 8 (72.7 %) 0.0042). On average, the EWB score decreases 5.9 points
>50 % completed college/university 3 (27.3 %) when a cohort has 10 % more patients being married.
Admittance status (n=3)
Purely outpatients 2 (66.7 %)
Both inpatients and outpatients 1 (33.3 %) Discussion
Previous chemotherapy (n=14)
≤50 % 5 (11.5 %) In the early stage cancer population, QOL is a secondary
>50 % 9 (11.5 %) treatment endpoint and is also used to assess the effects of
2272 Support Care Cancer (2013) 21:2267–2277

Table 3 Weighted analysis of variance for binary sociodemographic and clinical parameters

Binary predictor Categories Sum of weight Weighted mean (SD) p valuea

Physical well-being
Primary cancer Heterogeneous 2,006 21.38 (21.62) 0.9976
Homogeneous 1,379 21.57 (25.43)
Country >$1,000 US 2,277 21.29 (24.90) 0.4950
<$1,000 US 1,108 21.80 (NA)
Average age <60 years 2,914 20.96 (21.54) 0.0022
≥60 years 450 24.76 (9.18)
Previous chemotherapy <50 % 325 23.74 (14.40) 0.0003
≥50 % 660 18.93 (13.40)
Previous radiotherapy <50 % 538 22.94 (18.22) 0.0048
≥50 % 455 18.71 (20.39)
Previous surgery <50 % 450 22.61 (22.28) 0.5652
≥50 % 569 21.82 (15.05)
Previous hormonal therapy <50 % 96 22.32 (7.34) 0.2976
≥50 % 254 20.33 (14.27)
Functional well-being
Primary cancer Heterogeneous 2,006 15.97 (81.26) 0.1093
Homogeneous 1,379 18.25 (28.45)
Country >$1,000 US 2,277 18.45 (27.58) <0.0001
<$1,000 US 1,108 13.70 (NA)
Average age <60 years 2,914 16.15 (37.34) 0.0074
≥60 years 450 21.60 (11.51)
Previous chemotherapy <50 % 325 21.21 (15.57) 0.0002
≥50 % 660 15.25 (15.86)
Previous radiotherapy <50 % 538 19.93 (23.65) 0.0082
≥50 % 455 15.31 (23.24)
Previous surgery <50 % 450 20.23 (21.94) 0.1727
≥50 % 569 18.16 (21.44)
Previous hormonal therapy <50 % 96 19.38 (4.89) 0.2235
≥50 % 254 16.93 (14.03)
Emotional well-being
Primary cancer Heterogeneous 2,006 15.86 (76.41) 0.2887
Homogeneous 1,379 16.78 (12.30)
Country >$1,000 US 2,277 17.47 (14.98) <0.0001
<$1,000 US 1,108 13.70 (NA)
Average age <60 years 2,914 16.05 (31.33) 0.3564
≥60 years 450 17.29 (9.91)
Previous chemotherapy <50 % 325 18.04 (6.48) 0.0183
≥50 % 660 15.98 (12.50)
Previous radiotherapy <50 % 538 17.36 (8.46) 0.0739
≥50 % 455 15.81 (15.70)
Previous surgery <50 % 450 17.88 (6.09) 0.0794
≥50 % 569 17.20 (5.50)
Previous hormonal therapy <50 % 96 18.70 (0.98) 0.0182
≥50 % 254 16.64 (4.30)
Social/family well-being
Primary cancer Heterogeneous 2,006 20.43 (45.50) 0.1326
Homogeneous 1,379 21.46 (12.81)
Country >$1,000 US 2,277 21.61 (14.15) 0.0002
<$1,000 US 1,108 19.30 (NA)
Support Care Cancer (2013) 21:2267–2277 2273

Table 3 (continued)

Binary predictor Categories Sum of weight Weighted mean (SD) p valuea

Average age <60 years 2,914 20.62 (20.88) 0.0767

≥60 years 450 22.42 (10.81)
Previous chemotherapy <50 % 325 22.45 (10.96) 0.0202
≥50 % 660 20.44 (10.21)
Previous radiotherapy <50 % 538 21.96 (12.40) 0.0695
≥50 % 455 20.75 (5.01)
Previous surgery <50 % 450 22.09 (10.81) 0.1896
≥50 % 569 21.01 (13.14)
Previous hormonal therapy <50 % 96 24.04 (3.43) 0.0060
≥50 % 254 20.61 (4.10)
FACT-G total score (excluding relationship with doctor subscale)
Primary cancer Heterogeneous 2,006 73.64 (186.95) 0.1745
Homogeneous 1,407 78.02 (67.18)
Country >$1,000 US 2,305 78.79 (66.31) 0.0003
<$1,000 US 1,108 68.50 (NA)
Average age <60 years 2,914 73.77 (89.13) 0.0072
≥60 years 478 85.45 (25.40)
Previous chemotherapy <50 % 325 85.37 (22.60) <0.0001
≥50 % 660 70.62 (39.53)
Previous radiotherapy <50 % 538 82.22 (51.68) 0.0063
≥50 % 455 70.51 (57.11)
Previous surgery <50 % 478 82.39 (44.96) 0.1885
≥50 % 569 78.27 (49.14)
Previous hormonal therapy <50 % 96 84.19 (14.19) 0.0548
≥50 % 254 74.65 (28.04)
Relationship with doctorb
Primary cancer Heterogeneous 1,108 4.80 (NA) 0.0088
Homogeneous 329 6.75 (3.55)
Country >$1,000 US 329 6.75 (3.55) 0.0088
<$1,000 US 1,108 4.80 (NA)
Average age <60 years 1,108 4.80 (NA) 0.0088
≥60 years 329 6.75 (3.55)
Previous radiotherapy <50 % 171 6.70 (NA) NA
≥50 % 62 7.30 (NA)
FACT-G total score (including relationship with doctor subscale)b
Primary cancer Heterogeneous 1,108 73.40 (NA) 0.0020
Homogeneous 329 93.33 (15.89)
Country >$1,000 US 329 93.33 (15.89) 0.0020
<$1,000 US 1,108 73.40 (NA)
Average age <60 years 1,108 73.40 (NA) 0.0020
≥60 years 329 93.33 (15.89)
Previous radiotherapy <50 % 171 94.10 (NA) NA
≥50 % 62 90.80 (NA)

SD standard deviation, NA not available for calculation

a
P value was obtained by weighted analysis of variance; natural log transformation was applied for FACT-G subscale and total scores
b
There are only four study arms with available information for the relationship with doctor subscale
Bolded p-values significant (p<0.05)

treatments on the well-being of patients [4]. It has been features may affect QOL in cancer patients [9–13]. The
described that certain sociodemographic and clinical purpose of the present study was to identify the
2274 Support Care Cancer (2013) 21:2267–2277

Table 4 Weighted linear re-

gression analysis for continuous Predictor Coefficienta SE p valueb
sociodemographic and clinical
predictors Physical well-being
Gender (% male) 0.1216 0.0650 0.0762
Marital status (% married) 0.1896 0.0944 0.0658
% of cohort completing high school −0.1701 0.1515 0.3044
% of cohort completing college/university 0.1769 0.1079 0.1322
Functional well-being
Gender (% male) −0.1284 0.1377 0.3622
Marital status (% married) −0.4491 0.2854 0.1395
% of cohort completing high school 0.6631 0.4991 0.2322
% of cohort completing college/university 0.8877 0.2270 0.0029
Emotional well-being
Gender (% male) −0.1604 0.0958 0.1098
Marital status (% married) −0.5858 0.1692 0.0042
% of cohort completing high school 0.6813 0.3558 0.1040
% of cohort completing college/university 0.4758 0.2345 0.0699
Social/family well-being
Gender (% male) −0.0658 0.0548 0.2443
Marital status (% Married) −0.1861 0.1197 0.1441
% of cohort completing high school 0.2421 0.2110 0.2948
SE standard error % of cohort completing college/university 0.3345 0.1009 0.0078
a
Positive weighted coefficient Total FACT-G score (excluding relationship with doctor subscale)
indicates positive relationship
between predictor and outcome Gender (% male) −0.0446 0.0703 0.5331
b
P value was obtained by Marital status (% married) −0.2200 0.1425 0.1468
weighted linear regression anal- % of cohort completing high school 0.3117 0.2543 0.2662
ysis; natural log transformation % of cohort completing college/university 0.4417 0.1147 0.0032
was applied for FACT-G sub-
Relationship with doctorc
scales and total scores
c Gender (% male) 0.0471 0.5741 0.9420
There are only four study arms
with available information for Marital status (% married) 1.1350 2.1521 0.6910
the relationship with doctor Total FACT-G score (including relationship with doctor subscale)c
subscale Gender (% male) −0.0044 0.4024 0.9922
Bolded p-values significant Marital status (% married) 1.0145 1.2691 0.5707
(p<0.05)

determinants of QOL in early stage cancer patients by using
the FACT-G assessment tool.
When per capita national healthcare expenditure was ana-
lyzed for its predictive value, all FACT-G total and subscale
scores, except PWB, were statistically significant, suggesting
that this predictor is an important determinant of QOL. A 2002
work by the World Health Organization hypothesized that
there is a higher standard of care, more extensive support
network of healthcare professionals, and timely access to
treatment found more readily in countries with higher per
capita health expenditures [http://www.who.int/healthinfo/
paper51.pdf]. In turn, this has helped patients in developed
areas have access to more sophisticated and targeted treatment
regimens [http://www.who.int/healthinfo/paper51.pdf],
which, in turn, may have improved QOL.
It was found that when age was analyzed as a binary
predictor, cohorts with older patients had significantly
higher FACT-G domains of PWB, FWB, global QOL with-
out RWD, RWD, and global QOL with RWD. The positive
relationship between age and QOL has previously been
documented in both general and carcinoma-specific settings
[20, 29–31]. An explanation by Mor et al. as to why greater
QOL disruption exists in younger patients is as follows: “it
is likely that the different expectations of disease and dis-
ability among older and younger persons leads to a more
profound sense of relative deprivation among younger per-
sons—that the disease has forfeited their future, causing
relatively more emotional distress among them than among
older persons with cancer” [31]. Interestingly, it was found
that EWB was not significantly different between the two
groups, indicating that age may impact QOL through phys-
ical and functional dimensions rather than through emo-
tions. To date, the authors are unaware of any correlation
proposed by the literature between PWB/FWB and age.
Support Care Cancer (2013) 21:2267–2277
However, we hypothesize that perceived PWB and FWB
may be hindered by cancer to a greater extent among youn-
ger patients when compared to older individuals. Before
being diagnosed with cancer, it is reasonable to assume
that, on average, younger patients enjoy a greater degree
of perceived daily functioning and physical health. After
diagnosis, functioning and physical health may be consid-
erably reduced in younger patients, while it may change
less in older patients. In turn, the greater reduction of
functioning and health in younger patients may reflect
lower scores on PWB and FWB scales when compared
to older individuals.
Analysis of frequency of previous radiotherapy revealed
its value as a determinant of QOL in this patient population.
Cohorts with fewer patients undergoing radiotherapy had
significantly better PWB, FWB, and global QOL excluding
RWD. Conflicts currently exist in the literature with respect
to previous radiotherapy as a determinant of QOL; certain
studies contradict our findings [32, 33], while another
supports them [34]. It should be noted, however, that previ-
ous analyses [32–34] have only focused on site-specific
samples.
Previous chemotherapy showed definite promise as a
determinant of QOL in patients with nonadvanced cancers.
Cohorts with smaller percentages of patients undergoing
previous chemotherapy had better PWB, FWB, EWB,
SWB, and global QOL without RWD. Previous studies have
concluded similar findings in the early stage cancer popula-
tion [18, 24, 35]; however, a work by Glimelius et al. found
that chemotherapy improved QOL in advanced pancreatic
and biliary cancer patients [36].
Cohorts with a smaller percentage of patients receiving
previous hormonal therapy had greater QOL disruption in
EWB and SWB. To our knowledge, no published evidence
exists supporting or contradicting these results or describing
the reasoning behind these statistically significant correlations.
The incidence of completion of postsecondary programs
was analyzed as a continuous predictor of QOL and posi-
tively correlated with the domains of FWB, SWB, and total
FACT-G score excluding RWD. Ross et al. note that since
education gives access to nonalienated work and eventually
wealth, education increases the sense of personal control
[37], which may influence FWB. In addition, education
gives access to stable social relationships that increase social
support [37], leading to statistically significant increases in
SWB.
Our study is not without limitations. First, analysis of
QOL predictors may have been skewed because although
most of the included patients had received previous treat-
ment, others had not; in turn, this may have had an impact
on QOL scores. The review is dominated by cohorts from
the United States (65.2 %). As the RWD is now a defunct
subscale, current analyses of the RWD domain or total
2275
FACT-G score with RWD have limited sample sizes. In
addition, the large majority of included cohorts contain
patients of all primary tumor stages; thus, some advanced
cancer patients included in our analysis weaken our find-
ings. FACT-G scores from older versions of the question-
naire were not converted due to a paucity of data needed
to conduct such an analysis; thus, scores from the 28-item
FACT-G (with RWD) and scores from the 27-item FACT-
G (without RWD and with an added EWB item) are both
used in our analysis. Further limitations are due to the fact
that it is impossible to be certain that relationships found
at the cohort level will hold true at the individual level;
further it is not possible to fully control for confounding
variables.
This review has served to validate previous direct
relationships between QOL and older individuals, pa-
tients who previously underwent chemotherapy, who
had completed postsecondary programs, and who were
not married [9, 13, 18, 20, 24, 29–31, 35]. With the
present review, QOL profiles of 3,413 patients
encompassing 23 study arms were used to confirm
previous associations that were largely derived through
individual clinical cohorts previously. In addition, the
review has produced novel correlations between certain
sociodemographic and clinical variables and QOL in
early stage cancer patients. In the cancer setting, there
has currently been no published research linking total
per capita healthcare expenditures with QOL, although
it has previously been determined that a positive corre-
lation exists between decreased family financial burdens
due to cancer and QOL of the patient [38]. In terms of
treatment, previous hormonal therapy correlated posi-
tively with QOL; this association has not been previ-
ously described and merits further investigation.
Future work may focus on confirming the novel relation-
ship between per capita healthcare expenditures and QOL
through testing with a greater variety of originating coun-
tries. Treatment-related work may aim to identify whether
previous hormonal therapy positively influences QOL at
the individual patient level and whether the currently
contradicting parameter of previous radiotherapy holds any
predictive value to QOL. Further, the scope of the current
project may be expanded to include sociodemographic and
clinical parameters that were not discussed in this study; for
example, the present review did not consider aspects like
religion and spirituality that may have an influence on QOL.
Finally, researchers may find worthwhile work in determining
the cause and effect implications of significant results outlined
in this work. In closing, the authors hope that current and
future research on clinical and sociodemographic variables
and their influence on QOL will culminate in a more focused
and relevant evaluation of QOL in patients with early stage
cancers.
2276
Acknowledgments We thank the generous support of Bratty Family
Fund, Michael and Karyn Goldstein Cancer Research Fund, Joseph
and Silvana Melara Cancer Research Fund, and Ofelia Cancer Re-
search Fund.
Conflicts of interest None.
References
1. Burhansstipanov L, Dignan M, Jones KL, Krebs LU, Marchionda
P et al (2012) A comparison of quality of life between native and
non-native cancer survivors: native and non-native cancer survi-
vors’ QOL. J Canc Educ 27(Suppl 1):S106–S113
2. Tobiasz-Adamczyk B (2012) Health-related quality of life in wom-
en after cancer treatment. Przegl Lek 69(2):67–71
3. Bottomley A, Coens C, Gotay C, Mauer ME (2008)
Prognostic factor analysis of health-related quality of life data
in cancer: a statistical methodological evaluation. Expert Rev
Pharmacoeconomics Outcome Res 8(2):179
4. McQuellon RP, Thaler HT, Cella DF, Moore DH (2006) Quality of
life (QOL) outcomes from a randomized trial of cisplatin versus
cisplatin plus paclitaxel in advanced cervical cancer: a Gynecologic
Oncology Group study. Gynecol Oncol 101(2):296–304
5. Cella DF, Tulsky DS, Gray G, Sarafian B, Linn E et al (1993) The
Functional Assessment of Cancer Therapy scale: development and
validation of the general measure. J Clin Oncol 11(3):570–579
6. Fumimoto H, Kobayashi K, Chang CH, Eremenco S, Fujiki Y et al
(2001) Cross-cultural validation of an international questionnaire,
the General Measure of the Functional Assessment of Cancer
Therapy scale (FACT-G), for Japanese. Qual Life Res 10(8):701–
709
7. Cella DF, Hernandez L, Bonomi AE, Corona M, Vaquero M et al
(1998) Spanish language translation and initial validation of the
functional assessment of cancer therapy quality-of-life instrument.
Med Care 36(9):1407–1418
8. Conroy T, Mercier M, Bonneterre J, Luporsi E, Lefebvre JL et al
(2004) French version of FACT-G: validation and comparison with
other cancer-specific instruments. Eur J Cancer 40(15):2243–2252
9. Gil KM, Gibbons HE, Jenison EL, Hopkins MP, von Gruenigen
VE (2007) Baseline characteristics influencing quality of life in
women undergoing gynecologic oncology surgery. Health and
Quality of Life Outcomes 5(25)
10. Visser MRM, Smets EMA (1998) Fatigue, depression and quality
of life in cancer patients: how are they related? Support Care Canc
6(2):101–108
11. Ganz PA, Lee JJ, Sim M, Polinsky ML, Schag CA (1992)
Exploring the influence of multiple variables on the relationship
of age to quality of life in women with breast cancer. J Clin
Epidemiol 45:473–485
12. Awadalla AW, Ohaeri JU, Gholoum A, Khalid AOA, Hamad
HMA (2007) Factors associated with quality of life of outpatients
with breast cancer and gynecologic cancers and their family care-
givers: a controlled study. BMC Canc 7:102
13. Movsas B, Scott C, Watkins-Bruner D (2006) Pretreatment factors
significantly influence quality of life in cancer patients: a Radiation
Therapy Oncology Group (RTOG) analysis. Int J Radiat Oncol
Biol Phys 65:830–835
14. Esper P, Mo F, Chodak G, Sinner M, Cella DF et al (1997)
Measuring quality of life in men with prostate cancer using the
Functional Assessment of Cancer Therapy-prostate instrument.
Urology 50(6):920–928
15. Yu CLM, Fielding R, Chan CLW, Tse VKC, Choi PHK et al
(2000) A validation of the Chinese version of the functional
Support Care Cancer (2013) 21:2267–2277
assessment of cancer therapy-general (FACT-G) scale. Cancer
88(7):1715–1727
16. Rao D, Debb S, Blitz D, Choi SW, Cella DF (2008) Racial/ethnic
differences in the health-related quality of life of cancer patients.
Pain Symptom Manag 36(5):488–496
17. Crippa S, Dominguez I, Rodriguez JR, Razo O, Thayer SP et al
(2008) Quality of life in pancreatic cancer: analysis by stage and
treatment. J Gastrointest Surg 12(5):783–794
18. Paull DE, Thomas ML, Meade GE, Updyke GM, Arocho MA et al
(2006) Determinants of quality of life in patients following pul-
monary resection for lung cancer. Am J Surg 192(5):565–571
19. Monga U, Kerrigan AJ, Thornby J, Monga TN, Zimmermann KP
(2005) Longitudinal study of quality of life in patients with local-
ized prostate cancer undergoing radiotherapy. J Rehabil Res Dev
42(3):391
20. von Gruenigen VE, Gil KM, Frasure HE, Jenison EJ, Hopkins MP
et al (2005) The impact of obesity and age on quality of life in
gynecologic surgery. Am J Obstet Gynecol 193(4):1369–1375
21. Koinberg IL, Langius-Eklöf A, Holmberg L, Fridlund B (2006)
The usefulness of a multidisciplinary educational programme after
breast cancer surgery: a prospective and comparative study. Eur J
Oncol Nurs 10(4):273–282
22. Granda-Cameron C, Hanlon AL, Lynch MP, Houldin A (2011)
Experience of newly diagnosed patients with sarcoma receiving
chemotherapy. Oncol Nurs Forum 38(2):160–169
23. Crew KD, Capodice JL, Greenlee H, Apollo A, Jacobson JS
(2007) Pilot study of acupuncture for the treatment of joint symp-
toms related to adjuvant aromatase inhibitor therapy in postmeno-
pausal breast cancer patients. J Canc Survivorship 1(4):283–291
24. Taira N, Shimozuma K, Shiroiwa T, Ohsumi S, Kuroi K et al
(2011) Associations among baseline variables, treatment-related
factors and health-related quality of life 2 years after breast cancer
surgery. Breast Canc Res Treat 128(3):735–747
25. Head BA, Schapmire TJ, Keeney CE, Deck SM, Studts JL et al
(2012) Use of the distress thermometer to discern clinically rele-
vant quality of life differences in women with breast cancer. Qual
Life Res 21(2):215–223
26. Milne HM, Wallman KE, Gordon S, Courneya KS (2008) Effects
of a combined aerobic and resistance exercise program in breast
cancer survivors: a randomized controlled trial. Breast Canc Res
Treat 108(2):279–288
27. Ramsey SD, Andersen MR, Etzioni R, Moinpour C, Peacock S et
al (2000) Quality of life in survivors of colorectal carcinoma.
Cancer 88(6):1294–1303
28. Yoo HJ, Ahn SH, Kim SB, Kim WK, Han OS (2005) Efficacy of
progressive muscle relaxation training and guided imagery in reduc-
ing chemotherapy side effects in patients with breast cancer and in
improving their quality of life. Support Care Canc 13(10):826–833
29. Ganz PA, Lee JJ, Sim MS, Polinsky ML, Schag CAC (1992)
Exploring the influence of multiple variables on the relationship
of age to quality of life in women with breast cancer. J Clin
Epidemiol 45(5):473–485
30. Wenzel LB, Fairclough DL, Brady MJ, Cella DF, Garrett KM et al
(1999) Age-related differences in the quality of life of breast
carcinoma patients after treatment. Cancer 86(9):1768–1774
31. Mor V, Allen S, Malin M (1994) The psychosocial impact of
cancer on older versus younger patients and their families.
Cancer 74:2118–2127
32. Marijnen CAM, van de Velde CJH, Putter H, van den Brink M,
Maas CP et al (2005) Impact of short-term preoperative radiother-
apy on health-related quality of life and sexual functioning in
primary rectal cancer: report of a multicenter randomized trial. J
Clin Oncol 23(9):1847–1858
33. Härtl K, Janni W, Kästner R, Sommer H, Strobl B et al (2003) Impact
of medical and demographic factors on long-term quality of life and
body image of breast cancer patients. Ann Oncol 14(7):1064–1071
Support Care Cancer (2013) 21:2267–2277
34. Taphoorn MJB, Schiphorst AK, Snoek FJ, Lindeboom J, Wolbers
JG et al (1994) Cognitive functions and quality of life in patients
with low-grade gliomas: the impact of radiotherapy. Ann Neurol
36(1):48–54
35. Ganz PA, Rowland JH, Meyerowitz BE, Desmond KA (1998)
Impact of different adjuvant therapy strategies on quality of
life in breast cancer survivors. Recent Results Canc Res
152:396–411
2277
36. Glimelius B, Hoffman K, Sjoden PO, Jacobsson G, Sellstrom H et
al (1996) Chemotherapy improves survival and quality of life in
advanced pancreatic and biliary cancer. Ann Oncol 7:593–600
37. Ross CE, Willigen MV (1997) Education and the subjective qual-
ity of life. J Heal Soc Behav 38(3):275–297
38. Yun YH, Rhee YS, Kang IO et al (2005) Economic burdens and
quality of life of family caregivers of cancer patients. Oncology
68(2–3):107–114
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.