European Journal of Cancer 163 (2022) 128e139

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.ejcancer.com

Original Research

Real-world reference scores for EORTC QLQ-C30 and

EORTC QLQ-BR23 in early breast cancer patients

Maria M. Karsten a,*, Robert Roehle b,c, Sarah Albers a, Therese Pross a,
Anna M. Hage a, Karoline Weiler a,d, Felix Fischer e, Matthias Rose e,
Friedrich Kühn a,1, Jens-Uwe Blohmer a,1

a
Charité e Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt Universität zu Berlin,
Department of Gynecology with Breast Center, Berlin, Germany
b
Berlin Institute of Health at Charité e Universitätsmedizin Berlin, Berlin Germany
c
Charité e Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt Universität zu Berlin,
Institute of Biometry and Clinical Epidemiology, Berlin, Germany
d
Klinikum Dritter Orden, Munich, Germany
e
Charité e Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt Universität zu Berlin,
Department of Psychosomatic Medicine, Center for Internal Medicine and Dermatology, Berlin, Germany

Received 14 October 2021; received in revised form 16 December 2021; accepted 16 December 2021
Available online 20 January 2022

KEYWORDS Abstract Purpose: To deliver patient-reported outcome (PRO) reference data for breast can-
Health-related quality cer and various other breast diseases to facilitate the interpretation of PRO scores during
of life; routine breast cancer treatment.
Breast cancer; Methods: To determine reference baseline values for the PRO measures EORTC QLQ-C30
Reference values; and EORTC QLQ-BR23, PRO data captured in the breast cancer centre at Charité e Univer-
EORTC sitätsmedizin Berlin from 2016 to 2021 were evaluated. As part of the clinical routine, ambu-
latory patients were asked to answer a digital survey regarding their medical history, current
health status and health-related quality of life using the aforementioned questionnaires prior
to their doctor’s appointment in the outpatient breast clinic. Adjusted linear and variable
dispersion beta regression models were used to compare different diagnosis groups.
Results: A total of 3689 patients were included in the digital PRO program, of which 1478
were eligible for this study; 729 had invasive breast cancer or ductal carcinoma in situ, 270
patients were diagnosed with fibroadenoma and 479 patients had other breast diseases such
as cysts, mastopathy or abscesses. Overall, patients with breast cancer reported worse scores
in almost all domains except for role functioning, sexual functioning and body image.

* Corresponding author: Charité e Universitätsmedizin Berlin, Klinik für Gynäkologie mit Brustzentrum, Charitéplatz 1 Berlin, D e 10117,
Germany.
E-mail address: [email protected] (M.M. Karsten).
1
These authors contributed equally.

https://doi.org/10.1016/j.ejca.2021.12.020
0959-8049/ª 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139 129

Compared to previously published reference scores for early breast cancer, the current data
show a more pronounced impact on perceived emotional and cognitive functioning.
Conclusion: The results of this study are of high value for the interpretation of PROs and facil-
itate their use in clinical practice and clinical trials. The scores indicate an urgent need for psy-
chosocial support prior to treatment.
ª 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction individual patients whose self-reported health status

calls for specific attention.
Using patient-reported outcomes (PROs) in oncology To our knowledge, these are the first real-world PRO
care has been shown to improve patient satisfaction, reference data of this scale specifically for patients with
health-related quality of life and overall survival [1e3]. breast cancer prior to treatment.
They are defined as outcomes reported directly by the
patient without interpretation by an observer [4] and 2. Material and methods
provide health care professionals (HCPs) with valuable
feedback pertaining to the patient’s health status [5] and 2.1. Study design
psychosocial well-being [6]. However, PROs are still
primarily used in clinical trials and establishing them in In the breast centre at Charité e Universitätsmedizin
everyday cancer care comes with obstacles: due to a lack Berlin, standardised PRO assessment was established in
of experience, clinicians struggle to interpret them 2016 and is currently ongoing [18]. Since then, patients
properly [6e8]. Furthermore, it is still difficult for many have been asked to complete a survey containing their
HCPs to derive relevance from the values and determine medical history, demographics and PROs on a tablet
when they become actionable [8]. For one of the most computer before their appointment. Clinical data are
commonly used PRO measures (PROMs), the EORTC documented in the PRO-system (Heartbeat Medical,
QLQ-C30 [9], thresholds for clinical importance were Germany) using the ICHOM breast cancer standard
published only recently [10]. Moreover, the limited test [11], which has been adapted to our requirements.
length of traditional PROMs reduces their precision and The values presented in this study represent the first
may complicate accurate measurement. PRO assessment of the patients at the outpatient breast
Other than clinical trials, routine clinical documen- clinic.
tation often lacks standardised data acquisition. This 2.2. Patient characteristics
could facilitate the interpretability of PROs by making
comparisons between different patient populations A total of 3689 women were registered from November
possible. There are various initiatives addressing this 2016 to March 2021. They were 18 years and older and
issue, such as the International Consortium for Health had to speak German since the questionnaires were only
Outcomes Measurement (ICHOM), the EHDEN proj- provided in German language and signed consent for
ect and the European Health Outcomes Observatory the PRO assessment before they first answered the
[11e13]. In addition to aligning and harmonising data baseline questions at the Charité Breast clinic. After
acquisition, clinicians need to be provided with reference considering the exclusion criteria, 1727 women were
scores to make PROs useable for their decision making. eligible for inclusion. A total of 1478 of them (85.6%)
For some PROMs, comparative values from the general completed the PROMs. Of these, 729 (49.3%) patients
population or from clinical trials are available [14e17]. were diagnosed with breast cancer or ductal carcinoma
Clinical trials, however, prescribe strict inclusion in situ (DCIS), 270 (18.3%) patients had fibroadenoma
criteria, which impede their transfer to the very hetero- and 479 (32.4%) patients presented with other breast
geneous real-world patient population. What is missing diseases, such as mastopathy, cysts, lipoma, phyllodes
are reference values obtained before and during routine tumour or papilloma. 19 out of 668 (2.8%) patients with
care serving as an interpretation aid in everyday clinical breast cancer did not know their breast cancer diagnosis
practice. when they answered the baseline questions, so did 71/
The primary aim of this study is to deliver reference 270 (26.3%) patients with fibroadenoma and 27/479
data for common PROMs in breast cancer care to (5.6%) of the women in the group labelled as others.
support HCPs in interpreting the results before treat- For this report, we excluded patients as described in
ment (‘baseline’). We compared disease-specific group- Fig. 1.
level scores to published norm data from the general Patients with multiple diagnoses were not counted
population. Both will enable clinicians to identify more than once but allocated to one group according to
130 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

the following algorithm: breast cancer/DCIS > 2.4. Statistical analysis

fibroadenoma > other breast diseases. Initial diagnoses
from before and up to seven days after the survey were Continuous variables are presented as mean and stan-
considered. Questionnaire items included age, comorbid- dard deviation (SD) and median including the 25th and
ities, marital status and education. A complete list of 75th percentiles. Categorical variables are presented as
sociodemographic and clinical variables will be reported in absolute and relative frequencies.
an additional journal publication in Data in Brief. For the The influence of further variables on the QLQ-C30
analysis, education was classified as low (graduation from functional scores was evaluated using a fixed, predefined
secondary school or lower), high (high school graduation set of clinically relevant variables (diagnosis group, age,
or higher) or middle (anything in between). education, body mass index (BMI), marital status, pres-
The web-based PRO assessment was approved by the ence of heart disease or neurological disorders, depres-
local ethics committee (EA4/127/16). sion, previous breast surgery, previous cancer in the last
five years and family history of breast or ovarian cancer)
as covariates. Multiple linear and variable dispersion beta
2.3. Instruments regression models were used depending on the distribu-
tion of the respective score and the underlying assump-
The EORTC QLQ-C30 [9] comprises 30 questions and tions of the models [21e23]. Variable dispersion beta
covers the functional scales of physical (PF), role (RF), regression is a regression approach for variables varying
emotional (EF), cognitive (CF) and social functioning between 0 and 1 (mean model), as is the case for the PRO
(SF); the symptom scales of fatigue, nausea/vomiting, scores incorporating variable-dependent overdispersion
pain, dyspnoea, insomnia, appetite loss, constipation and (precision model). An estimate >0 in the beta regression
diarrhoea; financial difficulties and an assessment of the indicates higher scores under the influence of the respec-
global health status/QoL (all ranging from 0 to 100). tive covariate. The magnitude of the estimate can be used
Its breast cancer-specific supplement (EORTC QLQ- to make relative comparisons to other covariates
BR23 [19]) consists of 23 questions pertaining to the regarding their effect on the score. The precision model
functional scales of body image, future perspective, was developed by including all variables from the mean
sexual functioning and sexual enjoyment. The symptom model and retaining only those with p < 0.05. This
scales of systemic therapy side-effects, upset by hair loss, approach was supported by likelihood ratio tests
arm and breast symptoms are also assessed. comparing the model with full and reduced precision
Scoring was performed by building the raw sum score models. The logit link for the mean model was then
of the respective items and their linear transformation chosen by investigating the Akaike and Bayesian infor-
[20]. This results in scores ranging from 0 to 100, with mation criterion in comparison to log-log and comple-
higher scores on the functional scales representing a high mentary log-log-link.
level of functioning and higher scores on the symptom Patients with high leverage or a high Cook’s distance
scales implying a stronger symptom burden. relative to the remaining patients were excluded from

Fig. 1. Flow chart showing the inclusion process. EORTC, European Organisation for Research and Treatment of Cancer; QLQ, quality
of life questionnaire; C30, core module; BR23 breast cancer supplement; DCIS, ductal carcinoma in situ.
 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139 131

Fig. 2. Boxplot with scores of EORTC QLQ-C30 functional scales per disease group. QoL, Quality of life/Global health status; QLQ,
Quality of life questionnaire; C30, Core module; DCIS, Ductal carcinoma in situ.

model building. Furthermore, as a sensitivity analysis, being divorced (78/723 (10.8%) versus 14/276 (5.1%) and
bias correction and reduction were applied to the models 38/483 (7.9%)) or widowed (63/723 (8.7%) versus 2/276
[24,25]. (0.7%) and 5/483 (1%)), which matches their higher age,
The whole nature of this study was exploratory, and and a greater proportion of these patients had lower
no formal sample size calculation was conducted. All p- educational attainment (92/745 (12.3%) versus 20/289
values from this investigation have to be considered (6.9%) and 27/504 (5.3%).
non-confirmatory, and thus, no adjustment for multiple
testing was carried out. Analyses were performed using
R version 4.0.2 [26] and the betareg package [27]. 3.2. EORTC QLQ-C30 scores
During the digital surveys, we did not allow skipping
items of the PROMs to avoid missing responses. How- The scores for the QLQ-C30 functional scales are pre-
ever, due to the manual assignment of the question- sented in Fig. 2 and Table 1. Compared to patients with
naires at the beginning of the project, some patients fibroadenoma and patients with other breast diseases,
(n Z 4) did not receive all of them. the median scores were lower for patients with breast
cancer in all domains except for RF, where the median
3. Results was 100 in all groups. The difference was most obvious
in EF, SF (17 points) and CF (11 points).
3.1. Demographics and medical history (Table 1) For PF, 9 patients and 12 patients each for SF and
RF were excluded due to a high Cook’s distance or
On average, patients with breast cancer were older than leverage. Higher PF-scores were found in the adjusted
patients with fibroadenoma and those with other entities beta regression (Table 3) for patients with fibroadenoma
(mean (SD): 56 (13) versus 34 (10) and 41 (11) years). (estimate 0.27; 95% confidence interval (95% CI)
They also had a slightly higher BMI (25.5 (5.6) versus 0.01e0.53; p Z 0.0440) and other breast conditions
22.6 (3.8) and 22.9 (4) kg/m2) and more often presented (0.26; 95% CI 0.06e0.47; p Z 0.0118) than for patients
with comorbidities (414/746 (55.5%) versus 68/290 with breast cancer. Depression (
0.79; 95% CI
0.99 to
(23.4%) and 163/508 (32.1%)). In addition, the patients
0.58; p < 0.0001), heart diseases (
0.49; 95% CI
0.80
with breast cancer in our cohort more often reported to
0.18; p Z 0.0023) and a history of cancer in the last
132 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

Table 1
Demographic features and medical history.
Characteristic Breast cancer þ Fibroadenoma Other Total
DCIS (n Z 850) (n Z 313) (n Z 564) (n Z 1727)
Age [years] n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6)
Mean (SD) 55.73 (12.9) 34.29 (10.08) 40.92 (10.7) 47.01 (14.72)
Median (IQR) 55 (47; 65) 33 (27; 40.8) 40 (33; 48) 46 (36; 57)
Range 23; 87 16; 68 15; 78 15; 87
BMI [kg/m2] n (%) 744 (87.5) 289 (92.3) 504 (89.4) 1537 (89)
Mean (SD) 25.48 (5.58) 22.58 (3.77) 22.96 (3.98) 24.11 (4.97)
Median (IQR) 24.5 (22; 28) 22 (20; 24) 22 (20; 25) 23 (21; 26)
Range 15; 55 15; 38 16; 44 15; 55
Education n (%) 745 (87.6) 289 (92.3) 506 (89.7) 1540 (89.2)
Low 92 (12.3) 20 (6.9) 27 (5.3) 139 (9)
Middle 198 (26.6) 50 (17.3) 107 (21.1) 355 (23.1)
High 455 (61.1) 219 (75.8) 372 (73.5) 1046 (67.9)
Marital status n (%) 723 (85.1) 276 (88.2) 483 (85.6) 1482 (85.8)
Single 87 (12) 89 (32.2) 116 (24) 292 (19.7)
Married/in a relationship 495 (68.5) 171 (62) 324 (67.1) 990 (66.8)
Divorced/separated 78 (10.8) 14 (5.1) 38 (7.9) 130 (8.8)
Widowed 63 (8.7) 2 (0.7) 5 (1) 70 (4.7)
Any comorbidity n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 332 (44.5) 222 (76.6) 345 (67.9) 899 (58.2)
Yes 414 (55.5) 68 (23.4) 163 (32.1) 645 (41.8)
Heart disease* n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 702 (94.1) 287 (99) 499 (98.2) 1488 (96.4)
Yes 44 (5.9) 3 (1) 9 (1.8) 56 (3.6)
History of stroke n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 739 (99.1) 290 (100) 508 (100) 1537 (99.5)
Yes 7 (0.9) 0 (0) 0 (0) 7 (0.5)
Neurological n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
comorbidity**
No 732 (98.1) 284 (97.9) 497 (97.8) 1513 (98)
Yes 14 (1.9) 6 (2.1) 11 (2.2) 31 (2)
Depression n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 682 (91.4) 272 (93.8) 471 (92.7) 1425 (92.3)
Yes 64 (8.6) 18 (6.2) 37 (7.3) 119 (7.7)
Previous breast surgery n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 620 (83.1) 277 (95.5) 487 (95.9) 1384 (89.6)
Yes 126 (16.9) 13 (4.5) 21 (4.1) 160 (10.4)
Previous cancer (last 5 years) n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
No 622 (83.4) 283 (97.6) 492 (96.9) 1397 (90.5)
Yes 124 (16.6) 7 (2.4) 16 (3.1) 147 (9.5)
Previous breast cancer n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
(last 5 years)
No 682 (91.4) 283 (97.6) 492 (96.9) 1457 (94.4)
Yes 64 (8.6) 7 (2.4) 16 (3.1) 87 (5.6)
Family history of breast n (%) 746 (87.8) 290 (92.7) 508 (90.1) 1544 (89.4)
or ovarian cancer
No 458 (61.4) 188 (64.8) 318 (62.6) 964 (62.4)
Yes 288 (38.6) 102 (35.2) 190 (37.4) 580 (37.6)
DCIS ductal carcinoma in situ, BMI body mass index, SD standard deviation, IQR interquartile range.
*e.g. myocardial infarction, congestive heart failure.
**e.g. multiple sclerosis, Parkinson’s disease.

five years (
0.32; 95% CI
0.57 to
0.07; p Z 0.0112) scores (
0.68; 95% CI
0.95 to
0.40; p < 0.0001).
had a negative effect on the PF score. Furthermore, SF was found to be better in patients with
Compared to breast cancer, fibroadenoma and fibroadenoma (0.76; 95% CI 0.49e1.03; p < 0.0001) and
other breast diseases had almost no effect on RF other breast diseases (0.80; 95% CI 0.58e1.01;
(estimates<0.05; p > 0.5). In contrast, being married p < 0.0001) than in patients with breast cancer.
was associated with higher scores in that domain than Regarding EF and CF, the adjusted linear regression
being single (0.47; 95% CI 0.24e0.70; p Z 0.0001). As (Table 3) indicated higher scores for patients with
already shown for PF, beta regression indicated that RF fibroadenoma (EF: 17.26 points; 95% CI 12.93e21.59;
scores decreased with depression (
0.89; 95% CI
1.18 p < 0.0001; CF: 12.56; 95% CI 8.91e16.22; p < 0.0001)
to
0.59; p < 0.0001). The latter also applied for SF and patients with other conditions (EF: 14.8; 95% CI
Table 2
EORTC QLQ-C30 and EORTC QLQ-BR23 scores.
EORTC QLQ-C30 scale Breast cancerd Fibroadenoma Other (n Z 564) Total (n Z 1727) German general Breast cancer/DCIS
DCIS (n Z 850) (n Z 313) population norm patients reporting
data (Nolte baseline
et al.) [14] scores worse
than the TCIc
Physical functioninga n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 233 (32)
Mean (SD) 84.42 (20.52) 93.01 (12.4) 91.76 (13.05) 88.37 (17.49) 82.0 (21.5)
Median (IQR) 93.33 (73.3; 100) 100 (93.3; 100) 100 (86.7; 100) 93.33 (86.7; 100)
Range 6.7; 100 6.7; 100 20; 100 6.7; 100
TCIc 83

M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

Role functioninga n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 96 (13.2)
Mean (SD) 85.85 (25.45) 91.11 (20.35) 90.99 (21.06) 88.48 (23.34) 80.3 (27.4)
Median (IQR) 100 (83.3; 100) 100 (100; 100) 100 (100; 100) 100 (83.3; 100)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 58
Emotional functioninga n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 496 (68)
Mean (SD) 55.85 (26.61)d 69.23 (27.06) 67.4 (25.09) 62.04 (26.9) 75.1 (24.2)
Median (IQR) 58.33 (33.3; 75) 75 (50; 91.7) 66.67 (50; 91.7) 66.67 (41.7; 83.3)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 71
Cognitive functioninga n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 463 (63.5)
Mean (SD) 64.49 (22.88)d 74.2 (23.38) 73.74 (21.28) 69.26 (22.94) 85.4 (21.1)
Median (IQR) 66.67 (50; 83.3) 77.78 (61.1; 94.4) 77.78 (61.1; 88.9) 72.22 (55.6; 88.9)
Range 0; 100 5.6; 100 0; 100 0; 100
TCIc 75
Social functioninga n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 160 (21.9)
Mean (SD) 76.77 (28.01) 89.69 (20.47) 87.58 (21.28) 82.63 (25.37) 85.1 (25.5)
Median (IQR) 83.33 (66.7; 100) 100 (83.3; 100) 100 (83.3; 100) 100 (66.7; 100)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 58
Fatigueb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 211 (28.9)
Mean (SD) 29.52 (27.19) 24.57 (23.5) 25.08 (24.49) 27.18 (25.78) 31.4 (27.7)
Median (IQR) 22.22 (11.1; 44.4) 22.22 (0; 33.3) 22.22 (0; 33.3) 22.22 (0; 44.4)
Range 0; 100 0; 88.9 0; 100 0; 100
TCIc 39
Nausea/vomitingb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 153 (21)
Mean (SD) 5.67 (13.5) 5.19 (11.76) 5.22 (12.27) 5.44 (12.8) 5.2 (15.7)
Median (IQR) 0 (0; 0) 0 (0; 0) 0 (0; 0) 0 (0; 0)
Range 0; 100 0; 100 0; 66.7 0; 100
TCIc 8
Painb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 276 (37.9)
Mean (SD) 24.35 (27.72) 15.06 (21.64) 20.7 (26.14) 21.47 (26.4) 28.3 (31.1)
Median (IQR) 16.67 (0; 33.3) 0 (0; 16.7) 16.67 (0; 33.3) 16.67 (0; 33.3)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 25
(continued on next page)

133
 134
Table 2 (continued )
EORTC QLQ-C30 scale Breast cancerd Fibroadenoma Other (n Z 564) Total (n Z 1727) German general Breast cancer/DCIS
DCIS (n Z 850) (n Z 313) population norm patients reporting
data (Nolte baseline
et al.) [14] scores worse
than the TCIc
Dyspnoeab n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 238 (32.6)
Mean (SD) 15.82 (26.08) 10 (19.57) 10.37 (23.05) 12.99 (24.18) 19.6 (27.8)
Median (IQR) 0 (0; 33.3) 0 (0; 0) 0 (0; 0) 0 (0; 33.3)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 17
Insomniab n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 274 (37.6)

M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

Mean (SD) 39.64 (34.85)d 24.57 (29.89) 27 (30.63) 32.79 (33.34) 28.9 (33.6)
Median (IQR) 33.33 (0; 66.7) 0 (0; 33.3) 33.33 (0; 33.3) 33.33 (0; 66.7)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 50
Appetite lossb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 96 (13.2)
Mean (SD) 16.19 (27.24) 9.38 (20.18) 10.44 (23.07) 13.08 (24.93) 9.3 (22.2)
Median (IQR) 0 (0; 33.3) 0 (0; 0) 0 (0; 0) 0 (0; 33.3)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 50
Constipationb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 28 (3.8)
Mean (SD) 6.54 (17.07) 5.56 (16.94) 6.68 (16.33) 6.41 (16.81) 8.9 (21.6)
Median (IQR) 0 (0; 0) 0 (0; 0) 0 (0; 0) 0 (0; 0)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 50
Diarrhoeab n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 146 (20)
Mean (SD) 9.05 (20.16) 6.67 (16.91) 7.65 (19.33) 8.16 (19.34) 9.7 (22.2)
Median (IQR) 0 (0; 0) 0 (0; 0) 0 (0; 0) 0 (0; 0)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 17
Financial difficultiesb n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6) 151 (20.7)
Mean (SD) 10.47 (23.28) 5.06 (15.84) 5.57 (18.46) 7.89 (20.73) 10.4 (24.1)
Median (IQR) 0 (0; 0) 0 (0; 0) 0 (0; 0) 0 (0; 0)
Range 0; 100 0; 100 0; 100 0; 100
TCIc 17
Global health status/ n (%) 729 (85.8) 270 (86.3) 479 (84.9) 1478 (85.6)
Quality of lifea Mean (SD) 62.7 (23.26) 72.53 (21.22) 71.82 (20.11) 67.45 (22.4) 65.9 (22.2)
Median (IQR) 66.67 (50; 83.3) 75 (66.7; 83.3) 75 (58.3; 83.3) 66.67 (50; 83.3)
Range 0; 100 0; 100 8.3; 100 0; 100
EORTC QLQ-BR23 scale
Body Imagea n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 82.98 (22.87) 88.89 (16.88) 86.7 (21.67) 85.27 (21.61)
Median (IQR) 91.67 (75; 100) 100 (83.3; 100) 100 (83.3; 100) 91.67 (77.1; 100)
Range 0; 100 16.7; 100 0; 100 0; 100
Sexual functioninga n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 29.13 (29.35) 43.83 (29.12) 42.54 (28.08) 36.17 (29.71)
 Median (IQR) 16.67 (0; 50) 33.33 (16.7; 66.7) 33.33 (16.7; 66.7) 33.33 (0; 66.7)
Range 0; 100 0; 100 0; 100 0; 100
Sexual enjoymenta n (%) 391 (46) 185 (59.1) 334 (59.2) 910 (52.7)
Mean (SD) 69.99 (27.06) 77.12 (25.28) 75.45 (24.97) 73.44 (26.1)
Median (IQR) 66.67 (66.7; 100) 66.67 (66.7; 100) 66.67 (66.7; 100) 66.67 (66.7; 100)
Range 0; 100 0; 100 0; 100 0; 100
Future perspectivea n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 31.86 (32.72) 62.59 (31.42) 55.72 (31.99) 45.23 (34.9)
Median (IQR) 33.33 (0; 66.7) 66.67 (33.3; 100) 66.67 (33.3; 66.7) 33.33 (0; 66.7)
Range 0; 100 0; 100 0; 100 0; 100
Systemic therapy side-effectsb n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 17.58 (16.34) 14.43 (12.43) 15.48 (15.56) 16.32 (15.48)

M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

Median (IQR) 14.29 (4.8; 23.8) 9.52 (4.8; 23.8) 9.52 (4.8; 23.8) 14.29 (4.8; 23.8)
Range 0; 81 0; 57.1 0; 85.7 0; 85.7
Breast symptomsb n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 17.62 (20.54) 13.7 (16.54) 18.67 (22.97) 17.24 (20.78)
Median (IQR) 8.33 (0; 25) 8.33 (0; 22.9) 8.33 (0; 25) 8.33 (0; 25)
Range 0; 100 0; 83.3 0; 100 0; 100
Arm symptomsb n (%) 726 (85.4) 270 (86.3) 478 (84.8) 1474 (85.4)
Mean (SD) 14.92 (20.9) 7.9 (14.99) 11.27 (18.12) 12.45 (19.23)
Median (IQR) 0 (0; 22.2) 0 (0; 11.1) 0 (0; 11.1) 0 (0; 22.2)
Range 0; 100 0; 88.9 0; 100 0; 100
Upset by hair lossb n (%) 226 (26.6) 94 (30) 193 (34.2) 513 (29.7)
Mean (SD) 35.4 (33.64) 26.6 (30.75) 37.13 (34.15) 34.44 (33.48)
Median (IQR) 33.33 (0; 33.3) 33.33 (0; 33.3) 33.33 (0; 66.7) 33.33 (0; 66.7)
Range 0; 100 0; 100 0; 100 0; 100
DCIS ductal carcinoma in situ, SD standard deviation, IQR interquartile range, TCI threshold for clinical importance.
a
Higher scores in the functional scales represent a higher level of functioning.
b
Higher scores in the symptom scales imply greater symptom burden.
c
The thresholds for clinical importance were published by Giesinger et al. [10] Patients who report values worse than the TCI might need more intensive clinical attention.
d
These scores are clinically worse (i.e. > minimal clinically important difference of 10 points according to Snyder et al. [28]) than the norm data.

135
Table 3

136
Regression models.
Covariate Variable dispersion beta regressiona Adjusted linear regression
Physical functioning Role functioning Social functioning Emotional functioning Cognitive functioning
Estimate (95%-CI) p-value Estimate (95%-CI) p-value Estimate (95%-CI) p-value Estimate (95%-CI) p-value Estimate (95%-CI) p-value
Fibroadenoma (versus 0,27 (0.01; 0.53) 0.0440
0,02 (
0.21; 0.17) 0.8183 0,76 (0.49; 1.03) <0.0001 17,26 (12.93; 21.59) <0.0001 12,56 (8.91; 16.22) <0.0001
breast cancer)
Other (versus breast 0,26 (0.06; 0.47) 0.0118 0,04 (
0.11; 0.20) 0.5594 0,80 (0.58; 1.01) <0.0001 14,8 (11.37; 18.24) <0.0001 11,68 (8.78; 14.57) <0.0001
cancer)
Age [10 years]
0,10 (
0.15;
0.05) 0.0001
0,03 (
0.08; 0.03) 0.3205 0,00 (
0.06; 0.06) 0.9996 2,27 (1.03; 3.51) 0.0003 1,83 (0.78; 2.88) 0.0006
BMI [kg/m2]
0,04 (
0.05;
0.03) <0.0001
0,03 (
0.05;
0.01) 0.0007 0,00 (
0.01; 0.01) 0.9728
0,03 (
0.32; 0.25) 0.8156
0,02 (
0.26; 0.22) 0.8823

M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

Education: middle 0,34 (0.06; 0.63) 0.0177 0,12 (
0.12; 0.36) 0.3149 0,15 (
0.09; 0.39) 0.2236 2,77 (
2.42; 7.96) 0.2950 3,59 (
0.79; 7.97) 0.1082
(versus low)
Education: high 0,50 (0.24; 0.76) 0.0002 0,14 (
0.08; 0.36) 0.2150 0,09 (
0.14; 0.31) 0.4476 0,94 (
3.92; 5.8) 0.7044 1,6 (
2.5; 5.71) 0.4437
(versus low)
Relationship:
0,03 (
0.17; 0.11) 0.6624 0,47 (0.24; 0.70) 0.0001 0,02 (
0.13; 0.17) 0.7987 0,25 (
3.19; 3.68) 0.8884 1,21 (
1.68; 4.11) 0.4124
married/living
together (versus no
relationship)
Relationship: divorced/
0,26 (
0.49;
0.04) 0.0231 0,12 (
0.24; 0.47) 0.5206
0,06 (
0.31; 0.19) 0.6389
0,56 (
6.05; 4.92) 0.8401
0,71 (
5.34; 3.91) 0.762
not living together
(versus no
relationship)
Relationship: widowed
0,42 (
0.73;
0.11) 0.0079 0,35 (
0.11; 0.81) 0.1371 0,29 (
0.05; 0.63) 0.0952 5,76 (
1.54; 13.05) 0.1220 4,58 (
1.57; 10.74) 0.1443
(versus no
relationship)
Comorbidity: heart
0,49 (
0.80;
0.18) 0.0023
0,35 (
0.83; 0.12) 0.1459 0,18 (
0.30; 0.66) 0.4732
3,47 (
10.83; 3.89) 0.3556
1,77 (
7.98; 4.44) 0.5756
disease (versus not
present)
Comorbidity: stroke
0,44 (
1.14; 0.53) 0.3754
0,60 (
2.09; 0.89) 0.4308
0,56 (
2.03; 0.92) 0.4586 13,89 (
4.99; 32.77) 0.1491 13,55 (
2.38; 29.47) 0.0955
(versus not present)
Comorbidity:
1,31 (
1.75;
0.87) <0.0001
1,64 (-2.15;
1.13) <0.0001
1,13 (-1.61;
0.66) <0.0001
11,27 (-20.44;
2.09) 0.0161
14,08 (-21.82;
6.34) 0.0004
neurological
disorder (versus not
present)
Comorbidity: depression
0,79 (
0.99;
0.58) <0.0001
0,89 (
1.18;
0.59) <0.0001
0,68 (
0.95;
0.40) <0.0001
23,85 (
28.82;-18.88) <0.0001
23,11 (
27.3;
18.92) <0.0001
(versus not present)
Previous surgery (versus
0,42 (
0.62;
0.23) <0.0001
1,28 (
1.54;
1.01) <0.0001
0,33 (
0.55;
0.11) 0.0028
3,84 (
8.4; 0.71) 0.0983
4,64 (
8.48;
0.79) 0.0181
no previous surgery)
Cancer other than breast
0,32 (
0.57;
0.07) 0.0112
0,13 (
0.41; 0.15) 0.3527
0,45 (
0.73;
0.17) 0.0018
6,49 (
12.34;
0.64) 0.0296
5,93 (
10.87;
1) 0.0185
in the last 5 years
(versus not present)
Family history of breast
0,01 (
0.12; 0.10) 0.9005
0,31 (
0.5.;
0.13) 0.0009
0,09 (
0.22; 0.03) 0.1285
3,39 (
6.08;
0.7) 0.0136
3,22 (
5.48;
0.95) 0.0055
or ovarian cancer
(versus not present)
Adjusted R2 of emotional functioning Z 0.13 and of cognitive functioning Z 0.15; bold p-values are below 0.05, BMI body mass index.
a
Variable dispersion beta regression is a regression approach for variables varying between 0 and 1 incorporating variable-dependent overdispersion. An estimate >0 in the beta regression indicates
higher scores under the influence of the respective covariate. The magnitude of the estimate can be used to make relative comparisons to other covariates regarding their effect on the score.
 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139 137

11.37e18.24; p < 0.0001; CF: 11.68, 95% CI such as ours are needed to guide clinicians in the inter-
8.78e14.57; p < 0.0001) than for patients with breast pretation of PROs.
cancer. Moreover, EF and CF scores were found to be Our data demonstrate that patients with breast cancer
decreased in patients with depression (EF:
23.85; 95% recruited in routine clinical practice report increased
CI -28.82 to
18.88; p < 0.0001; CF:
23.11; 95% CI functional strain especially in the domains of EF and CF,
-27.3 to
18.92; p < 0.0001). prior to the start of their treatment compared to patients
The complete results of the symptom scales are also with primarily benign entities. Other trials reported that
presented in Table 2. cancer itself might cause affective and cognitive impair-
ment even before treatment initiation [38]. Additionally,
3.3. EORTC QLQ-BR23 scores pretreatment worry has been shown to be associated with
a decline in cognitive functioning [39].
Table 2 contains the scores of the QLQ-BR23. It shows Where possible, we used the easily interpretable linear
similar results across all diagnosis groups for body regression to investigate the effect of the respective
image and sexual enjoyment but lower future perspec- covariates. However, for some scales, it would have
tive (median 34 points) and sexual functioning (median distorted the results due to the distribution of the values,
16 points) scores for breast cancer patients. which is why we chose to perform beta regression in
these cases. The adjusted R2 represents the proportion
of the dispersion, which can be explained by the linear
4. Discussion models and seems fairly small in our analyses. However,
the scales represent very complex constructs and the
4.1. Aim of the study sensitivity analyses indicate the robustness of the
models. Ultimately, we consider them sufficient to
There is an increasing availability of PRO data not compare the diagnosis groups.
only in clinical trials [29,30] but also in everyday We decided to include scores of all women seen at
clinical practice where its implementation [31] still the breast centre in this publication to highlight the
faces some hurdles [32]. One of the cited reasons at the differences between PRO values in women with cancer
level of healthcare providers is the difficulty to compared to benign breast disease and enable clini-
correctly interpret the available PRO values in a spe- cians to develop a better understanding of the range of
cific setting [6,7,31,33,34]. While there are several PRO values among different diseases afflicting the
publications regarding PRO values in breast cancer breast. The availability of reference values for PROs
patients, many are lacking PRO values before the start ranging from general population norm data to benign
of treatment [35]. The aim of this study was to provide breast disease to malignant breast disease facilitates the
clinicians with reference values for the EORTC QLQ- understanding and therefore the use of PROs in clinical
C30 questionnaire in breast cancer patients at the time practice.
of their initial diagnosis and before the start of The group classified as ‘other breast diseases’ is very
treatment. heterogeneous, as it includes findings with unclear dig-
nity (phyllodes) and possible risk lesions (ductal papil-
4.2. Main findings loma), as well as benign diseases. The scores of the
fibroadenoma group were almost consistently better and
The following comparison refers to the German general even exceeded the normative scores, which is likely due
population norm data published by our group earlier to younger age and fewer comorbidities.
[14]. In accordance with Snyder et al., we chose a 10-
point score difference as the critical threshold [28]. 4.3. Limitations
Compared to the normative values, our breast cancer
patients reported clinically relevant poorer EF, which is Due to the unmet criteria for linear regression, we were
possibly due to uncertainty and anxiety regarding the not able to provide easily understandable estimates of
upcoming diagnosis and procedures [36]. These values score modifications at baseline for all functional do-
are furthermore well below the thresholds for clinical mains of the C30. Nevertheless, it showed the relative
importance [10] and even below the newly published importance of parameters in comparison to each other
reference scores for breast cancer [17]. The latter were via more appropriate beta regression models.
obtained in the AMOROS trial which included only
patients with early-stage unifocal tumours and without 5. Conclusions
previous oncological treatment [17,37]. Their knowledge
about the diagnosis and its favourable prognosis in the This study presents real-world reference scores, which
early and operable stages of the disease might explain will facilitate PRO usage in breast cancer trials and care.
the better EF. However, patient populations in real- Compared to the norm data, patients with breast cancer
world settings are much more heterogeneous, and data reported clinically worse EF and CF prior to treatment.
138 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139

In this situation, limited cognitive functioning collides [2] Denis F, Basch E, Septans A-L, Bennouna J, Urban T,
with the life decisions patients face when considering Dueck AC, et al. Two-year survival comparing web-based
symptom monitoring vs routine surveillance following treatment
their diagnosis. Follow-up scores obtained during for lung cancer. JAMA 2019;321:306e7.
routine treatment will be necessary to evaluate the tra- [3] Chen J, Ou L, Hollis SJ. A systematic review of the impact of
jectories, as well as the identification of disease-specific routine collection of patient reported outcome measures on pa-
thresholds. tients, providers and health organisations in an oncologic setting.
BMC Health Serv Res 2013;13:211.
[4] Calvert M, Blazeby J, Altman DG, Revicki DA, Moher D,
Author contributions Brundage MD, et al. Reporting of patient-reported outcomes in
randomized trials: the CONSORT PRO extension. JAMA 2013;
MMK: conceptualisation, methodology, investigation, 309:814e22.
writing e original draft/review & editing. RR: con- [5] Basch E, Stover AM, Schrag D, Chung A, Jansen J, Henson S,
ceptualisation, methodology, formal analysis, visual- et al. Clinical utility and user perceptions of a digital system for
electronic patient-reported symptom monitoring during routine
isation, writing e original draft/review & editing. SA: cancer care: findings from the PRO-TECT trial. JCO Clin Cancer
data acquisition, review & editing. TP: writing, review Inform 2020;4:947e57.
&editing. AMH: data acquisition, review & editing. [6] Nguyen H, Butow P, Dhillon H, Sundaresan P. A review of the
KW: methodology, review & editing. FF: methodology, barriers to using Patient-Reported Outcomes (PROs) and Patient-
review & editing. MR: methodology, review & editing. Reported Outcome Measures (PROMs) in routine cancer care. J
Med Radiat Sci 2021;68:186e95.
FK: conceptualisation, methodology, visualisation, [7] Boyce MB, Browne JP, Greenhalgh J. The experiences of pro-
writing e original draft/review & editing. JUB: con- fessionals with using information from patient-reported outcome
ceptualisation, investigation, writing e review & editing. measures to improve the quality of healthcare: a systematic review
of qualitative research. BMJ Qual Saf 2014;23:508.
[8] Graupner C, Breukink SO, Mul S, Claessens D, Slok AHM,
Trial registration
Kimman ML. Patient-reported outcome measures in oncology: a
qualitative study of the healthcare professional’s perspective.
Not applicable (evaluation of routine clinical data). Support Care Cancer 2021;29:5253e61.
[9] Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A,
Funding Duez NJ, et al. The European Organization for Research and
Treatment of Cancer QLQ-C30: a quality-of-life instrument for
use in international clinical trials in oncology. J Natl Cancer Inst
Funding for this work was provided by Charité e 1993;85:5253e61. https://doi.org/10.1007/s00520-021-06052-9.
Universitätsmedizin Berlin through a general research Epub 2021 Mar 2.PMID: 33655412.
fund. [10] Giesinger JM, Loth FLC, Aaronson NK, Arraras JI, Caocci G,
Efficace F, et al. Thresholds for clinical importance were estab-
lished to improve interpretation of the EORTC QLQ-C30 in
Conflict of interest statement
clinical practice and research. J Clin Epidemiol 2020;118:1e8.
[11] Ong WL, Schouwenburg MG, van Bommel ACM, Stowell C,
The authors declare that they have no known Allison KH, Benn KE, et al. A standard set of value-based pa-
competing financial interests or personal relationships tient-centered outcomes for breast cancer: the International
that could have appeared to influence the work reported Consortium for Health Outcomes Measurement (ICHOM)
initiative. JAMA Oncol 2017;3:677e85.
in this paper.
[12] Stamm T, Bott N, Thwaites R, Mosor E, Andrews MR,
Borgdorff J, et al. Building a value-based care infrastructure in
Acknowledgements Europe: the health outcomes observatory. NEJM Catal 2021.
https://catalyst.nejm.org/doi/full/10.1056/CAT.21.014.
[13] European Health Data & Evidence Network (EHDEN). https://
We want to thank Sarah Albers, Melissa Afshar-
www.ehden.eu/.
Bakshloo, Lea Pachal, Chiara Richter, Hivin Alsaied, [14] Nolte S, Waldmann A, Liegl G, Petersen MA, Groenvold M,
Emel Ertürk, Julia Barho, Trang Trinh, Laura-Antonia Rose M. Updated EORTC QLQ-C30 general population norm
Pohany, Ottilia Berninger and Franziska Abend for data for Germany. Eur J Cancer 2020;137:161e70.
documenting the clinical data in the PRO database. [15] Fischer F, Gibbons C, Coste J, Valderas JM, Rose M, Leplège A.
Measurement invariance and general population reference values
of the PROMIS Profile 29 in the UK, France, and Germany. Qual
Appendix A. Supplementary data Life Res 2018;27:999e1014.
[16] Mundy LR, Homa K, Klassen AF, Pusic AL, Kerrigan CL.
Supplementary data to this article can be found online Breast cancer and reconstruction: normative data for interpreting
at https://doi.org/10.1016/j.ejca.2021.12.020. the BREAST-Q. Plast Reconstr Surg 2017;139:1046ee55e.
[17] Mierzynska J, Taye M, Pe M, Coens C, Martinelli F, Pogoda K,
et al. Reference values for the EORTC QLQ-C30 in early and
References metastatic breast cancer. Eur J Cancer 2020;125:69e82.
[18] Karsten MM, Speiser D, Hartmann C, Zeuschner N, Lippold K,
[1] Basch E, Deal AM, Kris MG, Scher HI, Hudis CA, Sabbatini P, Kiver V, et al. Web-based patient-reported outcomes using the
et al. Symptom monitoring with patient-reported outcomes dur- international consortium for health outcome measurement data-
ing routine cancer treatment: a randomized controlled trial. J Clin set in a major German University Hospital: observational study.
Oncol: Off J Am Soc Clin Oncol 2016;34:557e65. JMIR Cancer 2018;4:e11373.
 M.M. Karsten et al. / European Journal of Cancer 163 (2022) 128e139 139

[19] Sprangers MA, Groenvold M, Arraras JI, Franklin J, te Velde A, [31] Trautmann F, Hentschel L, Hornemann B, Rentsch A,
Muller M, et al. The European Organization for Research and Baumann M, Ehninger G, et al. Electronic real-time assessment of
Treatment of Cancer breast cancer-specific quality-of-life ques- patient-reported outcomes in routine care-first findings and ex-
tionnaire module: first results from a three-country field study. J periences from the implementation in a comprehensive cancer
Clin Oncol 1996;14:2756e68. center. Support Care Cancer 2016;24:3047e56.
[20] Fayers PM, Aaronson NK, Bjordal K, Groenvold M, Curran D, [32] Snyder CF, Blackford AL, Wolff AC, Carducci MA,
Bottomley A. In: The EORTC QLQ-C30 scoring manual. 3rd ed. Herman JM, Wu AW. Feasibility and value of PatientViewpoint:
Brüssel: European Organisation of Research and Treatment of a web system for patient-reported outcomes assessment in clinical
Cancer; 2001. practice. Psycho Oncol 2013;22:895e901.
[21] Ferrari S, Cribari-Neto F. Beta regression for modelling rates and [33] Mouillet G, Falcoz A, Fritzsch J, Almotlak H, Jacoulet P,
proportions. J Appl Stat 2004;31:799e815. Pivot X, et al. Feasibility of health-related quality of life
[22] Smithson M, Verkuilen J. A better lemon squeezer? Maximum- (HRQoL) assessment for cancer patients using electronic patient-
likelihood regression with beta-distributed dependent variables. reported outcome (ePRO) in daily clinical practice. Qual Life Res
Psychol Methods 2006;11:54e71. 2021;30:3255e66.
[23] Simas AB, Barreto-Souza W, Rocha AV. Improved estimators for [34] Santana MJ, Haverman L, Absolom K, Takeuchi E, Feeny D,
a general class of beta regression models. Comput Stat Data Anal Grootenhuis M, et al. Training clinicians in how to use patient-
2010;54:348e66. reported outcome measures in routine clinical practice. Qual
[24] Kosmidis I, Firth D. A generic algorithm for reducing bias in Life Res 2015;24:1707e18.
parametric estimation. Electron J Stat 2010;4:1097e112. 16. [35] Lagendijk M, van Egdom LSE, van Veen FEE, Vos EL,
[25] Grün B, Kosmidis I, Zeileis A. Extended beta regression in R: Mureau MAM, van Leeuwen N, et al. Patient-reported outcome
Shaken, stirred, mixed, and partitioned. J Stat Software 2012;48:25. measures may add value in breast cancer surgery. Ann Surg Oncol
[26] R: a language and environment for statistical computing. Vienna, 2018;25:3563e71.
Austria: R Foundation for Statistical Computing; 2020. [36] Okazaki S, Iwamitsu Y, Masaru K, Todoroki K, Suzuki S,
[27] Cribari-Neto F, Zeileis A. Beta regression in R. J Stat Software Yamamoto K, et al. The psychological responses of outpatient
2010;34:24. breast cancer patients before and during first medical consulta-
[28] Snyder CF, Blackford AL, Sussman J, Bainbridge D, Howell D, tion. Palliat Support Care 2009;7:307e14.
Seow HY, et al. Identifying changes in scores on the EORTC- [37] Straver ME, Meijnen P, Tienhoven Gv, Velde CJHvd,
QLQ-C30 representing a change in patients’ supportive care Mansel RE, Bogaerts J, et al. Role of axillary clearance after a
needs. Qual Life Res 2015;24:1207e16. tumor-positive sentinel node in the administration of adjuvant
[29] Efficace F, Fayers P, Pusic A, Cemal Y, Yanagawa J, Jacobs M, therapy in early breast cancer. J Clin Oncol 2010;28:731e7.
et al. Quality of patient-reported outcome reporting across cancer [38] Mampay M, Flint MS, Sheridan GK. Tumour brain: pretreat-
randomized controlled trials according to the CONSORT patient- ment cognitive and affective disorders caused by peripheral can-
reported outcome extension: a pooled analysis of 557 trials. cers. Br J Pharmacol 2021;178:3977e96. https:
Cancer 2015;121:3335e42. //doi.org/10.1111/bph.15571.
[30] Tevis SE, James TA, Kuerer HM, Pusic AL, Yao KA, Merlino J, [39] Berman MG, Askren MK, Jung M, Therrien B, Peltier S,
et al. Patient-reported outcomes for breast cancer. Ann Surg Noll DC, et al. Pretreatment worry and neurocognitive responses
Oncol 2018;25:2839e45. in women with breast cancer. Health Psychol 2014;33:222e31.