Key words
breast cancer - metastatic setting - patient-reported outcomes - acceptance of technology-based
surveys - obstacles
Schlüsselwörter
Brustkrebs - metastasierte Situation - Patient-reported Outcomes - Akzeptanz von technikbasierten
Erhebungen - Hürden
Introduction
Despite recent advances in surgery, chemotherapy, and other forms of support, metastatic
breast cancer remains a challenge for gynecologic oncology [1], [2], [3]. The mean life expectancy of these patients is 3 years and depends on tumor biology
and the degree and site of metastasis, and the 5-year survival rate is 25 to 30 %
[4], [5], [6], [7], [8], [9], [10], [11]. This means that the prognosis for these patients is significantly worse than that
of patients in the adjuvant setting. The mean 5-year survival rate for all patients
with breast cancer is currently around 80 % [12], [13], [14], [15], [16]. For patients with metastases, palliative care may often be the only choice, with
the remaining therapy options aiming to extend the patientʼs survival time while remaining
largely free of tumor-related symptoms. This is why health-related quality of life
should always be included in therapy planning [17], [18], [19], [20], [21], [22], [23], [24], [25]. In the medium or longer term, systemic chemotherapy may also reduce the health-related
quality of life of patients receiving adjuvant treatment [26], [27], [28].
Measuring the health-related quality of life of patients with metastatic breast cancer
is very relevant, and not merely for healthcare research. According to the German
Act on the Reform of the Market for Medicinal Products (Arzneimittel-Neuordnungsgesetz [AMNOG]), the proof of the benefit of therapeutic interventions must be based on
patient-relevant endpoints and should include various aspects which show “how the patient feels, how he is able to perform functions and activities, and whether
he survives” (AMNOG 2010; § 35a SGB V and Code of Procedure of the Federal Joint Commission [G-BA-Verfahrensordnung], § 13) [29]. The measured variables used to assess the benefit (e.g. health-related quality
of life) are referred to as patient-reported outcomes (PRO) and reflect the patientʼs
subjective perception of her own state of health without evaluation by a third party
[30], [31]. Particularly in oncology patients, the patientʼs subjective perception of her own
state of health is considered an important indication of the efficacy of a specific
therapy [32], [33], [34].
Paper-based surveys still predominate; electronic methods to collect PROs (e.g. with
tablet PCs) have only begun to be used in recent years. But knowledge about the acceptance
and practicability of this method of data collection remains limited [32], [35], [36], [37], [38], [39], [40], [41], [42]. To date, there has been no study which has attempted to identify possible barriers
to the use of electronic surveys which could arise from the side-effects of therapy
or be due to aspects of the patientʼs biography. This lack result in a consistent
bias in such surveys, as it is unclear whether insufficient computer skills, therapy-related
barriers or other prerequisites necessary to complete an electronic survey might alter
the response or even result in a refusal to participate in the survey, i.e., whether
there are barriers which could influence the survey outcome. This study looked at
whether previous experience of using a tablet or the internet (computer skills), disease
status, patientsʼ health-related quality of life and the sociodemographic variables
“level of education” and “age” influenced the willingness of breast cancer patients
with metastasis or receiving adjuvant treatment to use electronic surveys. The hypothesis
was that older patients in a poorer state of health and a lower health-related quality
of life faced greater barriers and required more support compared to younger patients
in a better state of health, which is why particularly older and more ill patients
may be less willing to complete ePRO questionnaires (e.g. using a tablet).
Patients/Material and Methods
Patients/Material and Methods
Sample and study design
In summer 2015, 96 breast cancer patients with metastasis or receiving adjuvant therapy
and treated consecutively at the University Gynecological Hospital in Tübingen completed
a survey using a paper-based questionnaire. A total of 120 patients were asked to
participate in the survey, resulting in a response rate of 80 %. 65 patients (68 %)
had metastases and 31 patients (32 %) were receiving adjuvant therapy. The data collected
from the two patient groups were combined for statistical analysis. Patients completed
the questionnaire during an outpatient visit to the hospital under the supervision
of an attending physician. Patients were informed prior to completing the questionnaire
about the aims of the study and that participation in the study was voluntary. The
ethics committee gave its prior consent to the study (project number 196/2015B02).
All female breast cancer patients aged more than 18 years who either had metastasis
or were undergoing adjuvant treatment and who additionally had sufficient knowledge
of German to answer the questionnaire were included in the study.
Questionnaire
The survey consisted of three parts. The first part focused on the patientsʼ socio-economic
variables. The second part asked patients about their health-related quality of life
and overall state of health. The survey questionnaires EORTC QLQ-C30 and EQ VAS (EQ-5D-5L)
were used for this second part [43], [44], [45], [46]; the questionnaires had already been previously approved for use in a parallel study
(project number 234/2014BO1). The third part consisted of an additional questionnaire
consisting of validated “partial” questionnaires which had been developed to examine
our study questions. In this third part of the survey patients were asked to provide
information about their use of electronic technology at home, to evaluate their knowledge
and understanding of computers and the internet and comment on their general attitude
toward electronically-based surveys. EORTC QLQ-C30 is a disease-specific questionnaire,
the use of which has already been validated in research. It consists of 5 subscales,
various symptom scales, and individual items which aim to capture the patientsʼ quality
of life on a multidimensional level [43], [44]. The generic EQ-5D-5L questionnaire evaluates quality of life in five dimensions
using a five-step scale and the EQ VAS scale, with the current state of health recorded
as a number (0 = worst imaginable state of health, 100 = best imaginable state of
health) [45], [46]. To evaluate quality of life, patients were surveyed using the EQ VAS scale which
was combined with two questions from the EORTC QLQ-C30, and responses to questions
about the patientʼs current health status and current quality of life were recorded
using a seven-step Likert scale (from 1 = very poor to 7 = excellent). Calculations
of mean values were done in accordance with the official EORTC guidelines which require
a separate score to be calculated for each scale, with scores taking any value between
0 and 100 [47].
The additional questions on the patientʼs computer skills and needs based on their
prior experience of digital media consisted of: the modules on the private use of
technology from the KBF-BK questionnaire [48]; published, validated items of a survey on the acceptance and reliability of electronic
psycho-oncologic screening [40]; and additional questions, developed by the authors, on the aspects “technical barriers”
and “potential (technological) support structures which would take the patientʼs prior
experience of technological research tools into account” [49], [50].
Statistical analysis
A frequency analysis was done with MS Excel 2010 and IBM SPSS 21 to determine the
descriptive characteristics of the collected data. Differences were identified using
unpaired bilateral t-tests. A bilateral p-value of < 0.05 was considered statistically
significant in all analyses (α = 0.05). Pearsonʼs correlation coefficient was calculated
to show correlations between the variables “age”, “level of education”, “quality of
life”, “disease status” and “computer skills”. All calculations were based on the
assumption that data were normally distributed, and the Shapiro-Wilk test was used
prior to the evaluation of data to verify the normal distribution of data.
Results
Sociodemographic variables
[Table 1] shows the sociodemographic characteristics of the patient cohort. Mean age of the
patients was 56.68 years (minimum: 20 years, maximum: 85 years). 30 % of patients
had higher educational qualifications (entrance qualification for an advanced technical
college or for university), 42 patients (34 %) were working despite disease (at least
part time).
Table 1 Sociodemographic characteristics of the patient cohort.
|
Sociodemographic variables
|
Total
|
|
Age
|
|
|
|
56.68 (54)
|
|
|
12.38 (60 [20; 85])
|
|
Highest level of education achieved
|
|
|
|
n = 1 (1 %)
|
|
|
n = 31 (32 %)
|
|
|
n = 28 (29 %)
|
|
|
n = 15 (16 %)
|
|
|
n = 13 (14 %)
|
|
|
n = 8 (8 %)
|
|
Currently working
|
|
|
|
n = 11 (12 %)
|
|
|
n = 31 (32 %)
|
|
|
n = 43 (45 %)
|
|
|
n = 11 (11 %)
|
|
Disease/therapy status
|
|
|
|
n = 65 (68 %)
|
|
|
n = 31 (32 %)
|
Health-related quality of life and disease status
[Table 2] shows patientsʼ health-related quality of life and current state of health. The
median value for the patientʼs health status on the EQ VAS scale was 64.67, with 60
as the most commonly reported value. The median state of health using the EORTC QLQ-C30
was 56, while the mean score for quality of life for the overall patient cohort was
58.
Table 2 Quality of life and state of health of the total patient cohort.
|
|
EQ VAS scale (n = 96)
|
EORTC QLQ-C30 (current state of health) (n = 74)
|
EORTC QLQ-C30 (current quality of life) (n = 74)
|
|
|
|
Item value
|
Score value (in %)
|
Item value
|
Score value (in %)
|
|
Mean (median)
|
64.67 (70)
|
4.43 (4)
|
56.16 (50)
|
4.54 (5)
|
57.97 (66.67)
|
|
Standard deviation
|
18.15
|
1.32
|
23.56
|
1.31
|
23.50
|
|
Range (min; max)
|
90 (5; 95)
|
6 (1; 7)
|
99 (0;100)
|
6 (1;7)
|
99 (0;100)
|
Computer skills: previous experience of digital media
34 (35 %) patients stated that they had advanced or professional computer skills while
47 (49 %) patients reported having poor to moderate computer skills. Five patients
reported that they did not use either a computer or the internet. When asked about
their use of tablets, 33 patients (34 %) used or had used tablets, 33 (34 %) did not
use them, and 30 (31 %) did not specify their usage. [Table 3] shows the technology skills for the total patient cohort together with their disease-related
use of computers.
Table 3 Computer skills and willingness to complete electronic PRO questionnaires in the
total patient cohort.
|
Computer skills
|
Total
|
|
Computer skills (self-assessment by the patient)
|
|
Mean (standard deviation)
|
2.33 (0.75)
|
|
|
n = 10 (10 %)
|
|
|
n = 37 (39 %)
|
|
|
n = 30 (31 %)
|
|
|
n = 4 (4 %)
|
|
|
n = 15 (16 %)
|
|
Computer use (years)
|
|
Mean (median)
|
16.73 (15)
|
|
Standard deviation (range [min; max])
|
8.25 (34 [2;36])
|
|
Internet use (years)
|
|
Mean (median)
|
11.84 (10)
|
|
Standard deviation (range [min; max])
|
6.53 (24 [1;25])
|
|
Use of tablets
|
|
|
Mean (standard deviation)
|
1.91 (1.02)
|
|
|
n = 33 (34 %)
|
|
|
n = 10 (10 %)
|
|
|
n = 19 (20 %)
|
|
|
n = 4 (4 %)
|
|
|
n = 30 (31 %)
|
|
Could you imagine completing an electronic questionnaire on your subjective perception
of your own state of health?
|
|
|
n = 52 (55 %)
|
|
|
n = 35 (37 %)
|
|
|
n = 8 (8 %)
|
|
Do you think that the introduction of electronic surveys will …
|
|
|
n = 45 (47 %)
|
|
|
n = 11 (11 %)
|
|
|
n = 40 (42 %)
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … less suitable
(= 1), more suitable (= 5)
|
|
Mean (median)
|
3.34 (3)
|
|
Standard deviation
|
1.30
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … more tiring
(= 1), less tiring (= 5)
|
|
Mean (median)
|
3.22 (3)
|
|
Standard deviation
|
1.21
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … more difficult
(= 1), less difficult (= 5)
|
|
Mean (median)
|
3.06 (3)
|
|
Standard deviation
|
1.21
|
Willingness to use electronic PRO questionnaires (ePRO)
Patients were asked whether they could potentially imagine participating in electronic
PRO surveys, whether they were of the opinion that using electronic surveys to record
PRO would improve or worsen hospital care, and whether compared to paper-based questionnaires
electronic questionnaires were – in their view – more suitable or less suitable, more
exhausting or less exhausting, and more difficult or less difficult to complete ([Table 3]). Overall, slightly more than half of all participants reported that they could
imagine completing electronic surveys, while 37 % stated that they did not wish to
take part in such surveys. The question whether electronic surveys could have a positive
impact on care was answered in the affirmative by 45 (47 %) patients. No differences
between electronic questionnaires and paper-based questionnaires were found with regard
to suitability, how tiring it was to complete the survey, or the surveyʼs degree of
difficulty.
Correlations between willingness to use technology and the variables “age”, “level
of education”, “quality of life”, “health status” and “computer skills”
Patients were divided into one of two subgroups to identify possible relationships
between patientsʼ computer skills, disease status, health-related quality of life
and the sociodemographic factors “level of education” and “age” and patientsʼ willingness
to use electronic surveys. [Table 4] shows the statistical differences between patients who rejected electronic surveys
compared to those who were prepared to accept them. The patients in the subgroup who
were prepared to record their subjective perception of their state of health using
an electronic questionnaire were on average almost 9 years younger and had a higher
level of education (40 % either had an entrance qualification for an advanced technical
college or an entrance qualification for university). They also assessed the state
of their health (69.06 vs. 60.41 % and 60.09 vs. 53.79 %, resp.), their quality of
life (61.84 vs. 57.58 %) and their computer skills (2.56 vs. 1.94) as higher compared
to patients in the other subgroup. By comparison, patients who preferred a paper-based
survey were older (mean age: 62 years), had on average a lower level of education,
the state of their health was poorer and they had less previous experience with computers,
the internet, or tablets. The differences between the two groups were statistically
significant for the variables “age”, “level of education”, “state of health based
on the EQ VAS scale” and “computer skills”.
Table 4 Subgroup analysis according to the preferred method of survey (ePRO versus pPRO).
|
|
Electronic/tablet-based survey welcomed (n = 52)
|
Electronic/tablet-based survey not welcomed (n = 35)
|
Difference
|
95 % CI
|
p-value (α = 0.05)
|
|
|
Mean
|
SD
|
Mean
|
SD
|
|
|
|
|
Age
|
53.24
|
9.03
|
61.86
|
15.1
|
− 8.62
|
− 13.796; − 3.43
|
0.0014
|
|
State of health (EQ VAS)
|
69.06
|
17.35
|
60.41
|
18.13
|
8.65
|
0.73; 16.58
|
0.0327
|
|
State of health (EORTC QLQ-C30)
|
60.09
|
20.70
|
53.79
|
22.38
|
6.2998
|
− 5.14; 17.73
|
0.2747
|
|
Quality of life (EORTC QLQ-C30)
|
61.84
|
21.54
|
57.58
|
22.84
|
4.266
|
− 7.54; 16.08
|
0.4725
|
|
Level of education
|
3.44
|
1.05
|
2.6
|
0.88
|
0.84
|
0.41; 1.27
|
0.0002
|
|
|
n = 0
|
n = 1 (3 %)
|
|
|
|
|
|
n = 10 (19 %)
|
n = 19 (54 %)
|
|
|
|
|
|
n = 19 (37 %)
|
n = 9 (26 %)
|
|
|
|
|
|
n = 10 (19 %)
|
n = 5 (14 %)
|
|
|
|
|
|
n = 11 (21 %)
|
n = 1 (3 %)
|
|
|
|
|
|
n = 2 (4 %)
|
n = 0
|
|
|
|
|
Computer skills
|
2.56
|
0.66
|
1.94
|
0.7
|
0.61
|
0.29; 0.93
|
0.0003
|
|
|
n = 1 (2 %)
|
n = 7 (20 %)
|
|
|
|
|
|
n = 22 (42 %)
|
n = 14 (40 %)
|
|
|
|
|
|
n = 23 (44 %)
|
n = 6 (17 %)
|
|
|
|
|
|
n = 4 (8 %)
|
n = 0 (0 %)
|
|
|
|
|
|
n = 2 (4 %)
|
n = 8 (23 %)
|
|
|
|
Willingness correlated with state of health
The total patient cohort was divided into two subgroups according to the patientsʼ
assessment of their own state of health evaluated using the EQ VAS scale. Subgroup
1 consisted of all patients who – on a scale from 0 to 100 – had reported their state
of health as 60 or less; subgroup 2 consisted of patients in better health according
to their own assessment. As shown in [Table 5], the willingness to use electronic surveys was significantly lower for patients
in poorer health: only 40 % of the patients in this subgroup were willing to complete
a survey on the subjective perception of their own state of health electronically,
while 70 % of patients in better health were willing to do so. There were no other
significant differences between the two subgroups with respect to other surveyed items.
Table 5 Willingness to use electronic questionnaires: differences between subgroups.
|
|
State of health ≤ 60 (EQ VAS) (n = 40)
|
State of health > 60 (EQ VAS) (n = 50)
|
Difference
|
95 % CI
|
p-value (α = 0.05)
|
|
|
Mean
|
SD
|
Mean
|
SD
|
|
|
|
|
1 t-test; 2 χ2 test
|
|
Age1
|
54.81
|
13.81
|
56.72
|
10.58
|
− 1.92
|
− 7.02; 3.19
|
0.4574
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … less suitable
(= 1), more suitable (= 5)1
|
3.26
|
1.32
|
3.35
|
1.25
|
− 0.05
|
− 0.7; 0.6
|
0.8830
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … more tiring
(= 1), less tiring (= 5)1
|
3.27
|
1.15
|
3.11
|
1.24
|
0.16
|
− 0.46; 0.78
|
0.6034
|
|
Compared to a paper-based questionnaire, an electronic questionnaire is … more difficult
(= 1), less difficult (= 5)1
|
3.35
|
1.16
|
2.94
|
1.14
|
0.40
|
− 0.19; 0.995
|
0.1809
|
|
Computer skills1
|
2.19
|
0.79
|
2.45
|
0.72
|
− 0.26
|
− 0.61; 0.08
|
0.1347
|
|
|
n = 6 (15 %)
|
n = 4 (8 %)
|
|
|
|
|
|
n = 14 (35 %)
|
n = 20 (40 %)
|
|
|
|
|
|
n = 10 (25 %)
|
n = 19 (38 %)
|
|
|
|
|
|
n = 1 (3 %)
|
n = 3 (6 %)
|
|
|
|
|
|
n = 9 (23 %)
|
n = 3 (6 %)
|
|
|
|
|
Willingness to use technology-based surveys2
|
|
|
|
|
0.038
|
|
|
n = 16 (40 %)
|
n = 35 (70 %)
|
|
|
|
|
|
n = 18 (45 %)
|
n = 14 (28 %)
|
|
|
|
|
|
n = 6 (15 %)
|
n = 1 (2 %)
|
|
|
|
|
Do you think that the introduction of electronic surveys will …2
|
|
|
|
|
0.9144
|
|
|
n = 18 (45 %)
|
n = 25 (50 %)
|
|
|
|
|
|
n = 4 (10 %)
|
n = 6 (12 %)
|
|
|
|
|
|
n = 18 (45 %)
|
n = 19 (38 %)
|
|
|
|
Correlation analysis showed a moderate statistical correlation between the variable
“age” and the willingness to use electronic means to complete a survey (r = 0.321,
p = 0.002) but showed no significant correlation for any of the other variables ([Table 6]).
Table 6 Correlation analysis.
|
Variables
|
Correlation (Pearson)
|
Significance (α = 0,05)
|
|
Age vs. willingness to participate in an electronic survey
|
r = 0.321
|
p = 0.002
|
|
Level of education vs. willingness to participate in an electronic survey
|
r = 0.097
|
p = 0.348
|
|
State of health (EQ VAS) vs. willingness to participate in an electronic survey
|
r = − 0.006
|
p = 0.954
|
|
State of health (EORTC QLQ-C30) vs. willingness to participate in an electronic survey
|
r = − 0.084
|
p = 0.487
|
|
LQ (EORTC QLQ-C30) vs. willingness to participate in an electronic survey
|
r = − 0.022
|
p = 0.857
|
|
Computer skills vs. willingness to participate in an electronic survey
|
r = 0.116
|
p = 0.263
|
Discussion
In coming years, use of digital ePRO applications will become increasingly common
in research and thus also in routine clinical practice. The data collected in this
study show that, at present, it is primarily younger patients in better health who
are spontaneously willing to participate in electronic surveys while the barriers
to using electronic surveys are higher for older patients in a poorer state of health.
Almost half of the patients clearly had no idea what was meant by the term “tablet”.
This is in stark contrast to previous findings which had postulated that EPROs were
very feasible but without explicitly looking at existing computer skills [39], [40], [41], [42]. It could be that a need for support exists, but this has still to be substantiated
(publication in progress). Possible approaches could include training patients to
use the technology or support offered by study nurses or, in special cases, by members
of the patientʼs own family. The findings presented here expand the current understanding
of this issue. Oncologic studies have shown that electronic PRO reports are well received
by patients compared to paper-and-pencil versions when the assignment was randomized
[32], [35], [36], [37], [40]. However little attention has focused on the patientsʼ own preference for paper-based
or electronically-based questionnaires or on the acceptance of electronic questionnaires
if patients are free to choose between the two options. Schaeffeler et al. found that,
while levels of reliability and acceptance were high among patients with breast cancer,
the patientsʼ own preferences were not taken into consideration [40]. There were no previous studies of this type of patients with metastasis, and possible
correlations between socio-economic status or state of health and the willingness
to use ePRO were not much considered. This study offers some conclusions for clinical
practice which could help to improve PRO surveys in breast cancer patients with metastasis
or receiving adjuvant treatment. Thus, “age”, “level of education”, “state of health”
and “computer skills” have all been identified as variables which affect patientsʼ
willingness to use this form of survey. The results of the study emphasize the need
to take a detailed (social) history of patients as this will allow those patients
where the context and state of health indicate that there may be a barrier to using
electronic surveys to be identified in advance [51] and allow their need for support or preferences to be resolved early on. The findings
also emphasize the necessity of focusing on the user-friendliness of e-based surveys
and (after identifying the barriers) the importance of optimizing their ease of use.
The findings also offer some hints to supervising researchers or physicians about
the importance of taking individual needs and other influences into account, even
within the setting of research studies. Further studies will be necessary to elucidate
how to reach patients lacking a sufficient knowledge of German and with few or no
computer skills.
Conclusions for Practice
Currently, the majority of female patients with breast cancer would prefer ePRO surveys
to be done as part of routine clinical examination. Higher age and metastases were
identified as barriers to the prospective participation of patients in ePRO surveys.
If certain conditions with regard to age, educational level and current state of health
are present, support should be offered to ensure that patients are willing to participate
as this will underpin the validity of the survey. It would be useful to focus on the
ease of use of ePRO applications and design them to be more patient-oriented.
