ARTICLE
In recent years, improvements in the efficacy of therapeutic methods
in cancerology have been associated with stronger and stronger forms of
treatment which combine surgery, chemotherapy and radiotherapy, and which
often cause significant levels of toxicity. For the patient, the results
of the treatment received are measured not only in terms of their clinical
efficacy, the period of remission and the patient's life expectancy; they
are also measured in terms of their repercussions on daily life, as much
from the point of view of physical activities and psychological state
as from that of socio-professional relationships. The quality of life,
therefore, is a criterion which must be included in the evaluation of
therapeutic strategies, in particular during clinical trials and in the
estimation of the cost-utility ratio of these strategies.
Modifications in the quality of life result from the impact of the disease
and its form of treatment, and these two factors are felt in different
ways from one patient to another. Different instruments generally,
questionnaires filled out by the patients have been developed and
validated to obtain reliable measures of the quality of life of the patients,
but the vast majority of these instruments are in English. These days,
medical research and especially therapeutic trials are conducted in several
centres on an international scale. It is therefore necessary that validated
instruments be available in several languages. When we began our study,
there were very few scales on the quality of life which were internationally
recognised and validated in French, apart from the Nottingham Health Profile
(NHP) [1] and the QLQ-Q30 of the EORTC [2].
The former is an evaluation instrument which is not specific to a particular
disease and is therefore less discriminating. Only the QLQ-C30 has been
developed for cancer, but the interpretation of the results obtained is
sometimes difficult because it has no overall score. We thought it would
be valuable to have at our disposal an instrument which is easy to use,
which has validated psychometric qualities and which is adapted to French
culture.
From among the main instruments which were specifically developed for
cancer (Rotterdam Symptom Checklist [3], Quality of Life Index [4, 5],
Profile of Mood States [6]), we chose the Functional Living Index
Cancer (FLIC) developed by Schipper et al. [7] at the Manitoba
Cancer Treatment and Research Foundation, which is one of the most frequently
used instruments in the English-speaking world [8-10]. The FLIC explores
all the aspects associated with the quality of life: physical well-being,
psychological state, family and social relationships, and physical discomfort.
The quality of a measuring instrument lies mainly in its validity. The
process of validation draws on a complex procedure, the aim of which is
to verify the construct validity, internal consistency, relevance, reliability
and acceptability of the instrument [11-14]. Our paper presents the stages
of adaptation (translation and validation) of the French version of the
FLIC on the basis of the results observed in a sample of 200 patients
suffering from cancer.
Results
Description of the patients
From February to June 1994, 207 patients agreed to participate in this
study and 10 refused to fill in the questionnaires. Only 200 were kept
for analysis (88 in Besançon and 112 in Lille) as 7 of the FLIC
questionnaires had more than 4 answers missing.
The patients were aged between 31 and 88, with an average age of 57
(SD = 12 years). Seventy-one percent were women and 77% were married.
Twenty-six different tumour sites were observed, the most important being
breast cancer (47%) and colon-rectum cancer (10%). One hundred and twenty-four
patients (62%) were undergoing initial treatment and 126 (63%) were receiving
only chemotherapy. Only 5% of the patients were hospitalised and 50% had
a performance status equal to zero.
Acceptability of the questionnaire
The number of patients who refused to participate in the study was small
(10/217 = 4.5%), as was that of the number of unworkable questionnaires
(7/217 = 3.4%). Out of the 200 questionnaires analysed, 38 (17.5%) were
incompletely filled in, with an average of 1.7 unanswered items, which
usually were questions 10, 13, 20, 21 and 22. It took an average of 11
minutes (SD = 7 minutes) to fill in the FLIC questionnaire. This is comparable
to the time taken to fill in the QLQ-C30 of the EORTC, which is 9 minutes
(SD = 6 minutes). Approximately 10% of the patients reported that one
or several of the questions were unclear or difficult to answer and only
7% of them asked for complementary explanations. Some patients (3%) indicated
that certain questions were disturbing or indiscreet. Although 97 patients
expressed a preference for the QLQ-C30, which is in the form of a Likert
scale, only 23 patients preferred the FLIC. But the vast majority were
very happy to have the chance to express their own experience of the illness
and its treatment.
Internal consistency and construct validity
A factor analysis showed that 5 factors explain 62% of the overall variation
between the subjects. To facilitate interpretation, only values greater
than 0.40 are presented in table
I. Only one question was not associated with a single factor as
its loading was greater than 0.40 with respect to 2 factors. Seven items
were loaded on factor 1, which accounted for 55% of common variance. These
items appear to relate to activities of daily life living and physical
functioning. Five items loaded on the second factor and corresponding
to psychological functioning, accounted for 14% of the common variance.
The final 3 factors extracted accounted for approximately 10% of common
variance each. Respectively 5, 3 and 2 items are associated to factor
3 (current well-being), factor 4 (gastro-intestinal symptoms) and factor
5 (social functioning).
Cronbach's alpha coefficient is equal to 0.90 for the total number of
patients. Cronbach's alpha coefficients corresponding to the 5 factors
are indicated in table II.
The partial coefficients vary from 0.892 to 0.905.
Concurrent validity
The overall FLIC score is statistically correlated with all the scores
of the QLQ-C30 of the EORTC. Spearman's rank correlation coefficients
vary from 0.20 to 0.68 (table
III).
The highest values are obtained by the overall quality of life score
(r = 0.68) and the scores associated with fatigue
(r = 0.63) and social functioning (r = 0.63). The multitrait-multimethod
correlation matrix, based on the dimension of the FLIC defined by King
et al. [15] and the scores of QLQ-C30, is presented in table
III.
Clinical relevance
The mean FLIC scores according to the demographic and clinical characteristics
of the patients are presented in table
IV. There is a significant correlation between the tumour sites
(breast cancer as opposed to other cancers) and the overall FLIC score
(p = 0.03). Performance status, the main prognostic factor, is also correlated
with the overall FLIC score (p = 0.05).
The results of the Schipper factorial analysis show the existence of
a factor associated with physical well-being (PHYS). This factor corresponds
to 9 questions (4, 6, 7, 10, 11, 13, 15, 20 and 22). The correlation between
performance and PHYS scale score is significant (p = 0.012). In our study,
7 out of 9 items (4, 6, 7, 10, 13, 15 and 22) corresponded to this factor,
which also correlated significantly with the performance status (p = 0.0045).
Discussion and conclusions
The aim of this work was to develop a French version of the FLIC and
validate its psychometric characteristics among patients with different
tumour sites, including a high proportion of breast cancer.
The questionnaire was very well accepted by the patients, who took an
average of 11 minutes to fill it in without any particular difficulty.
A majority (56%) treated the answer line like an analogue scale. This
indicated that they had understood the concept of the measures of the
questionnaire, contrary to what Ganz et al. [9] observed during
a clinical trial carried out at random among patients suffering from lung
cancer with metastases. In this trial, a number of patients had some difficulty
in using the analogue scale correctly. They answered as though it were
a Likert scale.
The reliability of the French version of the FLIC is very satisfying
since the values of Cronbach's alpha coefficient are equal to 0.9. The
factorial analysis reveals 5 factors which represent 62% of total variance.
Our results are in agreement with those reported by Morrow et al.
[16], King et al. [15] and Schipper et al. [7] in his original
study, and confirm convergent validity for the global, rôle, emotional,
pain and nausea dimensions, but not the social dimension.
The correlations of the overall FLIC score with the different scores
of the QLQ-C30 are all significant. The highest value of the correlation
coefficient observed in the Global Health Status scale confirms that the
FLIC is an instrument which measures the overall functional quality of
life of patients suffering from cancer.
While it is necessary to ensure the acceptability and the structural
coherence of an instrument for measuring the quality of life, the essential
characteristic which these tools must have, if they are to be used for
evaluating therapeutic methods, is clinical relevance. Logical correlations
between the measures of the quality of life and the clinical characteristics
of the patients, particularly performance status, are indispensable. The
results obtained indicate that: a) the worse the performance status,
the lower the overall FLIC score; and b) breast cancer patients
have a higher score than the total number of patients with other localisations.
Most validation work on questionnaires on the quality of life shows
a correlation between the measures of the quality of life and the performance
status of the patient as evaluated by Karnofsky's index (KPS). We used
the WHO index, a shortened version of the KPS in five levels, suggested
by the Eastern Cooperative Oncology Group (ECOG). It is now used routinely
and in clinical trials instead of the KPS. This scale is less discriminating
than the KPS, especially when the population is heterogeneous. Moinpour
et al. [17], on the other hand, in her review of the literature,
indicates that the psychometric properties of the WHO index are inadequately
documented. Cella et al. [18] finds a moderate correlation between
the WHO index and the measures of the quality of life according to his
questionnaire (FACT-G2), which does not allow for any differentiation
between grades 2, 3 and 4. In Ganz's [9] trial, no significant relation
between the KPS and the quality of life measured by the FLIC was observed.
Furthermore, since the KPS is given by the doctor whereas the quality
of life is measured by the patients, the two assessments can be different.
Mercier et al. [19] also observed a discrepancy between the doctor's
and the patient's notation, which was greater for high low KPS values.
General questionnaires on the quality of life developed for cancer patients
have been evaluated for the most part with heterogeneous samples patients
who have one or several localisations at different stages of the disease
[2, 7, 19, 20]. This approach may seem more favourable to a generalisation
of the results, but, by increasing the variability of the measures, it
can influence the reliability and the validation of the instruments and,
in particular clinical validation. Scales of quality of life have been
developed mainly for clinical trials where the patients are selected according
to precise criteria and from a homogeneous group. It seems desirable,
therefore, that they be validated using several homogeneous sample groups
with different tumour localisations. This approach would make it possible
to assess the psychometric characteristics in the conditions where the
questionnaires are used and would facilitate the simultaneous evaluation
of specific modules. In fact, the current tendency in cancerology is to
combine a general questionnaire, whatever the type of cancer or the treatment,
with one or several specific questionnaires linked to the tumour localisation
or its form of treatment [19, 21, 22]. Measurement of the quality of life
is now considered to be a significant end point, the study of which is
as important as the study of life expectancy in clinical trials. In addition,
some studies [10, 23], have shown that the quality of life is a factor
in survival prognosis. Ganz [24] even suggests that it should be preferred
over performance status when a therapy is prescribed, because it would
be a better reflection of the disease status.
This poses the problem of obtaining an overall score which integrates
the different fields explored by the questionnaires and of deciding what
weight should be attributed to each of them. As is the case with the FLIC,
most instruments propose an unweighted overall score since choosing appropriate
weighting is very difficult in practice [25]. Certain questionnaires,
such as the QLQ-C30 of the EORTC, do not have an overall score, which
can sometimes make the comparison between several therapeutic methods
difficult, since the results of the scores associated with the different
fields explored do not allow us to conclude in favour of one of the forms
of treatment.
The information obtained on the basis of the measurement of the quality
of life is useful for the doctor in his daily practice and necessary in
the field of clinical research. It is therefore indispensable to collect
the data with great rigour and to use reliable and relevant instruments.
The results of our work show that the French version of the FLIC has
characteristics which are comparable to those of the original instrument,
and that it is a questionnaire filled in by the patient which is both
valid and reliable for measuring the overall quality of life of cancer
patients. Its reliability and relevance in clinical trials and longitudinal
studies must, however, be confirmed.
Patients and methods
Development of a French version of the questionnaire
The French version of the FLIC was developed according to the classic
method of translation and back-translation without changing the number
of questions [12]. The concepts expressed in the original questionnaire
had to be found in the translated questionnaire. The questions and suggested
answers had to be simple and unambiguous so that they could easily be
understood by everybody.
Firstly, the original English version was translated into French
by two groups, each of which consisted of two independent French-speaking
translators, one of whom was not advised of the problem and the characteristics
which were being looked for. After discussion, we obtained one translocation
for each group.
Two versions were then translated into English by two independent
English-speaking translators. These back-translations were then compared
to the original version by bilingual experts.
However, because of different cultural notions and linguistic
expressions, certain questions were difficult to translate either because
of the language or because of the concept. In order to find the most appropriate
turn of phrase, certain expressions were submitted to the opinion of about
twenty patients of different socio-cultural backgrounds, age and sex in
the course of a semi-structured interview with a psychologist.
A procedure whereby each item was progressively adjusted produced
a final French version which conformed the most to the concepts and the
expressions of the original version (see appendix).
Study patients and administration of the questionnaires
The French version of the FLIC was validated among patients suffering
from cancer and undergoing treatment in cancer wards in two hospitals:
the Centre Oscar-Lambret in Lille and the Centre hospitalier régional
in Besançon.
The QLQ-C30 questionnaire of the EORTC was given to each patient at
the same time as the FLIC questionnaire. The two questionnaires were filled
in in an arbitrary order on the day which corresponded to at least the
second session of chemotherapy or the second week (or sixth session) of
radiotherapy. A member of the medical team responsible for the study gave
the patients any explanations which were necessary for filling in the
questionnaires correctly, and they were handed in the same day to this
same person.
Information concerning socio-demographic factors (age, sex, marital
status) and clinical parameters (nature of the treatment, stage of the
disease, period of time passed since the beginning of treatment, performance
status evaluated by the WHO index) of each patient were also collected.
Form of the questionnaires
The FLIC questionnaire contains 22 questions in the form of a Likert
scale. The patient answers each question by tracing a vertical line which
corresponds the most to his state in relation to two extreme situations.
The score given is equal to the whole number (between 1 and 7) which is
closest to the line.
The overall score is obtained by adding the sum of the non-weighted
scores of the 22 items (taking into account positive and negative questions).
High numbers correspond to a better quality of life.
The QLQ-C30 of the EORTC contains 30 questions divided into 15 scales
consisting of one or several questions, the scores of which are expressed
by a number between 0 and 100. A high score in the five functional scales
and the overall quality of life scale represents a high level of functioning,
and a high score in the nine symptom scales represents a high level of
symptoms.
Validation of the French version of the FLIC
FLIC questionnaires, missing five or more answers (that is, 20% or more
of the total number of items) were excluded from the study. In other incomplete
FLIC questionnaires, the value for a missing item was replaced by the
average score of the patient for all the other items.
Acceptability of the tool, which reflects its validity as perceived
by the patient, was evaluated qualitatively by the remarks made by the
patients and quantitatively by the time taken to fill in the questionnaire
and the percentage of missing values.
Internal consistency, which allows the reliability of the scale
to be evaluated when only one measure per subject is available, was evaluated
by using Cronbach's alpha coefficient [26]. The partial coefficients,
obtained by taking out one of the 22 items, were calculated.
Construct validity, established conceptually by experts, was
studied by factorial analysis (FA) (after Varimax orthogonal rotation).
The more limited the number of factors retained, and the more each factor
can be associated with a particular field of the quality conceptually
expected in the questionnaire, the more coherent the internal structure.
Concurrent validity was estimated by the correlation between
the overall score of the FLIC questionnaire and the different scores of
the QLQ-C30 of the EORTC by using Spearman's rank correlation coefficients.
The QLQ-C30 was the only questionnaire in French specific to cancer.
Discriminative validity (also called predictive validity) was
expressed by the correlations between the measures of the scale (the overall
FLIC scores) and the prognostic variables of the disease (clinical parameters,
especially the performance status) by using a multiple factor variance
analysis (ANOVA). This reflects the clinical relevance of the instrument
which measures the quality of life.
Different statistical tests were conducted with a risk alpha = 5% according
to a bilateral approach using the Statistical Analysis System (SAS version
6.10) program.
REFERENCES
1. Mac Ewen J. The Nottingham health profile. In: Lancaster
SR, Walker and RM Rosen, eds. Quality of life assessment and application.
England MTP Press Ltd, 1987.
2. Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A, Duez
NJ, et al. The European Organisation for Research and Treatment
of Cancer QLQ-C30: a quality of life instrument for use in international
trials in oncology. J Natl Cancer Inst 1993; 85: 365-76.
3. De Haes JCJM, Van Knippenberg FCE, Neijt JP. Measuring psychological
and physical distress in cancer patients: structure and application of
the Rotterdam Symptom Checklist. Br J Cancer 1990; 61: 1034-8.
4. Padilla GU, Presant C, Grant MM, Metter G, Lipsett J, Heide
F. Quality of life index for patients with cancer. Res Nursing Health
1993; 6: 117-26.
5. Spitzer WO, Dobson JA, Hall J, Chesterman E, Levi J, Sheperd
R, et al. Measuring the quality of life of cancer patients. J
Chron Dis 1981; 34: 585-97.
6. Shumaker SA, Anderson RT, Czajkowski SM. Psychological tests
and scales. In: Spilker B. Quality of life assessment in clinical
trials. New York: Raven Press, 1990; 94-113.
7. Schipper H, Clinch J, Mac Murray A, Levitt M. Measuring the
quality of life of cancer patients: the functional living index
cancer: development and validation. J Clin Oncol 1984; 2: 472-83.
8. Cassileth BR, Lusk EJ, Guerry D, Blake AD, Walsh W, Kascius
L, et al. Survival and quality of life among patients receiving
unproven as compared with conventional cancer therapy. N Engl J Med
1991; 324: 1180-5.
9. Ganz PA, Haskell CM, Figlin RA, La Soto N, Siaus J. Estimating
the quality of life in a clinical trial of patients with metastatic lung
cancer using the Karnofsky performance status and the functional living
index cancer. Cancer 1988; 61: 849-56.
10. Ganz PA, Lee JJ, Siau J. Quality of life assessment. An independent
prognostic variable for survival in lung cancer. Cancer 1991; 67:
3131-5.
11. Coste J, Fermanian J, Venot A. Methodological and statistical
problems in the construction of composite measurement scales: a survey
of six medical and epidemiological journals. Statistics in Medicine
1995; 14: 331-45.
12. Guillemin F, Bombardier C, Beaton D. Cross-cultural adaptation
of health-related quality of life measures: literature review and proposed
guidelines. J Clin Epidemiol 1993; 46: 1417-32.
13. Moret L, Mesbah M, Chwallow J, Lellouch J. Internal validation
of a scale: relationship between principal component analysis, Cronbach's
alpha and reliability. Epidem and Pub Health 1993; 41: 179-86.
14. Van Knippenberg FCE, De Haes JCJM. Measuring the quality
of life of cancer patients: psychometric properties of instruments. J
Clin Epidemiol 1988; 41: 1043-53.
15. King MT, Dobson AJ, Harnett PR. A comparison of two quality
of life questionnaires for cancer clinical trials: the Functional Living
Index Cancer (FLIC) and the quality of life questionnaire core
module (QLQ-C30). J Clin Epidemiol 1996; 49: 21-9.
16. Morrow GR, Lindke J, Black P. Measurement of quality of life
in patients: psychometric analysis of the Functional Living Index
Cancer (Flic). Quality of Life Res 1992: 1: 287-96.
17. Moinpour CM, Feigl P, Metch B, Hayden KA, Meyskens FL, Crowley
J. Quality of life end points in cancer clinical trials: review and recommendations.
J Natl Cancer Inst 1989; 81: 485-95.
18. Cella DF, Tulsky DS, Gray G, Sarafian B, Linn E, Bonomi A,
et al. The functional assessment of cancer therapy scale: development
and validation of the general measure. J Clin Oncol 1993; 11: 570-9.
19. Mercier M, Schraub S, Bourgeois P. Quality of life. Lancet
1987; 8551:161-2.
20. Osoba D, Zee B, Pater J, Warr D, Kaizer L, Latreille J. Psychometric
properties and responsiveness of the EORTC quality of life questionnaire
(QLR-C30) in patients with breast, ovarian and lung cancer. Quality
Life Res 1994; 3: 353-64
21. Aaronson NK, Cull A, Kaasa S. The EORTC modular approach
to quality of life assessment in oncology. Int J Ment Health 1994;
23: 75-96.
22. Bjordal K, Kaasa S. Psychometric validation of the EORTC
core quality of life questionnaire, 30-items version and a diagnosis
specific module for head and neck cancer patients. Acta Oncologica
1992; 31: 311-21.
23. Coates A, Gebski V, Signorini D, Murray P, Mac Neil D, Byrne
M, et al. Prognostic value of quality of life scores during chemotherapy
for advanced breast cancer. J Clin Oncol 1992; 10:1833-8.
24. Ganz PA. Quality of life and the patient with cancer. Individual
and policy implications. Cancer 1994; 74: 1445-52.
25. Hays RD, Stewart AL, Sherbourne CD, Marshall GN. The states
versus weights dilemma in quality of life measurement. Quality
of Life Res 1993; 2: 167-8.
26. Cronbach LJ. Coefficient alpha and the internal structure
of tests. Psychometrika 1951; 16: 297-334.
|
Printable version
