Validation of the French translation-adaptation of the impact of cancer questionnaire version 2 (IOCv2) in a breast cancer survivor population


The IOCv2 questionnaire

The Impact of Cancer version 2 (IOCv2) is a 47-item questionnaire organized into 4
positive (altruism and empathy (AE), health awareness (HA), meaning of cancer (MOC),
positive self-evaluation (PSE)) and 4 negative (appearance concerns (AC), body change
concerns (BCC), life interference (LI) and worry (W)) impact dimensions 13] corresponding to the first 37 items. The questionnaire also includes 10 additional
items constituting conditional dimensions applicable to subsets of survivors assessing
employment concerns (EC), relationship concerns for individuals with a partner (P),
and relationship concerns for those without a partner (NP). All items are scored on
a five-point scale from 1?=?strongly disagree to 5?=?strongly agree. Each dimension
score is computed as the mean of the responses for the items constituting the dimension.
A higher score on a dimension implies stronger endorsement of that content area.

The original validation study of the IOCv2 in breast cancer survivors yielded high
factor loadings (0.58 to 0.94), high internal consistency reliability estimates (Cronbach’s
alpha coefficients between 0.78 and 0.99), and good discriminatory ability (Ferguson’s
delta statistic values between 0.91 and 0.99). The negative impact dimensions scores
were positively correlated with the Center for Epidemiologic Studies-Depression score.
They were also positively correlated with the Breast Cancer Prevention Trial Symptom
Checklist total score that assesses physical effects of medical interventions to prevent
and treat breast cancer.

Translation

The IOCv2 was originally translated into French by a project group within the European
Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Group using
standardized, forward-backward procedures 15]. This first translation of the IOCv2 was administered in April-May 2012 to a sample
of breast cancer survivors (n?=?371) followed in Gustave Roussy, a large cancer center
in Villejuif, France, but the psychometric analysis of this translation identified
two problematic items. First, the item “I do not take my body for granted since I
had cancer” had been translated to “Je ne considère plus mon corps comme quelque chose
d’acquis depuis le cancer”. But the analysis revealed that this item was weakly correlated
with the score of its dimension Health Awareness and moderately correlated with the
scores of Worry and Body Change Concerns. It was assumed that the problem lay in the
translation of the idiom “take for granted”. With the aim of being consistent with
the concept measured through the original dimension, the item was reformulated into
a statement exploring the concern about loss of confidence in the body: “Depuis le
cancer je ne fais plus la même confiance à mon corps”.

Second, the item, “I consider myself to be a cancer survivor”, was initially translated
into “Je me considère comme un survivant du cancer”, but this item was weakly correlated
with its initial dimension of Positive Self-Evaluation in the French version. This
item was then submitted to an internet focus group of after-cancer patients, thanks
to the contribution of a dedicated cancer website, cancercontribution.fr. In French,
surviving evokes mainly a feeling of loss and trauma due to a catastrophic event,
and the word “survivant” in French is rather empty of pride or honor. As this item
is part of the positive impact domain of IOC, we adopted a new formulation, “La traversée
du cancer m’a rendue plus forte”, to express that the experience of cancer makes the
subject stronger. This formulation avoids focussing on negative aspects and emphasizes
the positive aspect of having been confronted with cancer.

The first French translation of IOCv2 has thus been modified by integrating these
two adapted items (see Appendix A). This study examines the psychometric properties
of this second version of the French translation and adaptation of the IOCv2 in a
breast cancer survivor sample.

Patient sample

Female breast cancer survivors were recruited at Gustave Roussy, a large national
comprehensive cancer centre, where approximately 3000 women are treated annually for
breast cancer. The CANTO project including the validation study of the French translation-adaptation
of the IOCv2 was approved by a research ethics committee (CPP n°11-039, Kremlin Bicetre)
in October 2011. Patients attending their follow-up consultation were approached by
a psychologist in April-May 2013. Patients were eligible if they were at least 18 years
old, and were diagnosed with breast cancer more than 1 year earlier. Patients with
mental disorders or unable to fill in the questionnaire without assistance were excluded.
All consecutive eligible patients were invited to participate in the study. After
providing written, informed consent, the women were asked to complete a series of
questionnaires including the IOCv2, the MOS SF-12, the PostTraumatic Growth Inventory
(PTGI) and the Fear of Cancer Recurrence Inventory (FCRI). Sociodemographic and clinical
information were also reported by the patients. Questionnaires were completed in the
waiting room of the outpatient clinic.

Measures

The SF-12 16], 17] is an abbreviated version of the SF-36. The information from the 12 items is summarized
in physical (PCS) and emotional (MCS) component summary scores. Both scores are standardized
(mean?=?50, SD?=?10) to the 1998 general U.S. population. A higher score indicates
a better health. The high correlations between SF-12 and SF-36 summary scores (between
0.94 and 0.96 in France) support good reproduction of the SF-36 summary scores by
the SF-12 17]. The Cronbach’s alpha coefficients for the physical (PCS) and emotional (MCS) component
summary scores reach 0.89 showing good reliability in our sample.

The PostTraumatic Growth Inventory (PTGI) 18] measures significant positive change with 21 items and is composed of five dimensions:
relating to others, new possibilities, personal strength, spiritual change and appreciation
of life and a total score. All scores result from the sum of the responses of the
corresponding items. A higher score on a dimension implies a higher change encountered
in this domain. The Cronbach’s alpha coefficients in our sample for the PTGI dimensions
range from 0.69 to 0.93.

The Fear of Cancer Recurrence Inventory (FCRI) 19] is a 42-item scale assessing 7 dimensions of the fear of cancer recurrence (FCR),
such as potential triggers activating FCR, the presence and severity of intrusive
thoughts associated with FCR, psychological distress and functioning impairments as
potential consequences of FCR, self-criticism towards FCR, and a variety of coping
strategies that can be used to cope with and may influence FCR. All scores result
from the sum of the responses of the corresponding items. A higher score indicates
increased higher levels of FCR or of other constructs associated with the FCR. The
validation study of the FCRI 19] shows good internal consistency and correlations between FCRI and three questionnaires
assessing fear of cancer recurrence (between 0.68 to 0.78) and between FCRI and two
questionnaires assessing psychological distress (between 0.43 to 0.66) support construct
validity. The Cronbach’s alpha coefficients in our sample for the FCRI dimensions
range from 0.71 to 0.95.

Statistical analysis

Confirmatory factor analysis (CFA) based on the original IOCv2 13] 8-factor solution (reflecting the 8 hypothesized positive and negative impact dimensions
that form the core of the questionnaire) was performed using covariance-based structural
equation modelling with maximum likelihood estimation. Good (acceptable) fit was indicated
by the following criteria: RMSEA???0.05 (0.08), SRMR???0.05 (0.10), CFI???0.97 (0.95)
and NNFI???0.97 (0.95) 20]. Additional exploratory CFA’s were also conducted to evaluate the 3 conditional dimensions
(employment and relationship concerns, with or without a partner) and to explore an
higher-order factor structure, a positive domain consisting of the dimensions Altruism
and Empathy, Health Awareness, Meaning Of Cancer and Positive Self-Evalutation and
a negative domain consisting of the dimensions Appearance Concerns, Body Change Concerns,
Life Interferences and Worry.

Convergent validity of each dimension was evaluated by examining the item-rest score
correlations between the items and the rest scores of its hypothesized dimension (i.e.
the score computed from the items of the dimension deleting that item). The convergent
validity of the dimension was considered good when more than 90 % of the items of
the dimension had an item-rest score correlation greater than 0.4. This indicates
that the items composing the dimension are likely to be related to the same construct.
Divergent validity was evaluated by examining the item-score correlations between
the items and the scores of the other dimensions. The divergent validity of the dimension
was considered good when more than 80 % of the items of the dimension had an item-rest
score correlation with its own dimension higher than the item-score correlations with
the other dimensions. This indicates that the items composing the dimension are not
likely to be related to another construct.

Concurrent validity was evaluated by forming a priori hypotheses about patterns of association between the IOCv2 scores and the SF-12,
PTGI and FCRI scores. We expected that the PTGI subscale scores would be correlated
with the IOCv2 positive dimensions whereas some FCRI subscale scores (i. triggers,
severity and psychological distress, and ii. functioning impairments) would be correlated
with the negative ones (i. Worry, and ii. Body Change Concerns and Life Interference
respectively). As the IOCv2 measures a somewhat different concept than health status,
we expected only moderate associations between IOCv2 scores and SF-12 summary scores
(correlation coefficient’s absolute value between 0.3 and 0.5). The correlation coefficients
between dimensions of the IOCv2 were also examined. Positive dimensions (negative
dimensions respectively) were expected to be highly positively correlated with one
another.

Internal consistency reliability of the IOCv2 dimensions was assessed using the Cronbach’s
alpha coefficient 21]. Dimensions were considered reliable if ???0.70. The ability of each dimension to
discriminate among individuals was evaluated with the Ferguson’s ? statistic 22], 23]. Loevinger’s H scalability coefficients 24] evaluate the scalability, i.e. the quality of the scale as a unidimensional cumulative
scale and the degree to which the set of items is consistent within a dimension. The
scales’ and items’ Loevinger’s coefficients (H and H
i
respectively) were considered acceptable if H??0.3 (high degree of homogeneity of
the set of items) and H
i
??0.3 for all i (item consistent with the set of items). All analyses were performed
with Stata 12 (Stata Corp, College Station, Texas). CFA was carried out using LISREL
8.80.

Differences in IOCv2 dimensions as a function of sociodemographic and treatment characteristics
(age, time since diagnosis, treatment and partner status) were examined using Pearson
correlation coefficients and Student’s t tests. IOCv2 scores were expected to vary across treatments, age and time since diagnosis.
These additional elements of validity are to be compared with the results on the original
IOCv2 validation study 13] where the correlation coefficients of the IOCv2 scores with age were weakly negative.
No association was observed between the IOCv2 scores and time since diagnosis or hormonal
therapy. IOCv2 scores were significantly higher for all dimensions for patients who
had received chemotherapy. Finally, Altruism and Empathy, Health Awareness and Positive
Self-Evaluation were scored higher for patients with a partner.