questionnaire [8, 14, 19]. Finally, in our study in particular,
the two analyses were not performed at the same level of
the questionnaire’s construct. The CTT analyses were
performed at the scales’ level as well as at the domains’
level, whereas the IRT analyses were only performed at the
domains’ level. The problems identiﬁed by the CTT anal-
yses concerned the scales structure of the questionnaire.
Similar CTT analyses at the domains’ level did not ﬁnd
default in the questionnaire’s construct. Nevertheless, the
items misﬁt observed from the Rasch analysis (items 10, 32
and 34) broadly corresponded to anomalies observed from
the multitrait scaling analysis conducted for the three
domains, in CTT. Indeed, a more thorough examination
brought to light the items 10, 32 and 34, which presented
the smallest correlation coefﬁcients between items with
their own domain (with an important difference with the
correlation coefﬁcients of the other items). These misﬁts,
slightly present in the CTT, were identiﬁed more clearly in
the IRT analyses. Thus, the results of the CTT analysis and
the IRT analysis of the OUT-PATSAT35 mostly agreed. A
good internal consistency of the domains was found by the
Moreover, the results of the dimensionality or factor
structure assessment, using both CTT and IRT analyses,
differ. However, caution should be also exercised when
comparing the results of the EFA in CTT analysis and the
results of the residuals analysis in IRT. Indeed, the aim of
the EFA computed on a whole questionnaire is to assess
dimensionality of the whole questionnaire (to identify
factors within a correlation matrix), whereas for IRT ana-
lysis computed on each domain, the aim is to identify
whether multidimensionality exists in the residuals once
the unidimensional structure has been removed in each
domain (and not the whole questionnaire) and whether the
responses to the item pool can be explained by a multidi-
mensional latent trait [34, 35]. On another point, the IRT
identiﬁed more clearly problems in the structure of the
OUT-PATSAT35 questionnaire where none was identiﬁed
by the CTT, even in the previous studies, such as the
redundancy between items, problems in the response cat-
egory and misﬁtting items. Thus, the IRT offers many
advantages over CTT, mostly in the development, reﬁne-
ment and evaluation of a questionnaire or a reduced form,
but without replacing the CTT.
The ﬁndings of this study indicate that a reﬁnement of the
questionnaire could be necessary. One may wonder how
misﬁtting items should be treated. Removing items from the
domains, which do not ﬁt the IRT model, is advocated and
often may not have impact on the measurement properties of
the questionnaire or improve it [34, 36, 37]. Furthermore,
studies in the Rasch literature have investigated the impact
on clinical utility of the questionnaire to remove these mis-
ﬁtting items (impact on item locations, on the ability of the
revised questionnaire to detect signiﬁcant changes in scores
between different patient groups in clinical trial) [34, 38]. No
signiﬁcant impact was found, but caution was advised.
To conclude, the OUTPAT-SAT35 questionnaire does
not meet expectations of the IRT measurement model. The
Rasch analysis revealed misﬁtting and redundant items.
Taking the above problems into consideration, it could be
interesting to reﬁne the questionnaire in a future study.
Acknowledgments The authors thank all physicians from the cen-
ters participating in the study who agreed to invite patients to par-
ticipate in this study. We thank the clinical research assistants in the
two centers who participated in the data collection. This work was
supported by the Regional French Hospital Clinical Research
Conﬂict of interest The authors have no potential conﬂict of
1. Burke, L. (2006). Guidance for industry: Patient-reported out-
come measures: Use in medical product development to support
labeling claims: draft guidance. Health and Quality of Life Out-
2. Rubin, H. R., Gandek, B., Rogers, W. H., Kosinski, M.,
McHorney, C. A., & Ware, J. E., Jr. (1993). Patients’ ratings of
outpatient visits in different practice settings. Results from the
Medical Outcomes Study. JAMA, 270(7), 835–840.
3. Borras, J. M., Sanchez-Hernandez, A., Navarro, M., Martinez,
M., Mendez, E., Ponton, J. L., et al. (2001). Compliance, satis-
faction, and quality of life of patients with colorectal cancer
receiving home chemotherapy or outpatient treatment: A ran-
domised controlled trial. BMJ, 322(7290), 826.
4. Bredart, A., Bottomley, A., Blazeby, J. M., Conroy, T., Coens, C.,
D’Haese, S., et al. (2005). An international prospective study of
the EORTC cancer in-patient satisfaction with care measure
(EORTC IN-PATSAT32). European Journal of Cancer, 41(14),
5. Poinsot, R., Altmeyer, A., Conroy, T., Savignoni, A., Asselain,
B., Leonard, I., et al. (2006). Multisite validation study of ques-
tionnaire assessing out-patient satisfaction with care question-
naire in ambulatory chemotherapy or radiotherapy treatment.
Bulletin du Cancer, 93(3), 315–327.
6. Arraras, J. I., Illarramendi, J. J., Viudez, A., Lecumberri, M. J., de
la Cruz, S., Hernandez, B., et al. (2012). The cancer outpatient
satisfaction with care questionnaire for chemotherapy, OUT-
PATSAT35 CT: A validation study for Spanish patients. Sup-
portive Care in Cancer, 20(12), 3269–3278.
7. Arraras, J. I., Rico, M., Vila, M., Chicata, V., Asin, G., Martinez,
M., et al. (2010). The EORTC cancer outpatient satisfaction with
care questionnaire in ambulatory radiotherapy: EORTC OUT-
PATSAT35 RT. Validation study for Spanish patients. Psych-
ooncology, 19(6), 657–664.
8. Hays, R. D., Morales, L. S., & Reise, S. P. (2000). Item response
theory and health outcomes measurement in the 21st century.
Medical Care, 38(9 Suppl), II28–II42.
9. Nguyen, T. V., Bosset, J. F., Monnier, A., Fournier, J., Perrin, V.,
Baumann, C., et al. (2011). Determinants of patient satisfaction in
ambulatory oncology: A cross sectional study based on the OUT-
PATSAT35 questionnaire. BMC Cancer, 11, 526.
Qual Life Res