Content uploaded by Paolo Riva
Author content
All content in this area was uploaded by Paolo Riva on Oct 23, 2017
Content may be subject to copyright.
Digestive and Liver Disease 42 (2010) 718–723
Contents lists available at ScienceDirect
Digestive and Liver Disease
journal homepage: www.elsevier.com/locate/dld
Liver, Pancreas and Biliary Tract
A short version of a HRQoL questionnaire for Italian and Japanese patients
with Primary Biliary Cirrhosis夽
Lorenzo Montalia,1, Atsushi Tanakab,1, Paolo Rivaa, Hiroki Takahashic, Claudio Cocchid,
Yoshiyuki Uenoe, Massimo Migliorettia, Hajime Takikawab, Luca Vecchioa, Alessandra Frigerioa,
Ilaria Bianchif,g, Roberta Jorgensenh, Keith D. Lindorh, Mauro Poddaf,g, Pietro Invernizzif,∗, the
Italian-Japanese PBC Study Group2
aDepartment of Psychology, University of Milano-Bicocca, Milan, Italy
bDepartment of Medicine, Teikyo University School of Medicine, Tokyo, Japan
cDivision of Gastroenterology and Hepatology, Department of Internal Medicine, Jikei University School of Medicine, Tokyo, Japan
dDivision of Internal Medicine and Liver Unit, San Paolo Hospital School of Medicine, University of Milan, Milan, Italy
eDivision of Gastroenterology, Graduate School of Medicine, Tohoku University, Sendai, Japan
fDivision of Internal Medicine and Hepatobiliary Immunopathology Unit, IRCCS Istituto Clinico Humanitas, Via A. Manzoni 113, 20089 Rozzano, Italy
gDepartment of Translational Medicine, University of Milan, Milan, Italy
hDivision of Gastroenterology and Hepatology, Mayo Clinic and Foundation, Rochester, Minnesota, United States
article info
Article history:
Received 29 October 2009
Accepted 7 January 2010
Available online 16 February 2010
Keywords:
Factor structure
PBC-40
Principal component analysis
Quality of life
abstract
Background: The available self-report questionnaire for the quality of life in patients with primary biliary
cirrhosis (PBC-40) is currently validated only in the British population but it lacks an evaluation of its
dimensionality.
Aims: To validate the Italian and Japanese versions of PBC-40 and to assess the dimensionality of the orig-
inal structure of PBC-40 by a confirmatory factor analysis. PBC-40 was translated to Italian and Japanese
using the forward–backward method and then reviewed in focus groups in the framework of a large
multicentric study.
Methods: A sample of 290 patients with PBC (125 Italian and 165 Japanese) was administered two
questionnaires previously validated for PBC-specific (PBC-40) and general quality of life (SF-36).
Results: The confirmatory model failed to fit adequately the original hypothesized structure. A principal
component analysis led to a seven-factor structure, with exclusion of 13 items characterized by lower
load; PBC-27 questionnaire was the final instrument. The validity of the PBC-27 was supported by its
strong correlation with the SF-36 scores.
Conclusion: We here propose an alternative structure of the quality of life questionnaire for PBC, namely
PBC-27, which appears to be effective in detecting the impact of PBC on quality of life in Italian and
Japanese patients.
© 2010 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1. Introduction
Primary biliary cirrhosis (PBC) is a progressive, chronic liver
disease characterized by the immune-mediated damage to the bil-
iary epithelial cells lining the small intrahepatic bile ducts [1]. The
夽Grant support: Supported by Executive Program of Cooperation in the Field of
Science and Technology between the Government of Italy and the Government of
Japan.
∗Corresponding author. Tel.: +39 02 8224 5128; fax: +39 02 8224 5191.
E-mail address: pietro.invernizzi@humanitas.it (P. Invernizzi).
1These authors equally contributed to this work.
2Members of the Italian-Japanese PBC Study Group contributed equally and are
listed in Appendix A.
course of the disease is generally slow and often asymptomatic
[2]. However, most patients suffer from elusive symptoms, such
as fatigue and pruritus, known to reduce their individual health
related quality of life (HRQoL) and well-being [3,4] particularly
at early stages of liver disease (i.e. before the appearance of liver
cirrhosis and its complications).
In the past two decades HRQoL has become an important read-
out in clinical research evaluating secondary treatment outcomes.
The understanding of factors related to HRQoL in chronic disor-
ders is becoming increasingly relevant in clinical practice, with the
recent emphasis on the comprehensive management of patients.
The World Health Organization states that quality of life is a
complex concept resulting from the individual physical health,
psychological state, level of independence, social relationship, and
1590-8658/$36.00 © 2010 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.dld.2010.01.004
L. Montali et al. / Digestive and Liver Disease 42 (2010) 718–723 719
salient environmental features [5]. HRQoL is a subset relating only
to the health domain of that existence.
It is widely recognized that evaluation of the HRQoL is a par-
ticularly complex issue, since it is influenced by several social and
possibly geographical factors. HRQoL in specific patient population
can be measured by generic and/or disease-specific questionnaires
[6]. Generic questionnaires, such as the SF-36, are designed to be
applicable to populations with a wide variety of conditions. Their
major advantage is that they have been validated and widely used
to measure the HRQoL in various conditions so that they provide
a global assessment and allow comparisons with other conditions
[7]. However, such questionnaires have less sensitivity to small, but
clinically relevant, changes in patient HRQoL over time, a major
problem in case of rare diseases due to floor or ceiling effects
[8]. On the contrary, domain-specific and disease-specific ques-
tionnaires are more sensitive based on their “custom-design” to
focus on disease-specific issues, and are ultimately more reliable in
assessing the patient subjective well-being, effectiveness of inter-
ventions, or extent of disease progression [9].
To date, few studies have examined HRQoL in patients with PBC
and such assessment has not routinely entered clinical trial use
or normal clinical practice, despite the numerous studies suggest-
ing the impact of the disease symptoms [3]. A group from UK has
recently addressed this limitation and developed the first disease-
specific HRQoL measure for patients with PBC, named PBC-40 [10],
covering six hypothesized domains (Cognitive, Itch, Fatigue, Social,
Emotional and other Symptoms), later reduced to a five-domain
structure with the collapse of the social and emotional ones [11].
Nevertheless, the PBC-40 has been validated only in English with
British patients. Moreover we could not find in the literature a sta-
tistical analysis of the PBC factor structure, which represents the
only way to assess the dimensionality of a scale [12,13].
Based on the suggested population and cultural variations in
symptom relevance and impact for PBC [14], we herein validated
an Italian and Japanese version of a PBC-specific HRQoL question-
naire. By applying a confirmatory factor analysis (CFA) to evaluate
its psychometric properties, we also propose a shortened PBC-40
version, namely PBC-27, which provides a better fit in Italian and
Japanese patients with PBC.
2. Materials and methods
2.1. Study population and design
290 patients affected by PBC were consecutively enrolled at
one liver unit in Milan, Italy and six liver units in Japan between
June 2007 and June 2008. The diagnosis of PBC was based on
the presence of two out of three internationally accepted cri-
teria, i.e. detectable serum anti-mitochondrial antibodies (titre
>1:40), increased enzymes indicating cholestasis (i.e. alkaline
phosphatase) for more than six months, and a compatible or diag-
nostic liver histology [1]. One hundred and twenty-five patients
were Italian (116 females; mean age 62 years, range 39–84) and
165 Japanese (143 females; mean age 61 years, range 30–83).
Serum biochemical tests including aminotransferases, gamma-
glutamlytransferase, alkaline phosphatase, albumin, total bilirubin,
lipids, immunoglobulins, hepatitis B surface antigen, antibody to
hepatitis B core antigen, and antibody to hepatitis C virus were
assessed by routine laboratory methods in all patients upon enrol-
ment. Similarly, anti-mitochondrial, anti-nuclear, and anti-smooth
muscle antibodies were available in all patients using indirect
immunofluorescence and/or ELISA methods [15]. The presence of
symptoms was defined as the occurrence of pruritus, jaundice, or
major complications of portal hypertension: i.e. ascites, gastroin-
testinal bleeding, portal-systemic encephalopathy. The Mayo Score
was used as an overall measure of disease severity [16]. Disease
duration was calculated as the time between the date of the earli-
est suspected evidence of liver disease and the date of enrolment in
the study. The histological picture of PBC was classified according
to Ludwig et al. [17].Table 1 illustrates the characteristics of this
PBC population. Ursodeoxycholic acid was being administered to
156 (95%) of the Japanese and 83 (66%) of the Italian patients as the
only treatment for liver disease at the time of enrolment.
We designed a three-phases study which included (i) the
development of the Italian and Japanese versions of PBC-40;
(ii) the evaluation of the psychometric properties of the Italian
and Japanese versions of PBC-40; and (iii) the correlation of the
PBC-specific HRQoL questionnaire with SF-36. The study proto-
col conforms to the ethical guidelines of the 1975 Declaration of
Helsinki (6th revision, 2008) as reflected in a priori approval by the
institution’s human research committee. This project received eth-
ical approval from the local IRB in each involved hospital and all
subjects entering the protocol provided written informed consent
after receiving a complete description of the study and having the
opportunity to ask questions.
2.2. Questionnaires
The PBC-40 is a disease-specific HRQoL measure derived and
validated for self-completion use in PBC [10]. The PBC-40 has six
hypothesized domains: Fatigue (11 items), Cognitive (6 items),
Social (10 items), Emotional (3 items), Itch (3 items) and other
Symptoms (7 items). Items are rated on an ordinal scale ranging
from 1 to 5 (with high scores denoting the greater symptom impact
and the worse HRQoL). The total score is obtained by averaging the
40 items.
The SF-36 is a widely used and validated generic questionnaire
adopted to measure the HRQoL of various conditions. It includes 36
items divided into eight domains, which can be aggregated into two
summary scores: a “mental component summary” and a “physical
component summary”. These indices include Physical Functioning,
Role Physical (role limitations as a result of physical health), Bodily
Pain, General Health, Vitality, Social Functioning, Role-Emotional
(role limitations as a result of mental problems) and Mental Health.
SF-36 scores on the individual scales range between 0 and 100. SF-
36 was found to have the best performance in terms of internal
consistency and test–retest reliability as the generic measures of
HRQoL for PBC patients.
Both the Italian and the Japanese version of PBC-40 were devel-
oped by translating and then back-translating the questionnaire to
determine possible discrepancies with the English original version.
The resulting questionnaires were reviewed by a team of physicians
who usually provide care to patients with PBC.
2.3. Questionnaire administration
All patients with PBC attending a regular outpatient visit were
asked to fill out two self-report questionnaires, the PBC-40 and
the SF-36. Eight patients (3 Italians and 5 Japanese) declined to
take part in the study (7 claiming ‘lack of time’, 1 for ‘excessive
stress’). Demographic questions were also included in the forms
and all questionnaires were self-administered in the presence of
an instructed psychologist or physician in quiet rooms within the
liver unit facilities. On average, the completion of the questionnaire
took about 20 min.
2.4. Statistical analyses
All the analyses were performed in all subjects and subsequently
for each language separately. Cronbach’s ˛and CFA were first
utilized on the six-domain model of PBC-40 documented in the
720 L. Montali et al. / Digestive and Liver Disease 42 (2010) 718–723
Table 1
Characteristics of the study population.
Japanese (n= 165) Italian (n= 125) All subjects (n= 290)
Age (years) 61 ±10 62 ±10 62 ±10
Duration of disease (years) 11 ±812±811±8
Mayo Score 4.9 ±1.1 5.2 ±.6 5.1 ±.8
Alkaline phosphatase (IU/L) (n.v. <279) 355 ±224 353 ±292 354 ±260
Aspartate aminotransferase (IU/L) (n.v. <50) 39 ±26 38 ±25 39 ±25
Total bilirubin (mg/dL) (n.v. <1.0) .8 ±.6 .7 ±.3 .7 ±.5
Albumin (g/dL) (n.v. >3.5) 4.1 ±.4 4.3 ±.4 4.2 ±.4
Immunoglobulin G (mg/dL) (n.v. <1700) 15.80 ±456 12.37 ±362 13.84 ±438
Immunoglobulin A (mg/dL) (n.v. <450) 273 ±124 277 ±157 276 ±144
Immunoglobulin M (mg/dL) (n.v. <280) 281 ±217 305 ±225 294 ±221
n(%) n(%) n(%)
Women 143 (87) 116 (93) 259 (89)
Asymptomatic patients 24 (15) 26 (21) 50 (17)
Presence of cirrhosis 37 (22) 44 (35) 81 (28)
Presence of major portal hypertension complicationsa8 (5) 4 (3) 12 (4)
Positive for AMA 155 (93) 112 (90) 267 (92)
Positive for ANA 47 (28) 56 (46) 103 (36)
Positive for SMA 3 (2) 9 (7) 12 (4)
Associated autoimmune diseases
With Sjögren’s syndrome 11 (7) 3 (2) 14 (5)
With systemic sclerosis 0 (0) 5 (4) 5 (2)
Others 12 (7)b10 (10)c24 (8)
Mean values ±standard deviation unless otherwise stated.
Abbreviations: AMA, anti-mitochondrial antibodies; ANA, anti-nuclear antibodies; SMA, anti-smooth-muscle antibodies.
aMajor portal hypertension complications (ascites, gastrointestinal bleeding, and portal-systemic encephalopathy) were only observed in patients with advanced histo-
logical stages (III and IV).
bHemolytic anemia in 5 patients, rheumatoid arthritis in 3, and systemic lupus erythematosus, multiple sclerosis, sarcoidosis, hypopituitarism each in 1 patient.
cCREST in 6 patients, and rheumatoid arthritis, systemic lupus erythematosus, Werlhof’s disease, psoriasis each in 1 patient.
original study [10], in order to assess its consistency and dimension-
ality. Alpha coefficients greater than .60 are considered indicative of
an acceptable level of internal consistency. The CFA was performed
using the Lisrel 8.80 software (Scientific Software International Inc.,
Lincolnwood, IL) which calculates several practical indices, includ-
ing the Comparative Fit Index (CFI), the Goodness of Fit Index (GFI),
the Adjusted Goodness of Fit Index (AGFI), as well as the Root
Mean Square Error of Approximation (RMSEA) and the Consistent
Akaike’s Information Criterion (CAIC) [18]. These indices compare
the observed sample covariance matrix with the matrix estimated
from the model relative to a null model. Goodness of fit indices (GFI,
AGFI, and CFI) of .90 or greater, and RMSEA of less than .05, support
a good fit. The CAIC allows to compare the global fit of different
models, and the smallest CAIC value suggests the best fit. A second
type of analysis, i.e. a principal component analysis, was performed.
Finally, the convergent validity of the constructs measured by this
questionnaire was assessed by comparison between PBC-27 and
SF-36 scores, with Pearson’s correlation coefficients.
3. Results
Data screening demonstrated that most variables manifest a
variable degree of skewing. In particular, since the raw data dis-
tributions were skewed towards the lower end of the range, data
were log transformed to normalize their distribution and allow a
parametric analysis.
3.1. Original factor structure evaluation
We calculated Cronbach’s ˛to estimate the internal reliabil-
ity of the six domains (Table 2). Then we utilized CFA to assess
the dimensionality of the scale. The CFA with the original PBC-
40 domain-structure indicated a poor fit between the proposed
models and the present data. Table 3 illustrates the goodness of fit
indices for the whole sample and for the sample divided according
to the language (Italian or Japanese). The chi-square/degrees-of-
Table 2
Internal consistency of the six PBC-40 domains as measured by Cronbach’s coeffi-
cient ˛.
All subjects (n= 290) Italian (n= 125) Japanese (n= 165)
Symptoms .704 .667 .724
Itch .825 .728 .860
Fatigue .941 .931 .952
Cognition .906 .893 .924
Emotional .783 .745 .803
Social .866 .852 .885
freedom ratio indicates that the model was not an optimal fit to
the gathered data. Moreover, this finding was confirmed accord-
ing to the other general rule of thumb for acceptance of model fit
(GFI, AGFI, CFI > .90 and RMR <.05). These results suggest the need
to further examine the factor structure model of the PBC-40.
Table 3
Fit indices for the original six-factor model.
All subjects (n= 290) Italian (n= 125) Japanese
(n= 165)
Degrees-of-
freedom
725 725 725
Minimum fit
function chi-square
1858.42 1121.47 1359.66
Goodness of Fit
Index
.76 .69 .71
Adjusted Goodness
of Fit Index
.72 .65 .67
Comparative Fit
Index
.96 .96 .96
Standardized root
mean square
residual (RMR)
.075 .10 .082
Root mean square
error of
approximation
(RMSEA)
.074 .066 .073
L. Montali et al. / Digestive and Liver Disease 42 (2010) 718–723 721
Table 4
Fit indices for the seven-factor model.
All subjects
(n= 290)
Italian
(n= 125)
Japanese
(n= 165)
Degrees-of-freedom 302 302 302
Minimum fit
function chi-square
463.08 332.07 410.43
Goodness of Fit Index .89 .83 .84
Adjusted Goodness
of Fit Index
.87 .79 .80
Comparative Fit
Index
.99 .98 .98
Standardized root
mean square residual
(RMR)
.050 .060 .057
Root mean square
error of
approximation
(RMSEA)
.043 .028 .047
3.2. Exploring an alternative factor structure
Since using the original scales the CFA model did not provide
an optimal fit, we investigated the structure underlying the PBC-
40 and a principal component analysis with a promax rotation
was conducted on the PBC-40 scores. Initially, eight factors were
extracted, each with values >1.0. Items with multiple loads of .40
or greater and items without a single load of .40 or greater (items
1, 3, 29, 35, 38, and 39) were not retained. Further, a second princi-
pal component analysis was performed on the remaining 34 items.
Seven-factor were extracted according to three criteria: Kaiser’s
criterion (with eigen values greater than or equal to 1), a scree
test, and the interpretability of resulting factor structures [19].
The obtained structure was similar to the original PBC-40, in that
the six factors (fatigue, cognitive, social, emotional, itch, and other
symptoms) corresponded to the original domains. The main dif-
ference was found in the symptoms domain, which appeared to
be split into two dimensions: a generic symptoms domain (items
2, 4, and 7) and a dryness one (items 5 and 6). This seven-factor
structure explained 66.87% of variance. Moreover, the principal
component analysis revealed another difference between this new
factor structure and the original measure, since the item 32 (“I
feel guilty that I can’t do what I used to do because of having
PBC”) loaded on the emotional rather than on the social fac-
tor.
3.3. Revised factor structure evaluation
CFA was used to test whether the data fitted a seven-factor
model and to evaluate the adequacy of each item. Items with poor
multiple square correlation coefficients (items 18, 20, 21, 23, 30,
31, and 40) were excluded in order to improve the model fit. The
resulting version was composed of 27 selected items distributed
on a seven domains model: symptoms (3 items), dryness (2 items),
itch (3 items), fatigue (8 items), cognitive (5 items), social (3 items)
and emotional (3 items). This seven-factor CFA was tested and
fit indices were: 2(302) = 463.08; p< .05; RMSEA = .043; CFI = .99;
2/df = 1.53 thus indicating a reasonable fit of this model. Table 4
illustrates the fit indices while Table 5 shows that each factor load
is beyond the .40 level, both for the total sample and for the Italian
and Japanese subgroups. Cronbach’s ˛was calculated to estimate
the consistency of the seven factors. Alpha coefficients reached
acceptable levels for all seven subscales (Table 6). As additional con-
trol, we compared the hypothesized six-factor structure fit against
the seven-factor model one. Because the models were not nested,
we chose to compare them examining the CAIC values, with lower
values indicating a more parsimonious and thus preferable model.
Table 5
Factor loads obtained from CFA of the PBC-27.
All subjects
(n= 290)
Italian
(n= 125)
Japanese
(n= 165)
Factor 1: symptoms
Q2. Felt bloated/ate or drank .62 .45 .76
Q4. Right side discomfort .59 .59 .55
Q7. Aches long bones .75 .73 .75
Factor 2: dryness
Q5. Dry eyes .63 .56 .69
Q6. Dry mouth .80 .81 .81
Factor 3: itch
Q8. Itching/sleep .66 .56 .75
Q9. Scratched so much .87 .84 .85
Q10. Felt embarrassed .81 .67 .86
Factor 4: fatigue
Q11. Had to force myself/out of bed .70 .67 .73
Q12. Had to have a sleep .62 .61 .64
Q13. Daily routine .86 .86 .88
Q14. Felt worn out .85 .84 .89
Q15. Felt tired/force myself .88 .87 .90
Q16. Felt tired/go to bed early .74 .71 .79
Q17. Fatigue hit me .79 .76 .83
Q19. Long time to do anything .73 .72 .74
Factor 5: cognitive
Q22. Effort/remember things .70 .72 .70
Q24. Concentration span .85 .81 .89
Q25. Keeping un with conversation .77 .76 .80
Q26. Difficult concentrate .83 .78 .88
Q27 Remember/what I wanted to do .69 .70 .74
Factor 6: emotional
Q28. I get more stressed .82 .84 .80
Q33. Worry about the future .50 .44 .54
Q32. Feel guilty .78 .78 .76
Factor 7: social
Q34. I can’t go out/enjoy myself .81 .79 .83
Q36. Can’t plan holidays .87 .79 .93
Q37. Social life stopped .82 .83 .82
Seven-factor model had a lower CAIC value (969.99) than the six-
factor model (999.54).
The Italian and Japanese version of PBC-27 may be obtained on
request from the corresponding author.
3.4. Correlation between PBC-27 and SF-36
To examine the convergent validity of the PBC-27, we calculated
Pearson’s correlation between PBC-27 scores and the scores of SF-
36 for both Italian and Japanese sample (Table 7). For this analysis,
items for each PBC-27 factor were taken from the results of the CFA
that are shown in Table 5. Similar to previous studies in PBC, we
expected that some of the SF-36 scales correlated with PBC-specific
factors, given the overlap of the concept. Specifically, the moderate
to high correlation between the PBC-associated fatigue factor and
the vitality scale in SF-36 was confirmed. Other minor correlations
were found between the PBC social factor and the social function-
Table 6
Internal consistency of the seven PBC-27 domains as measured by Cronbach’s ˛
coefficient.
All subjects (n= 290) Italian (n= 125) Japanese (n= 165)
Symptoms .693 .600 .709
Dryness .671 .619 .711
Itch .825 .728 .860
Fatigue .920 .911 .932
Cognition .884 .872 .904
Emotional .741 .714 .745
Social .871 .845 .890
722 L. Montali et al. / Digestive and Liver Disease 42 (2010) 718–723
Table 7
Pearson’s correlation between the PBC-27 and SF-36.
PBC-27 factor SF-36 Pearson’s correlation
coefficient—Italian sample (n= 125)
Pearson’s correlation
coefficient—Japanese sample (n= 165)
Fatigue Energy/vitality −.667** −.695**
Social Social functioning −.453** −.524**
Cognitive Mental component summary −.548** −.592**
Emotional Mental health −.471** −.579**
Emotional Role emotional −.497** −.469**
Symptoms Physical functioning −.434** −.318**
Symptoms Physical pain −.647** −.592**
Itch Physical functioning −.194*−.230**
Dryness Physical functioning −.299** −.163*
Dryness Physical role −.402** −.414**
*Correlation significant at the .05 level (two tailed).
** Correlation significant at the .01 level (two tailed).
ing scale of the SF-36 and between the PBC cognitive factor and the
mental component of the SF-36. The PBC emotional factor also cor-
related moderately with both mental health and role-emotional of
SF-36 while the PBC symptoms factor correlated as predicted with
physical functioning and also with the physical pain of the SF-36
scales. Moreover, similarly to what previously observed elsewhere
[10] and due to its specific nature, the itch factor had a negligible
correlation with physical functioning. Finally, we observed that the
dryness factor correlated slightly with the physical functioning and
moderately with the physical role SF-36 scales. Similar correlations
were found between the PBC-40 factors and SF-36 (Table 8).
3.5. Correlation with demographic and clinical features
We failed to find a possible effect of participants’ age on the
questionnaire scores performing an ANOVA between three differ-
ent groups: adults (age ≤50 years), middle aged (50<age < 65) and
older patients (age ≥65). We did not find a correlation between the
results of the questionnaire and the severity of the disease, as mea-
sured by the Mayo Score. Concerning ursodeoxycholic acid therapy
we found that it has no effect on the results of the questionnaire
(data not shown).
4. Discussion
We herein report the validation and evaluation of the first Italian
and Japanese PBC-specific HRQoL questionnaires. The aim of this
study was to assess the dimensionality of the original hypothesized
six-domain structure of PBC-40 by a CFA. Our comprehensive eval-
uation of the psychometric properties of the Italian and Japanese
questionnaires suggested to modify the theoretical factor structure
of this tool. Accordingly, we developed, validated and now pro-
pose a modified questionnaire, namely PBC-27, to assess the HRQoL
impact in Italian and Japanese patients with PBC.
The reliability of the Italian and Japanese version of PBC-40
indicated that the internal consistency of the six domains was sat-
isfactory, with Cronbach’s ˛ranging from .70 to .94. However, it
is a common misconception to interpret a high degree of inter-
nal consistency as an index of the uni-dimensionality of a domain
[12,13]. When CFA was used to test the dimensionality of the origi-
nal PBC-40 factor structure, for the total sample and for the Japanese
and Italian questionnaires separately, the chi-square/degrees-of-
freedom ratio indicated that the model was not an optimal fit to
the data obtained in our patients. This finding was confirmed by
other general rules of thumb for acceptance of model fit, including
the CFI, GFI, AGFI, and RMSEA [18]. For this reason we modified
the original questionnaire in order to obtain better psychometric
properties.
The most rigorous strategy was utilized to suggest adequate
modifications to the PBC-40. First, principal component analysis
was performed for all the subjects, and subsequently for each lan-
guage separately, in order to obtain a factor model that optimally
accounted for the data. Second, the analysis was performed through
two steps. The first step was the item-exclusion step, in which
items were excluded if negligible from a psychometric standpoint,
whereas the second step consisted in the evaluation of the domain
structure of the questionnaire subsequent to the exclusion of inap-
propriate items. Such analysis yielded a seven-factor structure,
composed by the six domains of the original model and the new
domain of dryness. Third, CFA was then re-applied to test whether
the data fitted a seven-factor model and to evaluate the adequacy
of each item. Fourth, the items with poor multiple square correla-
tion coefficients were excluded and a final version composed by
27 selected items was obtained. Finally, the convergent validity
of the PBC-27 was assessed by calculating the Pearson correlation
between this new questionnaire scores and the scores of SF-36 for
both the Italian and the Japanese sample.
A general problem with disease-specific HRQoL questionnaires
is that they are rarely subject to a rigorous evaluation of their
Table 8
Pearson’s correlation between the PBC-40 and SF-36.
PBC-40 factor SF-36 Pearson’s correlation
coefficient—Italian sample (n= 125)
Pearson’s correlation
coefficient—Japanese sample (n= 165)
Fatigue Energy/vitality −.695** −.711**
Social Social functioning −.625** −.499**
Cognitive Mental component summary −.550** −.581**
Emotional Mental health −.500** −.524**
Emotional Role emotional −.480** −.365**
Symptoms Physical functioning −.396** −.225**
Symptoms Physical pain −.603** −.553**
Itch Physical functioning −.194*−.230**
Note: Correlations are negative because high scores on SF-36 measure a better health condition while high scores on PBC denoting the greater symptom and the worse HRQoL.
** Correlation significant at the .01 level (two tailed).
*Correlation significant at the .05 level (two tailed).
L. Montali et al. / Digestive and Liver Disease 42 (2010) 718–723 723
psychometric properties as generic instruments. Specifically, the
examination of the questionnaire’s factor structure is fundamental
to get a full evaluation of the impact of symptoms on the HRQoL
and consequently it is a crucial clinical endpoint to complement
common major outcomes. Here, we demonstrated that the theo-
retical factor structure at the base of the British version does not
fit well for the Italian and Japanese samples and propose a mod-
ified questionnaire, namely PBC-27. Our shorted version presents
two main advantages compared to the PBC-40 questionnaire. First
it is more informative, since it indicates that the dryness symptom
has a direct and autonomous impact on patients HRQoL. Secondly,
it is more economic and this is valuable for surveys and clinical
assessment where a smaller number of items is desired. A possi-
ble limitation of our study is that the PBC-27 questionnaire was
not submitted to a new sample, but this is often the case when a
shortened version of a questionnaire is proposed starting from an
existing one [20–23] particularly in the setting of a rare disease.
In summary, our results provide evidence to reduce the original
PBC-40 scale and to reconceptualise it in a seven-domain model.
Moreover, the convergent validity of the PBC-27 was assessed by
comparing its scores with the SF-36 scores. It is clear that future
larger cross-cultural studies are needed to make comparisons and,
hopefully, to strengthen our results. In particular, to provide a final
examination of PBC-27 it would be necessary to submit it to an
English speaking sample.
Conflict of interest statement
No declared.
List of abbreviations
PBC, primary biliary cirrhosis; HRQoL, health related quality
of life; CFA, confirmatory factor analysis; CFI, comparative fit
index; GFI, goodness of fit index; AGFI, adjusted goodness of
fit index; RMSEA, root mean square error of approximation;
CAIC, consistent Akaike’s information criterion.
Appendix A. Appendix A
Members of the Italian-Japanese PBC Study Group (in
alphabetical order): Pier Maria Battezzati (Ospedale San Paolo,
Milano), Andrea Crosignani (Ospedale San Paolo, Milano), Eriko
Hayami (Teikyo University Mizonokuchi Hospital, Kanagawa),
Naomi Hosoya (Teikyo University Mizonokuchi Hospital, Kana-
gawa), Osamu Kido (Tohoku University, Sendai), Kentaro Kikuchi
(Teikyo University Mizonokuchi Hospital, Kanagawa), Hiroshi
Miyakawa (Teikyo University Mizonokuchi Hospital, Kanagawa),
Kyoko Monoe (Fukushima Medical University, Fukushima), Saeko
Nezu (Kyorin University, Tokyo), Hiromasa Ohira (Fukushima Med-
ical University, Fukushima), Shuhei Okuyama (Kyorin University,
Tokyo), Carlo Selmi (IRCCS Istituto Clinico Humanitas, Roz-
zano), Akitaka Shibuya (Kitazato University, Kanagawa), Atsushi
Takahashi (Fukushima Medical University, Fukushima), Shin-
ichi Takahashi (Kyorin University, Tokyo), Atsuko Takai (Teikyo
University Mizonokuchi Hospital, Kanagawa), Junko Yokokawa
(Fukushima Medical University, Fukushima), Mikio Zeniya (Jikei
University, Tokyo), Massimo Zuin (Ospedale San Paolo, Milano).
References
[1] Kaplan MM, Gershwin ME. Primary biliary cirrhosis. N Engl J Med
2005;353:1261–73.
[2] Younossi ZM, Boparai N, Price LL, et al. Health-related quality of life in chronic
liver disease: the impact of type and severity of disease. Am J Gastroenterol
2001;96:2199–205.
[3] Selmi C, Gershwin ME, Lindor KD, et al. Quality of life and everyday activities
in patients with primary biliary cirrhosis. Hepatology 2007;46:1836–43.
[4] Rannard A, Buck D, Jones DE, et al. Assessing quality of life in primary biliary
cirrhosis. Clin Gastroenterol Hepatol 2004;2:164–74.
[5] Camfield L, Skevington SM. On subjective well-being and quality of life. J Health
Psychol 2008;13:764–75.
[6] Martin LM, Dan AA, Younossi ZM. Measurement of health-related quality of life
in patients with chronic liver disease. Liver Transpl 2006;12:22–3.
[7] Coons SJ, Rao S, Keininger DL, et al. A comparative review of generic quality-
of-life instruments. Pharmacoeconomics 2000;17:13–35.
[8] Bourke SC, McColl E, Shaw PJ, et al. Validation of quality of life instruments in
ALS. Amyotroph Lateral Scler Other Motor Neuron Disord 2004;5:55–60.
[9] Bondini S, Kallman J, Dan A, et al. Health-related quality of life in patients with
chronic hepatitis B. Liver Int 2007;27:1119–25.
[10] Jacoby A, Rannard A, Buck D, et al. Development, validation, and evaluation of
the PBC-40, a disease specific health related quality of life measure for primary
biliary cirrhosis. Gut 2005;54:1622–9.
[11] Jones DE, Newton JL. An open study of modafinil for the treatment of daytime
somnolence and fatigue in primary biliary cirrhosis. Aliment Pharmacol Ther
2007;25:471–6.
[12] Gardner PL. Measuring attitudes to science: unidimensionality and internal
consistency revisited. Res Sci Educ 1995;25:283–9.
[13] Gardner PL. The dimensionality of attitude scales: a widely misunderstood idea.
Int J Sci Educ 1996;18:913–9.
[14] Bjornsson E, Simren M, Olsson R, et al. Fatigue is not a specific symp-
tom in patients with primary biliary cirrhosis. Eur J Gastroenterol Hepatol
2005;17:351–7.
[15] Invernizzi P, Lleo A, Podda M. Interpreting serological tests in diagnosing
autoimmune liver diseases. Semin Liver Dis 2007;27:161–72.
[16] Dickson ER, Grambsch PM, Fleming TR, et al. Prognosis in primary biliary cir-
rhosis: model for decision making. Hepatology 1989;10:1–7.
[17] Ludwig J, Dickson ER, McDonald GSA. Staging of chronic nonsuppurative
destructive cholangitis (syndrome of primary biliary cirrhosis). Virchows
Archiv a—Pathol Anat Histopathol 1978;379:103–12.
[18] Hosmer DW, Hosmer T, Le Cessie S, et al. A comparison of goodness-of-fit tests
for the logistic regression model. Stat Med 1997;16:965–80.
[19] Floyd FJ, Widaman KF. Factor analysis in the development and refinement of
clinical assessment instruments. Psychol Assess 1995;7:286–99.
[20] Merlijn VP, Hunfeld JA, van der Wouden JC, et al. Shortening a quality of life
questionnaire for adolescents with chronic pain and its psychometric qualities.
Psychol Rep 2002;90:753–9.
[21] Chu LW, Tam S, Kung AW, et al. A short version of the ADAM Question-
naire for androgen deficiency in Chinese men. J Gerontol A Biol Sci Med Sci
2008;63:426–31.
[22] Ide R, Mizoue T, Yamamoto R, et al. Development of a shortened Japanese ver-
sion of the Oral Health Impact Profile (OHIP) for young and middle-aged adults.
Community Dent Health 2008;25:38–43.
[23] Toyota H, Morita T, Taksic V. Development of a Japanese version of
the emotional skills and competence questionnaire. Percept Mot Skills
2007;105:469–76.
