Clin Rheumatol (2012) 31:113121

DOI 10.1007/s10067-011-1788-0

ORIGINAL ARTICLE

Psychometric properties of self-administered Lequesne

Algofunctional Indexes in patients with hip and knee
osteoarthritis: an evaluation using classical test theory
and Rasch analysis
Franco Franchignoni & Fausto Salaffi &
Andrea Giordano & Alessandro Ciapetti &
Marina Carotti & Marcella Ottonello

Received: 22 March 2011 / Revised: 19 May 2011 / Accepted: 27 May 2011 / Published online: 14 June 2011
# Clinical Rheumatology 2011

Abstract The aim of this study is to perform a psychometric

analysis of the Lequesne Algofunctional Indexes (LAI) for the
severity of osteoarthritis (OA) of the hip (LAI-hip) and knee
(LAI-knee), using classical test theory (CTT) and Rasch
analysis. Questionnaires were completed by 1,214 patients
with symptomatic OA of the knee (n=697) and hip (n=517).
Internal consistency was evaluated using Cronbachs alpha
and an item-to-total correlation. Dimensionality was investigated with a factor analysis. Raw scores underwent Rasch
analysis. Cronbachs alpha was 0.84 for LAI-hip and 0.82
for LAI-knee. LAI-hip resulted in unidimensionality

F. Franchignoni : A. Giordano : M. Ottonello

Department of Physical and Rehabilitation Medicine,
Salvatore Maugeri Foundation,
Veruno, Novara, Italy
e-mail: franco.franchignoni@fsm.it

according to the factor analysis, while LAI-knee supported both a single and a two-factor solution (items 1
6b and 710, respectively). At Rasch analysis, the rating
categories of item maximum distance walked did not
comply with the criteria for category functioning in
either LAI-hip or LAI-knee. A test of the residual
correlation showed item dependency in both LAI-hip
and LAI-knee. Misfitting items were present in both the
scales. According to both CTT and Rasch analysis, in our
two samples representing a wide spectrum of both hip
and knee OA severity the LAI-hip and LAI-knee showed
a series of drawbacks, which rendered both questionnaires inadequate in relation to their metric properties
and severely limit their ability to perform, as a composite
measure, in line with the main aims of their developers.
Keywords Lequesne Algofunctional indexes .
Osteoarthritis . Outcome measure . Rasch analysis

A. Giordano
e-mail: andrea.giordano@fsm.it

Introduction
M. Ottonello
e-mail: marcella.ottonello@fsm.it
F. Salaffi (*) : A. Ciapetti (*)
Department of Rheumatology,
Politechnic University of the Marche,
Ancona, Italy
e-mail: fsalaff@tin.it
e-mail: ciapetti.a@libero.it
M. Carotti
Department of Radiology,
Politechnic University of the Marche,
Ancona, Italy
e-mail: marina.carotti@gmail.com

Osteoarthritis (OA) is the most common form of arthritis, and

its prevalence increases with age [1]. Knee and hip OA are
common causes of lower extremity pain and disability in the
general population [2]. Patients with lower extremity OA
exhibit deterioration in functions concerning mobility, transfer
and activities of daily living [3]. Pain and disability have a
significant impact on health-related quality of life and social
costs [4]. In evaluating the severity of OA of the lower limbs,
guidelines for outcome measurement in OA highlight the
need to evaluate at least three dimensions: pain, global
disease status and functional impairment [59]. International

114

organisations like the Fifth World Health Organisation/

International League Against Rheumatism Task Force [10]
and the OMERACT group [11] have examined the currently
available condition-specific health status measures in terms of
their properties and made recommendations on their use. In
the last decade, the disease-specific questionnaires Western
Ontario and McMaster Universities OA index [12] or the
derived Hip Disability and Osteoarthritis Outcome Score [13]
and Knee injury and Osteoarthritis Outcome Score [14] and
the Lequesne Algofunctional Indexes (LAI) [15] have been
the most recommended and most widely used outcome
measures for OA of the hip and knee in clinical trials [5, 16].
The Lequesne Algofunctional Indexes for severity of OA
of the hip (LAI-hip) and knee (LAI-knee) each consist of three
sections with a total of 11 questions: severity of pain (five
items), walking ability (two items) and physical function (four
items). In spite of positive psychometric validations performed according to the classical test theory (CTT) [1719],
some concerns have been raised about the internal consistency and validity of both LAI versions [17, 2022]. Among
them, the study on LAI-hip by Dawson et al. [21] is the only
one that has used Rasch analysis as a statistical approach. A
Rasch analysis evaluates the psychometric properties of a
questionnaire that are not analysed by CTT techniques, e.g.
how well an item performs in terms of its relevance or
usefulness for measuring the underlying construct, the
amount of the construct targeted by each question, the
possible redundancy of an item relative to other items in the
scale and the appropriateness of the response categories [23
25]. Rasch analysis is being increasingly used in the
development and evaluation of clinical tools for health care,
and the advantages of this approach have been documented
in the literature [26, 27]. The aim of this study was to
perform a comprehensive psychometric analysis of both
LAI-hip and LAI-knee, using both CTT and Rasch analysis,
in order to examine in detail their psychometric properties.

Clin Rheumatol (2012) 31:113121

included were those who had pain for more than 25 days of
the past 30 days and at least two of the following three
criteria: erythrocyte sedimentation rate <20 mm/h, osteophytes
or obliteration of joint space on local X-ray examination.
The exclusion criteria were as follows: concurrent
systemic inflammatory rheumatic disease, medical comorbidity that would render the patient unable to participate
fully in study procedures (e.g. terminal conditions such as
end-stage renal disease, heart failure or malignancy),
alcohol abuse, psychiatric disorder, or the patients having
undergone arthroplasty of the joint in question.
All OA patients had radiographs of their affected joints.
Radiographs of the knees were anteroposterior (AP)
weight-bearing, semiflexed view, whereas an AP pelvis
radiograph was collected on hip patients. The radiographs
used in this study were generally obtained within 1 year
from the date of the questionnaire assessments.
Measures
The Lequesne Algofunctional Indexes of severity for OA of
the hip (LAI-hip) and knee (LAI-knee) include three sections
(which are not graded separately) and take a few minutes to
complete [15] (Tables 1 and 2). They were developed using
an interview format with a total of ten questions, although
question 6 maximum distance walked could be split into
two, in that the use of one or two walking aids (which increases
the score by one or two points) is arguably an additional item.
Indeed, the self-administered version includes the use of
walking aids as an 11th question (item 6b in Tables 1 and 2)
[17, 18]. For the purposes of this study we chose to use the
self-administered format because in clinical trials it allows for
mailing to patients. Both Lequesne indexes are scored as the
sum of all items, resulting in a total score ranging from 0 to
24. Higher scores denote greater OA severity.
Statistical analysis

Patients and methods

A stage process was used to investigate the basic properties

of LAI-hip and LAI-knee:

Patients and data collection

Internal consistency and dimensionality
In this cross-sectional survey, the disease-specific LAI-hip and
LAI-knee questionnaires were completed by 1,214 patients
with symptomatic OA of the knees (697 patients) or hips (517
patients), enrolled by general practitioners participating in the
MI.D.A. Study [28]. According to the American College of
Rheumatology guidelines, the inclusion criteria [29, 30] for
patients with knee OA were as follows: (1) knee pain for more
than 25 days of the past 30 days, morning stiffness of less
than 30 min, and crepitation in the knee or (2) pain for more
than 25 days of the past 30 days and osteophytes on X-ray
examination of the knees [16]. Patients with hip OA who were

The items were examined for internal consistency calculating:

1. Cronbachs coefficient alpha. Alpha values >0.70 are
recommended for group level comparisons, whereas a
minimum of 0.850.90 is desirable for individual
judgments [31].
2. The Spearman rank correlation coefficients (rs), to
examine to what degree each item was correlated with
the total score, omitting that item from the total (item
total correlation, corrected for item overlap). The usual

Clin Rheumatol (2012) 31:113121

115

Table 1 The Lequesne Algofunctional Indexes of severity for

osteoarthritis of hip (LAI-hip) [15]

Table 2 The Lequesne Algofunctional Indexes of severity for

osteoarthritis of knee (LAI-knee) [15]

LAI-hip

Items

Pain or discomfort
1. During nocturnal bed rest
None or insignificant
Only on movement or on certain positions
With no movement
2. Morning stiffness or regressive pain after rising
1 min or less
More than 1 but less than 15 min
15 min or more
3. After standing for 30 min
No
Yes
4. While ambulating

Points

0
1
2
0
1
2
0
1

None
Only after ambulating some distance, or after
initial ambulation, not increasingly
Early after initial ambulation and increasingly with
continued ambulation
5. With prolonged sitting (2 h)

0
1

None
Yes
Maximum distance walked (may walk with pain)
6a. Maximum distance walked
Unlimited
More than 1 km but limited
About 1 km or about 15 min
From 500 to 1,000 m or 815 min
From 300 to 500 m
From 100 to 300 m
Less than 100 m
6b. Use of a walking aid
None
One walking stick or crutch
Two walking sticks or crutches
Activities of daily living
7. Put on socks by bending forward
8. Pick up an object from the floor

0
1

9. Climb up and down a standard flight of stairs

10. Get into and out of a car or a deep armchair
Without difficulty
With some difficulty
With moderate difficulty
With important difficulty
Unable

0
1
2
3
4
5
6
0
1
2

0
0.5
1
1.5
2

rule of thumb is that an item should correlate with the

total score with r>0.30.

Pain or discomfort
1. During nocturnal bed rest
None or insignificant
Only on movement or on certain positions
With no movement
2. Morning stiffness or regressive pain after rising
1 min or less
More than 1 but less than 15 min
15 min or more
3. After standing for 30 min
No
Yes
4. While ambulating

Points

0
1
2
0
1
2
0
1

None
Only after ambulating some distance or after initial
ambulation, not increasingly
Early after initial ambulation and increasingly with
continued ambulation
5. When getting up from sitting without the help of arms

0
1

None
Yes
Maximum distance walked (may walk with pain)
6a. Maximum distance walked
Unlimited
More than 1 km but limited
About 1 km or about 15 min
From 500 to 1,000 m or 815 min
From 300 to 500 m
From 100 to 300 m
Less than 100 m
6b. Use of a walking aid
None
One walking stick or crutch
Two walking sticks or crutches
Activities of daily living
7. Able to climb up a standard flight of stairs?
8. Able to climb down a standard flight of stairs?

0
1

9. Able to squat or bend on the knees?

10. Able to walk on uneven ground?
Without difficulty
With some difficulty
With moderate difficulty
With important difficulty
Unable

0
1
2
3
4
5
6
0
1
2

0
0.5
1
1.5
2

Dimensionality was investigated, given the uncertain

factorial structure of the scales, following an explorato-

116

ry factor analytical approach, with the programme

FACTOR [32]:
1. A diversified approach [33]utilizing parallel analysis
(PA) [34] and Minimum Average Partial method (MAP)
[35] was applied to identify the number of dimensions present in the data; the result sought after is a
single dimension.
2. 2. An exploratory factor analysis (EFA) for ordinal data
was then used to evaluate the contribution of each item
to the factor; a loading >0.3 was considered as
significant, given the sample size [36].
Rasch analysis
After the above analysis, the matrix of single raw
scores for each subject underwent Rasch analysis
(partial credit model) through the WINSTEPS software,
v. 3.68.2 [37]. As a first step, we investigatedthrough
rating scale diagnosticswhether the rating scales of
both LAI-hip and LAI-knee were being used in the
expected manner, according to the criteria suggested by
Linacre [25, 38]. If necessary, according to the findings
of the rating scale diagnostic, categories were collapsed
following specific guidelines, comparing several different patterns of categorization, looking not only at the
above indicators of category diagnostics but also at best
reliability indices and clinical meaning [24]. Reliability
was evaluated in terms of separation (G), defined as the
ratio of the true spread of the measures with their
measurement error [24]. The separation indices give an
estimate (in standard error units) of the spread or
separation of items or persons along the measurement
construct. These indices reflect the number of strata of
measures which are statistically discernible. A separation
of 2.0 is considered good and enables the distinction of
three groups or strata, defined as segments whose centres
are separated by distances greater than can be accounted
for by measurement error alone [number of distinct strata=
(4 G+1)/3]. A related index is the reliability of these
separation indices, providing the degree of confidence that
can be placed in the consistency of the estimates (range, 01;
coefficients >0.80 are considered as good, and >0.90
excellent) [25].
A principal component analysis (PCA) on the standardized
residuals was used to investigate:
1. The presence of subdimensions, as an independent
confirmation of the unidimensionality of the scale.
Unidimensionality in this context assumes thatafter
removal of the trait that the scale intended to
measure (the Rasch factor)the residuals will be
uncorrelated and normally distributed (i.e. there are

Clin Rheumatol (2012) 31:113121

no principal components). The following criteria

were used to determine whether additional factors
were likely to be present in the residuals: (a) a cutoff
of 50% of the variance explained by the Rasch factor
and (b) eigenvalue of the first residual factor smaller
than 3 [37].
2. The local independence of items. High correlation
(>0.30) of residuals for two items indicates that they
may not be locally independent, either because they
duplicate some feature of each other or because they
both incorporate some other shared dimension [38].
After a review of the rating scale categories, the
validity was analysed by evaluating the fit of individual
items of each index (LAI-hip and LAI-knee) to the latent
trait as per the Rasch modelling (unidimensionality) and
examining if the pattern of item difficulties was consistent with the model expectation. The Rasch model
estimates goodness-of-fit (or simply fit) of the real
data to the modelled data. The information-weighted
(infit) and outlier-sensitive (outfit) mean-square statistics
(MnSq) for each item were calculated (similar to a chisquare analysis) to test if there were items which did not
fit with the model expectancies [23]. We considered as an
indicator of acceptable fit MnSq >0.8 and <1.2, items
outside this range were considered underfitting (MnSq>
1.2) or overfitting (MnSq<0.8) [37]. The following step
was to calculate the level of difficulty achieved by each
item (item difficulty) and where each individual subject
fits along the continuum (subject ability) [23]. We also
assessed the possible differences in item functioning (DIF)
linked to gender or age. An item bias can occur when
different subsets of respondentsdespite equal levels of
the underlying characteristic being measuredrespond in
a different manner to an individual item, so that item
characteristics (e.g. their location along the measurement
construct) vary across these groups [24].

Results
Characteristics of patients
The mean age of the sample was 72.1 years (standard
deviation (SD), 10.5) for hip OA and 71.9 years (SD, 9.8)
for knee OA. Of the patients, 71.8% (n=872) were female
(n=371 with hip OA, n=501 with knee OA). Of the 1,214
enrolled subjects, 909 (74.8%) reported one or more
medical comorbidities, mostly cardiovascular (50.7%),
endocrinological (13.2%), gastrointestinal (11.8%) and
respiratory (11.5%) disorders. The LAI-hip median score
(interquartile range (IQR)) was 12 (IQR=6), the LAI-knee
median score (IQR) was 11.5 (IQR=7.5).

Clin Rheumatol (2012) 31:113121

117

Internal consistency and dimensionality

The Cronbachs alpha was 0.84 for LAI-hip and 0.82 for
LAI-knee. The item total correlation coefficients ranged
from 0.29 (pain or discomfort with prolonged sitting) to
0.71 (pick up an object from the floor) in LAI-hip and
from 0.28 (pain or discomfort when getting up from sitting
without the help of arms) to 0.70 (able to climb up a
standard flight of stairs?) in LAI-knee. LAI-hip resulted
unidimensional according to both PA and MAP, the single
factor accounting for 66.3% of the variance and all items
loading meaningfully to the factor (loadings between 0.45
and 0.88). For LAI-knee, MAP identified a single factor,
while PA suggested a two-factor solution. The single factor
accounted for 62% of the variance, and all items loaded
meaningfully to the factor, with loadings between 0.37 and
0.85, while adding a second factor increased the explained
variance to 73.7%. The two factors were composed of
items 16b (pain or discomfort+walking ability) and 7
10 (activities of daily living), loading between 0.38 and
0.93 after promax oblique rotation with an inter-factor
correlation of 0.63.
Rasch analysis
The rating categories in item maximum distance walked
did not comply with the set criteria for category functioning
in either LAI-hip or LAI-knee. Pre-set criteria were met
when the seven categories were collapsed into a 0112234
scheme, obtaining a new five-level rating scale: 0 =
unlimited distance; 1=1 km (taking 15 min or more) but
limited; 2=300900 m; 3=100300 m; 4=less than 100 m
(see Fig. 1a, b for LAI-knee). For both scales, a principal
component analysis of the residuals suggested a single trait
structure (variance explained by the Rasch factor LAI-knee
54.2%, LAI-hip 55.6%; the eigenvalue of first residual was
2.1 for LAI-knee, 2.0 for LAI-hip). To confirm this finding,
in partial disagreement with the results of the EFA for LAIknee, we split the items into two subsets based on positive
and negative loadings on the first residual contrast, obtained
an estimate of the measure of each subject performing
separate Rasch analyses on the two subsets and correlated
the measures. Pearsons correlation coefficient was 0.6
(0.83 if corrected for attenuation) with unity slope,
confirming a substantially unidimensional test. Coherently
with the structure obtained with the EFA, the item
distribution according to the sign of the loadings on the
first residual contrast was items 16b (pain or discomfort+
walking ability) versus 710 (activities of daily living). The
residual correlation test showed an inter-item correlation
higher than 0.30 between the following couples of items:
(a) in LAI-hip, put on socks by bending forward with
pick up an object from the floor (0.32); (b) in LAI-knee,

Fig. 1 Category probability curves of item 6a (maximum walking

distance) of LAI-knee. The y-axis represents the probability (0 to 1) of
responding to one of the rating categories and the x-axis represents the
different performance values [person measure minus the item
measure], in logits. The 0 curve declines as the subjects ability
increases; the crossing point (where 0 and 1 are equally probable) is
the first threshold. The same applies for the other curves. The ideal
plot should lookas in blike an ordered even succession of hills,
with an emerging crest where each category is modal over a certain
range. a Original scale with seven categories (Table 2). b Revised
scale after collapsing into five categories, according to the model
0112234 (see text). Similar results were found for LAI-hip

able to climb up a standard flight of stairs? with able to

climb down a standard flight of stairs? (0.4) and (c) able
to squat or bend on the knees? with able to walk on
uneven ground? (0.37).
In LAI-hip, items pain or discomfort during nocturnal bed
rest, maximum distance walked and use of a walking aid
resulted as underfitting, while climb up and down a standard
flight of stairs and can get into and out of a car presented
overfit (Table 1). Figure 2a shows the distribution map of
subject ability and item difficulty of the LAI-hip. The item
measures of LAI-hip ranged from 1.45 (after standing for
30 min) to +3.09 logits (use of a walking aid), and person
ability measures ranged from 5.54 to +3.99 logits. The item

118

Clin Rheumatol (2012) 31:113121

Fig. 2 Person ability and item difficulty map of both LAI-hip (left side)
and LAI-knee (right side). The vertical line represents the measure of
OA severity, with the units of measurement on the scale (logits, the
natural logarithm of the odds of mutually exclusive alternatives, e.g. pass
vs. fail or higher response vs. lower response). The left-hand column
locates the person ability measures along the variable, while the right-

hand column locates the item difficulty measures along the variable.
Each item is indicated by its number (see Table 1). By convention, the
average difficulty of items in the test is set at 0 logits (and indicated with
M), so that items with a positive/negative sign are harder than average,
and those with a negative/positive sign are easier than average.
Accordingly, a candidate with average ability is indicated by an M

separation index was 13.21 (item separation reliability=0.99)

and the person separation index 2.37 (person separation
reliability=0.85).
In LAI-knee, items pain or discomfort during nocturnal bed rest and morning stiffness or regressive pain
after rising resulted underfitting, while able to climb up
a standard flight of stairs? and able to climb down a
standard flight of stairs? showed overfit. Figure 2b

shows the distribution map of subject ability and item

difficulty of LAI-knee, according to the Rasch model. Item
difficulty measures ranged from 1.38 (while ambulating)
to +3.20 logits (use of a walking aid), and person ability
ranged from 5.47 to + 3.04 logits. The item separation
index was 15.35 (item separation reliability=1), the person
separation index 2.14, and the person separation reliability
0.82. The DIF analysis showed no difference in responses

Clin Rheumatol (2012) 31:113121

to the items in LAI-hip and LAI-knee due to gender, and

some influence of age on responses was present in items
Pain or discomfort during nocturnal bed rest and Use of
walking aid.

Discussion
The purpose of this study was to perform a comprehensive psychometric analysis of LAI-hip and LAI-knee,
using both CTT and Rasch analysis in order to better
understand the strengths and weaknesses of the two
instruments, and provide a rational basis for improving
their metric quality (it was not our intention to propose
revised versions). Hence, we focused on analysing
dimensionality and rating scale diagnostics, and identifying those items most useful for measuring the intended
construct (model fit). The Cronbach alpha (0.820.84)
and item-to-total statistics indicateaccording to CTT
an adequate homogeneity of the items of both scales.
This finding is consistent with previous studies reporting
Cronbachs alpha values ranging from 0.58 [22] to 0.84
for LAI-hip [21] and from 0.75 [39] to 0.82 [17] for LAIknee. A similar reliability was found with the Rasch
analysis (person separation reliability), with results ranging from 0.82 (LAI-knee) to 0.85 (LAI-hip). The internal
consistency of both scales seems adequate only for making
group comparisons [33, 40] and does not allow to
standardize decisions in individuals (e.g. regarding the
need for hip or knee replacement) [21]. In addition, the use
of the raw score seems inappropriate because items have
very different rating points, not properly weighted.
As for unidimensionality, no standard approach exists to
determine whether an outcome measure is sufficiently
unidimensional or when the presence of secondary dimensions threatens the accuracy of unidimensional Rasch
calibrations. The most used approaches are derived from
the factor analysis: EFA is used to study the relevant latent
factors (after some form of estimation of their number) and
the contribution of individual items to those traits in new or
modified models (e.g. following language adaptation) [26].
In the present study, the factor analysis hinted at two factors
in LAI-knee, the first one largely prevailing with a high
inter-factor correlation; these results, along with subsequent
Rasch analysis (PCA of residuals and split-sample correlation), supported an underlying responses structure sufficiently
unidimensional to allow further analysis using Rasch methods.
It seems reasonable that analyses based on different clinical
samples and extraction methods could lead to somewhat
different statistical solutions, also in terms of subscale content
[21]. In any case, we think that an instrument for examining
the severity of hip and knee OA should separately rate
functional difficulty and functional pain [41].

119

It should be also borne in mind that raw scores can be

misleading and numerical codes (0, 1, 2,) do not
necessarily imply proportionality among the measures
[42]. Moreover, the item weighting seem arbitrary. Whereas
traditional psychometric approaches mainly focus on an
instruments total score, Item Response Theory models such
as the Rasch measurement model are founded on an explicit
mathematical model that narrows the unit of measure of an
instrument to the item level. It describes the relationship
between responses to the instruments items and the notdirectly-measurable construct (latent trait) to be measured.
In this framework, it is possible to evaluate a series of
psychometric properties of a questionnaire that cannot be
analysed by CTT techniques [2325].
Concerning the Rasch analysis of the rating scales, both
questionnaires showed disordered thresholds in the item
Maximum distance walked. The category thresholds are
the ability levels at which the response to either of the two
adjacent categories is equally likely (see Fig. 1); disordered
thresholds occur when respondents have difficulty consistently discriminating between response options (e.g.
because there are too many response options or the
labelling of the options is confusing). Our findings
indicate that subjects were able to distinguish only five
categories (instead of the original seven levels), and thus,
it might be appropriate to simplify the item format.
Similarly, Dawson et al. pointed out the presence of
disordered thresholds in the same item (+25% of
respondents leaving this item blank) [21] and Bae
mentioned that among all the questions this was the least
comprehensible [39]. As for the fit of individual items to
the latent trait, items Pain or discomfort during nocturnal
bed rest, Maximum distance walked and Use of
walking aid in LAI-hip, and items Pain or discomfort
during nocturnal bed rest and Duration of morning
stiffness or pain after getting up in LAI-knee were
underfitting. This indicates that both LAI scales have
items that either do not tap the same underlying construct
or are poorly written or too sensitive to confounding
factors. For example, it is reasonable that unexpected
responses to the item Use of walking aid could be due to
various personal and environmental factors.
Problems with the internal consistency of pain items in
both scales were already pointed out by other studies using
CTT psychometric methods [17, 22]. Moreover, two items
of the LAI-hip (Climb up and down a standard flight of
stairs and Get into and out of a car or a deep armchair)
and two of the LAI-knee (Able to climb up a standard
flight of stairs and Able to climb down a standard flight of
stairs) were overfitting, i.e. contributing little extra
information over that provided by other items in the scale,
and with a response pattern too predictable from the overall
pattern of responses to other items. In addition, the test of

120

residual correlation showed inter-item correlations higher

than 0.30 in two pairs of items for LAI-knee and in one pair
for LAI-hip. This inter-item dependency indicates that
either they duplicate some feature of each other or they
both incorporate some other shared dimension.
Overall, these results show that several items did not
conform to theoretical Rasch requirements for scale
development and construct validation, and thus item
selection is suboptimal in both LAI-hip and LAI-knee.
Moreover, if some items are removed, this could reduce the
breadth, precision and reliability of both scales.
Our study has a number of limitations. First, in line with
other authors [17, 18, 21], we used the self-administered
format of the questionnaires, even if Lequesne suggests an
interview format given by an observer with sufficient
training [15]. Secondly, self-reported knee or hip pain or
limited physical function are a common complaint of
elderly people, and may be due not only to a local OA
but also to other comorbidities, thus confounding the
analyses [43]. Third, the low frequency of persons using
two walking sticks (11 in LAI-hip and six in LAI-knee)
does not allow a precise and reliable estimation of the
threshold calibration for the item use of walking aid.
In summary, based on both CTT and Rasch analysis
methods performed in two samples representing a wide
spectrum of both hip and knee OA severity, the LAI-hip
and LAI-knee did show a series of drawbacks, which render
both questionnaires inadequate in relation to their metric
properties and severely limit their ability to perform, as a
composite measure, in line with the main aims of their
developers [15, 44]. Specifically, we recommendin
developing new measures of physical function in patients
with knee osteoarthritisto use modern psychometric
approaches (like the Item Response Theory and Computed
Adaptive Test methods) to create instruments optimizing
the validity and technical quality with the fewest number of
items [41, 45, 46].
Acknowledgements The authors wish to thank all the members of
the MI.D.A Study Group [28] for their contribution to the acquisition
of data, as follows: Adami S., Arioli Giovanni, Avossa Marco,
Bazzicchi Laura, Begh Franco, Beltrametti Paolo, Bentivenga
Crescenzo, Benucci, Maurizio, Bernini Luigi, Bertolucci Daniela,
Blasetti Patrizia, Bombardieri S., Bordin Giorgio, Bortolotti Giuseppe,
Bozzolan Fabiola, Broggini Marco, Bucci Romano, Calcagnile Fabio,
Calligaro Antonella, Cammelli Daniele, Cantatore F.P., Cimmino M.
A., Iannone Francesco, Carignola Renato, Carlino Giorgio, Casari
Silvia, Casilli Oriana Elena, Castelnuovo Aurelio, Cecchetti Riccardo,
Cesaro Gianni, Ciancio Giovanni, Colombo Fulvio, Consonni Luigi,
Covelli, Michele, Cozzi Luisella, Cozzolongo Anna Carla, Crafa
Silvana, Davoli Camillo, Del Ross Teresa, Del Vino Piergiorgio, Di
Giacinto Giovanni, Di Giuseppe Paolo, Filardi Piergiuseppe, Francioni
Cinzia, Fusaro Enrico, Giorgio Gandolini, Gorla Roberto, Govoni
Marcello, Grattagliano Vito, Lagan Angela, Lapadula G, Lazzarin
Paolo, Leucci Pierfrancesco, Levi Marina, Limonta Massimiliano,
Lombardini Francesco, Longhi Marco, Lopez Vincenzo, Lubrano

Clin Rheumatol (2012) 31:113121

Ennio, Malatesta Renato, Manfredini Monica, Manganelli Paolo,
Mannoni Alessandro, Marchetta Antonio, Marin Gabriella, Marsico
Antonio, Mascia Maria Teresa, Massarotti Marco, Mastaglio Claudio,
Minosi Armando, Miserocchi Fabio, Moreno Mauro, Muratore
Maurizio, Murgo Antonella, Ortolani Sergio, Olivieri Ignazio, Paolazzi
Giuseppe, Pellerito Raffaele, Peronato Giovanni, Pianon Margherita,
Punzi L., Ramonda Roberta, Rastelli Emilio, Reta Massimo, Rizzi
Massimo, Rocchetta Pier Andrea, Rossi Fulvia, Rossini Maurizio,
Sabadini Luciano, Santo Leonardo, Sarzi-Puttini Piercarlo, Saviola
Gianantonio, Scendoni Pietro, Sconosciuto Carmelo, Semmola Maria
Vittoria, Tamburrino Vitalba, Terlizzi Nicola, Viardi Luigi, Volante
Daniela, Zuccaro Carmelo.
Disclosures None

References
1. Salaffi F, De Angelis R, Grassi W, MArche Pain Prevalence,
INvestigation Group (MAPPING) study (2005) Prevalence of
musculoskeletal conditions in an Italian population sample: results
of a regional community-based study I. The MAPPING study.
Clin Exp Rheumatol 23:819828
2. Salaffi F, De Angelis R, Stancati A, Grassi W, on behalf of
MArche Pain Prevalence INvestigation Group (MAPPING) study
(2005) Health-related quality of life in multiple musculoskeletal
conditions: a cross-sectional population based epidemiological
study II. The MAPPING study. Clin Exp Rheumatol 23:829839
3. Salaffi F, Carotti M, Grassi W (2005) Health-related quality of life
in patients with hip or knee osteoarthritis: comparison of generic
and disease-specific instruments. Clin Rheumatol 24:2937
4. Leardini G, Salaffi F, Caporali R, Canesi B, Rovati L, Montanelli
R et al (2004) Direct and indirect costs of osteoarthritis of the
knee. Clin Exp Rheumatol 22:699706
5. Bellamy N, Kirwan J, Boers M, Brooks P, Strand V, Tugwell P et
al (1997) Recommendations for a core set of outcome measures
for future phase III clinical trials in knee, hip, and hand
osteoarthritis: consensus development at OMERACT III. J
Rheumatol 24:799802
6. World Health Organization, Regional Office for Europe (1985)
Guidelines for the clinical investigation of drugs used in
rheumatic diseases: European drug guidelines. Series 5. European
League Against Rheumatism, Copenhagen
7. Turk DC, Dworkin RH, Allen RR, Bellamy N, Brandenburg N, Carr
DB et al (2003) Core outcome domains for chronic pain clinical
trials: IMMPACT recommendations. Pain 106:337345
8. Dworkin RH, Turk DC, Peirce-Sandner S, Baron R, Bellamy N,
Burke LB et al (2010) Research design considerations for
confirmatory chronic pain clinical trials: IMMPACT recommendations. Pain 149:177193
9. Altman R, Brandt T, Hochberg M, Moskovitz R (1996) Design
and conduct of clinical trials of patients with osteoarthritis.
Recommendations from a task force of the Osteoarthritis Research
Society. Osteoarthr Cartil 4:217243
10. Lequesne M (1993) ILAR guidelines for testing slow acting drugs
in osteoarthritis (SYSADOAs). Rev Esp Rheumatol 20:220221
11. Bellamy N (1995) Outcome measurement in osteoarthritis clinical
trials. J Rheumatol Suppl 43:4951
12. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW
(1988) Validation study of WOMAC: a health status instrument
for measuring clinically important patient relevant outcomes to
antirheumatic drug therapy in patients with osteoarthritis of the
hip or knee. J Rheumatol 15:18331840
13. Nilsdotter AK, Lohmander LS, Klassbo M, Roos EW (2003) Hip
disability and osteoarthritis outcome score (HOOS) e validity and

Clin Rheumatol (2012) 31:113121

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

responsiveness in total hip replacement. BMC Musculoskelet

Disord 4:10
Roos EM, Toksvig-Larsen S (2003) Knee injury and Osteoarthritis
Outcome Score (KOOS)validation and comparison to the
WOMAC in total knee replacement. Health Qual Life Outcomes
1:17
Lequesne MG, Mery C, Samson M, Gerard P (1987) Indexes of
severity for osteoarthritis of the hip and knee. Scand J Rheumatol
Suppl 65:8589
Altman R, Brandt K, Hochberg M, Moskowitz R, Bellamy N,
Bloch DA et al (1996) Design and conduct of clinical trials in
patients with osteoarthritis: recommendations from a task force of
the Osteoarthritis Research Society. Results from a workshop.
Osteoarthr Cartil 4:217243
Stucki G, Sagha O, Stucki S, Michel BA, Tyndall A, Dick W et al
(1998) Comparison of the WOMAC (Western Ontario and
McMaster Universities) osteoarthritis index and a self-report
format of the self-administered Lequesne-algofunctional index in
patients with knee and hip osteoarthritis. Osteoarthr Cartil 6:7986
Theiler R, Sangha O, Schaeren S, Michel BA, Tyndall A, Dick W
et al (1999) Superior responsiveness of the pain and function
sections of the Western Ontario and McMaster Universities
Osteoarthritis Index (WOMAC) as compared to the LequesneAlgofunctional Index in patients with osteoarthritis of the lower
extremities. Osteoarthr Cartil 7:515519
McConnell S, Kolopack P, Davis AM (2001) The Western Ontario
and McMaster Universities Osteoarthritis Index (WOMAC): a
review of its utility and measurement properties. Arthritis Rheum
45:453461
Faucher M, Poiraudeau S, Lefevre-Colau MM, Rannou F,
Fermanian J, Revel M (2002) Algo-functional assessment of knee
osteoarthritis: comparison of the testretest reliability and construct validity of the WOMAC and Lequesne indexes. Osteoarthr
Cartil 10:602610
Dawson J, Linsell L, Doll H, Zondervan K, Rose P, Carr A et al
(2005) Assessment of the Lequesne index of severity for
osteoarthritis of the hip in an elderly population. Osteoarthr Cartil
13:854860
Basaran S, Guzel R, Seydaoglu G, Guler-Uysal F (2010)
Validity, reliability, and comparison of the WOMAC osteoarthritis index and Lequesne algofunctional index in Turkish
patients with hip or knee osteoarthritis. Clin Rheumatol
29:749756
Tesio L (2003) Measuring behaviours and perceptions: Rasch
analysis as a tool for rehabilitation research. J Rehabil Med
35:105115
Bond TG, Fox CM (2007) Applying the Rasch model: fundamental measurement in the human sciences, 2nd edn. Lawrence
Erlbaum Associates, Mahwah
Wolfe EW, Smith EV Jr (2007) Instrument development tools and
activities for measure validation using Rasch models: part I
instrument development tools. J Appl Meas 8:97123
Franchignoni F, Giordano A, Sartorio F, Vercelli S, Pascariello B,
Ferriero G (2010) Suggestions for refinement of the Disabilities of
the Arm, Shoulder and Hand Outcome Measure (DASH): a factor
analysis and Rasch validation study. Arch Phys Med Rehabil
91:13701377
Belvedere SL, de Morton NA (2010) Application of Rasch
analysis in health care is increasing and is applied for variable
reasons in mobility instruments. J Clin Epidemiol 63:12871297

121
28. Cimmino MA, Salaffi F, Olivieri I, Trotta F, Frizziero L, Sarzi Puttini
P et al (2004) Pain patterns in Italian patients with osteoarthritis:
preliminary results of the MI.D.A. Study (Misurazione del Dolore
nellArtrosi). Reumatismo 56:253261
29. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K et al
(1986) Development of criteria for the classification and reporting
of osteoarthritis: classification of osteoarthritis of the knee.
Arthritis Rheum 29:10391049
30. Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D,
Brandt K et al (1991) The American College of Rheumatology
criteria for the classification of osteoarthritis of the hip. Arthritis
Rheum 34:505514
31. Streiner DL, Norman GR (1995) Health measurement scales. A
practical guide to their development and use, 2nd edn. Oxford
University Press, Oxford, pp 163180
32. Lorenzo-Seva U, Ferrando PJ (2006) FACTOR: a computer
program to fit the exploratory factor analysis model. Behav Res
Methods Instrum Comput 38:8891
33. Coste J, Boue S, Ecosse E, Leplge A, Pouchot J (2005)
Methodological issues in determining the dimensionality of
composite health measures using principal component analysis: case
illustration and suggestions for practice. Qual Life Res 14:641654
34. Horn JL (1965) A rationale and test for the number of factors in
factor analysis. Psychometrika 30:179185
35. Velicer WF (1976) Determining the number of components from
the matrix of partial correlations. Psychometrika 41:321327
36. Guadagnoli E, Velicer WF (1988) Relation to sample size to the
stability of component patterns. Psychol Bull 103:265275
37. Linacre JM (2009) A users guide to WINSTEPS-MINISTEP:
Rasch-model computer programs. Program manual 3.68.0. Chicago, IL: WINSTEPS.com. http://www.winsteps.com/a/winsteps.
pdf. Accessed 24 Sept 2009
38. Linacre JM (2002) Optimizing rating scale category effectiveness.
J Appl Meas 3:85106
39. Bae SC, Lee HS, Yun HR, Kim TH, Yoo DH, Kim SY (2001) Crosscultural adaptation and validation of Korean Western Ontario and
McMaster Universities (WOMAC) and Lequesne osteoarthritis
indices for clinical research. Osteoarthr Cartil 9:746750
40. Bland JM, Altman DG (1997) Cronbachs alpha. BMJ 314(7080):572
41. Jette AM, McDonough CM, Ni P, Haley SM, Hambleton RK,
Olarsch S et al (2009) A functional difficulty and functional pain
instrument for hip and knee osteoarthritis. Arthritis Res Ther 11:R107
42. Merbitz C, Morris J, Grip JC (1989) Ordinal scales and
foundations of misinference. Arch Phys Med Rehabil 70:308312
43. Hopman-Rock M, Odding E, Hofman A, Kraaimaat FW, Bijlsma
JWJ (1997) Differences in health status of older adults with pain
in the hip or knee only and with additional mobility restricting
conditions. J Rheumatol 24:24162423
44. Lequesne MG (1997) The algofunctional indices for hip and knee
osteoarthritis. J Rheumatol 24:779781
45. Perruccio AV, Stefan Lohmander L, Canizares M, Tennant A,
Hawker GA, Conaghan PG et al (2008) The development of a
short measure of physical function for knee OA KOOS-Physical
Function Shortform (KOOS-PS)an OARSI/OMERACT initiative. Osteoarthr Cartil 16:542550
46. Davis AM, Perruccio AV, Canizares M, Tennant A, Hawker GA,
Conaghan PG et al (2008) The development of a short measure of
physical function for hip OA HOOS-Physical Function Shortform
(HOOS-PS): an OARSI/OMERACT initiative. Osteoarthr Cartil
16:551559