OsteoArthritis and Cartilage (2005) 13, 854e860

ª 2005 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.joca.2005.05.006

International
Cartilage
Repair
Society

Assessment of the Lequesne index of severity for osteoarthritis

of the hip in an elderly population
J. Dawson D.Phil., Reader in Health Services Researchy*, L. Linsell M.Sc., Medical Statisticianz,
H. Doll M.Sc., Medical Statisticianz, K. Zondervan D.Phil., MRC Fellow and Epidemiologistx,
P. Rose M.B., B.Chir., F.R.C.G.P., General Practitioner and University Lecturerk,
A. Carr F.R.C.S. (Ortho), Professor of Orthopaedic Surgery{,
T. Randall M.A., M.R.C.G.P., General Practitioner and Senior Research Fellowk and
Professor R. Fitzpatrick Ph.D.z
y School of Health and Social Care, Oxford Brookes University, Marston Road Campus,
Jack Straws Lane, Oxford OX3 0FL, UK
z Department of Public Health, University of Oxford, Old Road, Oxford OX3 7LF, UK
x Wellcome Trust Centre for Human Genetics, University of Oxford, UK
k Department of Primary Health Care, University of Oxford, Old Road, Oxford OX3 7LF, UK
{ Nuffield Orthopaedic Centre, Windmill Road, Oxford OX3 7LD, UK

Summary
Objective: To assess the measurement properties of the Lequesne index of severity for osteoarthritis of the hip (LISOH) together with its
overall usefulness with reference to the original stated aims.
Method : Postal questionnaire was sent to a random sample of 5500 Oxfordshire residents, aged 65 and above. Respondents with hip
symptoms at baseline (but without verification of a diagnosis) were sent an identical follow-up questionnaire 12 months later. The
questionnaire included a general health section, including the Short Form-36 survey, and a hip section which began with a screening question
about hip pain. Respondents who reported having a prolonged episode of hip pain were asked to complete the LISOH.
Results: At baseline, response rate of 66.3% (3341/5039) was obtained from eligible participants; 19.2% (610/3175) of respondents reported
having hip pain. Internal reliability (Cronbach’s alpha) was 0.84 (95% CI: 0.81e0.86) for all 11 items of the LISOH; however, factor analysis
identified two factors (sub-scales): ‘function and mobility’ and ‘pain and discomfort’. Rasch analysis revealed that the two factors were only
unidimensional when applied to sub-groups of respondents. Convergent validity of the LISOH was questionable, as the ‘function and mobility’
factor was more highly correlated with SF-36 bodily pain score than was the ‘pain and discomfort’ factor. The assessment of sensitivity over
time was problematic due to changing patterns of symptomatic weight-bearing joints over time.
Conclusions: The current study identifies major limitations with the LISOH e particularly if used as a single composite measure.
ª 2005 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Key words: Hip, Osteoarthritis, Health status, Outcomes.

Introduction of severity have increasingly been devised that focus on

The ability to accurately assess severity and change in
peoples’ symptoms, in a standard, reliable and valid way, is
essential to the evaluation of the need for, and response to,
health care interventions. Recognition of this fact, together
with increasing societal demands for evidence of the
benefits (or otherwise) of health care interventions, has
led to an enormous demand for standard, validated health
status measures1.
In order that they might be particularly responsive to
changes in a person’s clinical status, standard assessments
*Address correspondence and reprint requests to: Dr Jill Dawson,
Reader in Health Services Research, School of Health and Social
Care, Oxford Brookes University, Marston Road Campus, Jack
Straws Lane, Oxford OX3 0FL, UK. Tel: 44-1865-227136; Fax: 44-
1865-485297; E-mail: jdawson@brookes.ac.uk
Received 15 March 2005; revision accepted 29 May 2005.
specific diseases, conditions, or interventions (e.g., myas-
thenia gravis2, arthritis3, hip replacement4), and the effects
that these have on peoples’ everyday functioning. Such
specific methods of assessment include standard clinical
assessments e devised and conducted by clinicians, and
patient-based questionnaires, which, in their purest form, are
generated from interviews with patients. Each of these
methods tends to reflect different perspectives, to a greater
or lesser extent, as patients and clinicians may differ in terms
of their priorities5.
In order for questionnaires to be considered suitable for
evaluating health care interventions it is now generally
agreed that they should be demonstrated to be reliable,
valid, sensitive to clinical change, and to attract a high rate
of completion6. In addition, if an instrument is to perform
satisfactorily as a single composite scale, ideally, such
a scale should be demonstrably unidimensional i.e., tap

854
Osteoarthritis and Cartilage Vol. 13, No. 10
a single underlying construct, and not consist of a number
of distinct sub-scales7.
This paper evaluates a particular instrument in order to
assess its suitability according to above criteria, as well as
in relation to its original stated aims.
Devised by rheumatologists during the 1980s, the
Lequesne index of severity for osteoarthritis of the hip
(LISOH) is an 11-item questionnaire designed to obtain
information of a subjective nature, from patients, about their
diseased hip. Patients’ responses are then used to
generate a single composite scale8e10. Since its develop-
ment, the LISOH has been widely used. Most often it has
been used as part of an interview/assessment by clinicians.
However, versions to facilitate self-completion have also
been employed (and assessed) in a number of studies11,12.
At the time of its development, the measurement properties
of the LISOH received only very limited attention, although
more recently, some additional evaluations have been
conducted within small scale studies12,13.
The LISOH has three stated purposes10: (1) to facilitate
comparisons based on levels of severity of patients’ hip
symptoms in the trials of new drugs, (2) to evaluate long-
term treatment effects (particularly for hip osteoarthritis
(OA)), and (3) to standardise decision-making regarding the
need for hip replacement.
This paper presents results from an analysis of data
obtained from a large scale population study of hip
symptoms in elderly people14. The analysis assesses the
LISOH questionnaire’s item response and scale properties,
that is: factor structure, dimensionality, internal reliability
and convergent validity. The overall usefulness of the
LISOH is then considered with reference to its stated aims.
Methods
Local research ethics committee approval was obtained
for the study (Applied and Qualitative Research Ethics
Committee (AQREC) reference A01.060).
STUDY POPULATION
A random sample of 5500 Oxfordshire residents, aged 65
and above, was obtained from the Oxfordshire Health
Authority register representing January 2002. A postal
questionnaire and covering letter was sent out to everyone
within a 2-week period in April 2002 and followed up with
two postal reminders (including a second copy of the
questionnaire). Respondents who reported hip symptoms at
baseline were sent an identical follow-up questionnaire 12
months later.
QUESTIONNAIRE
The questionnaire contained a general section which
consisted of a small number of demographic items and the
Anglicised version of the Short Form-36 general health
questionnaire15,16. The SF-36 contains 36 items and is
widely used as a generic health status instrument. It
provides scores on eight dimensions: physical functioning,
role limitations due to physical problems, bodily pain, social
functioning, general mental health, role limitations due to
emotional problems, energy/vitality and general health
perceptions representing the last 4 weeks. Scores for each
dimension range from 0 (poor health) to 100 (good health).
We selected the SF-36 as a measure of general health
855
status because it has been extensively validated, including
in relation to postal administration to elderly populations17,18.
The questionnaire also included a hip section which
began with a screening question that has been used by
others19: ‘‘During the past 12 months, have you had pain in
or around either of your hips on most days for one month or
longer?’’ A separate knee section began with an identically
worded question except that the word ‘knee’ replaced ‘hip’.
Respondents who reported having hip symptoms were
asked to complete the LISOH which had been formatted to
aid self-completion.
The LISOH is a composite measurement score ranging
from 1 to 24 points based on summed responses to 11
items. The score is intended to be calculated separately for
each hip. The first five items are concerned with the
presence of pain and discomfort when remaining in certain
positions or performing particular movements. These in-
clude: pain in bed at night, morning stiffness, walking,
standing for 30 min and sitting for 2 h. Each of these items
offers a maximum of three response categories and relate
to individual joints. As is customary, we therefore included
separate response boxes for the right and left leg. The
remaining six items relate to functional status and include
maximum walking distance and use of a walking aid. These
offer three response categories each, apart from ‘maximum
walking distance’, which offers five.
In addition to the LISOH, a separate item was added,
which asked patients to rate the pain severity in their
symptomatic hip(s), during the last 4 weeks, on a scale of
1e6 ranging from ‘none’ to ‘very severe’.a The 12-month
follow-up questionnaire was identical to the baseline
questionnaire.
STATISTICAL ANALYSIS
All analyses involving the LISOH have been based on
people who reported having unilateral hip pain (n Z 471),
since we were unable to obtain a single, overall LISOH
score for people with more than one symptomatic hip. The
completion rates of the LISOH and the SF-36 were
calculated to examine the proportion of respondents for
whom a total score could not be calculated due to a missing
response to one or more items within the scale.
Internal reliability of the LISOH was assessed using
Cronbach’s alpha coefficient22 and 95% confidence inter-
vals were calculated using the bootstrap method with
Normal approximation23. This coefficient measures the
extent to which items in a scale correlate with each other
and hence their degree of consistency in measuring the
same underlying construct.
To examine the underlying dimensions of severity
measured by the items of the LISOH, an exploratory factor
analysis (principal components method) was performed,
followed by an oblique (promax) rotation of retained
factors24. Examination of the dimensionality of the LISOH,
and the functioning and fit of individual items, was
undertaken by fitting a Rasch unidimensional measurement
model in RUMM201025. The Rasch model assesses the
unidimensionality of items in a scale based upon the
assumption that, as a person’s disability or symptoms
a
A precise rating of pain severity per se is not included within the
LISOH, and Lequesne et al. employed a separate measure in the
form of a visual analogue scale (VAS). We used a rating scale
instead because there is some evidence to suggest that many
elderly people experience difficulties with conceptualising and
providing responses using the VAS format 20,21.
856 J. Dawson et al.: Assessment of the LISOH

increase, the probability of a maximum score on the item Table I

increases26,27. Response rates for the Lequesne hip index compared to the SF-36
Convergent validity was assessed by calculating Spear- dimensions amongst people with one symptomatic hip (n Z 471)
man’s rank correlation coefficient for LISOH with the eight Completion rate* (%)
dimensions of the SF-36 and the self-rated pain severity
Number Unilateral hip cases
item. This was repeated using the component factors of the
of items
LISOH from the factor analysis. We hypothesized a strong
correlation of the LISOH with the physical components of Lequesne hip index 11 71.1
the SF-36 and the self-rated pain severity item, and SF-36 dimensions
a weaker correlation with the mental components of the Physical function 10 76.0
SF-36. All analyses were conducted using Stata 8.0 unless Role limitation (physical) 4 87.1
otherwise stated. The significance level throughout was set Pain 2 96.2
at two-sided P ! 0.05. Energy/vitality 4 86.0
Social function 2 94.3
Role limitation (emotional) 3 88.1
Mental health 5 86.8
Results General health perception 5 86.2
STUDY POPULATION Total SF-36 scale 36 57.5
From the original random sample of 5500 people, 3341 *The Lequesne index and separate dimensions of the SF-36
completed and returned a questionnaire, giving a response were only computed where people provided responses to every
rate of 66.3% (3341/5039) of those eligible for study constituent item.
participation. At baseline, 19.2% (610/3175) respondents
reported having hip pain on most days for 1 month or longer
during the last year. The overall response rate obtained for (promax) rotation. Two factors with an eigenvalue greater
symptomatic individuals followed up at 1 year was 81.6% than or equal to 1.0 were extracted and the results of this
(498/610). Further details of the study design, procedures analysis are shown in Table III.
and sample have been provided elsewhere28. Most of the six items (items 6e11) relating to the first
For the purposes of this analysis, we were unable to ‘function and mobility’ factor had high loadings (O0.6) while
calculate the LISOH for 21.8% (133/610) of people who the five items (items 1e5) concerned with the second ‘pain
reported hip pain at baseline since they had more than one and discomfort’ factor had at least moderately high loadings
symptomatic hip.b We, therefore, excluded from the (O0.5).
analysis people who had bilateral hip pain and calculated
the LISOH for the 471 (77.2%) who had reported only
unilateral hip pain (six people did not provide details on the RASCH ANALYSIS
number of hips affected and were excluded from the The Rasch analysis was run on three patient groups: (a)
analysis). This comprised 262 (56%) people who had all 471 people reporting a unilateral symptomatic hip with or
reported having a symptomatic knee as well as a symptom- without a symptomatic knee, (b) 209 people with a unilateral
atic hip and 209 (44%) who only had the single symptomatic hip problem only, and (c) 262 people with a unilateral hip and
hip (without any knee symptoms). The median (IQR) age of a symptomatic knee joint. Initially, a unidimensional Rasch
the unilateral hip cases (n Z 471) was 73 (68e79) years model was fitted to all 11 items of the LISOH (see Table IV).
and 60.5% were females. In all three groups of patients, the full 11-item Lequesne
questionnaire appeared not to be unidimensional, showing
COMPLETION AND ITEM RESPONSE RATES a highly statistically significant deviation from a unidimen-
sional model in each case (P ! 0.0001). The Rasch model
The completion rates for the LISOH and the SF-36 scales
for the 471 cases are presented in Table I. The overall
completion rate for the SF-36 was lower (57.5% SF-36 vs Table II
71.1% LISOH, P ! 0.001) due to the larger number of items Item-total correlation and summary Cronbach’s alpha coefficient (a)
(36 vs 11) that had to be completed to permit a total score to for items of the LISOH amongst people with one symptomatic hip
be calculated. However, the completion rates varied across (n Z 471)
the different SF-36 sub-scales, all of which obtained better Item Description of item n Item-total
response rates than the 11-item LISOH. correlation
1 Pain or discomfort in bed at night 390 0.84
SCALE PROPERTIES 2 Morning stiffness or pain 387 0.83
3 Pain when walking 386 0.83
The Cronbach’s alpha for all 11 items of the LISOH was 4 Increase in pain when remain standing 382 0.83
0.84 (95% CI: 0.81e0.86). The reliability for the 11 test 5 Pain or discomfort when sitting 388 0.83
scales, each of which representing the overall alpha 6 Maximum walking distance 355 0.82
excluding each item in turn, is reported in Table II and 7 Use of a walking aid 385 0.83
show that the removal of any one item did not improve the 8 Can put on socks 386 0.82
9 Can pick up object from floor 387 0.81
overall alpha to any important extent.
10 Can go up and down stairs 388 0.81
The 11 items were entered into an exploratory factor 11 Can get into and out of car 390 0.81
analysis with principal components extraction and oblique
Overall a 0.84
95% Confidence interval* (0.81e0.86)
b
There are no instructions on how to derive an overall index for
each person based on these separate scores. *Bootrap method with Normal approximation (1000 reps).
Osteoarthritis and Cartilage Vol. 13, No. 10 857

Table III was next applied to each of the two factors separately. In
Principal factor loadings for items of the LISOH amongst people the total sample (N Z 471), there was significant deviation
with one symptomatic hip (n Z 471) from unidimensionality within both the Factor 1 ‘function and
Item Description of item Factor loading mobility’ (P Z 0.033) and Factor 2 ‘pain and discomfort’
(P Z 0.004) items, suggesting that the items were not
1 2 ‘tapping’ the same underlying construct. Within Factor 1, the
1 Pain or discomfort in bed at night
0.07 0.51 thresholds of item 6 (maximum walking distance) were not
2 Morning stiffness 0.09 0.52 ordered, suggesting that the responses to this item were not
or pain related to the underlying severity of the condition. In both of
3 Pain when walking 0.09 0.60 the subsamples, item 6 again showed disordered thresh-
4 Increase in pain 0.03 0.56 olds. However, there was no significant deviation from
when remain standing unidimensionality in either factor in either subsample.
5 Pain or discomfort
0.07 0.55
when sitting
6 Maximum walking distance 0.68 0.04 CONVERGENT VALIDITY
7 Use of a walking aid 0.71
0.10
8 Can put on socks 0.68 0.02 Table V shows the Spearman rank correlation coeffi-
9 Can pick 0.81
0.03 cients related to the overall LISOH (11-item) scale and each
up object from floor of the two sub-scales (factors) with each dimension of the
10 Can go up and down stairs 0.72 0.09 SF-36 and the separate self-rated pain severity item. All
11 Can get into 0.59 0.18
and out of car
correlation coefficients shown in Table V were significantly
different from zero (P ! 0.05). The correlation between

Rasch analysis of LISOH amongst (a) people with one symptomatic hip with or without a symptomatic knee (n Z 471), (b) people with one
symptomatic hip and no symptomatic knee (n Z 209) and (c) people with one symptomatic hip and a symptomatic knee (n Z 262)
(a) N Z 471 (hip G knee)
Items not properly ordered
Item misfit
Residual O j2j
(b) N Z 209 (hip only)
Items not properly ordered
Item misfit
Residual O j2j
(c) N Z 262 (hip & knee)
Items not properly ordered
Item misfit
Residual O j2j
Table IV
All 11 items overall fit to Factor 1 (items 6e11) Factor 2 (items 1e5)
unidimensional model
c2 Z 188.95, df Z 77, c2 Z 60.44, df Z 42, c2 Z 48.10, df Z 25,
P ! 0.000001 P Z 0.0325 P Z 0.0036
Item 6 (
0.076, 0.235, Item 6 (
0.704, 0.068, None

0.204,
0.439, 0.485)
0.149,
0.217, 1.002)
Item 1 c2 Z 48.2, P ! 0.000001 Item 8 c2 Z 13.69, P Z 0.057 Item 1 c2 Z 15.4, P Z 0.009
Item 3 c2 Z 18.3, P Z 0.0108 Item 4 c2 Z 11.7, P Z 0.039
Item 9 c2 Z 22.1, P Z 0.0025 Item 5 c2 Z 9.83, P Z 0.080
Item 10 c2 Z 28.8, P Z 0.0002
Item 11 c2 Z 48.2, P Z 0.0011
Item 1 residual Z 5.454 None None
Item 2 residual Z 4.046
Item 9 residual Z
3.558
Item 10 residual Z
3.979
Item 11 residual Z
2.530
c2 Z 104.33, df Z 22, c2 Z 18.79, df Z 12, c2 Z 10.85, df Z 10,
P ! 0.000001 P Z 0.094 P Z 0.370
Item 6 (0.116, 0.403, Item 6 (
0.537,
0.032, None

0.377,
0.780, 0.638)
0.361,
0.330, 1.259)
Item 1 c2 Z 18.7, P Z 0.00009 Item 6 c2 Z 4.92, P Z 0.085 None
Item 2 c2 Z 11.0, P Z 0.0041
Item 3 c2 Z 6.09, P Z 0.0477
Item 9 c2 Z 10.2, P Z 0.0061
Item 10 c2 Z 27.9, P Z 0.000001
Item 11 c2 Z 21.1, P Z 0.000028
Item 1 residual Z 3.454 None None
Item 2 residual Z 2.232
Item 9 residual Z
2.464
Item 10 residual Z
3.469
Item 11 residual Z
2.854
c2 Z 72.66, df Z 33, c2 Z 21.85, df Z 18, c2 Z 16.01, df Z 15,
P Z 0.000083 P Z 0.239 P Z 0.381
Item 6 (
0.243, 0.125,
0.103, Item 6 (
0.836, 0.123, None

0.229, 0.449)
0.031,
0.153, 0.898)
Item 1 c2 Z 17.7, P Z 0.0005 Item 9 c2 Z 7.51, P Z 0.057 None
Item 7 c2 Z 6.76, P Z 0.0799
Item 8 c2 Z 13.6, P Z 0.0035
Item 1 residual Z 4.058 None None
Item 2 residual Z 3.593
Item 9 residual Z
2.395
Item 10 residual Z
2.094
858 J. Dawson et al.: Assessment of the LISOH

Table V We were unable to examine the sensitivity of the LISOH

Correlation of the LISOH with pain severity (self-rated) and the to change over time because we could only compare the
SF-36 amongst people with one symptomatic hip (n Z 471) baseline and follow-up scores in respondents who had
Spearman’s rho correlation coefficient a one-to-one mapping of scores on the same joint; in other
words, there was an insufficient number of unilateral cases
Lequesne Factor 1: Factor 2: who reported joint pain in the same unilateral joint 1 year
index Function Pain &
later, and who also provided complete data for the LISOH
(all 11 items) & mobility discomfort
(items 6e11) on both occasions. Thus, of the unilateral hip cases, at
baseline, only 185 (39.3%) had joint pain in the same hip 1
Pain severity (self-rated)* 0.57 0.45 0.59 year later and just over half of these 106 (57.3%) had
SF-36 scoresy a complete set of item responses for the LISOH.
Physical functioning
0.78
0.83
0.43
Role limitation
0.47
0.50
0.27
(physical)
Pain score
0.65
0.61
0.48 Discussion
Energy/vitality
0.49
0.46
0.36 The LISOH was originally designed to be both a site- and
Social functioning
0.61
0.61
0.38
a condition-specific measure. One stated aim of the LISOH
Role limitation
0.36
0.37
0.21
(emotional) was that it should be used as an outcome measure in
Mental health
0.28
0.29
0.17 clinical trials of new drugs to treat OA hip. Indeed, it has
General health
0.39
0.43
0.20 been recommended by Outcome Measures in Rheumatol-
perception ogy international network for this purpose29. In order for
a questionnaire to be considered suitable for such a purpose
*High positive correlation implies agreement as, in both scales, it should be demonstrated to have particular measurement
higher scores denote greater severity. properties. We undertook an analysis of the LISOH to
yHigh negative correlation implies agreement, as higher assess these properties.
Lequesne scores denote greater severity while lower SF-36 scores The LISOH is scored to produce a single composite scale e
denote greater severity. per hip e although around half of the questions are quite
generic in their focus. In the current study, problems
immediately arose in applying the LISOH in any analysis,
LISOH and self-rated pain severity was positive (0.57), because no instructions are given regarding methods to
indicating that the higher (more severe) the respondents’ combine the scores obtained for both hips that might take
LISOH, the higher they tended to rate their pain severity. account of the increased impact on health status that arises
Correlations with the SF-36 dimensions were all negative, where two hips are symptomatic (compared with just one)14.
implying that the higher (more severe) the LISOH, the Studies that have previously used the LISOH have not raised
worse the respondent’s general health status. The stron- this issue. A less site-specific instrument e such as the
gest negative correlations were obtained with the physical Arthritis Impact Measurement scales3 or the Western Ontario
function score (
0.78) and the SF-36 bodily pain score and McMaster Universities (WOMAC)30, would likely be less
(
0.65). The weakest negative correlations were with the problematic if used for evaluating systemic forms of treatment
mental health score (
0.28) and the emotional role for OA of weight-bearing joints.
limitation score (
0.36). For the particular purposes of this paper, in order to side-
The ‘function and mobility’ factor (Factor 1) was more step the practical difficulties of applying the LISOH in an
highly correlated with the SF-36 physical function and analysis, we used LISOH responses received only from
physical role limitation scores as compared with the ‘pain those people who reported having one symptomatic hip.
and discomfort’ factor (Factor 2), as might be expected. Within this subgroup, the overall completion rate for the
Factor 2 was more highly correlated with self-rated pain LISOH was fairly good (71%), given the elderly status of our
severity but, surprisingly, showed less correlation with the study population. It was, however, somewhat lower than for
SF-36 bodily pain score compared to Factor 1. each domain of the SF-36. One item, concerned with
maximum walking distance, was clearly particularly prob-
SENSITIVITY OVER TIME lematic with 25% of respondents leaving this item blank.
Assessed on all people with one symptomatic hip only,
Outcome data for the unilateral cases at 1-year follow-up while the internal consistency of the 11-item LISOH was
are presented in Table VI. found to be satisfactory (Cronbach’s a 0.84) for making
group comparisons e where levels of reliability above 0.70
are required, it was not adequate to justify the use of the
LISOH for individual patient assessment e where levels of
Table VI reliability above 0.90 have been recommended31e33. This
Outcome of people with one symptomatic hip (n Z 471) at 1-year clearly has implications for another of the stated aims of the
follow-up LISOH: to standardise decision-making regarding the need
Outcome at 1-year follow-up Baseline unilateral for hip replacement e in individuals. In fact other instru-
hip cases (n Z 471) ments e in particular, the New Zealand standard criteria34 e
n (%) have been developed specifically with this purpose in mind
and, while problems remain in using standard question-
Did not complete follow-up 86 (18.3) naires for this purpose, the New Zealand criteria have been
questionnaire or missing data reported as being helpful in assigning priority for hip
Bilateral hip joint pain 26 (5.5) replacement35.
No pain in the same unilateral joint 174 (36.9)
Joint pain in same unilateral joint 185 (39.3) Regarding other measurement properties of the LISOH,
Rasch analysis demonstrated that the index, comprising all
Osteoarthritis and Cartilage Vol. 13, No. 10
11 items, was not unidimensional and exploratory factor
analysis revealed the presence of two sub-scales. How-
ever, when the Rasch model was applied to each of these
sub-scales in turn, there still remained a significant de-
viation from unidimensionality within both of the factors
suggesting that the items were not ‘tapping’ the same
underlying construct. Further exploration of sub-groups,
using the Rasch model, did not confer any useful
clarification of this finding.
Assessment of the validity of the LISOH in relation to the
SF-36 dimensions revealed one anomalous finding that
appeared to undermine the validity of the LISOH.
We were unfortunately unable to examine the sensitivity/
responsiveness of the LISOH because there was an
insufficient number of unilateral cases who reported joint
pain in the same unilateral joint 1 year later, and who also
provided complete data for the LISOH on both occasions.
Nevertheless, a previous study has demonstrated the
WOMAC30, to be more responsive than the LISOH12,36.
Our study has a number of limitations14. However, the
main limitation in relation to this paper concerns our case
definition, as we could not verify that reported hip symptoms
were due to OA. So that, for instance, some peoples’
symptoms, while experienced as coming from the hip, could
have emanated from elsewhere in the body (e.g., the
spine). Nevertheless, our screening question will have at
least identified cases with symptoms that were not trivial
and the presence of such persistent hip pain in this age-
group, more often than not, represents OA37. Instruments
need to be validated in relation to the type(s) of population
to which they will be applied, and it may be the case that the
Lequesne is only truly suitable for application in a secondary
care specialist setting, where an unequivocal diagnosis of
OA hip can be confirmed. Nevertheless, if this were the
case, it would represent a serious limitation by comparison
with other more versatile instruments (e.g., WOMAC OA
index).
Overall, the LISOH was found to be inadequate in relation
to its measurement properties and this severely limits its
ability to perform in line with the originators’ main aims,
which were probably over ambitious and may even
compete. For instance, it is difficult for a measure to be
both site-specific (hip) and condition-specific (OA) as site-
specific instruments aim to minimise the effects of ‘noise’
(symptoms from other joints, in this instance), while
condition-specific instruments aim to measure some ‘noise’
directly38. In addition, while the LISOH was designed to
capture the patient’s perspective regarding their hip prob-
lem(s), arguably, questions that are designed to do this are
best generated with patients (by interview), rather than on
behalf of patients (by clinicians)39.
The conclusion of this paper is that investigators need to
choose the most appropriate outcome measures for their
particular purpose. The current study identifies major
limitations with the LISOH e particularly if used as
a composite measure. Investigators wishing to use this
instrument should give serious prior consideration to how
they intend to apply it and to whether it is the most
appropriate measure likely to meet with the specific aims of
their work.
Acknowledgements
Financial support was generously provided by grant from
the NHS Executive (South-East Region).
859
References
1. Jenkinson C. Measuring health and medical outcomes:
an overview. In: Jenkinson C, Ed. Measuring Health
and Medical Outcomes. London: UCL Press 1994:1e6.
2. Padua L, Evoli A, Aprile I, Caliandro P, Batocchi AP,
Punzi C, et al. Myasthenia gravis outcome measure:
development and validation of a disease-specific self-
administered questionnaire. Neurol Sci 2002;23:
59e68.
3. Wallston K, Brown G, Stein M, Dobbins C. Comparing
long and short versions of the Arthritis Impact
Measurement Scales. J Rheumatol 1989;16:1105e9.
4. Dawson J, Fitzpatrick R, Carr A, Murray D. Question-
naire on the perceptions of patients about total hip
replacement. J Bone Joint Surg Br 1996;78:185e90.
5. Rothwell PM, McDowell Z, Wong CK, Dorman PJ.
Doctors and patients don’t agree: cross sectional
study of patients’ and doctors’ perceptions and
assessments of disability in multiple sclerosis. BMJ
1997;314:1580e3.
6. Guyatt GH, Feeny DH, Patrick DL. Measuring health-
related quality of life. Ann Intern Med 1993;118:
622e9.
7. Coste J, Fermanian J, Venot A. Methodological and
statistical problems in the construction of composite
measurement scales: a survey of six medical and
epidemiological journals. Stat Med 1995;14:331e45.
8. Lequesne M, Mery C. European guidelines for clinical
trials of new antirheumatic drugs. EULAR Bull 1980;9:
171e5.
9. Lequesne M, Samson M. A functional index for hip
diseases. Reproducibility. Value for discriminating
drug’s efficacy. In: 15th International Congress of
Rheumatology. Paris: Expansion Scientifique Francais
1981.
10. Lequesne MG, Mery C, Samson M, Gerard P. Indexes
of severity for osteoarthritis of the hip and knee. Scand
J Rheumatol Suppl 1987;65:85e9.
11. Weigl M, Cieza A, Harder M, Geyh S, Amann E,
Kostanjsek N, et al. Linking osteoarthritis-specific
health-status measures to the International Classifica-
tion of Functioning Disability, and Health (ICF).
Osteoarthritis Cartilage 2003;11:519e23.
12. Stucki G, Sangha O, Stucki S, Michel BA, Tyndall A,
Dick W, et al. 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. Osteoarthritis Cartilage 1998;6:
79e86.
13. Theiler R, Sangha O, Schaeren S, Michel BA, Tyndall A,
Dick W, et al. Superior responsiveness of the pain
and function sections of the Western Ontario and
McMaster Universities Osteoarthritis Index (WOMAC)
as compared to the Lesquesne-algofunctional Index in
patients with osteoarthritis of the lower extremities.
Osteoarthritis Cartilage 1999;7:515e9.
14. Dawson J, Linsell L, Zondervan K, Rose P, Randall T,
Carr A, et al. Epidemiology of hip and knee pain and
its impact on overall health status in older adults.
Rheumatology 2004;43:497e504.
15. Ware-JE J, Sherbourne CD. The MOS 36-item short-
form health survey (SF-36). I. Conceptual framework
and item selection. Med Care 1992;30:473e83.
16. Brazier JE, Harper R, Jones NM, O’Cathain A, Thomas
KJ, Usherwood T, et al. Validating the SF-36 health
860
survey questionnaire: new outcome measure for
primary care. BMJ 1992;305:160e4.
17. Walters SJ, Munro JF, Brazier JE. Using the SF-36 with
older adults: a cross-sectional community-based
survey. Age Ageing 2001;30:337e43.
18. Hayes V, Morris J, Wolfe C, Morgan M. The SF-36
health survey questionnaire: is it suitable for use with
older adults? Age Ageing 1995;24:120e5.
19. Frankel S, Eachus J, Pearson N, Greenwood R, Chan P,
Peters PT, et al. Population requirement for primary
hip-replacement surgery: a cross-sectional study.
Lancet 1999;353:1304e9.
20. Herr KA, Mobily PR. Comparison of selected pain
assessment tools for use with the elderly. Appl Nurs
Res 1993;1:59e66.
21. Breivik E, Bjornsson G, Skovlund E. A comparison of
pain rating scales by sampling from clinical trial data.
Clin J Pain 2000;16:22e8.
22. Bland JM, Altman DG. Cronbach’s alpha. BMJ 1997;
314:572.
23. Efron B, Tibshirani R. Bootstrap measures for standard
errors, confidence intervals, and other measures of
statistical accuracy. Stat Sci 1986;1:54e77.
24. Fayers PM, Machin D. Factor analysis. Quality of Life.
Chichester: John Wiley & Sons Ltd 2000. pp. 91e116.
25. Rasch Unidimensional Measurement Models.
RUMM2010. 14 Dodonaea Court, Duncraig WA
6023, Australia, RUMM Laboratory Pty Ltd. 1998.
26. Hambleton RK, Jones RW. Comparison of classical
test theory and item response theory and their
applications to test development. Educ Meas Issues
Pract 1993;12:38e47.
27. Andrich D. Rasch Models for Measurement. London:
Sage Publications 1998.
28. Dawson J, Linsell L, Zondervan K, Rose P, Randall T,
Carr A, et al. Impact of persistent hip or knee pain on
overall health status in elderly people: a longitudinal
population study. Arthritis Care Res 2005;53:368e74.
29. Bellamy N, Kirwan J, Boers M, Brooks P, Strand V,
Tugwell P, et al. Recommendations for a core set of
J. Dawson et al.: Assessment of the LISOH
outcome measures for future phase III clinical trials in
knee, hip, and hand osteoarthritis. Consensus de-
velopment at OMERACT III. J Rheumatol 1997;24:
799e802.
30. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J,
Stitt LW. 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 1988;15:1833e40.
31. Nunnally JC, Bernstein IH. Psychometric Theory. New
York/Columbus, OH: McGraw-Hill 1978.
32. Streiner DL, Norman GR. Health measurement scales e
a practical guide to their development and use. New
York: Oxford University Press 1989.
33. Ware JE. SF-36 Health Survey Manual and Interpreta-
tion Guide. The Medical Outcomes Trust. Boston, MA:
Nimrod Press 1997.
34. Hadorn DC, Holmes AC. The New Zealand priority
criteria project. Part 1: Overview. BMJ 1997;314:
131e4.
35. Harry LE, Nolan JF, Elender F, Lewis JC. Who gets
priority? Waiting list assessment using a scoring
system. Ann R Coll Surg Engl 2000;82:186e8.
36. Bischoff HA, Conzelmann M, Lindemann D, Singer-
Lindpaintner L, Stucki G, Vonthein R, et al. Self-
reported exercise before age 40: influence on
quantitative skeletal ultrasound and fall risk in the
elderly. Arch Phys Med Rehabil 2001;82:801e6.
37. Bagge E, Bjelle A, Eden S, Svanborg A. Osteoarthritis
in the elderly: clinical and radiological findings in 79
and 85 year olds. Ann Rheum Dis 1991;50:535e9.
38. Dawson J, Fitzpatrick R, Murray D, Carr A. The
problem of ‘noise’ in monitoring patient-assessed
outcomes: generic, disease-specific and site-specific
instruments for total hip replacement. J Health Serv
Res Policy 1996;1:224e31.
39. Fitzpatrick R, Davey C, Buxton MJ, Jones DR.
Evaluating patient-based outcome measures for use
in clinical trials. Health Technol Assess 1998;2:1e74.