This article was downloaded by: [University of Western Ontario]

On: 07 October 2014, At: 22:53

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Ergonomics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/terg20

Development of a simple measurement scale to

evaluate the severity of non-specific low back pain for
industrial ergonomics
a b b
Yoshiyuki Higuchi , Hiroyuki Izumi & Mashaharu Kumashiro
a
School of Nursing, Fukuoka Prefectural University , Tagawa, Japan
b
Department of Ergonomics , University of Occupational and Environmental Health ,
Kitakyushu, Japan
Published online: 21 May 2010.

To cite this article: Yoshiyuki Higuchi , Hiroyuki Izumi & Mashaharu Kumashiro (2010) Development of a simple measurement
scale to evaluate the severity of non-specific low back pain for industrial ergonomics, Ergonomics, 53:6, 801-811, DOI:
10.1080/00140139.2010.489652

To link to this article: http://dx.doi.org/10.1080/00140139.2010.489652

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Ergonomics
Vol. 53, No. 6, June 2010, 801–811

Development of a simple measurement scale to evaluate the severity of non-specific low back pain
for industrial ergonomics
Yoshiyuki Higuchia*, Hiroyuki Izumib and Mashaharu Kumashirob
a
School of Nursing, Fukuoka Prefectural University, Tagawa, Japan; bDepartment of Ergonomics, University of Occupational and
Environmental Health, Kitakyushu, Japan
(Received 10 March 2009; final version received 28 February 2010)

This study developed an assessment scale that hierarchically classifies degrees of low back pain severity. This
assessment scale consists of two subscales: 1) pain intensity; 2) pain interference. First, the assessment scale devised
by the authors was used to administer a self-administered questionnaire to 773 male workers in the car
Downloaded by [University of Western Ontario] at 22:53 07 October 2014

manufacturing industry. Subsequently, the validity of the measurement items was examined and some of them were
revised. Next, the corrected low back pain scale was used in a self-administered questionnaire, the subjects of which
were 5053 ordinary workers. The hierarchical validity between the measurement items was checked based on the
results of Mokken Scale analysis. Finally, a low back pain assessment scale consisting of seven items was perfected.
Quantitative assessment is made possible by scoring the items and low back pain severity can be classified into four
hierarchical levels: none; mild; moderate; severe.

Statement of Relevance: The use of this scale devised by the authors allows a more detailed assessment of the
degree of risk factor effect and also should prove useful both in selecting remedial measures for occupational low
back pain and evaluating their efficacy.
Keywords: categorical component analysis; Mokken scale analysis; occupational low back pain; questionnaire;
severity evaluation

1. Introduction Nordic Questionnaire (SNQ) for analysis of

Low back pain (LBP) remains the most common work-
related musculoskeletal disorder and LBP among
workers is one of the main causes of sick leave (Huang
et al. 1998, Merlino et al. 2003). In order to solve the
problem of worker LBP, it is important to identify the
risk factors that exist in each workplace and devise
effective preventative measures for dealing with them
(Johanning 2000, Byrns et al. 2002). A great deal of
LBP research has been carried out in the past, but no
fundamental methodology has been established for its
prevention (Agius et al. 1994, Byrns et al. 2002). One of
the reasons for the lack of resolution of the LBP
problem in the workplace is the fact that its assessment
in the context of work-related LBP research is difficult
(Ferguson and Marras 1997, Leboeuf-Yde et al. 1997).
Most large-scale studies of work groups are carried
out using self-administered questionnaires, except
when existing, reliable medical records can be used to
gather information on LBP (e.g. van der Molen et al.
2009). The data most commonly used for such
questionnaires relate to the existence or lack of LBP
or discomfort within a specified period of time
(Ferguson and Marras 1997). The Standardised
musculoskeletal symptoms (Kuorinka et al. 1987) is
widely used as a questionnaire for self-administered
LBP or discomfort as an outcome (Hamberg-van
Reenen et al. 2008, Rehn et al. 2009, Tissot et al. 2009).
The SNQ is composed of a number of standardised
questions and can be used for general purposes but,
unfortunately, it does not present a method for
quantitative assessment of LBP.
Rather than identifying a case using one assessment
standard, such as the existence or lack of discomfort, it
is important, when conducting research that has as its
goals the identification of risk factors and assessing the
efficacy of preventative measures, to utilise a
combination of multiple items to assess the severity of
the case in a hierarchical manner (Ferguson and
Marras 1997, Hartvigsen et al. 2001, Hoozemans et al.
2002, Feng et al. 2007).
Combining multiple items relating to LBP allows a
hierarchical understanding of workers, ranging from
those who feel pain or discomfort in the lower back to
those who actually take sick leave or receive regular
treatment and the like, and this enables detailed
characterisation of the subject’s worker group.

*Corresponding author. Email: higuchi@fukuoka-pu.ac.jp

ISSN 0014-0139 print/ISSN 1366-5847 online

Ó 2010 Taylor & Francis
DOI: 10.1080/00140139.2010.489652
http://www.informaworld.com
 802 Y. Higuchi et al.
In addition, it should be possible to identify risk
factors and carry out more quantitative assessment of
intervention efficacy (Ferguson and Marras 1997,
Leboeuf-Yde et al. 1997).
In the past, Turner et al. (2004) presented a method
for hierarchical classification of numerically quantified
pain intensity in LBP patients in the context of
functional disability. Ozguler et al. (2002) presented a
severity classification technique based on scores gained
using the Dallas Pain Questionnaire for LBP in worker
groups. Further, von Korff et al. (1992) presented a
hierarchical classification tool using a combination of
numerically quantified pain intensity and functional
disability in relation to the severity of chronic pain
patients. Classification of numerical assessment into
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
categories corresponding to disability levels is useful
when clarifying LBP treatment goals (Turner et al.
2004). In addition, the method whereby a combination
of multiple assessments of LBP severities is integrated
should be useful for LBP research on worker group
subjects in the industrial context. The purpose of this
study is to develop an assessment scale of non-specific
LBP severity in ordinary worker groups.
The tool used to classify LBP severity developed in
this report can be used in epidemiological studies with
large groups as subjects, as well as in ergonomics
studies that have ordinary worker groups in the
workplace as subjects. When considering LBP pre-
ventative measures for the workplace, it is important to
inspect the working conditions and environment for
each workplace, focus on the problems and risk factors
and seek effective improvements appropriate to each
workplace. When doing so, classifying LBP severity
hierarchically allows identification of work-related risk
factors and examination of their effects as they relate to
each degree of severity. Assessing the effects of risk
factors is thought to be useful in prioritising measures
to be implemented and estimating their efficacy. In
addition, even when assessing the efficacy of LBP
preventative measures, a more detailed determination
can be made of the level of LBP that may be prevented
by a specific measure. The authors believe that their
severity classification can be of valuable use in planning
workplace and work improvements that correspond to
LBP severity and in prioritising those improvements.
2. Method
The authors combined the outcome measurements
used in common by past LBP studies and decided to
adopt hierarchical classification of LBP severity. The
hierarchical classification method of severity of non-
specific LPB in Japanese manufacturing workers was
examined. In this research, two studies were conducted
in stages. Study 1 was a preliminary study for
considering the classification method based on
severity of LBP. This study was conducted from April
to December 2005. Study 2 was a verification study,
conducted based on the results obtained from study 1,
using a more heterogeneous group to verify the validity
and reliability of the classification method. Study 2 was
conducted from April to December 2006.
There have been some reported LBP assessment
techniques that have quantitative assessment
calculation formulae and hierarchical classification
methods. However, not many of these assessment
methods indicate in detail the process by which the
particular formula or classification method was
determined. This study explains clearly the analytical
process followed in devising the scoring system and the
hierarchical classification method. This is significant in
the context of validity verification and should be useful
in analysing the results of actual research carried out
using this severity classification method.
2.1. Study 1: Preliminary study
2.1.1. Participants
As the basic demographic information, age, gender,
height, weight, type of work schedule and working
hours were obtained. The research subjects were
workers of three different automobile manufacturers
with production facilities in Japan. The occupational
physicians of each company explained the objective of
the study to their particular workers, provided
information and distributed questionnaires to 1099
workers. Of the 1099 employees to whom a
questionnaire was distributed, valid questionnaires
were collected from 933 participants, representing a
response rate of 84.9%. Participants were further
screened for male gender and a work cycle time of
3 min or less. Males were used as females represented
only 0.6% of the participants (6/933). Cycle time was
set at 3 min or less to enable analysis to be completed
for assembly line workers only and to reduce
population and risk factor variance (Wells et al. 1997).
A total of 16.5% of the participants (154/933) were
excluded using this criterion. This resulted in a total of
773 participant’s data being included in the analysis for
study 1.
2.1.2. Questionnaire
As the method for the classification of the severity of
LBP has been widely used as an outcome measurement
for non-specific LBP in current occupational LBP
studies, the measurement concerning two areas of pain
intensity and pain interference was adopted.
A questionnaire survey, written in Japanese, was
 Ergonomics
conducted for these two areas related to the severity
of LBP.
2.1.3. Pain intensity
Three items concerning pain intensity were evaluated on
a numerical rating scale (NRS) of 0–10 for current pain,
average pain during the last 6 months and worst pain
during the last 6 months. Zero corresponded to abso-
lutely no pain and 10 to the worst pain. The average of
the three pain intensity items was used as the overall
measurement of pain intensity (von Korff et al. 1992).
2.1.4. Pain interference
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
For the items regarding pain interference, four
modified items from the SNQ (Kuorinka et al. 1987)
were used. The original version by Kuorinka et al.
(1987) has seven dichotomous questions (answered by
yes or no) and two ordinal multiple-choice single
answer questions. These dichotomous questions are
suitable for fundamental epidemiological studies, but
to perform quantitative analysis it is better to use
hierarchical multiple-choice, single answer questions,
as was used in this study. The contents of the options
in each question were determined through consultation
with occupational physicians employed in the respec-
tive workplaces. Categorical principal components
analysis (CATPCA) was conducted to quantify the
multiple-choice for the respective four questions
regarding pain interference. The sum of quantified
items was used as the pain interference score. The
details of the four questions are shown below.
. Question 1: How often have you had low back
pain during the last 6 months?
Choices (five items): ‘almost daily’; ‘about 10
days a month’; ‘about 1 day a month’; ‘about
1 day during the last 6 months’; ‘none’.
. Question 2: Has your work activity been affected
by low back pain?
Choices (four items): ‘none’; ‘change of job
description or workplace’; ‘sick leave’; ‘affected
(except sick leave)’.
. Question 3: Has your activity, except work
activity, been affected by low back pain?
Choices (two items): ‘none’; ‘affected’.
. Question 4: Have you been seen by a doctor, phy-
siotherapist, chiropractor or other such person
because of low back trouble during the last 6
months?
Choices (five items): ‘none’; ‘I was hospitalised’; ‘I
only saw a physician’; ‘I saw an acupuncture
practitioner/masseuse/chiropractor’; ‘I saw both
of these’.
803
2.1.5. Statistical methods
The measurement values were converted to numerical
values to perform a quantitative LBP evaluation. The
authors assigned scores to the multiple-choice answers
for the four questions regarding pain interference and
used multivariate analysis to study the correlation
among the items. As the multiple-choice answers used
for study 1 were nominal variables, CATPCA was
used. The authors provisionally set the category scores
for each item taking into consideration the
standardised centroid coordinates of each category
obtained using CATPCA. This analysis was performed
using the SPSS1 15.0 (Japanese version) Category
15.0 statistical analysis software (SPSS Inc., Chicago,
IL, USA).
Mokken Scale Analysis (MSA) (Mokken 1971) is a
method for assessing the uni-dimensionality of a set of
items (Gillespie et al. 1987, Hosokawa et al. 1994). In
this study, MSA was used to verify the construct
validity of the calculated category score. Construct
validity was considered weak if the overall scalability
coefficients (H) were more than 0.3 and less than 0.4;
moderate if 0.4 5 H 5 0.5; and a scale is considered
strong if H 4 0.5 (van der Ark 2007, Kersten et al.
2009). For this analysis, the statistical analysis
software R Mokken package 1 (R Development Core
2007) was used.
2.2. Study 2: Verification study
2.2.1. Participants
During study 2, a sample from a wider range of
workplaces was collected to correct the contents of
the scale composition items and verify their validity
based on the results obtained from study 1. This
study was conducted in 224 workplaces in 11
companies with production facilities in Japan in the
automobile manufacturing, steel manufacturing,
metal products manufacturing, electrical equipment
manufacturing and conveyance equipment
manufacturing industries. The procedure used to
conduct the study was the same as for study 1. Of the
5667 participants obtained from this study, 634
participants were rejected because of incomplete
questionnaires, leaving 5033 participants for analysis.
The participants of study 1 were not included in
study 2.
2.2.2. Change of items regarding hindrance of activity
due to low back pain
Based on the results of study 1, part of the contents of
the multiple-choice questions about pain interference
 804 Y. Higuchi et al.

was changed. The modified questions are shown as

follows:
. Question 2: Has your work activity been affected
by low back pain?
Choices (three items): ‘none’; ‘affected (except
sick leave)’; ‘sick leave’.
. Question 4: Have you been seen by a doctor,
physiotherapist, chiropractor or other such
person because of low back trouble during the
last 6 months?
Choices (three items): ‘none’; ‘I received
temporary treatment’; ‘I received regular
treatment’.
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
3.1.2. Low back pain symptoms
Table 1 shows the variable of low back symptoms in
study 1. The prevalence rate during the last 6 months
was 57.2% (442/773). In regard to the effect of LBP on
work activity, 19 participants (2.5%) responded ‘sick
leave’, while 37 participants (4.7%) responded that
they visited a medical clinic for LBP. The pain intensity
was 4.6 (SD 1.9).
3.1.3. Scoring items regarding pain interference
CATPCA identified one component for which the
eigenvalue was 1 or more. The eigenvalue of the
identified component was 2.809. The variance rate

explained by the identified component was 56.2%. The

3. Results
3.1. Study 1
3.1.1. Participants
The average age of the participants in study 1 was 39.6
years (SD 11.1, range 18–59), average height was
168.8 cm (SD 6.0, range 149–188), average weight was
61.6 kg (SD 9.0, range 42–113) and average BMI was
21.6 kg/m2 (SD 2.7, range 16.2–37.8). Regarding the
work schedule, 97.2% (751/773) of the participants
worked in shifts. For the length of employment in the
current job, the mode was ‘2 to 5 years’ (19.5%,
151/773).
component loading for all four items was 0.6 or more.
The component loading for each question was 0.878 for
‘frequency of LBP’, 0.672 for ‘effect on work activity’,
0.638 for ‘effect on non-work activity’ and 0.641 for
‘received treatment’. This result supports the fact that
the four questions regarding pain interference used for
this research measured the same, single construct.
The value of the centroid coordinates for each item
category obtained using CATPCA (Table 1, column 5)
was rounded down to obtain the scale scores for the
multiple-choice of each question (Table 1, column 6).
The internal consistency (Cronbach a) for the four-
scored items was calculated to be 0.665.

Table 1. Summary of low back symptoms in study 1 (n ¼ 773).

Response % Pain intensity Centroid coordinates in CATPCA Scorea

Prevalence rate during last 6 months 442 57.2 4.6
Frequency of low back pain during the last 6 months
None 331 42.8 – 70.952 0
About 1 d 39 5.0 3.3 0.366 0
About 1 d per month 113 14.6 3.3 0.273 0
About 10 d per month 149 19.3 4.7 0.615 0
Almost daily 141 18.2 6.0 1.265 1
Affect on work activity
None 325 42.0 4.3 70.280 0
Affected (except sick leave) 88 11.4 5.2 1.418 1
Sick leave 19 2.5 7.1 2.308 2
Changed the content of work 10 1.3 5.4 1.514 1
Affect on non-work activity
None 342 44.2 4.3 70.246 0
Affected 100 12.9 5.8 1.654 1
Received treatment
None 342 44.2 4.3 70.245 0
Alternative medicine 63 8.2 5.8 1.482 1
Medical clinic 32 4.1 5.9 1.904 1
Both 5 0.6 6.8 2.148 2

CATPCA ¼ categorical principal components analysis.

a
The value of the centroid coordinates was rounded down to obtain the score of each choice.
 Ergonomics
3.1.4. Mokken Scale Analysis in study 1
MSA was conducted to verify the construct validity of
the four-scored items. The scalability coefficients (Hij)
among the items were all positive and satisfied the
validity first standard. The coefficients for the indivi-
dual items (Hi) were 0.308 for ‘frequency of LBP’,
0.431 for ‘effect on work activity’, 0.406 for ‘effect on
non-work activity’ and 0.395 for ‘received treatment’.
The obtained coefficients were all 0.3 or larger. The
scalability coefficient H for all four items regarding
pain interference was 0.387. This result was considered
weak as an overall scale.
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
3.1.5. Relationship with characteristic pain intensity
Table 1 shows the relationship between the choices for
each question as pain interference and pain intensity
(Table 1, column 4). In the scored pain interference
items, ‘sick leave’ for ‘effect of LPB on work activity’
was shown to have the highest average for pain
intensity. The next highest average was ‘I went to
both for those (medical doctor and alternative
medicine)’. Two points were given when scoring, based
on the result of CATPCA from these items.
The sum of the scores for the four questions of
pain interference was used as a pain interference
score and Figure 1 shows the relationship between
the pain interference score and pain intensity. The
majority of the participants with 2 points or more
pain interference score had moderate pain. In
addition, the average of pain intensity in the parti-
cipants with 4 points or more was 7 points or more.
The coefficient of correlation between pain interfer-
ence score and the characteristic pain intensity was
0.571 (p 5 0.001).
Figure 1. For each pain interference score, percentages of
pain intensity level are shown (study 1). Mild pain: pain
intensity 4 3.0; moderate pain: pain intensity 4 5.0; severe
pain 4 7.0.
805
3.2. Study 2
3.2.1. Participants
The average age of respondents in study 2 was 37.4
years (SD 11.3, range 18–69). Males made up 93.2%
(4708/5053) of the study participants, had an average
height of 169.8 cm (SD 6.1, range 137–199), an average
weight of 65.1 kg (SD 10.4, range 36–160) and an
average BMI of 22.6 kg/m2 (SD 3.2, range 15.2–54.7).
Females had an average height of 156.8 cm (SD 5.8,
range 143–172), an average weight of 51.2 kg (SD 8.5,
range 35–83) and average BMI of 20.8 kg/m2 (SD 3.3,
range 15.4–33.2). Regarding the work schedule, 1694
participants (33.5%) worked during the day and 3106
participants (61.5%) worked shifts. For the length of
employment in the current job, the mode was ‘10 years
or more’ (34.0%, 1719/5053).
Regarding the cycle time, 2259 participants
(44.7%) indicated that their cycle time was 4 min or
less. The cycle time for general assembly line work is
often 3–4 min. In this study, workers for which the
cycle time was set within 4 min or less were assumed to
be line workers and all others were assumed to be non-
line workers. In addition, the subjects were stratified by
gender and the analysis subjects for study 2 were
classified into the following four groups: male line
worker group (M_LW, n ¼ 2017); male non-line
worker group (M_NL, n ¼ 2691); female line worker
group (F_LW, n ¼ 242); female non-line worker
group (F_NL, n ¼ 103).
3.2.2. Low back pain symptoms
Table 2 shows the summary of low back symptoms for
each group. The group with the highest prevalence rate
during the last 6 months was M_LW (64.5%, 1301/
2017). The overall prevalence rate was 60.8% (3070/
5053). With regard to the effect on work activity, 179
participants (5.6%) responded ‘sick leave’. The group
with the highest rate was M_NL (6.5%, 176/2691).
Those who received temporary treatment for LBP
numbered 750 participants (14.8%) and those who
received regular treatment numbered 187 participants
(3.7%). The group with the highest rate of receiving
regular treatment was F_LW (5.0%, 12/242). The pain
intensity was 4.6 (SD 2.0) for overall. The group
showing the highest pain intensity was M_LW (4.8,
SD 2.0).
3.2.3. Summary of categorical principal components
analysis results
As with study 1, CATPCA was conducted for the four
questions relating to pain interference. In study 2,
 806 Y. Higuchi et al.

Table 2. Summary of low back symptoms in study 2.

M_LW n ¼ 2017 F_LW n ¼ 242 M_NL n ¼ 2691 F_NL n ¼ 103

Prevalence rate during last 6 months 64.5% 61.2% 58.5% 44.7%
Frequency of low back pain
About 1 d during the last 6 months 9.7% 6.2% 12.6% 7.8%
About 1 d per month 20.1% 24.0% 21.7% 10.7%
About 10 d per month 17.7% 18.6% 14.5% 12.6%
Almost daily 17.0% 12.4% 9.8% 13.6%
Affect on work activity
Affected (except sick leave) 21.9% 19.8% 19.1% 14.6%
Sick leave 5.4% 3.3% 6.5% 2.9%
Affect on non-work activity
Affected 20.3% 16.5% 19.3% 11.7%
Received treatment
Temporarily 15.5% 11.6% 14.6% 14.6%
Regularly 4.1% 5.0% 3.4% 1.9%
Downloaded by [University of Western Ontario] at 22:53 07 October 2014

Pain intensity (only people with low back pain)

Current pain (0–10) 4.1 (2.23)a 3.9 (2.24)a 3.6 (2.11)a 3.9 (2.32)a
Peak pain during last 6 months (0–10) 6.1 (2.44)a 6.0 (2.44)a 5.6 (2.45)a 5.9 (2.48)a
Average pain during last 6 months (0–10) 4.3 (2.02)a 4.2 (1.87)a 3.8 (1.92)a 4.4 (2.07)a
Average for the above three variables 4.8 (2.00)a 4.7 (1.91)a 4.4 (1.91)a 4.7 (2.04)a

M_LW ¼ male line worker; F_LW ¼ female line worker; M_NL ¼ male non-line worker; F_NL ¼ female non-line worker.
a
Values are shown as average (SD).

CATPCA was conducted for each of the four groups.
A summary of CATPCA is shown in Table 3. For all
groups, one component with an eigenvalue of one or
more was identified. The eigenvalues of identified
component were 2.322 to 2.457 (Table 3, row 3). The
variance rate explained by the identified component
was 58.0–61.4% (Table 3, row 4). The above results
indicate that for the respective groups the component
structures of the four items measured the same
construct. The component loading for each question
was 0.691–0.856 for ‘frequency of LBP’, 0.827–0.874
for ‘effect on work activity’, 0.755–0.791 for ‘effect on
non-work activity’ and 0.589–0.769 for ‘received
treatment’. The component loading was 0.7 or more
for all items except ‘frequency of LBP’ in F_LW and
‘received treatment’ in F_NL.
3.2.4. Scoring items regarding pain interference
As with study 1, the value of the centroid coordinates
for each choice for four questions obtained using
CATPCA was rounded down to obtain the score for
each choice (Table 3, column 6). The internal
consistency (Cronbach a) for the four-scored questions
was calculated to be 0.704–0.747. For the analysis
subjects overall, Cronbach a was 0.709.
3.2.5. Mokken Scale Analysis in study 2
The four items regarding pain interference were scored
based on the results of CATPCA. MSA was conducted
to verify the construct validity as a scale. The results
from calculating the scalability coefficients (Hij)
among each item of the four groups showed that the
coefficients were all positive and that the validity first
standard is satisfied. Most of the scalability coefficients
Hi for the respective groups were larger than 0.3,
showing that the validity second standard is satisfied.
The overall item scalability coefficient H was 0.450–
0.507. Based on these results, it was determined that
the scale is a moderate to strong scale model (Mokken
1971). The overall item scalability coefficient H for the
analysis subjects was 0.480.
3.2.6. Relationship with characteristic pain intensity
The sum of the scores for the four questions regarding
pain interference was used as the pain interference
score and Figure 2 shows the relationship between pain
interference score and pain intensity. The majority of
participants with a pain interference score of 3 or more
also indicated that they had moderate pain. The
average of pain intensity demonstrated a uniform
increase commensurate with the increase in pain
interference score. The average of pain intensity was
the highest when the pain interference score was 6
(average 7.2, SD 1.7). The coefficient of correlation for
the pain interference score and the characteristic pain
intensity was 0.548 (p 5 0.001). When MSA was used
to study the hierarchical relationship between the pain
interference score and the categorised characteristic
pain intensity (mild  3.0, moderate  5.0 and
 Ergonomics 807

Table 3. Quantification for pain interference items in study 2.

Centroid coordinates
M_LW F_LW M_NL F_NL Score
Total eigenvalue 2.322 2.457 2.380 2.404
Explained rate 58.0% 61.4% 59.5% 60.1%
Frequency of low back pain
About 1 d during the last 6 months 70.469 70.127 0.116 0.055 0
About 1 d per month 0.192 0.044 0.243 0.115 0
About 10 d per month 0.852 0.920 0.833 0.764 0
Almost daily 1.513 0.974 1.422 1.826 1
Affect on work activity
None 70.571 70.455 70.470 70.389 0
Affected (except sick leave) 1.178 1.279 1.087 1.608 1
Sick leave 2.175 2.909 2.164 2.980 2
Affect on non-work activity
None 70.392 70.352 70.376 70.274 0
Downloaded by [University of Western Ontario] at 22:53 07 October 2014

Affected 1.540 1.778 1.574 2.080 1

Received treatment
None 70.352 70.316 70.335 70.278 0
Temporarily 1.192 1.118 1.279 1.367 1
Regularly 2.406 2.719 2.596 1.710 2
Cronbach a 0.704 0.716 0.711 0.747 –

M_LW ¼ male line worker; F_LW ¼ female line worker; M_NL ¼ male non-line worker; F_NL ¼ female non-line worker.

severe (pain interference score  3). A pain

Figure 2. For each pain interference score, percentages of
pain intensity level are shown (study 2).
severe  7.0), the scalability coefficient H from the
respective combinations was 0.648, 0.639 and 0.551.
The best conformity was for mild pain (  30), but the
difference from moderate pain (5.0) was slight.
3.2.7. Hierarchical classification
The results of the MSA confirmed the hierarchical
relationship between the characteristic pain intensity
and pain interference score. The severity of LBP was
classified into the following three classes according to
the scheme used by von Korff et al. (1992): mild (pain
interference score 5 3, pain intensity 5 5.0); moder-
ate (pain interference score 5 3, pain intensity  5);
interference score of 3 was used as the first classification
criterion. This is based on the relationship between
non-scored choices and pain interference score
(Table 4). If the pain interference score is 3 or higher: 1)
most of these participants have LBP daily; 2) they have
almost all-activity limitation on work and non-work; 3)
they receive treatment. For the second criterion, of
those who had a small pain interference score (5 3),
those with a pain intensity ¼ 5.0 were used. This
classification is the same criteria used to identify the low
disability group by von Korff et al. (1992). When a
more detailed classification is necessary, a very severe
pain interference score is 5 points or more.
The results of applying the above classification to
the 5053 respondents were 1588 workers (31.4%) for
mild, 799 (15.8%) for moderate and 683 (13.5%) for
severe (Table 5). The 1983 (39.2%) who experienced no
LBP during the last 6 months were classified as ‘none’.
4. Discussion
Pain intensity and the degree of LBP effects (pain
interference) are considered important assessment
items in terms of non-specific LBP severity (Smith
1997, Hoozemans et al. 2002). In this study, the scale is
structured into two subscales: 1) that relating to pain
intensity; 2) that relating to pain interference. For the
item group relating to one of the two subscales of
which this scale consists, pain intensity, the three NRS
(current level of pain; average level of pain over the
 808 Y. Higuchi et al.

Table 4. Relationship between pain interference score and each item concerning pain interference.

Pain interference score

0 1 2 3 4 5 6
n 1119 722 546 340 214 100 29
Frequency of low back pain
About 1 d during the last 6 months 25.0% 15.4% 16.1% 14.7% 13.1% 1.0% 0%
About 1 d per month 50.3% 32.3% 30.0% 18.5% 13.6% 5.0% 0%
About 10 d per month 24.7% 26.5% 30.6% 28.2% 25.2% 20.0% 0%
Almost daily 0% 25.9% 23.3% 38.5% 48.1% 74.0% 100.0%
Affect on work activity
None 100.0% 65.1% 23.6% 10.3% 1.4% 0% 0%
Affected (except sick leave) 0% 34.9% 72.5% 67.9% 50.9% 31.0% 0%
Sick leave 0% 0% 3.8% 21.8% 47.7% 69.0% 100.0%
Affect on non-work activity
None 100.0% 83.9% 41.8% 28.5% 14.0% 10.0% 0%
Affected 0% 16.1% 58.2% 71.5% 86.0% 90.0% 100.0%
Downloaded by [University of Western Ontario] at 22:53 07 October 2014

Received treatment
None 100.0% 76.9% 64.7% 29.4% 2.8% 0% 0%
Temporarily 0% 23.1% 32.4% 62.6% 74.8% 33.0% 0%
Regularly 0% 0% 2.9% 7.9% 22.4% 67.0% 100.0%
Pain intensity 3.35 4.57 5.07 5.77 6.32 6.85 7.23

Table 5. Relationship between hierarchical classification of In order to preserve validity of the content, the same
the severity and each item. method used by von Korff et al. (1992) was adopted.
Severity classification For the other item group composing the scale, namely,
pain interference, four items derived from the SNQ
Mild Moderate Severe (Kuorinka et al. 1987) were used. These items relate to
n (%) 1588 799 683 frequency of pain felt in the lower back, effect on
(31.4) (15.8) (13.5) activities at work, effect on activities other than work
Frequency of low back pain and treatment received for LBP. These have been used
About 1 d during the 27.7% 4.9% 11.6% as items relating to LBP outcome on self-administered
last 6 months questionnaires in many past LBP studies (Merlino
About 1 d per month 47.5% 25.7% 14.2%
About 10 d per month 18.7% 42.2% 24.9% et al. 2003, Hartvigsen et al. 2005, Ghaffari et al. 2006,
Almost daily 6.0% 27.3% 49.3% Ramadan and Ferreira 2006, Feng et al. 2007). The
Affect on work activity original SNQ by Kuorinka et al. (1987) asks for
None 76.4% 63.2% 5.6% answers relating to time period for these items, such as
Affected (except 23.0% 35.4% 54.3% on an order scale (0 d, 1–7 d, . . . every day or more
sick leave)
Sick leave 0.6% 1.4% 40.1% than 30 d) and ‘yes or no’. For two of the four items,
Affect on non-work activity the forced choice method was used. When using the
None 84.6% 76.2% 20.1% forced choice method to obtain answers to questions,
Affected 15.4% 23.8% 79.9% the variables are treated as nominal variables or
Received treatment ordered variables. This means that the range for which
None 89.0% 76.7% 15.5% quantitative statistical analysis methods can be used
Temporarily 10.5% 22.3% 59.4%
Regularly 0.5% 1.0% 25.0% relative to those variables is often limited. In dealing
Pain intensity 3.07 6.18 6.16 with this problem, this study used non-linear
multivariate analysis, which allowed scores to be
assigned to each choice. In this way, it was possible to
handle the ordered variable relating to the effect of
past 6 months; level of most severe pain over the last 6 LBP on activities as a number. The fact that the results
months) on a scale of 0–10 was adopted (von Korff of study 2, where scores were assigned in multiple
et al. 1992). The use of a subjective measure for pain worker groups with differing genders and cycle times,
intensity is widely accepted (von Korff et al. 1992, were almost the same is evidence that supports the
Farrar et al. 2001, Fritz et al. 2001, Labus et al. 2003). structural validity of the scale. Assigning scores to each
 Ergonomics
item made it possible to assess numerically the effects
of LBP on activities.
Further, there appear to be several advantages to
using the forced choice method, instead of the NRS, to
obtain answers on the effect of non-specific LBP on
activities from ordinary worker groups using self-
administered questionnaires. One of the advantages of
using the forced choice method to get specific choices is
that the answers obtained do not rely only on absolute
self-assessment. For example, with regard to the effect
of LBP on work activities, the choice ‘I took sick leave’
is fact-related data and does not rely only on self-
assessment. When using a self-administered question-
naire, the limitations on the accuracy of the data
obtained vary, but the self-administered questionnaire
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
method is very often adopted in research programmes
with a large sample size.
The scale formulated in this study should prove
useful in carrying out self-administered questionnaire
projects relating to LBP with ordinary worker groups as
subjects. This report presents the method of classifying
the severity of LBP in an industrial setting in three
classes: mild; moderate; severe. The key in classifying
severity is to clarify the basis for the categorisation
criteria. The authors used the following three-categor-
isation criteria for the severity classification in this
study. The first criterion is a score of 3 or higher for pain
interference. When this criterion is not satisfied, the
effect of LBP is determined as being minor and when
this criterion is met, the category is ‘severe.’ This
categorisation is the most important and clear category
point in this hierarchical classification. This category
score was obtained from the relationship between the
item on LBP effect and pain interference score. Of those
classified as ‘severe’, 40.1% had reported taking sick
leave. In addition, 84.5% had received treatment for
LBP. The second categorisation criterion was pain
intensity  5.0. This is similar to the category used by
von Korff et al. (1992) and was used as an intensity
assessment to distinguish between ‘mild’ and ‘moderate’
in the group for which the effect of LBP on activities was
determined to be relatively small. The difference
between mild and moderate in ordinary worker groups
is considered to be the borderline between whether or
not there is an effect on job performance. Of those
classified as moderate, 36.8% reported limitations on
work activities; this figure was 23.6% for the mild
category. According to Turner et al. (2004), a varying
correlation with disability is suggested between pain
intensity levels of 0–4 on a scale of 0–10 and levels of 5–
10. In addition, the relationship between pain intensity
and sickness absence among ordinary workers has been
pointed out (Morken et al. 2003). The LBP severity
classification method presented in this report combines
disability and pain intensity and utilises the method of
809
assessing chronic pain severity (von Korff et al. 1992) as
a theoretical framework. This assumes that disability
and pain intensity have a hierarchical structure. The
results of the MSA presented in this study supported the
hierarchical structure of pain intensity and effects of
LBP in the context of the LBP severity classification
method examined by the authors.
The LBP severity classification method presented in
this report does not aim to classify LBP medically in
patient groups with specific organic abnormalities.
Rather, it aims to classify non-specific LBP severity in
order to assess the efficacy of interventions such as
improvement activities and ergonomics studies at the
workplace level for ordinary worker groups. In most
LBP cases, no physical abnormalities are discernible
(Manniche et al. 1994). The vast majority of
occupational LBP cases are non-specific (Tait and
Chibnall 2001). Medical diagnosis is possible in most
cases of unexpected accidental LBP that leads to long-
term absence from work, but in the case of chronic
non-accidental LBP, it is not uncommon to encounter
difficulties in making an accurate medical diagnosis.
Existing non-specific LBP studies have used a variety
of outcomes. Even studies that utilise two or more
outcomes often use only one of them as the dependent
variable for statistical analysis or they analyse each
outcome separately (Ferguson and Marras 1997). This
report presents a method whereby two outcomes
relating to LBP, which have frequently been used in
non-specific LBP studies, namely, pain intensity and
pain interference, are integrated into one outcome for
assessment purposes. This method can be used for
simple outcome measurement and, further, identifies
risk factors according to the classified severity. In
addition, these classification results can be used as
dependent variables for multivariate analysis in order
to construct a process model covering LBP progression
and increase in severity.
It is often the case that the number of measurement
items that can be used in large-scale epidemiological
studies and workplace intervention research is limited.
When ordinary worker groups rather than patients are
the study subjects, there is a risk that numerous
measurement items will lead to a decreased
participation rate and higher study costs. In particular,
when implementing a paper questionnaire it is
important to minimise defective answers as much as
possible and, toward this end, it is vital to simplify the
questions and make them easy to understand
intuitively. The scale developed by the authors is
composed of a mere seven items and can be
implemented in a short time, either in an interview or
by using a paper questionnaire.
There are some limits to these results. The subjects
in this report were not randomly selected. The results
 810 Y. Higuchi et al.
may be subject to the effects of selection bias, but
because the data were gathered from many different
manufacturing companies, these effects were expected
to be minimal. Further, this report does not include a
discussion of the consistency between severity
classification using this technique and the
classifications arrived at by using other methods.
Further research on greater reliability and validity
must be carried out with a focus on the above-
mentioned points.
5. Conclusion
In this study, a scale was developed for quantitative
assessment of LBP severity in workers, by combining
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
pain intensity and effects of LBP. The results of this
study indicate that the scale has sufficient structural
validity. This scale’s structure is adapted for a scoring
system and enables numerical assessment of LBP
severity. The use of this scale should make it possible
to evaluate the effects of risk factors in greater detail and
should also be useful in selecting improvement measures
for occupational LBP and assessing their effects.
Acknowledgements
The authors are very grateful for the cooperation from the
following occupational physicians from the respective
companies provided for this study: Dr Yasushi Kudo,
Dr Atsushi Funahashi, Dr Kazunari Kin, Dr Junichi Akatsu,
Dr Atsunari Fujii, Dr Isshin Suzuki, Dr Hajime Mikurube,
Dr Shuichiro Shazuki, Dr Hideki Suzuki, Dr Hideki Ito and
Dr Yuko Sakaeda. We would also like to express our
gratitude to the workers who participated in the study. This
study was supported by Health and Labour Sciences
Research Grants (Research on Occupational Safety and
Health; Research number H18-Rodo-Ippan-007; Principal
Investigator Masaharu Kumashiro), from the Japan Ministry
of Health, Labour and Welfare.
References
Agius, R.M., et al., 1994. Questionnaire for the identification
of back pain for epidemiological purposes. Occupational
and Environmental Medicine, 51, 756–760.
Byrns, G.E., et al., 2002. A new direction in low-back pain
research. AIAH Journal, 63, 55–61.
Farrar, J.T., et al., 2001. Clinical importance of changes in
chronic pain intensity measured on an 11-point numer-
ical pain rating scale. Pain, 94, 149–158.
Feng, C.K., Chen, M.L., and Mao, I.F., 2007. Prevalence of
and risk factors for different measures of low back pain
among female nursing aides in Taiwanese nursing homes.
BMC Musculoskeletal Disorders, 8, 52.
Ferguson, S.A. and Marras, W.S., 1997. A literature review
of low back disorder surveillance measures and risk
factors. Clinical Biomechanics, 12, 211–226.
Fritz, J.M., George, S.Z., and Delitto, A., 2001. The role of
fear-avoidance beliefs in acute low back pain: relation-
ships with current and future disability and work status.
Pain, 94, 7–15.
Ghaffari, M., et al., 2006. Low back pain among Iranian
industrial workers. Occupational Medicine, 56,
455–460.
Gillespie, M., Tenvergert, E.M., and Kingma, J., 1987. Using
Mokken scale analysis to develop unidimensional scales.
Quality and Quantity, 21, 393–408.
Hamberg-van Reenen, H.H., et al., 2008. Does
musculoskeletal discomfort at work predict future
musculoskeletal pain? Ergonomics, 51, 637–648.
Hartvigsen, J., et al., 2001. The association between physical
workload and low back pain clouded by the ‘healthy
worker’ effect: population-based cross-sectional and
5-year prospective questionnaire study. Spine, 26,
1788–1792.
Hartvigsen, J., et al., 2005. Intensive education combined
with low tech ergonomic intervention does not prevent
low back pain in nurses. Occupational and Environmental
Medicine, 62, 13–17.
Hoozemans, M.J.M., et al., 2002. Pushing and pulling in
association with low back and shoulder complaints.
Occupational and Environmental Medicine, 59, 696–702.
Hosokawa, T., et al., 1994. Assessment of functional status
with an extended ADL scale: (2) stroke patients living at
home [in Japanese]. The Japanese Journal of
Rehabilitation Medicine, 31, 475–482.
Huang, Q.M., Sato, M., and Thorstensson, A., 1998. Pulling
force in lateral lifting and lowering. Ergonomics, 41, 899–
908.
Johanning, E., 2000. Evaluation and management of
occupational low back disorders. American Journal of
Industrial Medicine, 37, 94–111.
Kersten, P., White, P.J., and Tennant, A., 2009. Construct
validity of the Holistic Complementary and Alternative
Medicines Questionnaire (HCAMQ) – An investigation
using modern psychometric approaches. Evidence-based
Complementary and Alternative Medicine. Available
from: http://ecam.oxfordjournals.org/cgi/reprint/
nep141v1 [Accessed 21 January 2010].
Kuorinka, I., et al., 1987. Standardised Nordic
questionnaires for the analysis of musculoskeletal
symptoms. Applied Ergonomics, 18, 233–237.
Labus, J.S., Keefe, F.J., and Jensen, M.P., 2003. Self-reports
of pain intensity and direct observations of pain
behavior: when are they correlated? Pain, 102, 109–124.
Leboeuf-Yde, C., Lauritsen, J.M., and Lauritzen, T., 1997.
Why has the search for causes of low back pain largely
been nonconclusive? Spine, 22, 877–881.
Manniche, C., et al., 1994. Low Back Pain Rating scale:
validation of a tool for assessment of low back pain.
Pain, 57, 317–326.
Merlino, L.A., et al., 2003. Symptoms of musculoskeletal
disorders among apprentice construction workers.
Applied Occupational and Environmental Hygiene, 18,
57–64.
Mokken, R.J., 1971. A Theory and Procedure of Scale
Analysis with applications in political research. The
Hague, the Netherlands: Mouton.
Morken, T., et al., 2003. Low back pain and widespread pain
predict sickness absence among industrial workers. BMC
Musculoskeletal Disorders, 4, 21.
Ozguler, A., et al., 2002. Using the Dallas pain questionnaire
to classify individuals with low back pain in a working
population. Spine, 27, 1783–1789.
R Development Core Team, 2007. A language and
environment for statistical computing. Vienna: R
Foundation for Statistical Computing.
 Ergonomics 811

Ramadan, P.A. and Ferreira, M.J., 2006. Risk factors
associated with the reporting of musculoskeletal symp-
toms in workers at a laboratory of clinical pathology.
Annals of Occupational Hygiene, 50, 297–303.
Rehn, B., et al., 2009. Neck pain combined with arm pain
among professional drivers of forest machines and the
association with whole-body vibration exposure. Ergo-
nomics, 52, 1240–1247.
Smith, M.J., 1997. Psychosocial aspects of working with
video display terminals (VDTs) and employee physical
and mental health. Ergonomics, 40, 1002–1015.
Tait, R.C. and Chibnall, J.T., 2001. Work injury manage-
ment of refractory low back pain: relations with ethni-
city, legal representation and diagnosis. Pain, 91, 47–56.
Tissot, T., Messing, K., and Stock, S., 2009. Studying the
relationship between low back pain and working
postures among those who stand and those who sit
most of the working day. Ergonomics, 52, 1402–1418.
Downloaded by [University of Western Ontario] at 22:53 07 October 2014
Turner, J.A., et al., 2004. The association between pain and
disability. Pain, 112, 307–314.
van der Ark, L.A., 2007. Mokken scale analysis in R. Journal
of Statistical Software, 20, 1–19.
van der Molen, H.F., Sluiter, J.K., and Frings-Dresen,
M.H.W., 2009. The use of ergonomic measures and
musculoskeletal complaints among carpenters and
pavers in a 4.5-year follow-up study. Ergonomics, 52,
954–963.
von Korff, M., et al., 1992. Grading the severity of chronic
pain. Pain, 50, 133–149.
Wells, R., et al., 1997. Assessment of physical work load in
epidemiologic studies: common measurement metrics for
exposure assessment. Ergonomics, 40, 51–61.

Appendix: Low back pain questionnaire in English

Questions about your low back pain

How often have you had low back pain during the last 6 months? Almost daily
About 10 days per month
About 1 day per month
About 1 day during the last 6 months
None
Has your work activity been affected by low back pain? Work activity Sick leave
Affected (except sick leave)
None
Except work Affected
None
Have you been seen by a doctor, physiotherapist, chiropractor or I received temporary treatment
other such person because of low back trouble during the last
6 months?
I received regular treatment
None
Currently, how bad is your low back pain? 0 1 2 3 4 5 6 7 8 9 10
No pain Most severe pain
In the past 6 months, how intense was your worst pain? 0 1 2 3 4 5 6 7 8 9 10
No pain Most severe pain
In the past 6 months, how intense was your average pain? 0 1 2 3 4 5 6 7 8 9 10
No pain Most severe pain