153

Pain, 19 (1984) 153-161

0 Elsevier

PAI 00647

An Experimental Basis for Revising the Graphic

Rating Scale for Pain
Marc W. Heft *S and Scott R. Parker **
*Neurobiology

and Anesthesiology Branch, National Institute of Dental Research, National Institutes of

Health, 9ooo Rockville Pike, Bethesda, MD 20205, and ** Department of Psychology, The American
Unioersity, Washington, DC 20016 (l.? S. A.)

(Received 24 October 1983, accepted 4 January 1984)

Summary

Seven subjects judged the differences between electrocutaneous shocks and words
from two category rating lists describing those sensations in each of two difference
estimation experiments. The electrocutaneous shocks used for the two experiments
were 10 supra~eshold
shock intensities determined separately for each subject.
There were two distinct 7-word category rating lists. Both lists shared 6 common
words; however, the seventh word made the rational ordering of the two lists
different.
Magnitude scales of meaning for the category rating words and sensory scales for
the electrocutaneous shock intensities were determined for each of the two experiments for each subject using conjoint measurement analysis. Comparisons of the
sensory scales for electrocutaneous shock between the two difference estimation
experiments for each subject showed that they judged the electrocutaneous shocks
similarly with the two word lists. This allowed for comparisons between the scales of
meaning for the words from the category rating lists. The two word lists were not
equivalent. There was substantial agreement among the subjects on characteristic
spacings of quantitative values for the category rating items. These results suggest
that clinical rating scales used for analgesimetry should not assume homogeneity of
spacing of category items. A scale incorporating our subjects common understanding is presented.

Present address: Laboratory of Behavioral Sciences, Gerontology Research Center, National Institute
on Aging, Baltimore City Hospitals, Baltimore, MD 21224, U.S.A.
Send aI1 correspondence to: Dr. Marc W. Heft, Laboratory of Behavioral Sciences, GRC. NIA, NIH,
Baltimore City Hospitals, Baltimore, MD 21224, U.S.A.

154

Introduction

The efficacy of an analgesic agent lies in its ability to relieve clinical pain. The
measures of therapeutic efficacy are determined from clinical trials comparing active
and control medications for the relief of clinical pain as judged from patients
reports of the severity of their pain. Since patients pain complaints are subjective
reports of an otherwise unmeasurable stimulus, the sensitivity of the pain measures
are constrained by the pain rating scale employed.
Category rating procedures have been used quite extensively to assess patients
pain levels in clinical analgesic studies. These methods require that patients rate their
pain by choosing a word from a category list of words describing a full range of
painful experience (for example, none, slight, moderate, severe, and very
severe). Graded scales allow the subject to communicate about the severity of his
clinical pain. Such procedures enable investigators to quantify patients pain levels
both prior to and after treatment. The change in these pain ratings provides a
measure of the effectiveness of the active drug or placebo treatment to alleviate the
pain.
In quantifying subjects responses, many investigators have treated each change in
pain intensity as a unit step [7,12,13]. For example, none is 0, slight is 1,
moderate is 2, severe is 3, and very severe is 4. Consequently, a drop in pain
from severe to slight is a pain relief score of 2, as is a drop from moderate to
none. This method assumes that word meanings serve only to order items on a
category rating list and that meanings do not contribute to the perceptual spacing of
the items. Lasagna [13] suggested that the assumption of equal-spaced categories
might not be valid and that these equally spaced pain relief scores do not correspond
to equal changes in pain levels of equal analgesic effects. He asked subjects to rank
in order of importance the following degrees of pain relief: very severe to severe,
severe to moderate, moderate to slight,and slight to none. The results of this
study indicated that, on the average, pain reduction from severe to moderate was
most important and from slight to none was least important. One possible
explanation for this finding is, as Lasagna suggested, that the assumption of
perceptually equal spacing among categories is not right. On the other hand, the
categories might be equally spaced, so that the actual pain reduction from severe to
moderate is the same as the reduction from slight to none, but the hedonic quality
importance depends on both degree of change and initial level with respect to these
categories. In this manner, pain and importance of pain might be interpreted as
two dimensions of pain experience, and be non-linearly related. Furthermore, since
there was clearly a preferred pain improvement among these theoretical changes in
pain (having the same pain relief scores), the category words must provide information about perceptual spacing between the category items in addition to their
inherent ordering (positions) on the category lists. The existence of a preferred pain
improvement indicates that there is either: (1) inhomogeneity of spacing between
categories, (2) non-correspondence of pain change and hedonic value (importance),
or (3) both inhomogeneity of spacing and non-correspondence.
A few studies have attempted to measure the category rating scale items and

155

explore the role of word meanings and position on perceptual scale values. Those
studies have, typically, measured the category rating items as stimuli to be rated and
found that both semantic meaning and position of descriptors on category lists
[5,18,23] contribute to the perceptual scale values for the descriptors.
The purpose of the present investigation is to assess the role of word meanings
and positions on category lists in determining the scale values for those items. We
employ procedures in which subjects judge magnitude differences between electrical
shock intensities and words from lists describing sensations associated with electrical
shock. One set of shock intensities is compared with items from two distinct category
ratings scales in two separate experiments. In this manner, the category rating scale
items are treated as stimuli whose perceptual magnitudes are to be compared with
the perceptual magnitudes associated with physical stimulus intensities.

Materials and Methods

Subjects
Seven adults (2 women and 5 men, ages 22-39 years, median age = 33) were
subjects for this study. Subjects were told that the purpose of the experiment was to
investigate the sensations produced by electrocutaneous shock and that the shock
intensities would range from just detectable to the rn~rn~
intensity they would
allow. Subjects then signed an informed-consent form that further explained the
procedures and stated that they were free to withdraw from the study at any time.
Each subject participated in two experiments over five 1 h sessions (1 preliminary
and 4 test sessions, 2 sessions/word list).
Stimulation
Electrocutaneous shock was administered by a high-impedance, battery-operated,
constant-current generator. The. electrical stimuli were 1 set trains of 100 Hz
monophasic square-wave pulses of 1 msec duration.
The test stimuli ranged from 1.5 to 20 mA and were delivered through an
electrolyte-paste sponge concentric bipolar electrode applied to the dorsal surface of
the wrist of the non-dominant arm. The skin under the electrode was first scrubbed
with electrode paste, reducing the skin impedance to 5000 52[21].
Test stimuli
Electrocutaneous shock. In a preliminary session, each subjects detection threshold
and tolerance levels were determined for the electrocutaneous shock. The ratio of
tolerance current/t~eshold
current for all 7 subjects was approximately the same
(10). Each subjects test stimuli for the two experiments were 10 current levels
between the detection threshold and tolerance levels determined for each individual
subject.
Cognitive stimuli. Table I shows the two category lists used in the two difference
estimation experiments. Words in lists Words I and Words II describe a range of
sensations from faint to severe (Words I) or intense (Words II). List Words I is

156

symmetrical about the word moderate in that there are 3 words defining intensities
less than and 3 words defining intensities greater than moderate. List Words II
includes words that describe a range of sensations from faint to intense* (which is
less than severe). This list is asymmetrical
about the word moderate in that there
are 4 words defining intensities less than moderate and only 2 words greater. Lists
Words I and Words II share 6 common words, but the ordinal positions of these
words are different on the two lists due to the differences in the seventh words:
severe on list Words I and very weak on list Words II. Thus, intense is the sixth
ranked category word on an ascending order of magnitude on Words I; however, it
is the seventh ranked category word on Words II. In the two difference estimation
experiments subjects judged perceptual differences between the category items (from
the two lists: Words I and Words II) and sensations
associated with electrical
shocks.
Procedure
Each subject participated
in two difference estimation
experiments.
In each of
these experiments, subjects were instructed to estimate perceptual differences within
pairs of stimuli. A pair of stimuli consisted of an electrocutaneous
shock and a
category word (from list Words I or Words II) describing perceptual magnitudes
associated with the electrical currents. The order of presentation
of word lists was
partially counter-balanced
across subjects. The electrical shock intensities were the
10 test stimuli described previously.
Since there were 10 levels of current and 7
cognitive stimuli in each of the two difference estimation experiments, there were 70
possible stimulus pairs per experiment.
There were 4 trials with each of the 70 stimulus pairs per experiment, or 280 trials
per experiment. Stimulus pairs were presented randomly and occurred every 20 sec.
An auditory ready cue preceded each trial by 5 sec. The electrocutaneous
shocks
were delivered to the non-dominant
arm, just after cognitive stimuli were projected
on a screen in front of the subject. Subjects judgments
included
estimates of
stimulus difference size and direction - when the shock was greater than the word
the difference was recorded as positive. The first session of the difference estimation
served as a training session and was not included in data analysis.

TABLE

CATEGORY

ITEMS

Words I

Words

Faint
Weak
Mild
Moderate
Strong
Intense
Severe

Faint
Very weak
Weak
Mild
Moderate
Strong
Intense

II

1.57

Data analysis

Subjects data from the two difference estimation experiments were analyzed
individually to determine the perceptual scales for the electrical shocks and for the
words. There were 4 trials with each stimulus pair presented over two experimental
sessions. The arithmetic mean of the 4 trials of difference judgments was computed
for each of the 70 stimulus pairs per experiment. Each of the 14 sets of 70 mean
judgments (2 experiments on 7 subjects) was analyzed by conjoint measurement
analysis [14], using Roskams computer program, UNICON [16], which provides
interval scale values for both the sensations associated with electrical stimulation
and for the sensations described by the category words.

Results
Electrical stimuli

The growth of perceived shock with intensity was consistent with previously
reported results for electrocutaneous shock to the wrist [2,3,10,19,20]. No systematic
difference in growth rate was observed between experiments 1 and 2. This consistency validates comparisons of perceptual scale values for the cognitive stimuli.
Cognitive stimuli
Within-subject

eom~~risons. The intrasubject comparisons served to assess the

importance of category item position as well as meaning on. the scale values for the
words. The role of category position was assessed by comparing the scale values for
the 7 ordered scale values (irrespective of semantic designation) for the two category
rating lists. This was achieved by ordering the 7 perceptual scale values for each of
the two lists from smallest to largest, irrespective of the word or category designation
(All). Intrasubject comparisons were then made on the basis of the scale values for
WORDS
. COihlN

Fig. 1. Seven ranked normalized scale values from Words II plotted as functions of the corresponding
ranked normalized scale values from Words I for 7 subjects.
Fig. 2. Normalized scale values for the common words from Words II plotted as functions of the
corresponding normalized scale values from Words I for 7 subjects.

TA3LE

II

SPACING

BETWEEN

Mean
S.E.M. a
a Standard

CATEGORY

ITEMS

FOR COMMON

WORDS

Faintweak

Weakmild

Mildmoderate

Moderatestrong

strongintense

0.296
0.075

0.302
0.080

0.648
0.052

0.682
0.071

0.593
0.077

error of the mean.

the first ranked items of Words I, for example, versus the scale value for the first
ranked item of Words II, the second ranked item of Words I versus the second
ranked item of Words II, and so on. On the other hand, the role of category word
meanings on the determination of the scale values, independent of the position on
the semantic hierarchy of the category rating lists, was assessed by pairing the scale
values for each of the 6 common words derived from the two experiments (Common).
For every subject, the correspondence between scale values was far better using
the 6 Common words than using All 7 identified by ordinal position. The group
average relations are shown in Fig. 1 and for the 6 Common words in Fig. 2.
Clearly, the 6 Common words retain their relative positions irrespective of the
identity of the seventh word.
The spacings among the Common category items are not homogeneous. It can
be seen in Table II that a change in sensation from faint to mild (a rise of two
category designations) was equal to 0.598 (0.296 -+ 0.302) - much less than a
change in sensation from moderate to strong (a rise of one category designation).
The degree of agreement among the individual subjects on these category spacings
was shown using the Kendall coefficient of concordance (W). These results suggest
that individuals share a common usage of the words (Words I Common, W = 0.29,
P -z0.08; Words II Common, W = 0.406, P < 0.05).

Discussion

The results of this study support the view that subjects do indeed make category
judgments on the basis of word meanings. Category ratings are not merely an
ordinal index. The lack of regular spacing among the derived scale values for the 6
common words indicates that the words do not serve to divide the perceptual
continuum into equal segments, and the meanings ascribed to the words are
contributing to the irregular spacing.
Other investigators who have measured pain descriptors have also found that
category words do not merely serve to divide the perceptual continuum into equal
segments. Most such studies, however, have evaluated the pain descriptors in the
absence of painful stimuli [5,18,22]. For example, several investigators have employed cross-rn~a~ty matching procedures, in which subjects use a common re-

159

sponse (such as handgrip force) to words and standard stimuli (line length), to
quantify pain intensity or unpleasantness words on the basis of meaning.
Sriwatanakul and his colleagues [18] requested subjects to rate the magnitude of pain
intensity words by locating them on visual analogue scales, lines whose endpoints
were represented by no pain and the worst pain I can imagine, with the same
objective.
Gracely and Wolskee [4], employing an experimental paradigm similar to the
present one, had subjects judge averages within stimulus pairs. Pairs consisted of
electrical tooth pulp stimuli and words describing sensations evoked by such stimuli.
They found that scale values for word meanings correlated highly with those
determined by cross-modality matching procedures. The present study has also
shown that subjects can judge relations between sensations evoked by electrical
stimulation and words that describe those sensations. Their results and ours indicate
that requesting subjects to make integrative judgments on sensory and verbal stimuli
is useful in evaluating verbal descriptors. Our study also measured individual
subjects abilities to perform the task (as reflected in the stability of the growth rates
for electrocutaneous shock). After the reliability of the judgments for shock intensity
was established, then we were able to measure the influence of category meanings
and positions on the derived scale values for the category items.
Although 4- and 5-point rating scales of pain have wide acceptance in the clinical
algesiometry literature, there has been much criticism of these rating methods
[8,9,11,15,17]. An advantage of these methods is their requiring minimal instructions
for use, since presumably the category designations can readily describe a subjects
pain intensity. However, these scales typically offer the subject the choice of 3 or 4
words to describe a full range of pain. Hardy et al. [6] suggested that there were 21
discernible pain levels between pain threshold and tolerance. Thus, 4- and 5-point
category rating scales would appear to lack sufficient sensitivity to measure the pain
experience.
Visual analogue scales have been employed in an effort to overcome this deficiency in sensitivity [l]. A visual analogue scale is a straight line whose endpoints
are defined as the extreme limits of the sensation to be measured; for example: no
pain and most painful you can imagine. This scale provides the advantage that
there are an infinite number of points between the extremes. However, the visual
analogue scale is ~fficult for the patient to use because there are no guides for
ratings other than the endpoints. Therefore, the judgments should be more variable
across subjects than those from the category rating scales; however, this variability
cannot be assessed because the ratings are of an unmeasurable, changing stimulus.
The development of graphic rating methods is a compromise between these two
methods. A graphic rating scale is a visual analogue scale with category word
designations on the line. It offers more sensitivity than a category rating scale, and is
easier to use than a visual analogue scale. Subjects tend to cluster their responses
near the verbal designators. Therefore, the location of these designators along the
scale is an important consideration.
We propose that the spacing of words on the graphic rating scale should reflect
the spacings between words as the subjects perceive them, An illustration of such a

160

FAINT

WEAK

Fig. 3. Graphic
category labels.

MILD
rating

MODERATE
scale of pain

STRONG
incorporating

INTENSE
subjects

understanding

of the spacings

between

scale based on the present study appears in Fig. 3. We can see that this combines the
virtues of the category rating and visual analogue scales. It also adapts the scale to
the common understanding
of the pain levels designated by the descriptors. Since we
do not know the distance from the lowest word to no pain, we suggest that
moderate should be designated as 0 and ratings greater than moderate should be
+ and less than moderate should be - when measuring the responses. We have
chosen moderate as the 0 point because its perceptual
scale value is roughly
halfway between the extreme words in this study, faint and intense; however, the
arbitrary selection of the 0 point will not affect the sensitivity of the scale. Similar
methods can be used to develop scales employing other words. Our point is that the
spacings of the words on the line should reflect the common understandings
of the
words.

Acknowledgements
The authors wish to express their appreciation
to Drs. Richard Gracely, Wilard
Larkin, and Anthony Riley for their critical reading and advice on the manuscript
and to Drs. Vjekoslav Miletich, Ray Dionne, and Ron Dubner for their comments
on a later draft.

References
1 Aitken, R.C.B., Measurement
of feelings using visual analogue scales, Proc. roy. Sot. B, 62 (1969)
17-24.
2 Babkoff, H., Electrocutaneous
psychophysical
input-output
functions and temporal integration,
Percept. Psychophys.,
23 (1978) 251-257.
scales and somatic evoked potentials
to the percutaneous
3 Beck, C. and Rosner, B.S., Magnitude
electrical stimulation,
Physiol. Behav., 3 (1968) 947-953.
4 Gracely, R.H. and Wolskee, P.J., Semantic functional
measurement
of pain: integrating
perception
and language, Pain, 15 (1983) 389-398.
P. and Dubner, R., Ratio scales of sensory and affective verbal pain
5 Gracely, R.H., McGrath,
descriptors,
Pain, 5 (1978) 5-18.
6 Hardy, J.D., Wolff, H.C. and Goodell, H., Pain Sensations and Reactions, Williams and Wilkins,
Baltimore, MD, 1952.
S.L. and Beaver, W.T., Clinical measurement
of pain. In: G. destevens
7 Houde, R.W., Wallenstein,
(Ed.), Analgetics, Academic Press, New York, 1965, pp. 75-122.
E.C., Measurement
of pain, Lancet, ii (1974) 1127-1131.
8 Husk&on,
9 Huskisson,
EC., Shenfield, G.M., Taylor, R.T. and Hart, F.D., A new look at ibuprofen,
Rheum.
phys. Med., Suppl. (1970) 88-92.
of painful and nonpainful
electrical shocks, Percept.
10 Jones, B., Algebraic models for integration
Psychophys.,
28 (1980) 572-576.

161

11 Joyce, C.R.B., Zutshi, D.W., Hrubes, V. and Mason, R.M., Comparison

of fixed interval and visual
analogue scales for rating chronic pain, Europ. J. clin. Pharmacol.,
8 (1975) 415-420.
12 Keele, K.D., The pain chart, Lancet, ii (1948) 6-8.
13 Lasagna, L., The clinical measurement
of pain, Ann. N.Y. Acad. Sci., 86 (1964) 28-37.
14 Lute, R.D. and Tukey, J.W., Simultaneous
conjoint measurement:
a new type of fundamental
measurement,
J. math. Psychol., 1 (1964) l-27.
15 Ohnhaus, E.E. and Adler, R., Methodological
problems in the measurement
of pain: a comparison
between the verbal rating scale and the visual analogue scale, Pain, 1 (1975) 379-384.
16 Roskam, E., A survey of the Michigan-Israel-Netherlands
integrated
series. In: J.C. Lingoes (Ed.),
Geometric Representation
of Relational Data, Mathesis Press, Ann Arbor, MI, 1977.
17 Scott, J. and Husk&son, E.C., Graphic representation
of pain, Pain, 2 (1976) 175-184.
18 Sriwatanakul,
K., Kelvie, W. and Lasagna, L., The quantification
of pain: an analysis of words used to
describe pain and analgesia in clinical trials, Clin. Pharmacol., Ther., 32 (1982) 143-148.
19 Stembach,
R.A. and Tursky, B., On the psychophysical
power function in electric shock, Psychon.
Sci., 1 (1964) 217-218.
20 Tashiro, T. and Higashiyama,
A., The perceptual properties of electrocutaneous
stimulation:
sensory
quality, subjective intensity, and intensity-duration
relation, Percept. Psychophys.,
30 (1981) 5799586.
21 Tursky, B., Physical, physiological,
and psychological
factors that affect pain reaction to electric
shock, Psychophysiology,
11 (1974) 95-112.
22 Tursky, B., The development
of a pain perception
profile: a psychophysical
approach.
In: M.
Weisenberg and B. Tursky (Eds.), Pain: New Perspectives in Therapy and Research, Plenum Press,
New York, 1976, pp. 171-194.
23 Wildt, A.R. and Mazis, M.B., Determination
of scale response: label versus position, J. Market. Res.,
15 (1978) 261-267.