Purn, 44 (19911 255-262 ‘2 1991 Elsevier Science Publishers AnOFJiS ~30439599l~og~i
255 B.V. 0304.3959/91/~03.50
PAIN 01752
Sex differences in the perception of noxious heat stimuli Jocelyne
S. Feine
‘, M. Catherine
Bushnell
n.h, Denis
Miron
” and
Ii Fan&6 de Mhdrcme Denrmre, and ’ Cenrre de Recherchr en Sciences ,V~z~r~}~o~19~~,.~, C!nrr*ersirc~de (Received
8 March
1990. revision
received
10 September
1990, accepted
Gary ,~[~~tr~t~l.
12 September
H. Duncan ~~~~F~I~~~J~,
a.’ Que. ~~ur~u~u)
1990)
This study compared pain perception in young male and female subjects, using experimental noxious Summary heat stimuli. During 2 sessions, each of 40 subjects rated the magnitude of 120 heat stimuli, ranging from 45°C to 50°C. The study included a comparison of visual analogue and magnitude matching rating procedures. as well as a test of simulated analgesia, in which the range of stimuli presented during the 2 experimental sessions was shifted by 1°C. We found that females rated noxious heat stimuli as more intense than did males, independent of the gender of the experimenter or the type of rating scale. In addition, the data suggest that females discriminate among the painful heat intensities better than males. For example, female subjects showed significant between-session discrimination of noxious heat stimuli, while male subjects did not, and females produced steeper within-session stimulus-response functions than did males. These observed differences in nociceptive discrimination between males and females indicate that the sex-related variation in pain perception is probably related to sensory factors rather than differences in attitude or emotional response. Key words: Pain;
Sex; Temperature;
Psychophysics
Epidemiologic studies of North Americans indicate that women report a higher incidence of both temporary and persistent pain than do men [15,4.53. Similarly, among patients seeking treatment for chronic pain, the majority are female [28,32]. These differences may be due to sociological factors which demand stoicism in males and allow expression of pain in females, or they may be due to physiologic or anatomic differences between the sexes. Women do not uniformly report higher levels of all types of pain. For example, some studies have found that males report at least as much back pain as do females [42,45]. However, for several other pain sites (head and face), women report pain more often than men [45]. These differences may indicate that environmental factors, such as type of work or exposure to stress, lead to differential pain patterns.
-Crrreqwnckwce IO: Joceiyne S. Feine. DCpartement de Restauration, FacultC de MCdecine Den&ire. Montrkal, iMontreal. Que. H3C 357, Canada.
de Dentisterie Universitk de
In an attempt to directly compare male and female pain perception in a controlled situation. numerous investigators have evaluated sex differences in the perception of experimental noxious stimuli. Using a variety of experimental pain stimuli, including pressure, heat, shock, and cold. many investigators have found that pain threshold and/or tolerance are lower for females than for males [2,5.6,11,16,36.38,41.46.47]. Studies evaluating subjects’ rutings of experimental pain similarly have found that females rate noxious stimuli higher than do males [l&47]. Nevertheless. these observed sex differences in the perception of experimental pain are not unequivocal. as other studies have not found a sex-related difference in pain threshold using noxious shock [34,35], heat [13], and cold [4], nor in pain tolerance using shock [34] and ischemia [lo]. In order to determine if observed differences in pain perception between males and females are related to discriminative capacity or to differences in willingness to report pain, several studies have employed sensory decision theory (SDT) analysis [25] to calculate separately a subject’s discriminative ability (d’) and his criterion for labeling stimuli as painful. Using such measures, results have been highly inconsistent across
256
studies. For example, some studies indicate that women discriminate better (larger d’) than men 16,221, while others report no difference in d’ between the sexes [I 11. Similarly, some studies conclude that women set higher criteria [12], while others report that men set higher criteria (111. In addition to producing inconsistent data, the SDT analysis has been criticized on theoretical grounds as a method for differentiating sensory differences in pain perception from differences in criterion [14,39,40]. Thus, such data have not led to a clear understanding of whether gender-related differences in pain perception are sensory or cultural in nature. One additional study, using classical psychophysical techniques (determination of the Weber fraction and exponent of stimulus-response functions), found no difference in the capacity of males and females to discriminate noxious electric shocks [41]. An important variable that has not been controlled in almost all of these studies is the sex of the experimenter, If sex-related differences in pain perception are even partially related to sociological factors, the sex of the experimenter could alter the subjects’ willingness to report pain. In only a few studies is the sex of the experimenter indicated, and in these the experimenter was female [11,41,46]. One could easily hypothesize that cultural demands would lead males to be less likely than females to report pain to a female investigator. Only one study has controlled for such interactions. by using both male and female investigators [36]. In the present study, we have attempted to clarify some of the ambiguities concerning gender differences in pain perception, by comparing the perception of a range of noxious heat stimuli by males and females, using both a male and a female experimenter, and utilizing response measures that have been previously validated to be sensitive to small differences in pain perception [17,18,37]. Some of these data have been previously reported in abstract form [21].
Methods Subjects Forty paid volunteers, 20 males (ages 19-30 years, mean 24.8 years, median 24 years) and 20 females (ages 20-36 years, mean 25.3 years, median 25 years) participated in this experiment. All read and signed a consent form acknowledging that the experimental procedures had been explained and that they could withdraw, without prejudice, from the experiment at any time. Ten of the 20 female subjects were taking oral contraceptives. Of those normally ovulating subjects (those not taking oral contraceptives), 25% participated in the experiment during the menstrual phase of their cycle, 63% during the post-menstrual phase, 0% during the
TABLF.
1
__~-.Low temperatures
_.-.-.-
--___
_.
Stimulus
o C‘
High temperature\ ~___... ._ _.. Stirnub ‘0f
Tl T2 T3 T4 T5
45 46 47 4x 49
-1’1 1‘2 T3 T4 ‘I‘5
--
.
ik 4’ * .,, 4‘; ”;a
n = 20 1st session - low temperatures 2nd session - high temperatures n = 20 1st session - high temperatures 2nd session - low temperatures
ovulatory phase.
phase.
and
12% during
the
premenstrual
Experimental design The 20 male and 20 female subjects were divided so that half of each group was assigned a female experimenter, and the other half a male experimenter. In order to further control for possible gender-related differences in the use of specific pain rating techniques, half of each group rated noxious thermal stimuli using visual analog scales PAS), a commonly used psychophysical testing technique, and half rated noxious thermal stimuli using the technique of magnitude matching [18]. Thus the gender of the subjects was counterbalanced relative to the gender of the experimenter and to the type of pain rating technique. All subjects participated in 2 experimental sessions (separated by 1 week), in which they rated a range of noxious heat stimuli. During one of these sessions the subjects were presented with 60 heat stimuli between 45°C and 49°C (12 presentations of each of 5 levels). while during the other session they were presented with temperatures between 46°C and 50°C (Table I). Half the subjects received the high range first and the low range second, while the other half received the reverse order. This design allowed us to evaluate subjects’ discriminative ability in 2 ways. First, by determining the slope of the stimulus-response function during a single session, one can assess the subjects’ discriminative sensitivity within the context of a specific range of temperatures 17,411. Secondly, by comparing responses between sessions, when the range of stimuli is shifted by 1°C. one can estimate the reliability of the subjects’ responses, as well as the subjects’ ability to discriminate small differences in stimulus range across time [ 20,24.40]. Rating techniques Visual unalog scales (VAS). Before each stimulus presentation, subjects assigned to the VAS group were given a printed scale consisting of a 10 cm horizontal line with verbal anchors of ‘no pain’ on the left side and
257
‘the most painful that one could imagine’ on the right side. The subjects were instructed to draw a vertical mark at the appropriate position on the visual analog scale to indicate the perceived painfulness of the stimulus relative to the two anchors. Magnitude matching. This psychophysical technique combines the freedom of unrestricted magnitude estimation with a no~alization procedure involving cmssmodality matching of experimental and control stimuli [1X,43). Subjects were instructed to rate the intensity of 60 painful heat stimuli and the brightness of 50 visual control stimuli on the same scale of magnitude, using any whole number or decimal that they thought appropriate. Zero would indicate that the heat stimulus was not at all painful or that the visual stimulus was not detected. No maximum number was specified. Ratings assigned to the visual control stimuli were averaged for each subject, and this mean was used to normalize the ratings of the experimental stimuli. These magnjtude matching procedures were identical to those described previously [ 18].
The standardized instructions describing each of the above rating techniques were memorized by the male and female experimenters and given verbally to the subject. In response to subjects’ questions, the instructions were explained a second time in paraphrased form. Both sessions were preceded by a practice period in which 3 sample heat stimuli were presented. Subjects could then ask additional questions or repeat the practice session if needed.
For all subjects, the thermal stimuli were applied by the investigator via a 1 cm diameter hand-held contact thermode, to 4 spots of skin above the subject’s upper lip. Neither the subject nor the experimenter knew the exact range of thermal stimuli nor the order of the intensities that were counterbalanced across skin locations. During these heat trials, the temperature of the thermode increased from a baseline of 39°C to 1 of 5 higher temperatures (45-49°C or 46-50°C; see Table
1). For those subjects using the technique of magnitude matching, thermal and visual stimuli were presented on alternate trials. On visual trials an opaque cue light, situated 40--50 cm in front of the subject, increased in intensity to 1 of 5 brightness levels. These light intensities were counterbalanced so that the intensity of the visual stimuli was not related to the intensity of the thermal stimuli in any systemic fashion. Following a ready signal on the computer monitor, the subjects initiated each trial by pressing the ‘return’ button on the keyboard. After each stimulus presentation, the subject rated the magnitude of the stimulus by
typing his numerical estimate on the keyboard (magnitude matching group), or by drawing a vertical line on a visual analog scale. For all subjects, intertrial intervals were adjusted to allow a minimum of 1 mm to elapse between presentation of consecutive thermal stimuli to the same spot of skin. We have previously shown that this period is sufficiently long to minimize response changes attributable to suppression or sensitization of nociceptive afferents [8,9].
Apparatus A minicomputer controlled all stimulus presentations, collected subjects’ numerical ratings and performed the statistical computations. Thermal stimuli were presented with a 1 cm diameter contact thermode which incorporated a servo-controlled heating element and active cooling via refrigerated, circulating water. The warming and cooling slopes of the thermode were appro~mately 6”C/sec. Visual stimuli were presented through a 3 cm diameter opaque white key by a tungsten lamp whose brightness was controIled by the digital-analog output of the computer. The luminosity of the experimental room was maintained at a constant level.
Data analysis and statistics Ratings collected using the magnitude matching technique were divided by the subject’s mean rating of the visual stimuli, in order to normalize the data across subjects (see Duncan et al. for details [lS]>. VAS ratings were calculated by meas~~ng the distance from the left end of the scale to the subject’s mark. For all analyses in which data were collapsed across the two rating techniques, data were normalized so that the subject’s rating was divided by the median rating of all subjects (male and female) who used that rating technique. A J-way ANOVA was used to compare experimenter gender, subject gender, and type of psychophysical test. A paired t test was used to analyze the difference in the subjects’ abilities to rate high vs. low temperatures. A repeated measures ANOVA and a multilevel variate analysis (MLVA) was used to determine differences in the slopes of thermal stimulus-response functions created by the male and female subjects. In addition, a Fischer’s exact chi-square analysis was used to evaluate sequence effects between the 2 experimental sessions.
Results Overall, female subjects rated the temperatures significantly higher than did male subjects. This is demonstrated in Fig. 1, in which the mean normalized ratings For all the temperatures from both sessions were collapsed and compared (P = 0.008, 3-way ANOVA). In contrast, there was no significant difference due to the
2.0 -
l’.ABLti
1.5-
M&Z\ -.___---“.-.._ crslrul an&g
If
-._.
_.
Females _._._~~~ ~~
vcoic~
High
44.x
High
Low
x1.9
Low
t>o -I
5ii.S
Paired f test P = 0.23
Patred i wst
I’ c= o.oii
MugnmJc High
High
>,‘i :
LOW
1.32
1.0-
LO\s
nmrching swfe 1.41 1.24
Paired t test P = 0.23
0.5 -
Paired f test P = 0.08
0.0 + Fig. 1. Normalized mean ratings of all subjects collapsed across all temperatures received during both sessions. Ratings of female subjects were significantly higher than those of males (P = 0.008, 3-way ANOVA). Because data were normalized using median ratings, which differed slightly from the means, the mean female rating (1.4) and the mean male rating (0.8) yield a grand mean of I .l.
rating technique (P = 0.20, 3-way ANOVA) nor to the gender of the experimenter (P = 0.70, 3-way ANOVA). Fig. 2 shows that females rated each temperature level higher than did males, independent of the type of rating scale. Separate repeated measures ANOVAs for the VAS and magnitude matching data, which compared males and females across the 5 temperature levels, confirmed that for both measures females rated the temperatures higher than did the males (magnitude matching F (1,18) = 5.02, P-c0.05; VAS F (1, 18) = 5.33 P -c0.05). In addition, using either measurement scale, subjects rated the 5 temperature levels differentially (magnitude matching F (4) = 44.51, P < 0.001; VAS F (4) = 222.9, P < 0.001). Finally, for the magni-
MAGNITUDE --* ---t
T,
MATCHING
100
FEMALE SUBJECTS MALE SUBJECTS
T2
T3
TEMPERATURES
tude matching data, but not for the VAS, there was an interaction between sex and temperature level (magnitude matching F (4. 72) = 2.83. P -c0.05:VAS I,‘ (4. 72) = 0.88. P = 0.48). To determine whether the subjects could discriminate between high and low temperature ranges between sessions. a normalized difference score was calculated for each subject bc subtracting the mean ratings for the IOU temperature session from the mean ratings for the high temperature session. and then dividing by the mean ratings assigned to the high temperatures. These normalized difference scores, combined across both rating scales, were then statistically compared for both the male and female subjects. Fig. 3 shows that the females rated the high temperatures higher than the low temperatures (P < 0.02, paired t test), while ratings assigned by the male subjects failed to reach significance (P zy 0.17. paired I test). Similar, but weaker, trends were obtained when the VAS and magnitude matching data were analyzed separately. Table II shows the mean ratings of the high
VISUAL ANALOG SCALE
80
T‘i
Ts
v
T,
TZ
T3
T4
TS
TEMPERATURES
Fig. 2. Mean ratings of males (filted circles) and females (open circles) for the 5 temperature levels presented in each of 2 sessions. Although the range was shifted between sessions (see Methods), the average rating for the lowest temperature of each session is represented as T1 and the highest as TS. The graph on the left shows magnitude matching ratings and the graph on the right VA.9 ratings.
High > Low
TABLE
111 Subjects .l__.-lFemales
MAlES
FEMALES
Fig. 3. Mean difference scores for female and male subjects. Females rated the high temperatures significantly higher than the low temperatures (paired I test, P = 0.02), while males did not (paired I test. P = 0.17).
temperature range and the Iow temperature range for males and females, using either the VAS or the magmtude matching scales. A paired t test revealed no statistical tendency for males to rate the high range as higher than the low range. However, for both scales, there was a tendency for females to rate the two ranges differently (P = 0.08 for each scale). As described above, when the number of subjects was doubled, by combining the two rating scales, these trends became significant differences for females (see Fig. 3). A comparison of ratings given to the 5 temperature levels within a single session also indicated that, although both mates and females differentiated among temperatures, females may have discriminated the intensities of heat better than did males. Because of the significant interaction that we observed between subject
OJ Ti
T2
T3
T4
T5
TEMPERATURES
Fig. 4. The mean normalized magnitude matching ratings assigned to each of the 5 temperatures by females (solid lines) and by males (dashed lines). The slope of the line representing the ratings of the female subjects (rating = 1.26 +O.f;4T) is significantly steeper than that of the msle subjects (rating = 0.61 +0.3OT) (MLVA. P c O.OOf).
Sessrot1s First Second
46..50°C 45-49OC
R@fiflK\High > low Low > high
11 I
Males
45.-49°C 46-49” C
4 4
46-50°C 45-49 o C‘
x 0
4S-4gQC 46-50° C
3 9
sex and temperature level using the magnitude matching procedure, we performed a slope analysis of those data. For each subject, we calculated an average rating of Tl, T2, T3, T4 and T5 of each session, to form a stimulusresponse curve for that subject. Based on these 5 points per subject, a regression function was obtained for males and for females. Fig. 4 shows that females produced a steeper stimulus-response function than did males. and these slopes were significantly different (MLVA. P < 0.001). These results suggest that females showed a more pronounced differentiation of the various temperature intensities presented within a single session. In order to determine whether experience with the experimental pain stimuli influenced subjects’ ratings of those stimuli, data were compared for a subject’s first and second session. Half of the subjects received the high range (4650°C) in the first session, while the other half received the high range in the second session. Table III shows that almost all subjects who received the high range before the row assigned higher ratings to the stimufi in the higher range. fn contrast, many subjects who received the Iow range first actually rated those lower stimuli as more intense than the subsequent stimuli in the high range. A chi-square analysis revealed a significant difference in ratings for male subjects receiving the higher stimulus range in the first session compared to those receiving this range in the second session (Fisher’s exact test, P = 0.002). A similar analysis for female subjects did not reach significance (P = 0.11).
These data show that female subjects rate experimental heat pain stimuli as more intense than do male subjects, independent of the gender of the experimenter or the nature of the rating scale. In addition, two aspects of our results suggest that females may discriminate various intensities of pain better than do males. First, female subjects showed a significant difference in their ratings of two series of heat pain stimuli, when the temperature range during one session was
760
45549°C and in the other session 46650°C. Although males showed a similar tendency, they did not significantly differentiate these two ranges. Second. when comparing an individual’s temperature ratings across the 5 temperatures presented within a single session. females produced steeper stimulus-response functions than did males. Our data also suggest that, although both males and females tended to rate stimuli as more intense during the first experimental session than during the second, this tendency was more pronounced for males than for females. The current findings that female subjects rate painful thermal stimuli significantly higher than do male subjects are in agreement with the majority of studies using experimental pain stimuli [2,5,6,11,16,36,38,41,46,47]. Our results and those of Otto and Dougher [36] indicate that the finding that females report higher pain than males is independent of the sex of the experimenter, and is thus not related to a tendency for males to act in a stoic manner in the presence of a female investigator. Our findings also indicate that the higher pain ratings on the part of females reflect a difference in sensory pain transmission, either at a peripheral or central level, rather than a difference in attitude or emotional response. Most previous studies have used either pain threshold or pain tolerance as the response measure. However, non-sensory factors, such as anxiety or willingness to report pain, can easily influence either of these measures [3,12]. In our study, by presenting a variety of heat pain intensities in a pseudorandom order and by changing the range of intensities between experimental sessions, we tried to render subjects unaware of the specific stimuli being presented. Thus, their ratings should have been less influenced by attitudinal variables than when threshold or tolerance measures are employed. Further support for the notion that sex differences in pain perception are sensory in nature is derived from our findings that females tended to discriminate among temperatures better than did males. Females, but not males, showed a significant difference in their ratings of the two temperature ranges (shifted by l°C) between experimental sessions. In addition, using the magnitude matching procedure, females produced steeper stimulus-response functions than males for the 5 temperatures presented within an experimental session, indicating that they were better able to differentiate among the various stimulus levels. We did not obtain these slope differences for the VAS measure, but previous data have indicated that VAS ratings can lead to ceiling effects that reduce the discriminative sensitivity of this measure, especially at the higher stimulus intensities
WI. Data from other sensory modalities indicate that females are more sensitive than males to a variety of non-noxious stimuli. Essick et al. [19] found that the
peak sensitivity of female subjects ill brushrng ~.li fir.,facial skin was significantly greater than that of male subjects. Further. females arc reported t
255 B.V. 0304.3959/91/~03.50
PAIN 01752
Sex differences in the perception of noxious heat stimuli Jocelyne
S. Feine
‘, M. Catherine
Bushnell
n.h, Denis
Miron
” and
Ii Fan&6 de Mhdrcme Denrmre, and ’ Cenrre de Recherchr en Sciences ,V~z~r~}~o~19~~,.~, C!nrr*ersirc~de (Received
8 March
1990. revision
received
10 September
1990, accepted
Gary ,~[~~tr~t~l.
12 September
H. Duncan ~~~~F~I~~~J~,
a.’ Que. ~~ur~u~u)
1990)
This study compared pain perception in young male and female subjects, using experimental noxious Summary heat stimuli. During 2 sessions, each of 40 subjects rated the magnitude of 120 heat stimuli, ranging from 45°C to 50°C. The study included a comparison of visual analogue and magnitude matching rating procedures. as well as a test of simulated analgesia, in which the range of stimuli presented during the 2 experimental sessions was shifted by 1°C. We found that females rated noxious heat stimuli as more intense than did males, independent of the gender of the experimenter or the type of rating scale. In addition, the data suggest that females discriminate among the painful heat intensities better than males. For example, female subjects showed significant between-session discrimination of noxious heat stimuli, while male subjects did not, and females produced steeper within-session stimulus-response functions than did males. These observed differences in nociceptive discrimination between males and females indicate that the sex-related variation in pain perception is probably related to sensory factors rather than differences in attitude or emotional response. Key words: Pain;
Sex; Temperature;
Psychophysics
Epidemiologic studies of North Americans indicate that women report a higher incidence of both temporary and persistent pain than do men [15,4.53. Similarly, among patients seeking treatment for chronic pain, the majority are female [28,32]. These differences may be due to sociological factors which demand stoicism in males and allow expression of pain in females, or they may be due to physiologic or anatomic differences between the sexes. Women do not uniformly report higher levels of all types of pain. For example, some studies have found that males report at least as much back pain as do females [42,45]. However, for several other pain sites (head and face), women report pain more often than men [45]. These differences may indicate that environmental factors, such as type of work or exposure to stress, lead to differential pain patterns.
-Crrreqwnckwce IO: Joceiyne S. Feine. DCpartement de Restauration, FacultC de MCdecine Den&ire. Montrkal, iMontreal. Que. H3C 357, Canada.
de Dentisterie Universitk de
In an attempt to directly compare male and female pain perception in a controlled situation. numerous investigators have evaluated sex differences in the perception of experimental noxious stimuli. Using a variety of experimental pain stimuli, including pressure, heat, shock, and cold. many investigators have found that pain threshold and/or tolerance are lower for females than for males [2,5.6,11,16,36.38,41.46.47]. Studies evaluating subjects’ rutings of experimental pain similarly have found that females rate noxious stimuli higher than do males [l&47]. Nevertheless. these observed sex differences in the perception of experimental pain are not unequivocal. as other studies have not found a sex-related difference in pain threshold using noxious shock [34,35], heat [13], and cold [4], nor in pain tolerance using shock [34] and ischemia [lo]. In order to determine if observed differences in pain perception between males and females are related to discriminative capacity or to differences in willingness to report pain, several studies have employed sensory decision theory (SDT) analysis [25] to calculate separately a subject’s discriminative ability (d’) and his criterion for labeling stimuli as painful. Using such measures, results have been highly inconsistent across
256
studies. For example, some studies indicate that women discriminate better (larger d’) than men 16,221, while others report no difference in d’ between the sexes [I 11. Similarly, some studies conclude that women set higher criteria [12], while others report that men set higher criteria (111. In addition to producing inconsistent data, the SDT analysis has been criticized on theoretical grounds as a method for differentiating sensory differences in pain perception from differences in criterion [14,39,40]. Thus, such data have not led to a clear understanding of whether gender-related differences in pain perception are sensory or cultural in nature. One additional study, using classical psychophysical techniques (determination of the Weber fraction and exponent of stimulus-response functions), found no difference in the capacity of males and females to discriminate noxious electric shocks [41]. An important variable that has not been controlled in almost all of these studies is the sex of the experimenter, If sex-related differences in pain perception are even partially related to sociological factors, the sex of the experimenter could alter the subjects’ willingness to report pain. In only a few studies is the sex of the experimenter indicated, and in these the experimenter was female [11,41,46]. One could easily hypothesize that cultural demands would lead males to be less likely than females to report pain to a female investigator. Only one study has controlled for such interactions. by using both male and female investigators [36]. In the present study, we have attempted to clarify some of the ambiguities concerning gender differences in pain perception, by comparing the perception of a range of noxious heat stimuli by males and females, using both a male and a female experimenter, and utilizing response measures that have been previously validated to be sensitive to small differences in pain perception [17,18,37]. Some of these data have been previously reported in abstract form [21].
Methods Subjects Forty paid volunteers, 20 males (ages 19-30 years, mean 24.8 years, median 24 years) and 20 females (ages 20-36 years, mean 25.3 years, median 25 years) participated in this experiment. All read and signed a consent form acknowledging that the experimental procedures had been explained and that they could withdraw, without prejudice, from the experiment at any time. Ten of the 20 female subjects were taking oral contraceptives. Of those normally ovulating subjects (those not taking oral contraceptives), 25% participated in the experiment during the menstrual phase of their cycle, 63% during the post-menstrual phase, 0% during the
TABLF.
1
__~-.Low temperatures
_.-.-.-
--___
_.
Stimulus
o C‘
High temperature\ ~___... ._ _.. Stirnub ‘0f
Tl T2 T3 T4 T5
45 46 47 4x 49
-1’1 1‘2 T3 T4 ‘I‘5
--
.
ik 4’ * .,, 4‘; ”;a
n = 20 1st session - low temperatures 2nd session - high temperatures n = 20 1st session - high temperatures 2nd session - low temperatures
ovulatory phase.
phase.
and
12% during
the
premenstrual
Experimental design The 20 male and 20 female subjects were divided so that half of each group was assigned a female experimenter, and the other half a male experimenter. In order to further control for possible gender-related differences in the use of specific pain rating techniques, half of each group rated noxious thermal stimuli using visual analog scales PAS), a commonly used psychophysical testing technique, and half rated noxious thermal stimuli using the technique of magnitude matching [18]. Thus the gender of the subjects was counterbalanced relative to the gender of the experimenter and to the type of pain rating technique. All subjects participated in 2 experimental sessions (separated by 1 week), in which they rated a range of noxious heat stimuli. During one of these sessions the subjects were presented with 60 heat stimuli between 45°C and 49°C (12 presentations of each of 5 levels). while during the other session they were presented with temperatures between 46°C and 50°C (Table I). Half the subjects received the high range first and the low range second, while the other half received the reverse order. This design allowed us to evaluate subjects’ discriminative ability in 2 ways. First, by determining the slope of the stimulus-response function during a single session, one can assess the subjects’ discriminative sensitivity within the context of a specific range of temperatures 17,411. Secondly, by comparing responses between sessions, when the range of stimuli is shifted by 1°C. one can estimate the reliability of the subjects’ responses, as well as the subjects’ ability to discriminate small differences in stimulus range across time [ 20,24.40]. Rating techniques Visual unalog scales (VAS). Before each stimulus presentation, subjects assigned to the VAS group were given a printed scale consisting of a 10 cm horizontal line with verbal anchors of ‘no pain’ on the left side and
257
‘the most painful that one could imagine’ on the right side. The subjects were instructed to draw a vertical mark at the appropriate position on the visual analog scale to indicate the perceived painfulness of the stimulus relative to the two anchors. Magnitude matching. This psychophysical technique combines the freedom of unrestricted magnitude estimation with a no~alization procedure involving cmssmodality matching of experimental and control stimuli [1X,43). Subjects were instructed to rate the intensity of 60 painful heat stimuli and the brightness of 50 visual control stimuli on the same scale of magnitude, using any whole number or decimal that they thought appropriate. Zero would indicate that the heat stimulus was not at all painful or that the visual stimulus was not detected. No maximum number was specified. Ratings assigned to the visual control stimuli were averaged for each subject, and this mean was used to normalize the ratings of the experimental stimuli. These magnjtude matching procedures were identical to those described previously [ 18].
The standardized instructions describing each of the above rating techniques were memorized by the male and female experimenters and given verbally to the subject. In response to subjects’ questions, the instructions were explained a second time in paraphrased form. Both sessions were preceded by a practice period in which 3 sample heat stimuli were presented. Subjects could then ask additional questions or repeat the practice session if needed.
For all subjects, the thermal stimuli were applied by the investigator via a 1 cm diameter hand-held contact thermode, to 4 spots of skin above the subject’s upper lip. Neither the subject nor the experimenter knew the exact range of thermal stimuli nor the order of the intensities that were counterbalanced across skin locations. During these heat trials, the temperature of the thermode increased from a baseline of 39°C to 1 of 5 higher temperatures (45-49°C or 46-50°C; see Table
1). For those subjects using the technique of magnitude matching, thermal and visual stimuli were presented on alternate trials. On visual trials an opaque cue light, situated 40--50 cm in front of the subject, increased in intensity to 1 of 5 brightness levels. These light intensities were counterbalanced so that the intensity of the visual stimuli was not related to the intensity of the thermal stimuli in any systemic fashion. Following a ready signal on the computer monitor, the subjects initiated each trial by pressing the ‘return’ button on the keyboard. After each stimulus presentation, the subject rated the magnitude of the stimulus by
typing his numerical estimate on the keyboard (magnitude matching group), or by drawing a vertical line on a visual analog scale. For all subjects, intertrial intervals were adjusted to allow a minimum of 1 mm to elapse between presentation of consecutive thermal stimuli to the same spot of skin. We have previously shown that this period is sufficiently long to minimize response changes attributable to suppression or sensitization of nociceptive afferents [8,9].
Apparatus A minicomputer controlled all stimulus presentations, collected subjects’ numerical ratings and performed the statistical computations. Thermal stimuli were presented with a 1 cm diameter contact thermode which incorporated a servo-controlled heating element and active cooling via refrigerated, circulating water. The warming and cooling slopes of the thermode were appro~mately 6”C/sec. Visual stimuli were presented through a 3 cm diameter opaque white key by a tungsten lamp whose brightness was controIled by the digital-analog output of the computer. The luminosity of the experimental room was maintained at a constant level.
Data analysis and statistics Ratings collected using the magnitude matching technique were divided by the subject’s mean rating of the visual stimuli, in order to normalize the data across subjects (see Duncan et al. for details [lS]>. VAS ratings were calculated by meas~~ng the distance from the left end of the scale to the subject’s mark. For all analyses in which data were collapsed across the two rating techniques, data were normalized so that the subject’s rating was divided by the median rating of all subjects (male and female) who used that rating technique. A J-way ANOVA was used to compare experimenter gender, subject gender, and type of psychophysical test. A paired t test was used to analyze the difference in the subjects’ abilities to rate high vs. low temperatures. A repeated measures ANOVA and a multilevel variate analysis (MLVA) was used to determine differences in the slopes of thermal stimulus-response functions created by the male and female subjects. In addition, a Fischer’s exact chi-square analysis was used to evaluate sequence effects between the 2 experimental sessions.
Results Overall, female subjects rated the temperatures significantly higher than did male subjects. This is demonstrated in Fig. 1, in which the mean normalized ratings For all the temperatures from both sessions were collapsed and compared (P = 0.008, 3-way ANOVA). In contrast, there was no significant difference due to the
2.0 -
l’.ABLti
1.5-
M&Z\ -.___---“.-.._ crslrul an&g
If
-._.
_.
Females _._._~~~ ~~
vcoic~
High
44.x
High
Low
x1.9
Low
t>o -I
5ii.S
Paired f test P = 0.23
Patred i wst
I’ c= o.oii
MugnmJc High
High
>,‘i :
LOW
1.32
1.0-
LO\s
nmrching swfe 1.41 1.24
Paired t test P = 0.23
0.5 -
Paired f test P = 0.08
0.0 + Fig. 1. Normalized mean ratings of all subjects collapsed across all temperatures received during both sessions. Ratings of female subjects were significantly higher than those of males (P = 0.008, 3-way ANOVA). Because data were normalized using median ratings, which differed slightly from the means, the mean female rating (1.4) and the mean male rating (0.8) yield a grand mean of I .l.
rating technique (P = 0.20, 3-way ANOVA) nor to the gender of the experimenter (P = 0.70, 3-way ANOVA). Fig. 2 shows that females rated each temperature level higher than did males, independent of the type of rating scale. Separate repeated measures ANOVAs for the VAS and magnitude matching data, which compared males and females across the 5 temperature levels, confirmed that for both measures females rated the temperatures higher than did the males (magnitude matching F (1,18) = 5.02, P-c0.05; VAS F (1, 18) = 5.33 P -c0.05). In addition, using either measurement scale, subjects rated the 5 temperature levels differentially (magnitude matching F (4) = 44.51, P < 0.001; VAS F (4) = 222.9, P < 0.001). Finally, for the magni-
MAGNITUDE --* ---t
T,
MATCHING
100
FEMALE SUBJECTS MALE SUBJECTS
T2
T3
TEMPERATURES
tude matching data, but not for the VAS, there was an interaction between sex and temperature level (magnitude matching F (4. 72) = 2.83. P -c0.05:VAS I,‘ (4. 72) = 0.88. P = 0.48). To determine whether the subjects could discriminate between high and low temperature ranges between sessions. a normalized difference score was calculated for each subject bc subtracting the mean ratings for the IOU temperature session from the mean ratings for the high temperature session. and then dividing by the mean ratings assigned to the high temperatures. These normalized difference scores, combined across both rating scales, were then statistically compared for both the male and female subjects. Fig. 3 shows that the females rated the high temperatures higher than the low temperatures (P < 0.02, paired t test), while ratings assigned by the male subjects failed to reach significance (P zy 0.17. paired I test). Similar, but weaker, trends were obtained when the VAS and magnitude matching data were analyzed separately. Table II shows the mean ratings of the high
VISUAL ANALOG SCALE
80
T‘i
Ts
v
T,
TZ
T3
T4
TS
TEMPERATURES
Fig. 2. Mean ratings of males (filted circles) and females (open circles) for the 5 temperature levels presented in each of 2 sessions. Although the range was shifted between sessions (see Methods), the average rating for the lowest temperature of each session is represented as T1 and the highest as TS. The graph on the left shows magnitude matching ratings and the graph on the right VA.9 ratings.
High > Low
TABLE
111 Subjects .l__.-lFemales
MAlES
FEMALES
Fig. 3. Mean difference scores for female and male subjects. Females rated the high temperatures significantly higher than the low temperatures (paired I test, P = 0.02), while males did not (paired I test. P = 0.17).
temperature range and the Iow temperature range for males and females, using either the VAS or the magmtude matching scales. A paired t test revealed no statistical tendency for males to rate the high range as higher than the low range. However, for both scales, there was a tendency for females to rate the two ranges differently (P = 0.08 for each scale). As described above, when the number of subjects was doubled, by combining the two rating scales, these trends became significant differences for females (see Fig. 3). A comparison of ratings given to the 5 temperature levels within a single session also indicated that, although both mates and females differentiated among temperatures, females may have discriminated the intensities of heat better than did males. Because of the significant interaction that we observed between subject
OJ Ti
T2
T3
T4
T5
TEMPERATURES
Fig. 4. The mean normalized magnitude matching ratings assigned to each of the 5 temperatures by females (solid lines) and by males (dashed lines). The slope of the line representing the ratings of the female subjects (rating = 1.26 +O.f;4T) is significantly steeper than that of the msle subjects (rating = 0.61 +0.3OT) (MLVA. P c O.OOf).
Sessrot1s First Second
46..50°C 45-49OC
R@fiflK\High > low Low > high
11 I
Males
45.-49°C 46-49” C
4 4
46-50°C 45-49 o C‘
x 0
4S-4gQC 46-50° C
3 9
sex and temperature level using the magnitude matching procedure, we performed a slope analysis of those data. For each subject, we calculated an average rating of Tl, T2, T3, T4 and T5 of each session, to form a stimulusresponse curve for that subject. Based on these 5 points per subject, a regression function was obtained for males and for females. Fig. 4 shows that females produced a steeper stimulus-response function than did males. and these slopes were significantly different (MLVA. P < 0.001). These results suggest that females showed a more pronounced differentiation of the various temperature intensities presented within a single session. In order to determine whether experience with the experimental pain stimuli influenced subjects’ ratings of those stimuli, data were compared for a subject’s first and second session. Half of the subjects received the high range (4650°C) in the first session, while the other half received the high range in the second session. Table III shows that almost all subjects who received the high range before the row assigned higher ratings to the stimufi in the higher range. fn contrast, many subjects who received the Iow range first actually rated those lower stimuli as more intense than the subsequent stimuli in the high range. A chi-square analysis revealed a significant difference in ratings for male subjects receiving the higher stimulus range in the first session compared to those receiving this range in the second session (Fisher’s exact test, P = 0.002). A similar analysis for female subjects did not reach significance (P = 0.11).
These data show that female subjects rate experimental heat pain stimuli as more intense than do male subjects, independent of the gender of the experimenter or the nature of the rating scale. In addition, two aspects of our results suggest that females may discriminate various intensities of pain better than do males. First, female subjects showed a significant difference in their ratings of two series of heat pain stimuli, when the temperature range during one session was
760
45549°C and in the other session 46650°C. Although males showed a similar tendency, they did not significantly differentiate these two ranges. Second. when comparing an individual’s temperature ratings across the 5 temperatures presented within a single session. females produced steeper stimulus-response functions than did males. Our data also suggest that, although both males and females tended to rate stimuli as more intense during the first experimental session than during the second, this tendency was more pronounced for males than for females. The current findings that female subjects rate painful thermal stimuli significantly higher than do male subjects are in agreement with the majority of studies using experimental pain stimuli [2,5,6,11,16,36,38,41,46,47]. Our results and those of Otto and Dougher [36] indicate that the finding that females report higher pain than males is independent of the sex of the experimenter, and is thus not related to a tendency for males to act in a stoic manner in the presence of a female investigator. Our findings also indicate that the higher pain ratings on the part of females reflect a difference in sensory pain transmission, either at a peripheral or central level, rather than a difference in attitude or emotional response. Most previous studies have used either pain threshold or pain tolerance as the response measure. However, non-sensory factors, such as anxiety or willingness to report pain, can easily influence either of these measures [3,12]. In our study, by presenting a variety of heat pain intensities in a pseudorandom order and by changing the range of intensities between experimental sessions, we tried to render subjects unaware of the specific stimuli being presented. Thus, their ratings should have been less influenced by attitudinal variables than when threshold or tolerance measures are employed. Further support for the notion that sex differences in pain perception are sensory in nature is derived from our findings that females tended to discriminate among temperatures better than did males. Females, but not males, showed a significant difference in their ratings of the two temperature ranges (shifted by l°C) between experimental sessions. In addition, using the magnitude matching procedure, females produced steeper stimulus-response functions than males for the 5 temperatures presented within an experimental session, indicating that they were better able to differentiate among the various stimulus levels. We did not obtain these slope differences for the VAS measure, but previous data have indicated that VAS ratings can lead to ceiling effects that reduce the discriminative sensitivity of this measure, especially at the higher stimulus intensities
WI. Data from other sensory modalities indicate that females are more sensitive than males to a variety of non-noxious stimuli. Essick et al. [19] found that the
peak sensitivity of female subjects ill brushrng ~.li fir.,facial skin was significantly greater than that of male subjects. Further. females arc reported t
