- Email: [email protected]
On the sensitivity of the tourniquet pain test
105 Pain, :3 (1977)105--119 © Elsevier/North-Hollan d Biomedical Press
ON THE SENSITIVITY OF THE TOURNIQUET PAiN TEST
RICHARD A. STE.RNBACH *, LYDIA M. DEEMS, GRETCHEN TIMMERMANS and LEIGHTON Y. HUEY
Pain Unit, Veterans Administration Hospital, San Diego, Calif. 92161 (U.S.A.) (Accepted October 4th, 1976)
SUMMAI~Y
Twenty-four chronic pain patients were given, on each of 4 successive days, oral doses of 60 mg morphine, 60 mg codeine, 60(1, mg aspirin and placebo, using a double-blind counterbalanced design. Two hours after ir.gestion, subjective pain estimates and tourniquet pain scores were obta~,ned. Variability o f the tourniquet pain scores was too great for differences in response to the analgesics to be significant. However, differences in pain estimates were also too smaU to discriminate among the drugs, and the lack of sensitivity may be a function of pain chronicity. The tourniquet techniques will continue to be useful until there is a purely objectiw~ measure of the severity of clinical pain.
INTRODUCTION
Some time ago we reported a method for measuring the seve~:ity of clinical pain, the tourniquet pain test [8]. It is an adaptation of the submaximum effort tourniquet technique developed by Smith and his colle~gues [2--4]. Two measures of ischemic pain are obtained from the patient: (a) the clinical pain level, which is the time (in seconds) required for the ischemic pain to match the severity or intensity of the patient's clinical p~dn; {b) the maximum pain tolerance, which is the duration (in seconds) which the patient can endure the ischemic pain. From these direct scores, an indirect measure is obtaiI,ed, (c) the tourniquet pain score, which is the percentage of (a)/(b) × 100. This score reflects the proportion of the patient's clinic~fl pain to his own pain tolerance. This pain score can then be compared with (d) the patient's pain .estimate, which is
* Present address: Pain Treatment Center, Scripps Clinic Medical Group, Inc., La Jo!la, Calif. 92037, U.S.A.
106
his previously given rating of the severity of his clinical pain, on a scale from 0 to I00. In the earlier paper, we reported the reliability of this method: clinical pain levels had a reliability ranging from 0.50 to 0.88, with an average o£ about 0.70. The; reliability of the maximum pain tolerance ranged from 0.65 to 0.94, with an average of about 0.80. The reliability of the ratio score, although clinically significant, was considerably lower than its components, ranging from 0.';8 to 0.72, with an average of about 0.47. As evidence for the validity of the tourniquet pain test, we showed that the scores varie~! in the appropriate direction and degree with psychological and surgical treatment for pain, and were correlated significantly (0.48) with the pain estimates [8]. We also cited the studies by Smith et al. [2--4] that ischemic pain was. sensitive not merely to 10 mg morphine i.m., but also to 600 mg aspirin p.o. Subsequently we have also shown that the patient's pain estimate is associated with the impact of the pain on his daily activities, whereas the tourniquet pain ratio score was associated with the patient's level of depression [10], the two scores clearly reflecting different factors [9,10]. In addition, the usefuhaess of the tourniquet pain test was shown in the following: the clinical pain level, but not the maximum pain tolerance, was reduced with the analgesia provided by transcutaneous neurostimulation [6]; however, the maximum pain tolerance, but not the clinical pain level, was increased with the administration of chlorimipramine [7]. Thus the clinical pain level component of the tourniquet pain test seems to reflect peripheral phenomena, wbile the m a x k m m pain tolerance component seems to reflect central phenomena. However, some doubt about the sensitivity of the submaximum effort tourniquet technique was rel,orted by Moore et al. [1 ], who co Llld not show any difference between 900 mg aspirin and placebo in a study ~sing healthy prison volunteers. Such a finding, if supported, would limit the usefulness of our modification of the submaximum effort tourniquet technique, as described. Accordingly, we have carefully tested the sen.~itivity of the tourniquet pain test as follows. METHOD
On each of 4 consecutive =lays, subjects received orally either 60 mg morphine, 60 mg codehie, 600 mg aspirin, or placebo. Subjects were 24 male patient volunteers from the pain ward of the Veterans Administration Hospital, San Diego, who .had completed their .'.~ourse of inpatient treatment [5] m~d were ready for discharge, but elected to remain for the additional 4 days necessary to complete this study. All patients had somatogenic pain and had been used to making daily pain estilnates and weekl:y toumiq':~et pa~n tests. No patient was used who had a known sensitivity to one of the a:~algesics, or who had a history of cardiovascular disease. Most were taking mild analgesics every 4 h, and were using
107 the neurostimulator daily. All patients gave fully informed voluntary consent. Procedure. Each patient received no analgesic (nor electrical neurostimulation) after 9.00 p.m. At 9.00 a.m. the following day he received a capsule whose contents were unknown to the patient and to the ward staff. Each hour from 8,00 a.m. to 12 noon, the patient made a pain estimate (on a 0-100 scale of intensity). At 11.00 a.m. he took the tourniquet pain test..~fter noon, he could resume his use of the neurostimulator and~or usual analgesics until 9.00 p.m., when the sequence of events was repeated. Thus at least 12 h free of analgesia, chemical or electrical, preceded the ingestion of the test capsule. There are 24 possible sequences of the 4 drugs, morphine, codeine, aspirin, and placebo. Eacb of ~he 24 patients followed a different one cf possible sequences, so that the order effect of the drugs was completely cQunterbalanced for the group. Thus the tourniquet pain test was being evaluated for its sensitivity to known, standard analgesics of different potency, using a double-blind counterbalanced design. Analysis was made of the data taken 2 h after ingestion of the analgesic, which is the usual time for the appearance of a maximum analgesic effect. Four figures were obtained from each patient each day" his pain estimate at 11.00 a.m., and the clinical pain level, maximum pain tolerance, and tourniquet pain ratio score also obtained at 11.00 a.m. each day. RESULTS
Variability of results in all scores was so great as to preclude achieving statistical significance, as sh~)wn m Fig. 1. The clinical pain level sc,ores (Fig. la) show no relationship to analgesic doses. The scores represent the duration of ischemia requf:ced to match clinical pain levels, and "should" be lowest for morphine, highest for placebo. Instead, it is highest for morphine, lowest for aspirin, with placebo intermediate. (It is, of course, possible that aspirin has a greater peripheral analgesic mechanism.) Analysis of variance for treatment e::fects is not significant (F = 0.872). The maximum pain tolerance scores (Fig. l b ) are properly ordered according to expected analgesic strength, with morphine producing the greatest increase in duration for tolerating ischemic pain, and ~;pirin and placebo equal and least effective. PTowever, the variability is so great that analysis of variance is n o t significant (F = 0 942). The tourniquet pain (ratio) scores (Fig. l c ) are virtually identical for all the analgesics, due to the similar slopes of the component scores in Fig. l a and lb. The mean scores far each of the analgesics are only slightly lower than for the placebo. Again, the differences are not significant (F = 0.182). The patients' pain estimates (Fig. ld) appropriately follow the analgesic dosages, being lowest for morphine and highest for placebo. The graph is virtually a mirror image of the changes in maximum pain t;olerance (Fig. 2b).
108 500
240 220
co 2 0 0 o
¢--> L~
460
T
180
160
co 420 z o :580 L~D IJ
I
340
[
1
300
140
ol
0
o o x
~
,~ ¢n
T
50
45 40
[
i
i
•
55 50
~ 45 ~'-
4O
~
35
z
35
,
d
lOO
c _J
55
,
MAXIMUMPAINTOLEF~NEE
CLINICAL PAINLEVEL I00
,
M C A P
[
30 o
M C A P TOURNIQUETRATIO
PAIN ESTIMATE
Fig. 1. Means and standard errors o f responses to oral administration of 60 mg morphine (M), 60 mg codeine (C), 600 mg acetylsaiicylic acid (aspirin) (A), and placebo (P). a: time (in sec) for the induced ischemic pain to match in intensity the patient's clinical pain level, b: time (in sec) for the ischemic pain to reach the maximum severity the patient could tolerate, c: derived tourniquet ratio scores reflecting the proportio~l o f the matched ischemic pain to the patient's pain tolerance, d" traditional pain estimates reflecting subjective magnitude production. None of the scores discriminate among the analgesics.
However, as with the other scores, the differences are not significant ( F 0.916). DISCUSSION
The failure of the tourniquet pain test to detect differences in the analgesic potency of the drugs used in this study is disappointing. The disappointrr.ent is relieved somewhat by the fact that patients' pain estimates, the traditional subjective measure used in clinical analgesic studies, also failed to discriminate among the analgesics. Since the analgesics used in this study are usually discriminable in acute pain studies in bot:h clinical and laboratory tests, we suspect that the failure may lie with our patients: most had been in chronic pain for several years, ~a~d had been using narcotic analgesics for a similar period. The 12-h "washout" period bet~vveen analgesics may have been inadequate. This chronicity may lave impaired their capacity to discriminate. De~:pite the failure of the tourniquet pain score to be sensitive to the dif-
109
ferent strengths o f analgesics, we note t h a t the direction of differences is as expected in the maximum pain tolerance scores (as they are in the subjective pain estimates). This suggests that it is this component of the tourniquet ratio score whichresponds to centrally acting agents [7]. The matching clinical pain level scores seem not to do so, but have responded to peripheral analgesic mechanisms to which the maximum pain tolerance scores were unresponsive [6]. These results tend to support the finding that the matching Gf ischemic pain t o the patients' clinical pain intensity (magnitude matching) yields a score which reflects something different from patients" subjective pain ratings (magnitude production). Subjective pain estimates seem to reflect the impact of pain on patients' daily activities, whereas the tom~iquet pain score (probably due to the maximum tolerance component) seems to reflect the level of depression, at least in chronic pain patients [ 10]. Thus we may conclude that in this study employing chronic pain patients, the variability of tourniquet pain test scores was so great that the technique failed to discriminate among analgesics of different strengths. This lack of sensitivity may have been due to an insensitivity of the chronic pain patients rather than a failure of the technique, because differences between the means in the subjective pain estimates are too small to achieve significance, despite relatively small variability in these scores (Fig. ld). Because of previous findings of its usefulness, the tourniquet technique ~ill continue to be useful until a purely objective physiological measure of clin~ical pain intensity is developed. ACKNOWLEDGEMENT
This work was supported by Vetelans Administration I~e~earch Grant MRIS-8131 to R.A. Sternbach. REFEEENCES 1 Moore, J.D., Weissman, L., Thomas, G. and Whitman, E.N., Response of experimental ischemic pain to analgesics in prisoner volunteers, J. clin. Pharmacol., 11 (1971) 433-439. 2 Smith, G.M. and Beecher, H.K., Experimental production of pain in man: sensitivity of a new method to 600 mg of aspirin, Clin. Pharmacol. Ther., 10 (1969) 213--216. 3 Smith, G.M., Egbert, L.D., Markowitz, R.A., Mosteller, F. and Beecher, H.K., An experimental pain method sensitive to morphine in man: the submaximum eff,~rt tourniquet technique, J. Pharmacol. exp. Ther., 154 (1966) 324--332. 4 Smith, G.M., Lowenstein, E., Hubbard, J.H. and Beecher, H.K., Experimental pain produced by the submaximum effort tourniquet technique: further evidence of validity, J. Pharmacol. exp. Ther., 163 (1968) 468--474. 5 Sternbach, R.A.,, Pain Patients: Traits and Treatment, Academic Press, New York, 1974, pp. 94--116. 6 Sternbach, R.A., Ignelzi, R.J., Deems, L.M. and Timmermans, G., Transcutaneous electrical analgesia: a follow-up anaiysis, Pain, 2 (1976) 35--41. 7 Sternbach, R.A., Janowsky, D.S., Huey, L.Y. and Segal, D.S., Effects of a!tering brain
110 serotonin activity on human chronic pain. In: J.J. Bonita and D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy, Vol. I, Raven Press, N e w York, 1976, pp. 601--606. 8 Sternbach, R.A., Murphy, R.W., Timmermans, G,, Greenhoot, J.H. and A k ~ o n , W.H., Mea~suring the severity of clinical pain. In: J.J. Bonica (Ed.), Advances in Neurology, Vol. 4, International Symposium on Pain, Raven Press, New York, 1974, pp. 281-288. 9 Timmermans, G. and Sternbach, R.A., Factors of hu~.~an chronic pain: an analysis of personality an~ pain reaction variables, Science, 184 (1974) 806--808. 10 Timmerman.% 1}. and Sternbach, R.A., Human chronic pain and personality: a canonical correlation analysis. ~ : J.J. Bonica and D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy, Vol. I, Raven Press, New York, 1976, pp. 307--310.
ON THE SENSITIVITY OF THE TOURNIQUET PAiN TEST
RICHARD A. STE.RNBACH *, LYDIA M. DEEMS, GRETCHEN TIMMERMANS and LEIGHTON Y. HUEY
Pain Unit, Veterans Administration Hospital, San Diego, Calif. 92161 (U.S.A.) (Accepted October 4th, 1976)
SUMMAI~Y
Twenty-four chronic pain patients were given, on each of 4 successive days, oral doses of 60 mg morphine, 60 mg codeine, 60(1, mg aspirin and placebo, using a double-blind counterbalanced design. Two hours after ir.gestion, subjective pain estimates and tourniquet pain scores were obta~,ned. Variability o f the tourniquet pain scores was too great for differences in response to the analgesics to be significant. However, differences in pain estimates were also too smaU to discriminate among the drugs, and the lack of sensitivity may be a function of pain chronicity. The tourniquet techniques will continue to be useful until there is a purely objectiw~ measure of the severity of clinical pain.
INTRODUCTION
Some time ago we reported a method for measuring the seve~:ity of clinical pain, the tourniquet pain test [8]. It is an adaptation of the submaximum effort tourniquet technique developed by Smith and his colle~gues [2--4]. Two measures of ischemic pain are obtained from the patient: (a) the clinical pain level, which is the time (in seconds) required for the ischemic pain to match the severity or intensity of the patient's clinical p~dn; {b) the maximum pain tolerance, which is the duration (in seconds) which the patient can endure the ischemic pain. From these direct scores, an indirect measure is obtaiI,ed, (c) the tourniquet pain score, which is the percentage of (a)/(b) × 100. This score reflects the proportion of the patient's clinic~fl pain to his own pain tolerance. This pain score can then be compared with (d) the patient's pain .estimate, which is
* Present address: Pain Treatment Center, Scripps Clinic Medical Group, Inc., La Jo!la, Calif. 92037, U.S.A.
106
his previously given rating of the severity of his clinical pain, on a scale from 0 to I00. In the earlier paper, we reported the reliability of this method: clinical pain levels had a reliability ranging from 0.50 to 0.88, with an average o£ about 0.70. The; reliability of the maximum pain tolerance ranged from 0.65 to 0.94, with an average of about 0.80. The reliability of the ratio score, although clinically significant, was considerably lower than its components, ranging from 0.';8 to 0.72, with an average of about 0.47. As evidence for the validity of the tourniquet pain test, we showed that the scores varie~! in the appropriate direction and degree with psychological and surgical treatment for pain, and were correlated significantly (0.48) with the pain estimates [8]. We also cited the studies by Smith et al. [2--4] that ischemic pain was. sensitive not merely to 10 mg morphine i.m., but also to 600 mg aspirin p.o. Subsequently we have also shown that the patient's pain estimate is associated with the impact of the pain on his daily activities, whereas the tourniquet pain ratio score was associated with the patient's level of depression [10], the two scores clearly reflecting different factors [9,10]. In addition, the usefuhaess of the tourniquet pain test was shown in the following: the clinical pain level, but not the maximum pain tolerance, was reduced with the analgesia provided by transcutaneous neurostimulation [6]; however, the maximum pain tolerance, but not the clinical pain level, was increased with the administration of chlorimipramine [7]. Thus the clinical pain level component of the tourniquet pain test seems to reflect peripheral phenomena, wbile the m a x k m m pain tolerance component seems to reflect central phenomena. However, some doubt about the sensitivity of the submaximum effort tourniquet technique was rel,orted by Moore et al. [1 ], who co Llld not show any difference between 900 mg aspirin and placebo in a study ~sing healthy prison volunteers. Such a finding, if supported, would limit the usefulness of our modification of the submaximum effort tourniquet technique, as described. Accordingly, we have carefully tested the sen.~itivity of the tourniquet pain test as follows. METHOD
On each of 4 consecutive =lays, subjects received orally either 60 mg morphine, 60 mg codehie, 600 mg aspirin, or placebo. Subjects were 24 male patient volunteers from the pain ward of the Veterans Administration Hospital, San Diego, who .had completed their .'.~ourse of inpatient treatment [5] m~d were ready for discharge, but elected to remain for the additional 4 days necessary to complete this study. All patients had somatogenic pain and had been used to making daily pain estilnates and weekl:y toumiq':~et pa~n tests. No patient was used who had a known sensitivity to one of the a:~algesics, or who had a history of cardiovascular disease. Most were taking mild analgesics every 4 h, and were using
107 the neurostimulator daily. All patients gave fully informed voluntary consent. Procedure. Each patient received no analgesic (nor electrical neurostimulation) after 9.00 p.m. At 9.00 a.m. the following day he received a capsule whose contents were unknown to the patient and to the ward staff. Each hour from 8,00 a.m. to 12 noon, the patient made a pain estimate (on a 0-100 scale of intensity). At 11.00 a.m. he took the tourniquet pain test..~fter noon, he could resume his use of the neurostimulator and~or usual analgesics until 9.00 p.m., when the sequence of events was repeated. Thus at least 12 h free of analgesia, chemical or electrical, preceded the ingestion of the test capsule. There are 24 possible sequences of the 4 drugs, morphine, codeine, aspirin, and placebo. Eacb of ~he 24 patients followed a different one cf possible sequences, so that the order effect of the drugs was completely cQunterbalanced for the group. Thus the tourniquet pain test was being evaluated for its sensitivity to known, standard analgesics of different potency, using a double-blind counterbalanced design. Analysis was made of the data taken 2 h after ingestion of the analgesic, which is the usual time for the appearance of a maximum analgesic effect. Four figures were obtained from each patient each day" his pain estimate at 11.00 a.m., and the clinical pain level, maximum pain tolerance, and tourniquet pain ratio score also obtained at 11.00 a.m. each day. RESULTS
Variability of results in all scores was so great as to preclude achieving statistical significance, as sh~)wn m Fig. 1. The clinical pain level sc,ores (Fig. la) show no relationship to analgesic doses. The scores represent the duration of ischemia requf:ced to match clinical pain levels, and "should" be lowest for morphine, highest for placebo. Instead, it is highest for morphine, lowest for aspirin, with placebo intermediate. (It is, of course, possible that aspirin has a greater peripheral analgesic mechanism.) Analysis of variance for treatment e::fects is not significant (F = 0.872). The maximum pain tolerance scores (Fig. l b ) are properly ordered according to expected analgesic strength, with morphine producing the greatest increase in duration for tolerating ischemic pain, and ~;pirin and placebo equal and least effective. PTowever, the variability is so great that analysis of variance is n o t significant (F = 0 942). The tourniquet pain (ratio) scores (Fig. l c ) are virtually identical for all the analgesics, due to the similar slopes of the component scores in Fig. l a and lb. The mean scores far each of the analgesics are only slightly lower than for the placebo. Again, the differences are not significant (F = 0.182). The patients' pain estimates (Fig. ld) appropriately follow the analgesic dosages, being lowest for morphine and highest for placebo. The graph is virtually a mirror image of the changes in maximum pain t;olerance (Fig. 2b).
108 500
240 220
co 2 0 0 o
¢--> L~
460
T
180
160
co 420 z o :580 L~D IJ
I
340
[
1
300
140
ol
0
o o x
~
,~ ¢n
T
50
45 40
[
i
i
•
55 50
~ 45 ~'-
4O
~
35
z
35
,
d
lOO
c _J
55
,
MAXIMUMPAINTOLEF~NEE
CLINICAL PAINLEVEL I00
,
M C A P
[
30 o
M C A P TOURNIQUETRATIO
PAIN ESTIMATE
Fig. 1. Means and standard errors o f responses to oral administration of 60 mg morphine (M), 60 mg codeine (C), 600 mg acetylsaiicylic acid (aspirin) (A), and placebo (P). a: time (in sec) for the induced ischemic pain to match in intensity the patient's clinical pain level, b: time (in sec) for the ischemic pain to reach the maximum severity the patient could tolerate, c: derived tourniquet ratio scores reflecting the proportio~l o f the matched ischemic pain to the patient's pain tolerance, d" traditional pain estimates reflecting subjective magnitude production. None of the scores discriminate among the analgesics.
However, as with the other scores, the differences are not significant ( F 0.916). DISCUSSION
The failure of the tourniquet pain test to detect differences in the analgesic potency of the drugs used in this study is disappointing. The disappointrr.ent is relieved somewhat by the fact that patients' pain estimates, the traditional subjective measure used in clinical analgesic studies, also failed to discriminate among the analgesics. Since the analgesics used in this study are usually discriminable in acute pain studies in bot:h clinical and laboratory tests, we suspect that the failure may lie with our patients: most had been in chronic pain for several years, ~a~d had been using narcotic analgesics for a similar period. The 12-h "washout" period bet~vveen analgesics may have been inadequate. This chronicity may lave impaired their capacity to discriminate. De~:pite the failure of the tourniquet pain score to be sensitive to the dif-
109
ferent strengths o f analgesics, we note t h a t the direction of differences is as expected in the maximum pain tolerance scores (as they are in the subjective pain estimates). This suggests that it is this component of the tourniquet ratio score whichresponds to centrally acting agents [7]. The matching clinical pain level scores seem not to do so, but have responded to peripheral analgesic mechanisms to which the maximum pain tolerance scores were unresponsive [6]. These results tend to support the finding that the matching Gf ischemic pain t o the patients' clinical pain intensity (magnitude matching) yields a score which reflects something different from patients" subjective pain ratings (magnitude production). Subjective pain estimates seem to reflect the impact of pain on patients' daily activities, whereas the tom~iquet pain score (probably due to the maximum tolerance component) seems to reflect the level of depression, at least in chronic pain patients [ 10]. Thus we may conclude that in this study employing chronic pain patients, the variability of tourniquet pain test scores was so great that the technique failed to discriminate among analgesics of different strengths. This lack of sensitivity may have been due to an insensitivity of the chronic pain patients rather than a failure of the technique, because differences between the means in the subjective pain estimates are too small to achieve significance, despite relatively small variability in these scores (Fig. ld). Because of previous findings of its usefulness, the tourniquet technique ~ill continue to be useful until a purely objective physiological measure of clin~ical pain intensity is developed. ACKNOWLEDGEMENT
This work was supported by Vetelans Administration I~e~earch Grant MRIS-8131 to R.A. Sternbach. REFEEENCES 1 Moore, J.D., Weissman, L., Thomas, G. and Whitman, E.N., Response of experimental ischemic pain to analgesics in prisoner volunteers, J. clin. Pharmacol., 11 (1971) 433-439. 2 Smith, G.M. and Beecher, H.K., Experimental production of pain in man: sensitivity of a new method to 600 mg of aspirin, Clin. Pharmacol. Ther., 10 (1969) 213--216. 3 Smith, G.M., Egbert, L.D., Markowitz, R.A., Mosteller, F. and Beecher, H.K., An experimental pain method sensitive to morphine in man: the submaximum eff,~rt tourniquet technique, J. Pharmacol. exp. Ther., 154 (1966) 324--332. 4 Smith, G.M., Lowenstein, E., Hubbard, J.H. and Beecher, H.K., Experimental pain produced by the submaximum effort tourniquet technique: further evidence of validity, J. Pharmacol. exp. Ther., 163 (1968) 468--474. 5 Sternbach, R.A.,, Pain Patients: Traits and Treatment, Academic Press, New York, 1974, pp. 94--116. 6 Sternbach, R.A., Ignelzi, R.J., Deems, L.M. and Timmermans, G., Transcutaneous electrical analgesia: a follow-up anaiysis, Pain, 2 (1976) 35--41. 7 Sternbach, R.A., Janowsky, D.S., Huey, L.Y. and Segal, D.S., Effects of a!tering brain
110 serotonin activity on human chronic pain. In: J.J. Bonita and D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy, Vol. I, Raven Press, N e w York, 1976, pp. 601--606. 8 Sternbach, R.A., Murphy, R.W., Timmermans, G,, Greenhoot, J.H. and A k ~ o n , W.H., Mea~suring the severity of clinical pain. In: J.J. Bonica (Ed.), Advances in Neurology, Vol. 4, International Symposium on Pain, Raven Press, New York, 1974, pp. 281-288. 9 Timmermans, G. and Sternbach, R.A., Factors of hu~.~an chronic pain: an analysis of personality an~ pain reaction variables, Science, 184 (1974) 806--808. 10 Timmerman.% 1}. and Sternbach, R.A., Human chronic pain and personality: a canonical correlation analysis. ~ : J.J. Bonica and D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy, Vol. I, Raven Press, New York, 1976, pp. 307--310.