Pain, 47 (1991) 183-189 :~~ 1991 Elsevier Science Publishers B.V. All rights reserved 0304-3959/91/$1)3.50
183
PAIN 01869
Psychosocial factors in chronic spinal cord injury pain Jay D. S u m m e r s ~, Michael A. R a p o f f h, G e o r g e V a r g h e s e b, K e n t P o r t e r c and R i c h a r d E. P a l m e r ~' " Department of Psychology, University of Kansas, Lawrence, KS 66045 (U.S.A.), and i, Department ~f Rehabilitation Medicine, Unit'ersity of Kansas Medical Center, Kansas City, KS 66103 (U.S.A.), and ' Kansas City VA Medical Center, Kansas City, MO 64128 (U.S.A.) (Received 1 October 1990, revision received 20 February 1991, accepted 28 February 1991)
Summary Chronic pain is a problem among patients with spinal cord injuries, but the psychosocial factors associated with spinal cord injury (SCI) pain are not well understood. To understand SCI pain further, 54 patients (19 with quadriplegia and 35 with paraplegia) completed the Beck Depression Inventory, State-Trait Anxiety Inventory, Profile of Mood States, Acceptance of Disability Scale and SCI Interference Scale. Forty-two patients stated they had SCI pain and completed the Multidimensional Pain Inventory and the Pain Experience Scale. Results revealed that anger and negative cognitions were associated with greater pain severity. Patients who reported pain in response to a general prompt experienced more severe pain than patients who reported pain only when directly questioned about the presence of pain, but these different reporting groups did not differ on emotional variables. Those who were less accepting of their disability reported greater pain severity. Additionally, patients who perceived a significant other expressing punishing responses (e.g., expressing anger at the patients or ignoring the patients) to their pain behaviors reported more severe pain. Level of lesion, completeness of injury, surgical fusion a n d / o r instrumentation and veteran status were not associated with pain severity. Finally, pain was associated with emotional distress over and above the distress associated with the SCI itself. Overall, psychosocial factors, not physiological factors, were most closely associated with the experience of pain. Multidimensional aspects of pain are used to explain these findings and suggest that treatment should be directed at the emotional and cognitive sequelae of chronic SCI pain. Key words: Chronic pain; SCI; Psychosocial factors
Introduction
Before World War II, the life expectancy of someone suffering a spinal cord injury (SCI) was estimated in months [3]. Because of advances in medical technology, patients are able to survive past the acute stage of the injury and live close to a normal life span [21]. Despite this more favorable prognosis, there are negative side effects, complications and difficulties in living with a SCI, including chronic pain. A patient with a SCI may experience pain long after healing of the damaged bone, muscle, connective and nervous tissue has taken place. Estimates on the num-
Correspondence to." Michael A. Rapoff, Department of Rehabilitation Medicine, University of Kansas Medical Center, Kansas City, KS 66103, U.S.A.
ber of patients affected by chronic SCI pain range from 14% to as high as 94% [3,4,7]. Pain that is disabling or interferes with daily functioning affects 27-30% of patients with SCI [3,7]. Most studies of chronic SCI pain have focused strictly upon the medical causes and clinical manifestations of pain [3,4,7,8,10,18,21]. In these studies, psychosocial factors are assigned a secondary rote relative to physiological factors, and it is not specified how psychosocial factors impact SCI pain. However, a few studies have examined the impact of psychosocial factors on chronic SCI pain [6,16,19,25]. Richards et al. [19] found that greater pain severity was associated with higher age, verbal intelligence and levels of anxiety as well as a more negative psychosocial environment. Also, those patients who reported that pain interfered with activities of daily living were
older, more depressed, more intelligent, experiencing greater levels of distress and living in a more negative psychosocial environment. Greater pain severity was associated with psychosocial variables and not physiological factors such as injury level, completeness of injury or type of surgery for the injury. Umlauf et al. [26] found no differences between SCI pain patients and chronic pain patients on pain severity, description and location as assessed by the McGill Pain Questionnaire. Chronic pain and non-SCl medical problems were uniquely associated with psychosocial impairment [16]. They concluded that pain is a significant problem for patients with spinal cord injuries, and pain is associated with greater psychosocial impairment in this population over and above its association with other medical factors. Also, contrary to other studies [3,4], they found a slightly greater incidence of pain associated with quadriplegia than with paraplegia. In another study, Cohen et al. [6] found that patients with complete SCIs reported significantly less severe pain than did pain clinic patients but did not differ from patients with incomplete lesions. Patients with complete SCIs and pain clinic patients showed a significantly more disturbed MMPI profile than did patients with incomplete SCIs. It was hypothesized that those patients with complete lesions view themselves as more functionally limited than patients with incomplete lesions, and completeness of the SCI may be more important in determining psychosocial adjustment than pain per se. Because a patient with a SCI does not view pain as a threat (as determined by adjectives checked on the McGill Pain Questionnaire) and the pain clinic patient does, Cohen and colleagues viewed pain as an annoyance and an insignificant additional problem to the numerous problems and limitalions caused by the SCI. Overall, the present study attempted to clarify the relative importance of psychosocial versus physiological factors in SCI pain. Specifically, we attempted to determine if: (a) emotional and cognitive factors, significant others" responses to pain and acceptance of disability were associated with pain severity; (b) patients differ in pain severity and emotional adjustment according to how the presence of pain was elicited; (c) physiological factors (level of lesion, completeness of SCI and fusion/instrumentation) were associated with pain severity; and (d) SCI pain is related to adjustment problems over and above the adjustment problems related to the SCI itself. In congruence with the general chronic pain literature [24] and the above cited studies showing affective distress to be associated with chronic SCI pain [16,19,25], it was predicted that general affective distress and negative cognitions about pain would be associated with more severe pain [16,19,24]. In a similar manner, less adaptation or acceptance
of the disability was expected to be associated with greater pain severity but not level or completeness of injury [28]. Acceptance of a disability is thc way that a patient with a physical disability adapts to the disability by making a series of value changes concerning the importance of physique relativc to other values, l:or example, a person with a SCI would come to appreciate personal qualities other than physique and begin to subordinate physique to these other wdues. The disability would be kept from spreading to other aspects of the self. Also, comparative status values would bc transformed in asset values whereby comparison with others would not be made on the basis of physical abilities that were lost but rather an emphasis would be made upon the person's own assets and abilities. Patients who voluntarily stated to the physician that they had pain were predicted to report more severe pain and greater psychosocial adjustment problems than those who reported pain only when directly questioned about the presence of pain. Consistent with the literature on chronic pain, solicitous responses of significant others as perceived by the patient (e.g., significant other gets pain medication for the patient) were expected to be associated with greater pain severity [1 I]. It was predicted that lower levels of lesion would be associated with greater pain severity [3,4,16]. Contrary to Guttman [13] and Burke [4], fusion/instrumentation was hypothesized to be associated with less intense pain because stabilization of the vertabral column would result in less pain around the site of lesion. Because of the discrepancies concerning the impact of pain on the life of someone with a SCI [6,16,19], the final question was whether or not pain associated with SC! is related to adjustment problems over and above the adjustment problems related to the SCI itself. It was predicted that interference of SC1 pain would be associated with adjustment problems over and above the adjustment problems related to the SCI itself.
Method Patients Seventy-two consecutive patients with spinal cord injuries who were attending a SCI clinic at a VA medical center or a university medical center for routine treatment of their SCI were approached to participate in this study. Only those patients who were at least one year post injury were recruited to ihsure that healing of the SCI had taken place. Seventy-one patients agreed to participate, and 54 (76%) completed questionnaire packets. Of the final sample of 54 patients, 34 (63%) were veterans and 20 (37%,) were non-veterans. Eight of the patients (15%) were female; only 1 (2%) of the patients was a female veteran. The
185 patients had a mean age of 40.6 (S.D. = 11.8) years and had been injured for an average of 12.0 (S.D. = 8.0) years. Forty-four (78%) of the patients had been experiencing pain associated with their injury for an average of 8.2 (S.D. = 6.2) years. Twenty-five (46%) of the patients were married, 13 (24%) of the patients were single, and 16 (30%) were separated or divorced. Nineteen patients (35%) had quadriplegia (injury at C8 or above), and 35 patients (65%) had paraplegia (injury at T1 or below).
Procedures Patients either completed the questionnaire packet while waiting to see a physician at the clinic, or they completed the packet at home and mailed it back to the investigator. Patients were independently interviewed by both a research assistant and a physiatrist in the clinic. The interviews differed in the way that the presence of pain was elicited. The research assistant asked the patient " D o you experience any pain associated with your injury?" and the physician asked the patient "Are you having any problems?" with no reference to the presence or absence of pain. The attending physiatrist examined the patient and completed a checklist that detailed the length of injury, surgical fusion of the vertebrae, level of injury and completeness of injury. If the patient complained of pain, the physiatrist recorded the length of pain.
Measures Patients completed the following self-report questionnaires and inventories. Beck Depression InL'entory (BDI). Twenty-one selfevaluative statements were weighted from 0 to 3 to indicate the severity of depression. A total score was attained by summing the individual items of the inventory. Adequate validity and reliability have been established for this inventory [1]. Profile of Mood States (POMS). This 65-item, 5-point adjective rating scale provides an index of 6 affective states by summing the items associated with a specific index. It was used in this study to provide a measure of anger/hostility and vigor. Vigor was chosen because it is considered a measure of positive affect whereas only negative emotions have been studied in relation to pain in the past. Adequate validity and reliability have been established for this inventory [17]. State-Trait Anxiety Im'entory (STAD. This 40-item inventory provided an index of state anxiety (STAI-S) as well as a comparable measure of trait anxiety (STAI-T) by summing the items associated with each subscale. Trait anxiety was used in this study to measure the general anxiety level of patients. Adequate validity and reliability have been established for this inventory [20].
Acceptance of Disabifity Scale (ADS). This scale was developed from the theory of Dembo, Leviton and Wright and expanded by Wright [28] that explains that acceptance of disability as a series of value changes [15]. Patients expressed agreement or disagreement with 50 items (e.g., " T h e r e are many things a person with my disability can do.") on a 6-point Likert-type scale to provide an index of acceptance of a disability. Scores were obtained by summing the responses. Pain Experience Scale (PES). This 19-item, 7-point Likert-type scale measures negative cognitive appraisal or self-statements associated with chronic pain. Two subscales, emotionality and worry, were found through factor analysis. Scores were obtained by taking the mean of the items on each of the 2 subscales. Both subscales have acceptable test-retest reliability [22,23]. Multidimensional Pain lm,entory (MPI). This 3-section inventory assesses several factors associated with chronic pain. Responses were made on a 7-point Likert-type scale, and scores were obtained by taking the mean of the items on each of the factors. The first section assesses the interference associated with pain, perceived support from others, pain severity, self-control, and negative mood. The second section examines patients' perception of punishing, solicitous, and distracting responses from a significant other when the patient complains about pain. The third section was not used in this study. The validity and reliability of this inventory are well established [14]. Spinal Cord Injury Interference Scale (SCIIS). This scale was developed for the present study by modifying the first section of the MPI. With this sample of patients, the SCI presents a confound when measuring the interference aspects of chronic pain. This scale was developed to provide an index of the interference, support, self-control and negative mood specifically associated with a SCI. The scale was devised by modifying those items from the MPI that were measuring pain interference, support, self-control and negative mood to measure SCI related interference, support, self-control and negative mood. Those patients who reported that they had pain associated with their SCI received all of the questionnaires. Those patients who reported that they did not have any pain received only the BD1, POMS, STAI, ADS, and SCIIS.
Results
Multivariate analysis of variance, univariate analysis of variance, multiple and simple regression, and multivariate multiple regression were used to analyze the data. Because of the exploratory nature of the study, all analyses used two-tailed tests of significance. Because there were no significant gender or veteran sta-
180 "1,.\BLIZ I I N T I . R ( "OR R [ . i L A T I O N S
I:'am ,',everity ( M P I ) Wont (PES) Anger (POMS) Vigor (POMS) "1 trot anxiety ( S T A I ) l)epression ( BDI )
OF MPI PAIN SEVERITY
AND EM()'I'IONS
Pain severil_,,
Worry
Anger
Vigor
Trait /\nxicty
l)cprcsmon
MPI
PES
P()MS
P()MS)
STAI
BI)I
1.00 0.743 * :~ 11.572 :' ~: + (.3z; "7 :~:
1.00 ().5.12 + ).q4~ ; +:'
1.0110 -- 0.,1_0 ~ +
I).301 0.278
(1.461 :+ ' 1).513 ~ '
0.585 :)...... ().4b) ....
1.0() () 794 ......... .7q2 ' * '
l.l)O 11.815 + i,,
1+~11)
+ /' - ( k 0 5 : ** P < 0.01: * * * P --" 0.001.
tus differences on any of the demographic or dependent measures, subjects were combined [br further analyses. A stepwise multiple regression was run with MPI Pain Severity as the criterion variable, and scores from the POMS/Anger-Hostility, P O M S / V i g o r , STAI-T and BDI as the predictor variables. Intcrcorrelations between the criterion and predictor variables are shown in "Fable I. Only P O M S / A n g c r - H o s t i l i t y and P O M S / V i g o r were significantly correlated with MPI Pain Severity (r = (i.572, beta - 0.52 and 1...... (I.337, beta ..... 0.17, respectively) t POMS/Anger-Hostility accounted for 32.77~ of the variance in MPI Pain Severity ( F (4, 3 4 ) = 18.03. P < 0.001). No other emotional variable accounted for significant additional variance beyond the variance a c c o u n t e d for by POMS/Anger-Hostility. PES Worry was entered into a regression equation after POMS/Anger-Hostility. PES Worry accounted for 26.49Cf of the variance in MPI Pare Severity after the variance accounted for by POMS/Anger-Hostility was removed ( F (2, 36)-23.40, 1'
183
PAIN 01869
Psychosocial factors in chronic spinal cord injury pain Jay D. S u m m e r s ~, Michael A. R a p o f f h, G e o r g e V a r g h e s e b, K e n t P o r t e r c and R i c h a r d E. P a l m e r ~' " Department of Psychology, University of Kansas, Lawrence, KS 66045 (U.S.A.), and i, Department ~f Rehabilitation Medicine, Unit'ersity of Kansas Medical Center, Kansas City, KS 66103 (U.S.A.), and ' Kansas City VA Medical Center, Kansas City, MO 64128 (U.S.A.) (Received 1 October 1990, revision received 20 February 1991, accepted 28 February 1991)
Summary Chronic pain is a problem among patients with spinal cord injuries, but the psychosocial factors associated with spinal cord injury (SCI) pain are not well understood. To understand SCI pain further, 54 patients (19 with quadriplegia and 35 with paraplegia) completed the Beck Depression Inventory, State-Trait Anxiety Inventory, Profile of Mood States, Acceptance of Disability Scale and SCI Interference Scale. Forty-two patients stated they had SCI pain and completed the Multidimensional Pain Inventory and the Pain Experience Scale. Results revealed that anger and negative cognitions were associated with greater pain severity. Patients who reported pain in response to a general prompt experienced more severe pain than patients who reported pain only when directly questioned about the presence of pain, but these different reporting groups did not differ on emotional variables. Those who were less accepting of their disability reported greater pain severity. Additionally, patients who perceived a significant other expressing punishing responses (e.g., expressing anger at the patients or ignoring the patients) to their pain behaviors reported more severe pain. Level of lesion, completeness of injury, surgical fusion a n d / o r instrumentation and veteran status were not associated with pain severity. Finally, pain was associated with emotional distress over and above the distress associated with the SCI itself. Overall, psychosocial factors, not physiological factors, were most closely associated with the experience of pain. Multidimensional aspects of pain are used to explain these findings and suggest that treatment should be directed at the emotional and cognitive sequelae of chronic SCI pain. Key words: Chronic pain; SCI; Psychosocial factors
Introduction
Before World War II, the life expectancy of someone suffering a spinal cord injury (SCI) was estimated in months [3]. Because of advances in medical technology, patients are able to survive past the acute stage of the injury and live close to a normal life span [21]. Despite this more favorable prognosis, there are negative side effects, complications and difficulties in living with a SCI, including chronic pain. A patient with a SCI may experience pain long after healing of the damaged bone, muscle, connective and nervous tissue has taken place. Estimates on the num-
Correspondence to." Michael A. Rapoff, Department of Rehabilitation Medicine, University of Kansas Medical Center, Kansas City, KS 66103, U.S.A.
ber of patients affected by chronic SCI pain range from 14% to as high as 94% [3,4,7]. Pain that is disabling or interferes with daily functioning affects 27-30% of patients with SCI [3,7]. Most studies of chronic SCI pain have focused strictly upon the medical causes and clinical manifestations of pain [3,4,7,8,10,18,21]. In these studies, psychosocial factors are assigned a secondary rote relative to physiological factors, and it is not specified how psychosocial factors impact SCI pain. However, a few studies have examined the impact of psychosocial factors on chronic SCI pain [6,16,19,25]. Richards et al. [19] found that greater pain severity was associated with higher age, verbal intelligence and levels of anxiety as well as a more negative psychosocial environment. Also, those patients who reported that pain interfered with activities of daily living were
older, more depressed, more intelligent, experiencing greater levels of distress and living in a more negative psychosocial environment. Greater pain severity was associated with psychosocial variables and not physiological factors such as injury level, completeness of injury or type of surgery for the injury. Umlauf et al. [26] found no differences between SCI pain patients and chronic pain patients on pain severity, description and location as assessed by the McGill Pain Questionnaire. Chronic pain and non-SCl medical problems were uniquely associated with psychosocial impairment [16]. They concluded that pain is a significant problem for patients with spinal cord injuries, and pain is associated with greater psychosocial impairment in this population over and above its association with other medical factors. Also, contrary to other studies [3,4], they found a slightly greater incidence of pain associated with quadriplegia than with paraplegia. In another study, Cohen et al. [6] found that patients with complete SCIs reported significantly less severe pain than did pain clinic patients but did not differ from patients with incomplete lesions. Patients with complete SCIs and pain clinic patients showed a significantly more disturbed MMPI profile than did patients with incomplete SCIs. It was hypothesized that those patients with complete lesions view themselves as more functionally limited than patients with incomplete lesions, and completeness of the SCI may be more important in determining psychosocial adjustment than pain per se. Because a patient with a SCI does not view pain as a threat (as determined by adjectives checked on the McGill Pain Questionnaire) and the pain clinic patient does, Cohen and colleagues viewed pain as an annoyance and an insignificant additional problem to the numerous problems and limitalions caused by the SCI. Overall, the present study attempted to clarify the relative importance of psychosocial versus physiological factors in SCI pain. Specifically, we attempted to determine if: (a) emotional and cognitive factors, significant others" responses to pain and acceptance of disability were associated with pain severity; (b) patients differ in pain severity and emotional adjustment according to how the presence of pain was elicited; (c) physiological factors (level of lesion, completeness of SCI and fusion/instrumentation) were associated with pain severity; and (d) SCI pain is related to adjustment problems over and above the adjustment problems related to the SCI itself. In congruence with the general chronic pain literature [24] and the above cited studies showing affective distress to be associated with chronic SCI pain [16,19,25], it was predicted that general affective distress and negative cognitions about pain would be associated with more severe pain [16,19,24]. In a similar manner, less adaptation or acceptance
of the disability was expected to be associated with greater pain severity but not level or completeness of injury [28]. Acceptance of a disability is thc way that a patient with a physical disability adapts to the disability by making a series of value changes concerning the importance of physique relativc to other values, l:or example, a person with a SCI would come to appreciate personal qualities other than physique and begin to subordinate physique to these other wdues. The disability would be kept from spreading to other aspects of the self. Also, comparative status values would bc transformed in asset values whereby comparison with others would not be made on the basis of physical abilities that were lost but rather an emphasis would be made upon the person's own assets and abilities. Patients who voluntarily stated to the physician that they had pain were predicted to report more severe pain and greater psychosocial adjustment problems than those who reported pain only when directly questioned about the presence of pain. Consistent with the literature on chronic pain, solicitous responses of significant others as perceived by the patient (e.g., significant other gets pain medication for the patient) were expected to be associated with greater pain severity [1 I]. It was predicted that lower levels of lesion would be associated with greater pain severity [3,4,16]. Contrary to Guttman [13] and Burke [4], fusion/instrumentation was hypothesized to be associated with less intense pain because stabilization of the vertabral column would result in less pain around the site of lesion. Because of the discrepancies concerning the impact of pain on the life of someone with a SCI [6,16,19], the final question was whether or not pain associated with SC! is related to adjustment problems over and above the adjustment problems related to the SCI itself. It was predicted that interference of SC1 pain would be associated with adjustment problems over and above the adjustment problems related to the SCI itself.
Method Patients Seventy-two consecutive patients with spinal cord injuries who were attending a SCI clinic at a VA medical center or a university medical center for routine treatment of their SCI were approached to participate in this study. Only those patients who were at least one year post injury were recruited to ihsure that healing of the SCI had taken place. Seventy-one patients agreed to participate, and 54 (76%) completed questionnaire packets. Of the final sample of 54 patients, 34 (63%) were veterans and 20 (37%,) were non-veterans. Eight of the patients (15%) were female; only 1 (2%) of the patients was a female veteran. The
185 patients had a mean age of 40.6 (S.D. = 11.8) years and had been injured for an average of 12.0 (S.D. = 8.0) years. Forty-four (78%) of the patients had been experiencing pain associated with their injury for an average of 8.2 (S.D. = 6.2) years. Twenty-five (46%) of the patients were married, 13 (24%) of the patients were single, and 16 (30%) were separated or divorced. Nineteen patients (35%) had quadriplegia (injury at C8 or above), and 35 patients (65%) had paraplegia (injury at T1 or below).
Procedures Patients either completed the questionnaire packet while waiting to see a physician at the clinic, or they completed the packet at home and mailed it back to the investigator. Patients were independently interviewed by both a research assistant and a physiatrist in the clinic. The interviews differed in the way that the presence of pain was elicited. The research assistant asked the patient " D o you experience any pain associated with your injury?" and the physician asked the patient "Are you having any problems?" with no reference to the presence or absence of pain. The attending physiatrist examined the patient and completed a checklist that detailed the length of injury, surgical fusion of the vertebrae, level of injury and completeness of injury. If the patient complained of pain, the physiatrist recorded the length of pain.
Measures Patients completed the following self-report questionnaires and inventories. Beck Depression InL'entory (BDI). Twenty-one selfevaluative statements were weighted from 0 to 3 to indicate the severity of depression. A total score was attained by summing the individual items of the inventory. Adequate validity and reliability have been established for this inventory [1]. Profile of Mood States (POMS). This 65-item, 5-point adjective rating scale provides an index of 6 affective states by summing the items associated with a specific index. It was used in this study to provide a measure of anger/hostility and vigor. Vigor was chosen because it is considered a measure of positive affect whereas only negative emotions have been studied in relation to pain in the past. Adequate validity and reliability have been established for this inventory [17]. State-Trait Anxiety Im'entory (STAD. This 40-item inventory provided an index of state anxiety (STAI-S) as well as a comparable measure of trait anxiety (STAI-T) by summing the items associated with each subscale. Trait anxiety was used in this study to measure the general anxiety level of patients. Adequate validity and reliability have been established for this inventory [20].
Acceptance of Disabifity Scale (ADS). This scale was developed from the theory of Dembo, Leviton and Wright and expanded by Wright [28] that explains that acceptance of disability as a series of value changes [15]. Patients expressed agreement or disagreement with 50 items (e.g., " T h e r e are many things a person with my disability can do.") on a 6-point Likert-type scale to provide an index of acceptance of a disability. Scores were obtained by summing the responses. Pain Experience Scale (PES). This 19-item, 7-point Likert-type scale measures negative cognitive appraisal or self-statements associated with chronic pain. Two subscales, emotionality and worry, were found through factor analysis. Scores were obtained by taking the mean of the items on each of the 2 subscales. Both subscales have acceptable test-retest reliability [22,23]. Multidimensional Pain lm,entory (MPI). This 3-section inventory assesses several factors associated with chronic pain. Responses were made on a 7-point Likert-type scale, and scores were obtained by taking the mean of the items on each of the factors. The first section assesses the interference associated with pain, perceived support from others, pain severity, self-control, and negative mood. The second section examines patients' perception of punishing, solicitous, and distracting responses from a significant other when the patient complains about pain. The third section was not used in this study. The validity and reliability of this inventory are well established [14]. Spinal Cord Injury Interference Scale (SCIIS). This scale was developed for the present study by modifying the first section of the MPI. With this sample of patients, the SCI presents a confound when measuring the interference aspects of chronic pain. This scale was developed to provide an index of the interference, support, self-control and negative mood specifically associated with a SCI. The scale was devised by modifying those items from the MPI that were measuring pain interference, support, self-control and negative mood to measure SCI related interference, support, self-control and negative mood. Those patients who reported that they had pain associated with their SCI received all of the questionnaires. Those patients who reported that they did not have any pain received only the BD1, POMS, STAI, ADS, and SCIIS.
Results
Multivariate analysis of variance, univariate analysis of variance, multiple and simple regression, and multivariate multiple regression were used to analyze the data. Because of the exploratory nature of the study, all analyses used two-tailed tests of significance. Because there were no significant gender or veteran sta-
180 "1,.\BLIZ I I N T I . R ( "OR R [ . i L A T I O N S
I:'am ,',everity ( M P I ) Wont (PES) Anger (POMS) Vigor (POMS) "1 trot anxiety ( S T A I ) l)epression ( BDI )
OF MPI PAIN SEVERITY
AND EM()'I'IONS
Pain severil_,,
Worry
Anger
Vigor
Trait /\nxicty
l)cprcsmon
MPI
PES
P()MS
P()MS)
STAI
BI)I
1.00 0.743 * :~ 11.572 :' ~: + (.3z; "7 :~:
1.00 ().5.12 + ).q4~ ; +:'
1.0110 -- 0.,1_0 ~ +
I).301 0.278
(1.461 :+ ' 1).513 ~ '
0.585 :)...... ().4b) ....
1.0() () 794 ......... .7q2 ' * '
l.l)O 11.815 + i,,
1+~11)
+ /' - ( k 0 5 : ** P < 0.01: * * * P --" 0.001.
tus differences on any of the demographic or dependent measures, subjects were combined [br further analyses. A stepwise multiple regression was run with MPI Pain Severity as the criterion variable, and scores from the POMS/Anger-Hostility, P O M S / V i g o r , STAI-T and BDI as the predictor variables. Intcrcorrelations between the criterion and predictor variables are shown in "Fable I. Only P O M S / A n g c r - H o s t i l i t y and P O M S / V i g o r were significantly correlated with MPI Pain Severity (r = (i.572, beta - 0.52 and 1...... (I.337, beta ..... 0.17, respectively) t POMS/Anger-Hostility accounted for 32.77~ of the variance in MPI Pain Severity ( F (4, 3 4 ) = 18.03. P < 0.001). No other emotional variable accounted for significant additional variance beyond the variance a c c o u n t e d for by POMS/Anger-Hostility. PES Worry was entered into a regression equation after POMS/Anger-Hostility. PES Worry accounted for 26.49Cf of the variance in MPI Pare Severity after the variance accounted for by POMS/Anger-Hostility was removed ( F (2, 36)-23.40, 1'
