- Email: [email protected]
Psychological Correlates of Pain Behavior in Patients With Chronic Low Back Pain
Psychological Correlates of Pain Behavior in Patients With Chronic Low Back Pain CHRIS DICKENS, PH.D. MALCOLM JAYSON, M.D. FRANCIS CREED, M.D.
Pain behaviors that are excessive for the degree of known physical disease are common in patients with chronic low back pain and are frequently assumed to arise from a comorbid depressive illness. Although some studies have confirmed an association between depression and excessive pain behavior, methodologic problems (such as the use of depression ratings that also recorded symptoms attributable to physical disease) make interpretation of this finding difficult. We recruited 54 consecutive patients with chronic (⬎6 months) low back pain from a hospital clinic. Subjects completed self-rated assessments of anxiety and depression (Hospital Anxiety and Depression Scale) designed to be minimally affected by physical symptoms, along with assessments of disability (ODQ), pain (visual analogue scale), pain behavior (Waddell checklist), and physical impairment. Seventeen subjects (31%) exhibited excessive pain behavior. Overall, they were no more depressed or anxious than the remainder, although men with excessive pain behavior showed a trend toward being more depressed. Patients with excessive pain behavior were more disabled (self-rated and observer-rated), reported greater pain, and were more likely to be female and to have pain of shorter duration. Pain behavior did not correlate with anxiety or depression, but correlated with measures of disability and pain intensity. Factor analysis revealed that physical disability, pain intensity, and pain behavior loaded heavily on the first factor. Anxiety and depression loaded together on a separate factor. We conclude that pain behaviors were not related to anxiety or depression in our group, although gender differences between groups could have contributed to our negative findings. Pain behaviors may influence other physical measures. Further studies are required to investigate the relation between depression and pain behavior while controlling for gender differences. (Psychosomatics 2002; 43:42–48)
T
he association between depression and medically unexplained pain has been investigated extensively. Depression has been shown to be positively associated with
Received March 30, 2001; revised September 18, 2001; accepted September 24, 2001. From the Department of Psychiatry, Manchester University, Manchester, United Kingdom. Address correspondence and reprint requests to Dr. Dickens, Department of Psychiatry, Rawnsley Building, Manchester Royal Infirmary, Oxford Rd, Manchester M13 9WL, United Kingdom. E-mail: [email protected] Copyright 䉷 2002 The Academy of Psychosomatic Medicine.
42
somatization and somatoform disorders, in which medically unexplained pain may arise.1–4 Numerous hypotheses have arisen to explain the mechanisms by which depression might play a role in the etiology of otherwise unexplained pain.1,5 The association between depression and pain behaviors, however, has received considerably less research interest. Pain behaviors were first described by Fordyce6 as denoting the verbal and nonverbal behaviors exhibited by pain sufferers that serve to communicate the fact that they are experiencing pain. Pain behaviors may, at times, appear Psychosomatics 43:1, January-February 2002
Dickens et al. inappropriate (i.e., excessive for the degree of detectable physical abnormality). Excessive pain behaviors are common in patients with chronic low back pain, being reported in 12% of patients with previously untreated back pain, in 33% of tertiary referrals to specialist back-pain centers, and in up to 50% of subjects undergoing work disability assessments for compensation.7 Such excessive pain behavior in patients with chronic low back pain indicates that an assessment of psychological (and social) factors is necessary and that conventional treatments such as physical therapies and surgery may have inadequate effectiveness.8,9 Yet, despite the recognition that excessive pain behaviors are associated with psychological problems, the exact relation between pain behaviors and psychological factors is unclear. A number of studies have indicated that depression is greater in patients with low back pain who display excessive pain behavior compared with those without such behavior.7,10–12 However, the magnitude of this difference between groups has been small. Furthermore, the use of self-rated depression scales, validated for use in subjects with psychiatric illness (e.g., the Minnesota Multiphasic Personality Inventory,7,12 the Beck Depression Inventory,10,12 and the Zung Depression Scale11), has been shown to give spuriously exaggerated scores when used in subjects with physical complaints and is likely to have increased the magnitude of this association, making the results difficult to interpret.13,14 In this study, we aimed to elucidate the relation between depression and excessive pain behavior in a consecutive series of patients with chronic low back pain. We used robust measures designed for use in populations with physical complaints. The hypothesis stated that subjects exhibiting excessive pain behavior would be more depressed than those who did not exhibit excessive levels of pain behavior. METHODS This study was performed in a teaching hospital in Salford, United Kingdom. Consecutive subjects attending the rheumatology clinic because of low back pain of greater than 6 months’ duration were considered suitable for the study. The following exclusion criteria were applied: concurrent severe physical illness (e.g., severe cardiovascular disease, progressive neurodegenerative disorders, malignancy), systemic inflammatory disorders, and alcohol dependency or abuse. Assessments were completed while subjects were waiting to see their physician. Pain behavior was measured with the checklist of nonPsychosomatics 43:1, January-February 2002
organic signs devised by Waddell et al.7 Waddell et al. designed this checklist to be independent of subjective assessment of the appropriateness of any behavior observed that might be vulnerable to observer bias. The checklist of nonorganic signs consists of the following items: 1) superficial or nonanatomic tenderness, 2) complaints of pain on axial loading of the spine or simulated rotation of the spine, 3) limited straight-leg raising that is significantly increased on distraction, 4) regional sensory loss or weakness that cannot be accounted for by recognized pathophysiologic processes, and 5) overreaction to the clinical examination. These items were selected because they were shown to have a high degree of interrater reliability (⬎80%). Each item was marked as either present or absent. Waddell found that scores from patient populations with chronic low back pain were biphasic in distribution, with subjects tending to score either zero or one sign or three or more signs. When compared with the opinion of experienced clinicians, the presence of three or more of these signs provided the best cutoff for identifying subjects with significant degrees of abnormal illness behavior. In keeping with Waddell’s findings, a score of three or more positive nonorganic signs was taken as the cutoff for indicating excessive pain behavior.7 Pain behavior was also rated using Waddell’s checklist of nonorganic symptoms.15 Subjects were asked whether they had ever experienced any symptoms from a list of seven. Unlike the checklist of nonorganic signs, the scores obtained from this questionnaire have never been validated against clinician assessment. Symptoms of anxiety and depression were measured using the Hospital Anxiety and Depression Scale (HAD).16 This self-report questionnaire rates the severity of seven symptoms of depression and seven symptoms of anxiety over the previous week. The HAD was designed for use in subjects with comorbid physical illnesses because it minimizes the recording of symptoms of the physical illness as indicative of psychological illness. Each symptom is scored from 0 to 3. Subjects rating more than 10 on the Anxiety or Depression scale were considered to be probable cases of psychiatric disorder.16 This method of assessing depression is quick to complete and acceptable to the majority of patients, and it enabled us to consider depression as both a diagnostic category and a continuous variable. We recognized that a standardized research interview would have provided a more valid and reliable method of diagnosing anxiety and depressive disorder; however, we considered that such a procedure would have 43
Psychological Correlates of Pain Behavior greatly lengthened the assessment and reduced its acceptability for patients. Disability was assessed using the Oswestry Disability Questionnaire (ODQ).17 This self-report questionnaire rates the subject’s ability on 10 tasks of daily living. Scores obtained are percentages, with 0% representing no disability and 100% representing total disability. Objective physical impairment was measured using the Physical Impairment Scale. This observer rating assesses the range of movement on seven objective tests of functional ability. Items on the scale were chosen to be as independent of the effects of abnormal illness behavior as possible.18 On each test, means for normal populations were recorded, and the cutoffs indicating an abnormal result on each test were taken as 1 standard deviation below this norm. Each item was scored as being normal (score of 0) or abnormal (score of 1), giving a total score of 0–7, with higher scores indicating increasing levels of physical disability. Pain severity (10-cm visual analogue scale), duration of pain (number of months), number of previous operations, and involvement of the subject in any legal dispute connected with their back pain were recorded. Subjects with a score of three or more nonorganic signs (excessive pain behavior) were compared with the remainder of the subjects recruited (i.e., those with fewer than three nonorganic signs). Comparison of continuous variables was performed using the Mann-Whitney U test. Categorical data were compared using the v2 test. Spearman correlation coefficients were calculated for measured variables. Factor analysis was performed on the data to examine the nature of the association between variables measured. This was achieved by performing principal component analysis on the data followed by varimax rotation to obtain the optimal factor solution. All data analyses were performed using the Statistics Package for Social Sciences (version 7.5.1). RESULTS Of the 55 consecutive outpatients who were suitable for the study, only 1 subject refused to complete the assessments. Complete data sets were collected on 54 subjects. Of the 54 study subjects, 51 were consulting the rheumatologists primarily for their low back pain; 3 subjects were consulting the rheumatologists primarily for pain or stiffness in other body regions but were noted to have chronic low back pain also. The median age of the subjects was 46 years (inter44
quartile range, 25th–75th percentile [IQR] 38–53.5 years). Twenty-six (48%) of the study patients were male. The median HAD Depression rating was 9 (IQR 6– 12), and the mean HAD Anxiety score was 11 (IQR 8–15). Scores of greater than 10 on either the Anxiety or Depression subscale of the HAD, indicating subjects with probable psychiatric disorder, were recorded by 17 subjects with probable depression (31%) and 29 subjects with probable anxiety (54%). Seventeen subjects (31%) showed three or more signs on Waddell’s checklist of nonorganic signs. These 17 subjects with excessive pain behavior were compared with the rest of the study population. There were no statistically significant differences between patients with and without excessive pain behavior with regard to age, although there was a nonsignificant trend for a greater proportion of women to exhibit excessive pain behavior (Table 1). The patients with excessive pain behavior reported significantly greater disability, pain intensity, and physical impairment with a shorter history of pain. There were no significant differences between the groups with regard to the severity of depressive symptoms or anxiety symptoms or the number of possible cases of depression or anxiety (Table 1). When depression scores for those with and without excessive pain behavior were repeated for men and women separately, men with excessive pain behavior were found to be significantly more anxious (medians 14 and 9, respectively, P ⳱ 0.04) and showed a trend toward being more depressed (medians 12 and 8, respectively, P ⳱ 0.08) than their counterparts without excessive pain behavior. No such trends were seen in women (P ⳱ 0.31 and P ⳱ 0.69, respectively), indicating gender differences in the association of anxiety and depression with excessive pain behavior. To examine the possible influence of pain duration on the relations of anxiety and depression with excessive pain behavior, we compared the magnitude of the correlation of anxiety and depression with the number of nonorganic signs with and without controlling for pain duration. On controlling for pain duration, we observed small increases in the correlation of depression with the number of nonorganic signs (0.17 to 0.19) and similarly for the correlation between anxiety and nonorganic signs (0.19 to 0.21), indicating that the duration of pain had a small effect on the relations of anxiety and depression with excessive pain behavior. Our results indicate that excessive pain behavior is Psychosomatics 43:1, January-February 2002
Dickens et al. highly correlated with measures of physical impairment, pain severity, and disability. Spearman correlation coefficients confirmed this (Table 2). There was no significant correlation between anxiety or depression scores and the number of nonorganic symptoms or signs or other indicators of pain, and only some weak correlations of depression with disability and physical impairment. There were, however, strong correlations between nonorganic signs, disability, physical impairment, and pain severity. To further explore the relations between these highly correlated variables, we performed factor analysis. A fourfactor solution was obtained accounting for 81% of the variance observed in the measured variables (Table 3). The largest factor (factor 1) accounted for 38% of the variance in the data. The nonorganic sign (pain behavior) score loaded most heavily on this factor. Other measures of pain behavior, pain intensity, and disability also loaded on factor 1. Except for the pain behavior symptoms, the variables loading on factor 1 failed to contribute significantly to other factors. Factor 2 contributed a further 17% to the overall variance of the data recorded. Anxiety and depression scores alone loaded heavily on this factor and not on any of the other factors. Contributors to factors 3 and 4 can be seen in Table 3. DISCUSSION This results of this study demonstrated that patients with chronic low back pain with and without excessive pain TABLE 1.
behavior had similar levels of anxiety and depression, which does not support our main hypothesis. When men were studied separately, however, those with excessive pain behavior were more anxious and depressed than those without excessive pain behavior. No such difference was seen when women were studied alone. Anxiety and depression scores were not correlated with the number of nonorganic symptoms or signs. In addition, on factor analysis, nonorganic symptoms and signs loaded on factors separate from those on which anxiety and depression loaded. We studied consecutive patients with chronic low back pain attending the outpatient department of a center with a special interest in low back pain problems. It cannot be assumed that our findings can be generalized to other care settings, where the prevalence rates of anxiety, depression, and excessive pain behaviors will differ, although with the large number of patients with chronic low back pain attending such clinics, this does not detract from the interest of our findings. All measures used in this study were either designed specifically for use in the population with chronic low back pain or have been widely used on medically ill populations in previous research. The HAD is not a reliable and valid way of diagnosing cases of anxiety and depression as compared with a standardized research interview. For this reason, we examined the relation between nonorganic signs and psychological factors in two ways: the total anxiety and depression scores as well as the number of subjects above and below the recommended cutoff scores.
Subjects with and without excessive pain behavior: background and mood variables No EPB
EPB
Number Male Female Mean age (y) Pain duration (mo) Physical impairment Pain severity Mean ODQ No. in legal dispute
37 21 16 46 (38–54) 36 (12–84) 4 (3–5) 34 (28–57) 40 (28–51) 4/37 (11%)
17 5 12 45 (38–52) 14 (12–36) 7 (6–7) 70 (54.5–80.5) 62 (56–69) 2/17 (11%)
Previous operation
3/37 (8%)
3/17 (18%)
Anxiety score Depression score Cases of anxiety
11 (7–14) 9 (6–10.5) 19 (51%)
11 (9–16.5) 9 (5–13) 10 (59%)
9 (24%)
8 (47%)
Cases of depression
Significance P⳱0.08 (Fisher’s exact test) P⳱0.51 P⳱0.11 P⬍0.00005 P⳱0.0009 P⬍0.00005 P⳱0.37 (Fisher’s exact test) P⳱1.0 (Fisher’s exact test) P⳱0.29 P⳱0.45 P⳱0.77 (Fisher’s exact test) P⳱0.12 (Fisher’s exact test)
Note. EPB ⳱ excessive pain behavior. Data are presented as median values (interquartile ranges: 25th-75th percentile) or as n (%). Continuous variables are compared using the Mann-Whitney U test. Noncontinuous variables are compared using the v2 test, unless stated otherwise.
Psychosomatics 43:1, January-February 2002
45
Correlation between main physical, psychological, and behavioral factors
Sex Anxiety Depression Non-organic signs Non-organic symptoms Disability Pain Duration Pain Severity
Psychosomatics 43:1, January-February 2002
Physical Impairment
Age
Sex
Anxiety score
Depression score
Nonorganic signs
Nonorganic symptoms
Disability
Pain duration
Pain severity
ⳮ0.14 P⳱0.34 ⳮ0.1 P⳱0.5 ⳮ0.96 P⳱0.5 0.11 P⳱0.5 ⳮ0.2 P⳱0.16 ⳮ0.06 P⳱0.7 0.13 P⳱0.36 ⳮ0.1 P⳱0.4 0.07 P⳱0.65
ⳮ0.2 P⳱0.14 ⳮ0.1 P⳱0.5 ⳮ0.09 P⳱0.5 ⳮ0.3 P⳱0.03 ⳮ0.2 P⳱0.1 ⳮ0.05 P⳱0.7 ⳮ0.2 P⳱0.16 ⳮ0.05 P⳱0.75
0.66 P⬍0.0005 0.14 P⳱0.3 0.18 P⳱0.18 0.25 P⳱0.06 ⳮ0.01 P⳱0.9 0.24 P⳱0.08 0.18 P⳱0.19
0.16 P⳱0.2 0.09 P⳱0.53 0.31 P⳱0.02 0.04 P⳱0.8 0.25 P⳱0.06 0.28 P⳱0.04
0.33 P⳱0.01 0.67 P⬍0.0005 ⳮ0.15 P⳱0.28 0.49 P⬍0.0005 0.71 P⬍0.0005
0.42 P⳱0.001 0.24 P⳱0.08 0.32 P⳱0.02 0.26 P⳱0.06
ⳮ0.01 P⳱0.9 0.52 P⬍0.0005 0.71 P⬍0.0005
0.02 P⳱0.8 ⳮ0.21 P⳱0.13
0.34 P⳱0.01
Note. Number of subjects ⳱ 54 in each calculation of Spearman correlation coefficient.
Psychological Correlates of Pain Behavior
46 TABLE 2.
Dickens et al. The prevalence of probable depression found in our study (31%) is in line with that found in other studies, lying approximately midway between the prevalence found in a population sample (18%) and that found in patients attending rehabilitation centers for chronic pain (46%).19–21 Using Waddell’s criteria for identifying subjects with excessive pain behavior, we found that the prevalence of significant levels of nonorganic signs was high (31%) but similar to that in Waddell’s sample (33%) in a different specialist back pain center.7,11 These figures indicate that our population of patients with chronic low back pain was not unlike those in other studies. Our findings have indicated a lack of association of anxiety and depression with excessive pain behavior. This contradicts previous findings of a weak relation between depression and pain behavior, which may have been due to problems with the method (e.g., inflated depression scores because of depression assessments that mistakenly rate physical symptoms).7,10–12 From the opposite perspective, our negative findings may have resulted from a lack of specificity of the HAD, which may have been improved by the use of a standardized research interview. However, the lack of physical items on the HAD is likely to make it more, not less, specific than other questionnaire assessments when used in this patient group. In addition, the fact that this scale detected depressed subjects in numbers comparable to those of other studies and has been shown to be specific in other groups indicates that this is unlikely to account for our negative findings.22 We did find gender differences in the association of anxiety and depression with excessive pain behaviors. Depression (trend only) and anxiety were greater in men with excessive pain behavior compared with those without such behavior. No such trends were apparent in women. These
TABLE 3.
gender differences have not been found in other studies7,12 and may have resulted from selection bias in our small groups. Because there tended to be more women in the excessive pain behavior group in our study, gender differences between groups may have contributed to the lack of association of anxiety and depression with excessive pain behavior. Despite the clinically significant difference in pain duration between those with and without excessive pain behavior, there was no apparent effect of the differences in pain duration on the relations of anxiety and depression with pain behaviors. An alternative explanation is that our negative finding may be the result of a type II error due to small group sizes. Without doubt, the recruitment of greater numbers of subjects would have increased the robustness of the statistical analyses in this study. However, the lack of any correlation (Table 2) and our finding that anxiety and depression loaded on a completely different factor from the measures of pain behavior suggest that a larger sample would have yielded the same results. The high degree of correlation between measures of physical status and measures of pain behavior were not expected. It is probable that this association is spurious because of items on the measures of physical state being rated positive because of the presence of pain behaviors. This association may, however, reflect a true association between observed pain behaviors that appear excessive and underlying physical pathology. Peripheral and central neurophysiologic sensitization, occurring as a result of peripheral tissue damage, may result in complaints of pain that are apparently excessive for the degree of physical stimulation. Reorganization of the dorsal root ganglion as a result of peripheral nerve injury may lead to persistence of the pain experience beyond gross healing of the peripheral
Factor analysis of measured variables
Cumulative variance accounted for Nonorganic signs Objective physical impairment Self-rated disability Pain severity (VAS) HAD Depression score HAD Anxiety score Pain duration (mo) Nonorganic symptoms Age
Factor 1
Factor 2
Factor 3
Factor 4
38% 0.88 0.88 0.85 0.61
55%
69%
81%
0.90 0.89 0.46
0.89 0.72 0.95
Note. VAS ⳱ visual analogue scale; HAD ⳱ Hospital Anxiety and Depression Scale. Rotated factor matrix: varimax rotation. Factor loadings of less than 0.40 have been excluded for clarity.
Psychosomatics 43:1, January-February 2002
47
Psychological Correlates of Pain Behavior nerve.23 Because of the lack of obvious cause for these pain behaviors, they may be interpreted as being excessive for the degree of physical damage. The clinical implications of our findings are that excessive pain behavior does not indicate underlying anxiety or depression, although further investigation of the effects of gender on pain behavior is required. Furthermore, our study indicates that care is needed in interpreting findings on any clinical measure of disease status because findings
are likely to be affected by the subjects’ pain behavior. Further research into the relation between pain behavior and mood is required, using structured interviews to identify psychiatric cases, larger numbers of patients, and measures of physical dysfunction that are independent of the effects of pain behavior. It may be necessary to accept, however, that “objective” measures of physical dysfunction are always influenced by pain behavior, the etiology of which remains obscure.
References
1. von Knorring L: Idiopathic pain and depression. Qual Life Res 1994; 3:S57–S68 2. Lipowski ZJ: Somatization and depression. Psychosomatics 1990; 31:13–21 3. Maier W, Falkai P: The epidemiology of comorbidity between depression, anxiety disorders and somatic diseases. Int Clin Psychopharmacol 1999; 14:S1–S6 4. Jorgensen CK, Fink F, Olesen F: Psychological distress among patients with musculoskeletal illness in general practice. Psychosomatics 2000; 41:321–329 5. Gamsa A: The role of psychological factors in chronic pain. I. A half century of study. Pain 1994; 57:5–15 6. Fordyce WE: Behavioral Methods for Chronic Pain Assessment. St Louis, MO, CV Mosby Co, 1976 7. Waddell G, McCulloch JA, Kummel E, et al: Nonorganic physical signs in low-back pain. Spine 1980; 5:117–125 8. Waddell G: Biopsychosocial analysis of low back pain. Baillieres Clin Rheumatol 1992; 6:523–557 9. Polantin PB, Cox B, Gatchel RJ: A prospective study of Waddell signs in patients with chronic low back pain. Spine 1997; 22:1618–1621 10. Keefe FJ, Wilkins RH, Cook WA Jr, et al: Depression, pain, and pain behavior. J Consult Clin Psychol 1986; 54:665–669 11. Waddell G, Pilowsky I, Bond MR: Clinical assessment and interpretation of abnormal illness behaviour in low back pain. Pain 1989; 39:41–53 12. Novy DM, Collins HS, Nelson DV, et al: Waddell signs: distributional properties and correlates. Arch Phys Med Rehabil 1998; 79:820–822
48
13. Pincus T, Callahan LF, Bradley LA, et al: Elevated MMPI scores for hypochondriasis, depression and hysteria in patients with rheumatoid arthritis reflect disease rather than psychological status. Arthritis Rheum 1986; 29:1456–1465 14. Pincus T, Williams A: Models and measurement of depression in chronic pain. J Psychosom Res 1999; 47:211–219 15. Waddell G, Main CJ, Morris EW, et al: Chronic low-back pain, psychologic distress, and illness behavior. Spine 1984; 9:209–213 16. Zigmond AS, Snaith RP: The hospital and depression scale. Acta Psychiatr Scand 1983; 67:361–370 17. Fairbank JC, Couper J, Davies JB, et al: The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 1980; 66:271–273 18. Waddell G, Somerville D, Henderson I, et al: Objective clinical evaluation of physical impairment in chronic low back pain. Spine 1992; 17:617–628 19. Kinney R, Gatchel R, Polantin P, et al: Prevalence of psychopathology in acute and chronic low back pain patients. J Occup Rehabil 1993; 3:95–103 20. Kramlinger KG, Swanson DW, Maruta T: Are patients with chronic pain depressed? Am J Psychiatry 1983; 140:747–749 21. Magni G: Chronic low-back pain and depression: an epidemiological survey. Acta Psychiatr Scand 1984; 70:614–617 22. Herrmann C: International experience with the Hospital Anxiety and Depression Scale—a review of validation and clinical results. J Psychosom Res 1997; 42:17–41 23. Dubner R, Basbaum AI: Spinal dorsal horn plasticity following tissue or nerve injury, in Textbook of Pain. Edited by Wall PD, Melzack R. Edinburgh, Churchill Livingstone, 1994, pp 225–242
Psychosomatics 43:1, January-February 2002
Pain behaviors that are excessive for the degree of known physical disease are common in patients with chronic low back pain and are frequently assumed to arise from a comorbid depressive illness. Although some studies have confirmed an association between depression and excessive pain behavior, methodologic problems (such as the use of depression ratings that also recorded symptoms attributable to physical disease) make interpretation of this finding difficult. We recruited 54 consecutive patients with chronic (⬎6 months) low back pain from a hospital clinic. Subjects completed self-rated assessments of anxiety and depression (Hospital Anxiety and Depression Scale) designed to be minimally affected by physical symptoms, along with assessments of disability (ODQ), pain (visual analogue scale), pain behavior (Waddell checklist), and physical impairment. Seventeen subjects (31%) exhibited excessive pain behavior. Overall, they were no more depressed or anxious than the remainder, although men with excessive pain behavior showed a trend toward being more depressed. Patients with excessive pain behavior were more disabled (self-rated and observer-rated), reported greater pain, and were more likely to be female and to have pain of shorter duration. Pain behavior did not correlate with anxiety or depression, but correlated with measures of disability and pain intensity. Factor analysis revealed that physical disability, pain intensity, and pain behavior loaded heavily on the first factor. Anxiety and depression loaded together on a separate factor. We conclude that pain behaviors were not related to anxiety or depression in our group, although gender differences between groups could have contributed to our negative findings. Pain behaviors may influence other physical measures. Further studies are required to investigate the relation between depression and pain behavior while controlling for gender differences. (Psychosomatics 2002; 43:42–48)
T
he association between depression and medically unexplained pain has been investigated extensively. Depression has been shown to be positively associated with
Received March 30, 2001; revised September 18, 2001; accepted September 24, 2001. From the Department of Psychiatry, Manchester University, Manchester, United Kingdom. Address correspondence and reprint requests to Dr. Dickens, Department of Psychiatry, Rawnsley Building, Manchester Royal Infirmary, Oxford Rd, Manchester M13 9WL, United Kingdom. E-mail: [email protected] Copyright 䉷 2002 The Academy of Psychosomatic Medicine.
42
somatization and somatoform disorders, in which medically unexplained pain may arise.1–4 Numerous hypotheses have arisen to explain the mechanisms by which depression might play a role in the etiology of otherwise unexplained pain.1,5 The association between depression and pain behaviors, however, has received considerably less research interest. Pain behaviors were first described by Fordyce6 as denoting the verbal and nonverbal behaviors exhibited by pain sufferers that serve to communicate the fact that they are experiencing pain. Pain behaviors may, at times, appear Psychosomatics 43:1, January-February 2002
Dickens et al. inappropriate (i.e., excessive for the degree of detectable physical abnormality). Excessive pain behaviors are common in patients with chronic low back pain, being reported in 12% of patients with previously untreated back pain, in 33% of tertiary referrals to specialist back-pain centers, and in up to 50% of subjects undergoing work disability assessments for compensation.7 Such excessive pain behavior in patients with chronic low back pain indicates that an assessment of psychological (and social) factors is necessary and that conventional treatments such as physical therapies and surgery may have inadequate effectiveness.8,9 Yet, despite the recognition that excessive pain behaviors are associated with psychological problems, the exact relation between pain behaviors and psychological factors is unclear. A number of studies have indicated that depression is greater in patients with low back pain who display excessive pain behavior compared with those without such behavior.7,10–12 However, the magnitude of this difference between groups has been small. Furthermore, the use of self-rated depression scales, validated for use in subjects with psychiatric illness (e.g., the Minnesota Multiphasic Personality Inventory,7,12 the Beck Depression Inventory,10,12 and the Zung Depression Scale11), has been shown to give spuriously exaggerated scores when used in subjects with physical complaints and is likely to have increased the magnitude of this association, making the results difficult to interpret.13,14 In this study, we aimed to elucidate the relation between depression and excessive pain behavior in a consecutive series of patients with chronic low back pain. We used robust measures designed for use in populations with physical complaints. The hypothesis stated that subjects exhibiting excessive pain behavior would be more depressed than those who did not exhibit excessive levels of pain behavior. METHODS This study was performed in a teaching hospital in Salford, United Kingdom. Consecutive subjects attending the rheumatology clinic because of low back pain of greater than 6 months’ duration were considered suitable for the study. The following exclusion criteria were applied: concurrent severe physical illness (e.g., severe cardiovascular disease, progressive neurodegenerative disorders, malignancy), systemic inflammatory disorders, and alcohol dependency or abuse. Assessments were completed while subjects were waiting to see their physician. Pain behavior was measured with the checklist of nonPsychosomatics 43:1, January-February 2002
organic signs devised by Waddell et al.7 Waddell et al. designed this checklist to be independent of subjective assessment of the appropriateness of any behavior observed that might be vulnerable to observer bias. The checklist of nonorganic signs consists of the following items: 1) superficial or nonanatomic tenderness, 2) complaints of pain on axial loading of the spine or simulated rotation of the spine, 3) limited straight-leg raising that is significantly increased on distraction, 4) regional sensory loss or weakness that cannot be accounted for by recognized pathophysiologic processes, and 5) overreaction to the clinical examination. These items were selected because they were shown to have a high degree of interrater reliability (⬎80%). Each item was marked as either present or absent. Waddell found that scores from patient populations with chronic low back pain were biphasic in distribution, with subjects tending to score either zero or one sign or three or more signs. When compared with the opinion of experienced clinicians, the presence of three or more of these signs provided the best cutoff for identifying subjects with significant degrees of abnormal illness behavior. In keeping with Waddell’s findings, a score of three or more positive nonorganic signs was taken as the cutoff for indicating excessive pain behavior.7 Pain behavior was also rated using Waddell’s checklist of nonorganic symptoms.15 Subjects were asked whether they had ever experienced any symptoms from a list of seven. Unlike the checklist of nonorganic signs, the scores obtained from this questionnaire have never been validated against clinician assessment. Symptoms of anxiety and depression were measured using the Hospital Anxiety and Depression Scale (HAD).16 This self-report questionnaire rates the severity of seven symptoms of depression and seven symptoms of anxiety over the previous week. The HAD was designed for use in subjects with comorbid physical illnesses because it minimizes the recording of symptoms of the physical illness as indicative of psychological illness. Each symptom is scored from 0 to 3. Subjects rating more than 10 on the Anxiety or Depression scale were considered to be probable cases of psychiatric disorder.16 This method of assessing depression is quick to complete and acceptable to the majority of patients, and it enabled us to consider depression as both a diagnostic category and a continuous variable. We recognized that a standardized research interview would have provided a more valid and reliable method of diagnosing anxiety and depressive disorder; however, we considered that such a procedure would have 43
Psychological Correlates of Pain Behavior greatly lengthened the assessment and reduced its acceptability for patients. Disability was assessed using the Oswestry Disability Questionnaire (ODQ).17 This self-report questionnaire rates the subject’s ability on 10 tasks of daily living. Scores obtained are percentages, with 0% representing no disability and 100% representing total disability. Objective physical impairment was measured using the Physical Impairment Scale. This observer rating assesses the range of movement on seven objective tests of functional ability. Items on the scale were chosen to be as independent of the effects of abnormal illness behavior as possible.18 On each test, means for normal populations were recorded, and the cutoffs indicating an abnormal result on each test were taken as 1 standard deviation below this norm. Each item was scored as being normal (score of 0) or abnormal (score of 1), giving a total score of 0–7, with higher scores indicating increasing levels of physical disability. Pain severity (10-cm visual analogue scale), duration of pain (number of months), number of previous operations, and involvement of the subject in any legal dispute connected with their back pain were recorded. Subjects with a score of three or more nonorganic signs (excessive pain behavior) were compared with the remainder of the subjects recruited (i.e., those with fewer than three nonorganic signs). Comparison of continuous variables was performed using the Mann-Whitney U test. Categorical data were compared using the v2 test. Spearman correlation coefficients were calculated for measured variables. Factor analysis was performed on the data to examine the nature of the association between variables measured. This was achieved by performing principal component analysis on the data followed by varimax rotation to obtain the optimal factor solution. All data analyses were performed using the Statistics Package for Social Sciences (version 7.5.1). RESULTS Of the 55 consecutive outpatients who were suitable for the study, only 1 subject refused to complete the assessments. Complete data sets were collected on 54 subjects. Of the 54 study subjects, 51 were consulting the rheumatologists primarily for their low back pain; 3 subjects were consulting the rheumatologists primarily for pain or stiffness in other body regions but were noted to have chronic low back pain also. The median age of the subjects was 46 years (inter44
quartile range, 25th–75th percentile [IQR] 38–53.5 years). Twenty-six (48%) of the study patients were male. The median HAD Depression rating was 9 (IQR 6– 12), and the mean HAD Anxiety score was 11 (IQR 8–15). Scores of greater than 10 on either the Anxiety or Depression subscale of the HAD, indicating subjects with probable psychiatric disorder, were recorded by 17 subjects with probable depression (31%) and 29 subjects with probable anxiety (54%). Seventeen subjects (31%) showed three or more signs on Waddell’s checklist of nonorganic signs. These 17 subjects with excessive pain behavior were compared with the rest of the study population. There were no statistically significant differences between patients with and without excessive pain behavior with regard to age, although there was a nonsignificant trend for a greater proportion of women to exhibit excessive pain behavior (Table 1). The patients with excessive pain behavior reported significantly greater disability, pain intensity, and physical impairment with a shorter history of pain. There were no significant differences between the groups with regard to the severity of depressive symptoms or anxiety symptoms or the number of possible cases of depression or anxiety (Table 1). When depression scores for those with and without excessive pain behavior were repeated for men and women separately, men with excessive pain behavior were found to be significantly more anxious (medians 14 and 9, respectively, P ⳱ 0.04) and showed a trend toward being more depressed (medians 12 and 8, respectively, P ⳱ 0.08) than their counterparts without excessive pain behavior. No such trends were seen in women (P ⳱ 0.31 and P ⳱ 0.69, respectively), indicating gender differences in the association of anxiety and depression with excessive pain behavior. To examine the possible influence of pain duration on the relations of anxiety and depression with excessive pain behavior, we compared the magnitude of the correlation of anxiety and depression with the number of nonorganic signs with and without controlling for pain duration. On controlling for pain duration, we observed small increases in the correlation of depression with the number of nonorganic signs (0.17 to 0.19) and similarly for the correlation between anxiety and nonorganic signs (0.19 to 0.21), indicating that the duration of pain had a small effect on the relations of anxiety and depression with excessive pain behavior. Our results indicate that excessive pain behavior is Psychosomatics 43:1, January-February 2002
Dickens et al. highly correlated with measures of physical impairment, pain severity, and disability. Spearman correlation coefficients confirmed this (Table 2). There was no significant correlation between anxiety or depression scores and the number of nonorganic symptoms or signs or other indicators of pain, and only some weak correlations of depression with disability and physical impairment. There were, however, strong correlations between nonorganic signs, disability, physical impairment, and pain severity. To further explore the relations between these highly correlated variables, we performed factor analysis. A fourfactor solution was obtained accounting for 81% of the variance observed in the measured variables (Table 3). The largest factor (factor 1) accounted for 38% of the variance in the data. The nonorganic sign (pain behavior) score loaded most heavily on this factor. Other measures of pain behavior, pain intensity, and disability also loaded on factor 1. Except for the pain behavior symptoms, the variables loading on factor 1 failed to contribute significantly to other factors. Factor 2 contributed a further 17% to the overall variance of the data recorded. Anxiety and depression scores alone loaded heavily on this factor and not on any of the other factors. Contributors to factors 3 and 4 can be seen in Table 3. DISCUSSION This results of this study demonstrated that patients with chronic low back pain with and without excessive pain TABLE 1.
behavior had similar levels of anxiety and depression, which does not support our main hypothesis. When men were studied separately, however, those with excessive pain behavior were more anxious and depressed than those without excessive pain behavior. No such difference was seen when women were studied alone. Anxiety and depression scores were not correlated with the number of nonorganic symptoms or signs. In addition, on factor analysis, nonorganic symptoms and signs loaded on factors separate from those on which anxiety and depression loaded. We studied consecutive patients with chronic low back pain attending the outpatient department of a center with a special interest in low back pain problems. It cannot be assumed that our findings can be generalized to other care settings, where the prevalence rates of anxiety, depression, and excessive pain behaviors will differ, although with the large number of patients with chronic low back pain attending such clinics, this does not detract from the interest of our findings. All measures used in this study were either designed specifically for use in the population with chronic low back pain or have been widely used on medically ill populations in previous research. The HAD is not a reliable and valid way of diagnosing cases of anxiety and depression as compared with a standardized research interview. For this reason, we examined the relation between nonorganic signs and psychological factors in two ways: the total anxiety and depression scores as well as the number of subjects above and below the recommended cutoff scores.
Subjects with and without excessive pain behavior: background and mood variables No EPB
EPB
Number Male Female Mean age (y) Pain duration (mo) Physical impairment Pain severity Mean ODQ No. in legal dispute
37 21 16 46 (38–54) 36 (12–84) 4 (3–5) 34 (28–57) 40 (28–51) 4/37 (11%)
17 5 12 45 (38–52) 14 (12–36) 7 (6–7) 70 (54.5–80.5) 62 (56–69) 2/17 (11%)
Previous operation
3/37 (8%)
3/17 (18%)
Anxiety score Depression score Cases of anxiety
11 (7–14) 9 (6–10.5) 19 (51%)
11 (9–16.5) 9 (5–13) 10 (59%)
9 (24%)
8 (47%)
Cases of depression
Significance P⳱0.08 (Fisher’s exact test) P⳱0.51 P⳱0.11 P⬍0.00005 P⳱0.0009 P⬍0.00005 P⳱0.37 (Fisher’s exact test) P⳱1.0 (Fisher’s exact test) P⳱0.29 P⳱0.45 P⳱0.77 (Fisher’s exact test) P⳱0.12 (Fisher’s exact test)
Note. EPB ⳱ excessive pain behavior. Data are presented as median values (interquartile ranges: 25th-75th percentile) or as n (%). Continuous variables are compared using the Mann-Whitney U test. Noncontinuous variables are compared using the v2 test, unless stated otherwise.
Psychosomatics 43:1, January-February 2002
45
Correlation between main physical, psychological, and behavioral factors
Sex Anxiety Depression Non-organic signs Non-organic symptoms Disability Pain Duration Pain Severity
Psychosomatics 43:1, January-February 2002
Physical Impairment
Age
Sex
Anxiety score
Depression score
Nonorganic signs
Nonorganic symptoms
Disability
Pain duration
Pain severity
ⳮ0.14 P⳱0.34 ⳮ0.1 P⳱0.5 ⳮ0.96 P⳱0.5 0.11 P⳱0.5 ⳮ0.2 P⳱0.16 ⳮ0.06 P⳱0.7 0.13 P⳱0.36 ⳮ0.1 P⳱0.4 0.07 P⳱0.65
ⳮ0.2 P⳱0.14 ⳮ0.1 P⳱0.5 ⳮ0.09 P⳱0.5 ⳮ0.3 P⳱0.03 ⳮ0.2 P⳱0.1 ⳮ0.05 P⳱0.7 ⳮ0.2 P⳱0.16 ⳮ0.05 P⳱0.75
0.66 P⬍0.0005 0.14 P⳱0.3 0.18 P⳱0.18 0.25 P⳱0.06 ⳮ0.01 P⳱0.9 0.24 P⳱0.08 0.18 P⳱0.19
0.16 P⳱0.2 0.09 P⳱0.53 0.31 P⳱0.02 0.04 P⳱0.8 0.25 P⳱0.06 0.28 P⳱0.04
0.33 P⳱0.01 0.67 P⬍0.0005 ⳮ0.15 P⳱0.28 0.49 P⬍0.0005 0.71 P⬍0.0005
0.42 P⳱0.001 0.24 P⳱0.08 0.32 P⳱0.02 0.26 P⳱0.06
ⳮ0.01 P⳱0.9 0.52 P⬍0.0005 0.71 P⬍0.0005
0.02 P⳱0.8 ⳮ0.21 P⳱0.13
0.34 P⳱0.01
Note. Number of subjects ⳱ 54 in each calculation of Spearman correlation coefficient.
Psychological Correlates of Pain Behavior
46 TABLE 2.
Dickens et al. The prevalence of probable depression found in our study (31%) is in line with that found in other studies, lying approximately midway between the prevalence found in a population sample (18%) and that found in patients attending rehabilitation centers for chronic pain (46%).19–21 Using Waddell’s criteria for identifying subjects with excessive pain behavior, we found that the prevalence of significant levels of nonorganic signs was high (31%) but similar to that in Waddell’s sample (33%) in a different specialist back pain center.7,11 These figures indicate that our population of patients with chronic low back pain was not unlike those in other studies. Our findings have indicated a lack of association of anxiety and depression with excessive pain behavior. This contradicts previous findings of a weak relation between depression and pain behavior, which may have been due to problems with the method (e.g., inflated depression scores because of depression assessments that mistakenly rate physical symptoms).7,10–12 From the opposite perspective, our negative findings may have resulted from a lack of specificity of the HAD, which may have been improved by the use of a standardized research interview. However, the lack of physical items on the HAD is likely to make it more, not less, specific than other questionnaire assessments when used in this patient group. In addition, the fact that this scale detected depressed subjects in numbers comparable to those of other studies and has been shown to be specific in other groups indicates that this is unlikely to account for our negative findings.22 We did find gender differences in the association of anxiety and depression with excessive pain behaviors. Depression (trend only) and anxiety were greater in men with excessive pain behavior compared with those without such behavior. No such trends were apparent in women. These
TABLE 3.
gender differences have not been found in other studies7,12 and may have resulted from selection bias in our small groups. Because there tended to be more women in the excessive pain behavior group in our study, gender differences between groups may have contributed to the lack of association of anxiety and depression with excessive pain behavior. Despite the clinically significant difference in pain duration between those with and without excessive pain behavior, there was no apparent effect of the differences in pain duration on the relations of anxiety and depression with pain behaviors. An alternative explanation is that our negative finding may be the result of a type II error due to small group sizes. Without doubt, the recruitment of greater numbers of subjects would have increased the robustness of the statistical analyses in this study. However, the lack of any correlation (Table 2) and our finding that anxiety and depression loaded on a completely different factor from the measures of pain behavior suggest that a larger sample would have yielded the same results. The high degree of correlation between measures of physical status and measures of pain behavior were not expected. It is probable that this association is spurious because of items on the measures of physical state being rated positive because of the presence of pain behaviors. This association may, however, reflect a true association between observed pain behaviors that appear excessive and underlying physical pathology. Peripheral and central neurophysiologic sensitization, occurring as a result of peripheral tissue damage, may result in complaints of pain that are apparently excessive for the degree of physical stimulation. Reorganization of the dorsal root ganglion as a result of peripheral nerve injury may lead to persistence of the pain experience beyond gross healing of the peripheral
Factor analysis of measured variables
Cumulative variance accounted for Nonorganic signs Objective physical impairment Self-rated disability Pain severity (VAS) HAD Depression score HAD Anxiety score Pain duration (mo) Nonorganic symptoms Age
Factor 1
Factor 2
Factor 3
Factor 4
38% 0.88 0.88 0.85 0.61
55%
69%
81%
0.90 0.89 0.46
0.89 0.72 0.95
Note. VAS ⳱ visual analogue scale; HAD ⳱ Hospital Anxiety and Depression Scale. Rotated factor matrix: varimax rotation. Factor loadings of less than 0.40 have been excluded for clarity.
Psychosomatics 43:1, January-February 2002
47
Psychological Correlates of Pain Behavior nerve.23 Because of the lack of obvious cause for these pain behaviors, they may be interpreted as being excessive for the degree of physical damage. The clinical implications of our findings are that excessive pain behavior does not indicate underlying anxiety or depression, although further investigation of the effects of gender on pain behavior is required. Furthermore, our study indicates that care is needed in interpreting findings on any clinical measure of disease status because findings
are likely to be affected by the subjects’ pain behavior. Further research into the relation between pain behavior and mood is required, using structured interviews to identify psychiatric cases, larger numbers of patients, and measures of physical dysfunction that are independent of the effects of pain behavior. It may be necessary to accept, however, that “objective” measures of physical dysfunction are always influenced by pain behavior, the etiology of which remains obscure.
References
1. von Knorring L: Idiopathic pain and depression. Qual Life Res 1994; 3:S57–S68 2. Lipowski ZJ: Somatization and depression. Psychosomatics 1990; 31:13–21 3. Maier W, Falkai P: The epidemiology of comorbidity between depression, anxiety disorders and somatic diseases. Int Clin Psychopharmacol 1999; 14:S1–S6 4. Jorgensen CK, Fink F, Olesen F: Psychological distress among patients with musculoskeletal illness in general practice. Psychosomatics 2000; 41:321–329 5. Gamsa A: The role of psychological factors in chronic pain. I. A half century of study. Pain 1994; 57:5–15 6. Fordyce WE: Behavioral Methods for Chronic Pain Assessment. St Louis, MO, CV Mosby Co, 1976 7. Waddell G, McCulloch JA, Kummel E, et al: Nonorganic physical signs in low-back pain. Spine 1980; 5:117–125 8. Waddell G: Biopsychosocial analysis of low back pain. Baillieres Clin Rheumatol 1992; 6:523–557 9. Polantin PB, Cox B, Gatchel RJ: A prospective study of Waddell signs in patients with chronic low back pain. Spine 1997; 22:1618–1621 10. Keefe FJ, Wilkins RH, Cook WA Jr, et al: Depression, pain, and pain behavior. J Consult Clin Psychol 1986; 54:665–669 11. Waddell G, Pilowsky I, Bond MR: Clinical assessment and interpretation of abnormal illness behaviour in low back pain. Pain 1989; 39:41–53 12. Novy DM, Collins HS, Nelson DV, et al: Waddell signs: distributional properties and correlates. Arch Phys Med Rehabil 1998; 79:820–822
48
13. Pincus T, Callahan LF, Bradley LA, et al: Elevated MMPI scores for hypochondriasis, depression and hysteria in patients with rheumatoid arthritis reflect disease rather than psychological status. Arthritis Rheum 1986; 29:1456–1465 14. Pincus T, Williams A: Models and measurement of depression in chronic pain. J Psychosom Res 1999; 47:211–219 15. Waddell G, Main CJ, Morris EW, et al: Chronic low-back pain, psychologic distress, and illness behavior. Spine 1984; 9:209–213 16. Zigmond AS, Snaith RP: The hospital and depression scale. Acta Psychiatr Scand 1983; 67:361–370 17. Fairbank JC, Couper J, Davies JB, et al: The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 1980; 66:271–273 18. Waddell G, Somerville D, Henderson I, et al: Objective clinical evaluation of physical impairment in chronic low back pain. Spine 1992; 17:617–628 19. Kinney R, Gatchel R, Polantin P, et al: Prevalence of psychopathology in acute and chronic low back pain patients. J Occup Rehabil 1993; 3:95–103 20. Kramlinger KG, Swanson DW, Maruta T: Are patients with chronic pain depressed? Am J Psychiatry 1983; 140:747–749 21. Magni G: Chronic low-back pain and depression: an epidemiological survey. Acta Psychiatr Scand 1984; 70:614–617 22. Herrmann C: International experience with the Hospital Anxiety and Depression Scale—a review of validation and clinical results. J Psychosom Res 1997; 42:17–41 23. Dubner R, Basbaum AI: Spinal dorsal horn plasticity following tissue or nerve injury, in Textbook of Pain. Edited by Wall PD, Melzack R. Edinburgh, Churchill Livingstone, 1994, pp 225–242
Psychosomatics 43:1, January-February 2002