Pain, 44 (1991) 51-56 D 1991 Elsevier Science Publishers ADONIS
51 B.V. (Biomedical
Division)
0304-3959/91/$03.50
030439599100055C
PAIN 01707
Pain and impairment beliefs in chronic low back pain: validation of the Pain and Impairment Relationship Scale (PAIRS) * Mark A. Slater, Holly F. Hall, J. Hampton Atkinson and Steven R. Garfin VA Medical Center and University of California, San Diego, CA (U.S.A.) (Received
23 February
1990, revision received
14 June 1990, accepted
26 June 1990)
Few validated instruments are available to assess beliefs and attitudes that patients have regarding pain, or ability to function S-arY despite discomfort. The Pain and Impairment Relationship Scale (PAIRS) was developed to tap these important beliefs and attitudes in chronic pain patients. Preliminary data indicate that the PAIRS is internally consistent and significantly related to impairment in a highly selected pain clinic sample of patients, including some chronic low back pain patients. The present study was designed to extend the validation of the PAIRS to a more general sample of chronic benign low back pain patients. Furthermore, additional tests supported the discriminant, convergent and divergent validity, as well as the reliability and relative independence from favorable self-report response bias of the PAIRS, by respectively demonstrating that: (1) the impairment beliefs assessed with the PAIRS were more prominent in chronic low back pain (CLBP) patients than in matched non-pain, healthy controls; (2) scores on the PAIRS were significantly related to measures of physical impairment, but not to physicians ratings of disease severity; (3) the impairment beliefs assessed with the PAIRS are readily distinguishable from cognitive distortions and emotional distress; (4) PAIRS scores for chronic low back pain patients are relatively consistent over time; and (5) PAIRS scores are not significantly associated with measures of favorable self-report response bias. We conclude that the PAIRS has demonstrated at least preliminary utility for applications by researchers and clinicians interested in chronic pain. Key words:
Chronic
pain;
Assessment;
Pain beliefs
Introduction
Attitudes and beliefs about pain and impairment are frequently the targets of cognitive-behavioral approaches to the treatment of chronic pain syndromes. In chronic benign low back pain (CLBP), behavioral and cognitive-behavioral treatment techniques attempt to: (1) reduce the degree to which patients attend to their pain, (2) increase patients’ functional capacities, and (3) reduce the emotional distress and suffering related to pain and impairment [7,11,24]. Treatment typically includes restructuring cognitive schema that may perpetuate excessive pain and disability [24]. Beliefs about the meaning of pain and one’s ability to function despite discomfort are important aspects of this cognitive schema that is typically a target for intervention.
* This research was supported in part by Medical from the Department of Veterans Affairs.
Research
Correspondence to: Mark A. Slater, Ph.D., Director, ment Program (116B), San Diego VA Medical Center, Village Drive, San Diego, CA 92161, U.S.A.
Grants
Pain Manage3350 La Jolla
Unfortunately, few validated instruments are available to assess beliefs and attitudes that patients have regarding their pain, or its effect on daily function. Thus, Riley et al. [19] developed the Pain and Impairment Relationship Scale (PAIRS) to identify these important beliefs and attitudes in chronic pain populations. Preliminary data indicate that the PAIRS is internally consistent and significantly related to functional impairment in a highly selected pain clinic sample of heterogeneous chronic pain patients [19]. The study presented here was designed to extend the previous validation of the PAIRS to a more general sample of chronic benign low back pain patients. Additionally, several important new tests of the validity of the PAIRS were conducted: (1) discriminant validity was assessed by comparing chronic low back pain (CLBP) patients to healthy non-pain controls on the PAIRS; (2) convergent validity was determined by evaluating the relationships of the PAIRS with standardized measures of physical impairment and disease severity; (3) divergent validity was evaluated by determining the relationship of the PAIRS to measures of cognitive distortion and emotional distress; (4) reliubility over time was assessed by comparing PAIRS scores in a test-retest sample; and
finally, (5) vulnerability to response bias was evaluated by assessing the relationship between the PAIRS and a measure of favorable self-report response bias.
Methods Subjects Subjects were 31 male chronic benign low back pain patients, recruited from a general orthopedic clinic, and 19 healthy non-pain male volunteer controls. Inclusion criteria for CLBP patients were: (1) 21-64-year-old male outpatients attending a general orthopedic clinic, (2) benign low back pain (T6 or below) present “on a daily basis” for the previous 6 months or longer, (3) no major co-existing medical illness (e.g., chronic obstructive pulmonary disease, diabetes mellitus, congestive heart failure), and no orthopedic or pain problem other than CLBP, (4) no major surgery within the previous 12 months, (5) no organic brain syndrome or psychosis, and (6) fluent and literate in English. Controls were healthy men without a significant history of back pain (i.e., no lifetime history of pain lasting more than 1 week in duration) who were matched socio-demographitally to index CLBP patients. All subjects were part of a larger study of psychosocial factors in chronic low back pain. Other aspects of the project will be published elsewhere when complete. A report on depression and stressful life events in CLBP [l] has been published from the larger project and includes a sample that overlaps with that employed in the present study. Procedure Each of the following measures was administered following the patient’s medical evaluation. Visual Analog Scale (VAS). Current and typical pain intensity were measured on a 100 mm visual analog scale, with “0” representing “no pain” and “100” representing “ the worst pain imaginable.” Sickness Impact Profile (SIP). The SIP [4,.5,12] is a self-rated global measure of functional impairment. It has acceptable reliability [18] and validity [5] for a variety of illness populations and specifically for use with chronic pain patients [9]. In addition to ratings of overall impairment, this scale contains 3 dimensions of psychosocial, and other impairment, i.e., physical, (primarily work and recreational) impairment. Test-re-
Pain and Impairment Relationship Scale (PAIRS). The PAIRS [19] assesses beliefs and attitudes associated with the experience of chronic pain and one’s ability to function despite pain. The PAIRS consists of 15 personal statements that reflect thoughts, attitudes and opinions about pain. The following 2 statements are samples from the PAIRS: “I can still be expected to fulfill my work and family responsibilities despite my pam “. , “As long as I am in pain, I’ll never be able to live as well as I did before.” Each statement is accompanied by a 7-point Likert rating scale for the patient to identify to what degree she/he agrees or disagrees with the given statement. High scores on the PAIRS indicate a greater tendency for the respondent to equate pain with impairment, and to restrict functioning in the presence of pain. An instrument development study demonstrated that the PAIRS has high internal consistency (Cronbach’s coefficient alpha = 0.82) and is correlated with measures of impairment [19]. Self-Perception Scale (SPS). The SPS is a standardized and validated measure of favorable self-report response bias [20]. The SPS indicates a subject’s propensity to deny negative attributes or falsely endorse positive attributes, and was used in this study as a validity check on self-report measures. Cognitive Errors Questionnaire (CEQ). The CEQ [17] is a measure of depression-related cog&ions based on Beck’s [2] cognitive theory of depression and has been validated for use with CLBP patients [17]. CEQ items assess pain-related cognitive distortions, as well as distorted thinking in relation to more general non-pain related situations. Four specific types of cognitive errors (catastrophizing, personalizing, over-generalizing, and selective abstraction) are described by Beck [2] and assessed with the CEQ. These errors have been associated with depression in CLBP and non-pain samples
u71. Beck Depression Inventory (BDI). The BDI [3] is a 21-item self-report instrument that measures cognitive and endogenous aspects of depression. The BDI provides a quantitative measure of depressive symptoms. The BDI has undergone extensive reliability and validation studies from numerous laboratories, providing strong support for the psychometrics of this instrument [3,151.
test reliability and internal consistency coefficients are 0.92, and 0.94, respectively [4].
Results
WaddeN Physical Impairment Index (WPII). The WPII is a standardized, validated, physician-administered protocol for assessing the severity of orthopedic disease in painful back conditions [25]. This instrument combines findings from the history and physical examination to estimate the percentage of the patient’s functioning that is impaired by the back condition.
Sample characteristics Demographic data for all 50 subjects involved in this study are summarized in Table I. A multivariate analysis of variance (MANOVA) was conducted with income, years of education, and age as dependent variables to evaluate whether the patient and control groups differed significantly in terms of demographics. No
53
TABLE MEANS GROUP
I (f S.D.)
FOR
DEMOGRAPHIC
VARIABLES
BY
a
Variable
CLBP (N = 31)
Control
Age (years) Family income
46.20 ( f 12.04)
39.60 ( f 13.05)
( x %lO,OOO/year) Education (years)
2.03 12.77
’ MANOVA between 46: P>O.19.
groups:
( f 1.05) ( f 2.03)
2.11 13.79
WiIks Lambda
(N = 19)
(f 1.45) ( f 2.42)
= 0.9; F = 1.64; df = 3,
significant differences were found between groups (Wilks Lambda = 0.90; approx. F = 1.64; df = 3, 46; P 7 0.19), indicating that the recruitment matching procedures produced comparable groups. Mean ( f S.D.) duration of pain for the CLBP group was 134.71 months (f 116.53). Mean current pain intensity on the VAS was 45.13 ( f 21.99), while mean typical pain was 53.10 (f 24.64). The primary orthopedic diagnoses of the CLBP patients were: degenerative disk disease (n = 15), herniated nucleus pulposus (n = 5), spinal stenosis (n = 6), post-surgical pain (n = 2) . Thus, on average, this sample was characterized by middleaged males with long standing low back pain, typically related to degenerative disease of the spine. Discriminant validity: can the PAIRS discriminate CLBP patients from healthy controls? In order for a test to be valuable it must be able to distinguish relevant patient groups from controls. Table II displays means and standard deviations for the PAIRS scores by group (i.e., CLBP patient vs. controls). Significant differences were found between the means of the two groups (t = 3.54; df = 48; P < O.OOl), with the CLBP group scoring significantly higher than the control group on the PAIRS. This result indicates that CLBP patients are more likely than the healthy control group to equate pain with impairment and restriction in functioning. Moreover, this finding supports the validity of the PAIRS by demonstrating that the instrument can distinguish an important pain patient group from controls in terms of pain and impairment beliefs. Convergent validity: is the PAIRS related to measures of impairment? The relationship of the PAIRS to the SIP and the WPII, within the CLBP sample, was evaluated by using TABLE
II
MEANS (k SD.) FOR THE PAIN AND TIONSHIP SCALE (PAIRS) BY GROUP Variable
CLBP (N = 31)
Control
PAIRS
71.32 ( f 13.31)
57.95 ( f 12.34)
IMPAIRMENT
(N = 19)
RELA-
t (48)
P
3.54
< 0.001
the Pearson product-moment correlation to assess the instrument’s convergent validity. Correlational analysis in the CLBP group revealed that the PAIRS scores are positively correlated with overall functional impairment, as measured by the SIP (r = 0.54; P c 0.001). Specifically, the PAIRS was most strongly correlated to the physical impairment dimension of the SIP (r = 0.59; P < 0.0001) and the “other” impairment dimension of the SIP (r = 0.49; P < 0.005). A smaller correlation was found between the PAIRS and the psychosocial impairment dimension (r = 0.33; P < 0.05) of the SIP. Regression analyses revealed that when pain intensity and duration were controlled statistically the PAIRS was still significantly related to overall impairment on the SIP ( R2 change = 0.25; P < 0.005). Correlational analyses between the PAIRS and the WPII revealed no significant correlation (r = 0.27; P > 0.20), indicating that the pain and impairment beliefs tapped by the PAIRS were not significantly related to physician ratings of disease severity. Correlational analyses between the SIP and the WPII also revealed no significant relationship (r = 0.26; P 7 O.lO), verifying the diversity between disease severity (WPII) and impairment in daily functioning (SIP). Regression analyses revealed that when disease severity (WPII) was controlled statistically the PAIRS was still highly correlated with the SIP ( R2 change = 0.24; P < 0.005). These analyses support the convergent validity of the PAIRS by demonstrating that its scores of health beliefs are related, in the expected direction, to a validated measure of impairment in daily function without being significantly related to pain intensity, duration, or severity of spine dysfunction. Divergent validity: can the beliefs measured by the PAIRS be differentiated from emotional distress and distorted thinking? Divergent validity was assessed by evaluating the relationship of the PAIRS to measures of cognitive distortions (CEQ) and emotional distress (BDI). Correlational analysis revealed no significant relationship between the PAIRS and the CEQ (r = - 0.035; P > 0.40), indicating that the cognitive schema measured by the PAIRS is independent of the cognitive set measured by the CEQ. Moreover, PAIRS scores were not related significantly to either the low back (r = 0.07; P 7 0.30) or general (r = - 0.07; P 7 0.15) cognitive error subscales of the CEQ. Given that the CEQ was developed to measure cognitive distortions associated with depression, a correlational analyses between the CEQ and the BDI was conducted. Results revealed a significant positive relationship between the two tests (r = 0.54; P < 0.001). Even after controlling for impairment (SIP) through regression analysis, the CEQ accounted for a significant amount of variance in the BDI (R2 change = 0.11;
54
P < 0.015) thus indicating that the CEQ is a strong predictor of depression. Analyses of the relations~p between the PAIRS and the BDI also revealed that the PAIRS was correlated positively with the BDI (Y = 0.42; P < 0,015). A closer look at the relationship between impairment (SIP) and depression (BDI), however, also revealed a strong, significant relations~p (r = 0.75; P < 0.0001). Therefore, further analyses were calculated to clarify the relationship between the PAIRS and the BDJ, since both depression and pain beliefs were significantly correlated with impairment (r = 0.75 and r = 0.54, respectively). A regression analysis of the PAIRS and BDI relationship was conducted to assess whether the PAIRS was related to emotional distress after level of impairment was taken into account. Results of the regression analysis revealed that the PAIRS was no longer related to emotional distress ( R2 change = 0.0003; P > 0.90) once level of impairment (SIP) was controlled. Fig. 1 grap~~ally summarizes the interrelationships among the PAIRS, CEQ, SIP and BDI. There is a strong relationship between impairment (SIP) and depression (BDI), but not between the cognitive schema assessed by the PAIRS and depression (BDI). This cognitive schema (PAIRS) is specifically related to impairment (SIP), whife the cognitive set assessed by the CEQ is specifically related to depression (BDI). These analyses support the divergent validity of the PAIRS by demonstrating that the instrument is more strongly related to levels of impairment (SIP) than to measures of psychological distress (BDI and CEQ). Thus, these re-
suits indicate that the PAIRS is specific to its intended area of measurement, rather then a8general measure of distorted thinking or emotional distress. Test-retest
reliability:
is the PAIRS
reliable over time?
Test-retest results were obtained through a mail-out follow-up re-administration, with 29 of 31 CLBP subjects returning completed retests. Retesting did not occur at a fixed point in time for each subject. Rather, retests were obtained after at least 4 weeks had elapsed from the original testing to obtain a range of intervals of administration. The number of weeks between administrations ranged from 4 to 39, with a mean of 21.8. Despite this long interval, test-retest results reveal a strong correlation between PAIRS scores at time 1 and time 2 (r = 0.66; P < 0.05). A paired t test was calculated to determine whether there was a significant difference between the mean scores at the two testings. Results were significant {mean score at time 1 = 71.4 (S.D. = 12.1) and at time 2 = 75.6 (SD. = 13.4); t (28) = 2.1; P < O.OS), indicating that the retest scores were significantly higher. Thus, in summary, PAIRS scores were relatively consistent (reliable) over an extended interval, with overall test scores increasing over time, suggesting that impairment maintaining beliefs become more prominent with continued chronicity of pain. Vulnerability to self-report response bias: to what degree is the PAIRS independent of favorable self-report response bias? S&e response bias is often an issue with self-report,
group performance on the Self-Perception Scale (SPS) was evaluated. The groups did not differ significantly with respect to either of the response biases tested by the SPS (i.e., the propensity to deny negative attitudes or to falsely endorse positive attributes, all Ps > 0.60). Further, correlational analyses between the SPS and the PAIRS revealed no significant correlations between the PAIRS and either type of response bias (all Ps > 0.10). These results suggest that the PAIRS is not related to favorable self-report bias, at least as measured by the SPS. These results also suggest that the obtained group differences on the PAIRS cannot be attributed to a differentially biased presentation by either the CLBP group or the healthy control group.
Discussion Pig. 1. Summary of regression analyses on the PAIRS, SIP, BDI, and CEQ. Area “A” illustrates that cognitive factors assessed on the PAIRS are specifically related to impairment (SIP). Area “B” shows that impairment (SIP) and depression (BDI) are strongly related. Area “c” represents the relationship between cognitive distortion (CEQ) and depression (BDI). See text for further discussion of the relationships among these variables and the implications for the convergent and divergent validity of the PAIRS.
These findings confirm and extend previous findings from other investigators. Riley et al. [19] demonstrated that the PAIRS has adequate internal consistency and construct validity when used with a specialty pain clinic sample. The study presented here extends this work to a sample that was more homogeneous in terms of gender (all male), and pain syndrome (CLBP). It is also im-
55
portant to note that the present sample was not collected from a specialty pain clinic, but rather from an outpatient orthopedic clinic. This may be an important difference given that psychological distress may be over represented in pain clinic samples [1,14,16]. In addition, we conducted several important tests of reliability and validity for the first time on the PAIRS. This study has demonstrated that the PAIRS reliably discriminates between pain and non-pain groups, and that the PAIRS is correlated in the expected direction with functional impairment. Moreover, it should be noted that the PAIRS was significantly related to impairment in function even after pain intensity, duration, and severity of spine dysfunction were controlled. A comparison of effect size estimates (R*) indicates that the beliefs tapped by the PAIRS account for more variance in impairment than do those traditional “disease” variables combined. These results, in combination with those from Riley et al. [19], indicate that the PAIRS is a valid instrument in both a general and a highly selected group of CLBP patients. Present results also indicate that beliefs and attitudes held by the CLBP group, and reflected by the PAIRS scores, were not related to cognitive distortions associated with depression or with depression itself. Additionally, scores on the PAIRS were not related to favorable self-report bias. These findings further substantiate the proposal that the cognitive schema measured by the PAIRS are independent of cognitive distortion, emotional distress or response bias, supporting the specificity of the instrument. Finally, the PAIRS was found to be reliable over extended time intervals. The PAIRS operationalizes the frequently described cognitive schema that characterizes the stereotypical clinical presentation of chronic pain patients [68,10,11,23,24]. This cognitive schema asserts that impairment is a necessary aspect of pain, that activity despite pain is dangerous, and that pain is an acceptable excuse for neglecting responsibilities [10,22]. These beliefs frequently may be appropriate in the case of acute pain, but often may be maladaptive in cases of chronic nonmalignant pain [23]. The cognitive schema measured by the PAIRS have major implications for the rehabilitation of the CLBP patient. For example, Follick and colleagues [lo] describe the importance of confronting maladaptive beliefs and attitudes regarding pain and disability. The PAIRS directly measures these important beliefs and attitudes. Other studies support the importance of maladaptive beliefs in rehabilitation [13,21] and have suggested that altering cognitive schemas to incorporate the possibility of living a meaningful life despite pain is a prominent factor in the success of rehabilitation [13]. In this respect, the PAIRS may be useful in providing targets for cognitive treatments, to improve rehabilitation potential and in predicting response to treatment.
Further studies are under way to determine the utility of the PAIRS in these direct clinical applications. In the interim, the current .data suggest that the PAIRS may be a particularly useful addition to pretreatment evaluation batteries. Elevated scores on the PAIRS identify patients with strong beliefs about their pain that may impede functioning to capacity, increase the likelihood of deconditioning, augment pain through fear, and decrease compliance with rehabilitation efforts, Impairment beliefs elicited by the PAIRS should be addressed directly through education, behavioral demonstration [ 111 and/or cognitive restructuring techniques [24] as a part of a comprehensive pain management program. The PAIRS may also be useful as a manipulation check to determine the impact of cognitive interventions. Repeated administration during or after pain treatment can be helpful in identifying whether treatment has been effective in altering critical pain/impairment beliefs, and in identifying cognitive factors that may lead to relapse following treatment. In conclusion, the initial validation of the PAIRS is encouraging. The PAIRS appears to contribute a useful new perspective to the assessment armamentarium of researchers and clinicians interested in chronic pain. Further validation studies on larger samples of various pain syndromes and settings are needed to enlarge the validation base of this measure. Additionally, investigation into the limitations of this measure are also warranted. Although further studies using the PAIRS have been initiated in our laboratories, the present results are sufficiently encouraging to suggest that the measure be employed by other investigators interested in assessing the beliefs and attitudes patients have about their functional abilities as measured by the PAIRS.
Acknowledgements
Support for this research was provided by grants from the Veterans Administration Medical Research Program. The authors gratefully appreciate JoAnn Grant, R.N., for her contributions to data collection on this projecct.
References Atkinson, J.H., Slater, M.A., Grant, I., Patterson, T.L. and Garfin, S.R., Depressed mood in chronic low back pain: relationship with stressful life events, Pain, 35 (1988) 47-55. Beck, A.T., Cognitive Therapy and Emotional Disorders, Intemational University Press, New York, 1976. Beck, A.T., Ward, C., Mendelson, M., Mock, J. and Erbaugh, J., An inventory for measuring depression, Arch. Gen. Psychiat., 4 (1961) 53-63. Bergner, M., Bobbitt, R.A., Carter, W.B. and Gilson, B.S., Sickness Impact Profile: development and final revision of health status measure, Med. Care, 19 (1981) 787-805.
5 Bergner, M., Bobbitt, R.A., Pollard, W.E., Martin, D.P. and Gilson, B.S., The sickness impact profile: validation of a health status measure, Med. Care, 14 (1976) 57-67. 6 Black, R.G., Chronic pain syndrome, Clin. Med., 82 (1975) 17-20. 7 Follick, M.J., Aberger. E.W., Ahern, D.K. and McCartney. J.R., Chronic low back pain syndrome: identification and management, RI Med. J., 67 (1984) 219-224. 8 Follick. M.J., Ahem, D.K. and Aberger, E.W., Development of an audiovisual taxonomy of pain behavior: reliability and discriminant validity, Hlth Psychol., 4 (1985) 555-568. 9 Follick. M.J., Smith, T.W. and Ahern, D.K.. The sickness impact profile: a global measure of disability in chronic low back pain, Pain, 21 (1985) 67-76. 10 Follick, M.J., Zitter, R.E. and Ahem, D.K., Failures in operant treatment of chronic pain. In: E.B. Foa and P. Emmelkamp (Eds.), Failures in Behavior Therapy, Wiley, New York, 1983, pp. 311334. 11 Fordyce, W.E., Behavioral Methods for Chronic Pain and Illness, Mosby, St. Louis, MO, 1976. 12 Gilson, B.S., Gilson, J., Bergner, M., Bobbitt, R., Kressel, L., Pollard. W.E. and Vesselagg, M., The sickness impact profile: development of an outcome measure of health care, Am. J. Publ. Hlth, 65 (1975) 1304-1310. 13 Herman, E. and Baptiste, S., Pain control: mastery through group experience, Pain, 10 (1981) 79-86. 14 Katon, W., Egan, K. and Miller, D., Chronic pain: lifetime psychiatric diagnosis and family history. Am. J. Psychiat., 142 (1985) 115661160. 15 Kendall. P.C., Hollen. S.D.. Beck, A.T., Hummen, C.L. and In-
16 17
18
19
20 21
22
23 24
25
gram, R.E.. Issues and recommendations regarding use of the Beck Depression Inventory, Cogn. Ther. Res., 11 (1987) 289-299. Kramlinger, D.G., Swanson, P.W. and Manna, T., Are patients with chronic pain depressed? Am. J. Psychiat., 140 (1983) 747-749. Lefebvre, M.F.. Cognitive distortion and cognitive errors in depressed psychiatric and low back pain patients, J. Consult. Clin. Paychol.. 49 (1981) 517-525. Pollard, W.E., Bobbitt, R.A., Bergner, M., Martin, C.P. and Gilson. B.S., The sickness impact profile: reliability of a health status measure, Med. Care, 14 (1976) 1466155. Riley, J.F., Ahem, D.K. and Follick, M.J.. Chronic pain and functional impairment: assessing beliefs about their relationship, Arch. Phys. Med. Rehab., 69 (1988) 579-582. Roth. D.L., Snyder, C.R. and Pace, L.M., Dimensions of favorable self-presentation, J. Pers. Sot. Psychol.. 51. (1986) 867-874. Schwartz, D.P., DeGood, D.E. and Shutty, M.S., Direct asssessment of beliefs and attitudes of chronic pain patients, Arch. Phys. Med. Rehab., 66 (1985) 806-809. Smith, T.W., Follick, M.J., Ahern, D.K. and Adams, A.E.. Cognitive distortion and disability in chronic low back pain, Cogn. Ther. Res., 10 (1986) 201-210. Sternbach, R.A., Chronic pain as a disease entity, Triangle, 20 (1981) 27-32. Turk, D.C., Meichenbaum, D. and Genest, M.. Pain and Behavioral Medicine: Cognitive-Behavioral Perspective, Guilford Press, New York, 1983. Waddell, G. and Main, C.J., Assessment of severity in low-back disorders, Spine, 9 (1984) 204-208.
51 B.V. (Biomedical
Division)
0304-3959/91/$03.50
030439599100055C
PAIN 01707
Pain and impairment beliefs in chronic low back pain: validation of the Pain and Impairment Relationship Scale (PAIRS) * Mark A. Slater, Holly F. Hall, J. Hampton Atkinson and Steven R. Garfin VA Medical Center and University of California, San Diego, CA (U.S.A.) (Received
23 February
1990, revision received
14 June 1990, accepted
26 June 1990)
Few validated instruments are available to assess beliefs and attitudes that patients have regarding pain, or ability to function S-arY despite discomfort. The Pain and Impairment Relationship Scale (PAIRS) was developed to tap these important beliefs and attitudes in chronic pain patients. Preliminary data indicate that the PAIRS is internally consistent and significantly related to impairment in a highly selected pain clinic sample of patients, including some chronic low back pain patients. The present study was designed to extend the validation of the PAIRS to a more general sample of chronic benign low back pain patients. Furthermore, additional tests supported the discriminant, convergent and divergent validity, as well as the reliability and relative independence from favorable self-report response bias of the PAIRS, by respectively demonstrating that: (1) the impairment beliefs assessed with the PAIRS were more prominent in chronic low back pain (CLBP) patients than in matched non-pain, healthy controls; (2) scores on the PAIRS were significantly related to measures of physical impairment, but not to physicians ratings of disease severity; (3) the impairment beliefs assessed with the PAIRS are readily distinguishable from cognitive distortions and emotional distress; (4) PAIRS scores for chronic low back pain patients are relatively consistent over time; and (5) PAIRS scores are not significantly associated with measures of favorable self-report response bias. We conclude that the PAIRS has demonstrated at least preliminary utility for applications by researchers and clinicians interested in chronic pain. Key words:
Chronic
pain;
Assessment;
Pain beliefs
Introduction
Attitudes and beliefs about pain and impairment are frequently the targets of cognitive-behavioral approaches to the treatment of chronic pain syndromes. In chronic benign low back pain (CLBP), behavioral and cognitive-behavioral treatment techniques attempt to: (1) reduce the degree to which patients attend to their pain, (2) increase patients’ functional capacities, and (3) reduce the emotional distress and suffering related to pain and impairment [7,11,24]. Treatment typically includes restructuring cognitive schema that may perpetuate excessive pain and disability [24]. Beliefs about the meaning of pain and one’s ability to function despite discomfort are important aspects of this cognitive schema that is typically a target for intervention.
* This research was supported in part by Medical from the Department of Veterans Affairs.
Research
Correspondence to: Mark A. Slater, Ph.D., Director, ment Program (116B), San Diego VA Medical Center, Village Drive, San Diego, CA 92161, U.S.A.
Grants
Pain Manage3350 La Jolla
Unfortunately, few validated instruments are available to assess beliefs and attitudes that patients have regarding their pain, or its effect on daily function. Thus, Riley et al. [19] developed the Pain and Impairment Relationship Scale (PAIRS) to identify these important beliefs and attitudes in chronic pain populations. Preliminary data indicate that the PAIRS is internally consistent and significantly related to functional impairment in a highly selected pain clinic sample of heterogeneous chronic pain patients [19]. The study presented here was designed to extend the previous validation of the PAIRS to a more general sample of chronic benign low back pain patients. Additionally, several important new tests of the validity of the PAIRS were conducted: (1) discriminant validity was assessed by comparing chronic low back pain (CLBP) patients to healthy non-pain controls on the PAIRS; (2) convergent validity was determined by evaluating the relationships of the PAIRS with standardized measures of physical impairment and disease severity; (3) divergent validity was evaluated by determining the relationship of the PAIRS to measures of cognitive distortion and emotional distress; (4) reliubility over time was assessed by comparing PAIRS scores in a test-retest sample; and
finally, (5) vulnerability to response bias was evaluated by assessing the relationship between the PAIRS and a measure of favorable self-report response bias.
Methods Subjects Subjects were 31 male chronic benign low back pain patients, recruited from a general orthopedic clinic, and 19 healthy non-pain male volunteer controls. Inclusion criteria for CLBP patients were: (1) 21-64-year-old male outpatients attending a general orthopedic clinic, (2) benign low back pain (T6 or below) present “on a daily basis” for the previous 6 months or longer, (3) no major co-existing medical illness (e.g., chronic obstructive pulmonary disease, diabetes mellitus, congestive heart failure), and no orthopedic or pain problem other than CLBP, (4) no major surgery within the previous 12 months, (5) no organic brain syndrome or psychosis, and (6) fluent and literate in English. Controls were healthy men without a significant history of back pain (i.e., no lifetime history of pain lasting more than 1 week in duration) who were matched socio-demographitally to index CLBP patients. All subjects were part of a larger study of psychosocial factors in chronic low back pain. Other aspects of the project will be published elsewhere when complete. A report on depression and stressful life events in CLBP [l] has been published from the larger project and includes a sample that overlaps with that employed in the present study. Procedure Each of the following measures was administered following the patient’s medical evaluation. Visual Analog Scale (VAS). Current and typical pain intensity were measured on a 100 mm visual analog scale, with “0” representing “no pain” and “100” representing “ the worst pain imaginable.” Sickness Impact Profile (SIP). The SIP [4,.5,12] is a self-rated global measure of functional impairment. It has acceptable reliability [18] and validity [5] for a variety of illness populations and specifically for use with chronic pain patients [9]. In addition to ratings of overall impairment, this scale contains 3 dimensions of psychosocial, and other impairment, i.e., physical, (primarily work and recreational) impairment. Test-re-
Pain and Impairment Relationship Scale (PAIRS). The PAIRS [19] assesses beliefs and attitudes associated with the experience of chronic pain and one’s ability to function despite pain. The PAIRS consists of 15 personal statements that reflect thoughts, attitudes and opinions about pain. The following 2 statements are samples from the PAIRS: “I can still be expected to fulfill my work and family responsibilities despite my pam “. , “As long as I am in pain, I’ll never be able to live as well as I did before.” Each statement is accompanied by a 7-point Likert rating scale for the patient to identify to what degree she/he agrees or disagrees with the given statement. High scores on the PAIRS indicate a greater tendency for the respondent to equate pain with impairment, and to restrict functioning in the presence of pain. An instrument development study demonstrated that the PAIRS has high internal consistency (Cronbach’s coefficient alpha = 0.82) and is correlated with measures of impairment [19]. Self-Perception Scale (SPS). The SPS is a standardized and validated measure of favorable self-report response bias [20]. The SPS indicates a subject’s propensity to deny negative attributes or falsely endorse positive attributes, and was used in this study as a validity check on self-report measures. Cognitive Errors Questionnaire (CEQ). The CEQ [17] is a measure of depression-related cog&ions based on Beck’s [2] cognitive theory of depression and has been validated for use with CLBP patients [17]. CEQ items assess pain-related cognitive distortions, as well as distorted thinking in relation to more general non-pain related situations. Four specific types of cognitive errors (catastrophizing, personalizing, over-generalizing, and selective abstraction) are described by Beck [2] and assessed with the CEQ. These errors have been associated with depression in CLBP and non-pain samples
u71. Beck Depression Inventory (BDI). The BDI [3] is a 21-item self-report instrument that measures cognitive and endogenous aspects of depression. The BDI provides a quantitative measure of depressive symptoms. The BDI has undergone extensive reliability and validation studies from numerous laboratories, providing strong support for the psychometrics of this instrument [3,151.
test reliability and internal consistency coefficients are 0.92, and 0.94, respectively [4].
Results
WaddeN Physical Impairment Index (WPII). The WPII is a standardized, validated, physician-administered protocol for assessing the severity of orthopedic disease in painful back conditions [25]. This instrument combines findings from the history and physical examination to estimate the percentage of the patient’s functioning that is impaired by the back condition.
Sample characteristics Demographic data for all 50 subjects involved in this study are summarized in Table I. A multivariate analysis of variance (MANOVA) was conducted with income, years of education, and age as dependent variables to evaluate whether the patient and control groups differed significantly in terms of demographics. No
53
TABLE MEANS GROUP
I (f S.D.)
FOR
DEMOGRAPHIC
VARIABLES
BY
a
Variable
CLBP (N = 31)
Control
Age (years) Family income
46.20 ( f 12.04)
39.60 ( f 13.05)
( x %lO,OOO/year) Education (years)
2.03 12.77
’ MANOVA between 46: P>O.19.
groups:
( f 1.05) ( f 2.03)
2.11 13.79
WiIks Lambda
(N = 19)
(f 1.45) ( f 2.42)
= 0.9; F = 1.64; df = 3,
significant differences were found between groups (Wilks Lambda = 0.90; approx. F = 1.64; df = 3, 46; P 7 0.19), indicating that the recruitment matching procedures produced comparable groups. Mean ( f S.D.) duration of pain for the CLBP group was 134.71 months (f 116.53). Mean current pain intensity on the VAS was 45.13 ( f 21.99), while mean typical pain was 53.10 (f 24.64). The primary orthopedic diagnoses of the CLBP patients were: degenerative disk disease (n = 15), herniated nucleus pulposus (n = 5), spinal stenosis (n = 6), post-surgical pain (n = 2) . Thus, on average, this sample was characterized by middleaged males with long standing low back pain, typically related to degenerative disease of the spine. Discriminant validity: can the PAIRS discriminate CLBP patients from healthy controls? In order for a test to be valuable it must be able to distinguish relevant patient groups from controls. Table II displays means and standard deviations for the PAIRS scores by group (i.e., CLBP patient vs. controls). Significant differences were found between the means of the two groups (t = 3.54; df = 48; P < O.OOl), with the CLBP group scoring significantly higher than the control group on the PAIRS. This result indicates that CLBP patients are more likely than the healthy control group to equate pain with impairment and restriction in functioning. Moreover, this finding supports the validity of the PAIRS by demonstrating that the instrument can distinguish an important pain patient group from controls in terms of pain and impairment beliefs. Convergent validity: is the PAIRS related to measures of impairment? The relationship of the PAIRS to the SIP and the WPII, within the CLBP sample, was evaluated by using TABLE
II
MEANS (k SD.) FOR THE PAIN AND TIONSHIP SCALE (PAIRS) BY GROUP Variable
CLBP (N = 31)
Control
PAIRS
71.32 ( f 13.31)
57.95 ( f 12.34)
IMPAIRMENT
(N = 19)
RELA-
t (48)
P
3.54
< 0.001
the Pearson product-moment correlation to assess the instrument’s convergent validity. Correlational analysis in the CLBP group revealed that the PAIRS scores are positively correlated with overall functional impairment, as measured by the SIP (r = 0.54; P c 0.001). Specifically, the PAIRS was most strongly correlated to the physical impairment dimension of the SIP (r = 0.59; P < 0.0001) and the “other” impairment dimension of the SIP (r = 0.49; P < 0.005). A smaller correlation was found between the PAIRS and the psychosocial impairment dimension (r = 0.33; P < 0.05) of the SIP. Regression analyses revealed that when pain intensity and duration were controlled statistically the PAIRS was still significantly related to overall impairment on the SIP ( R2 change = 0.25; P < 0.005). Correlational analyses between the PAIRS and the WPII revealed no significant correlation (r = 0.27; P > 0.20), indicating that the pain and impairment beliefs tapped by the PAIRS were not significantly related to physician ratings of disease severity. Correlational analyses between the SIP and the WPII also revealed no significant relationship (r = 0.26; P 7 O.lO), verifying the diversity between disease severity (WPII) and impairment in daily functioning (SIP). Regression analyses revealed that when disease severity (WPII) was controlled statistically the PAIRS was still highly correlated with the SIP ( R2 change = 0.24; P < 0.005). These analyses support the convergent validity of the PAIRS by demonstrating that its scores of health beliefs are related, in the expected direction, to a validated measure of impairment in daily function without being significantly related to pain intensity, duration, or severity of spine dysfunction. Divergent validity: can the beliefs measured by the PAIRS be differentiated from emotional distress and distorted thinking? Divergent validity was assessed by evaluating the relationship of the PAIRS to measures of cognitive distortions (CEQ) and emotional distress (BDI). Correlational analysis revealed no significant relationship between the PAIRS and the CEQ (r = - 0.035; P > 0.40), indicating that the cognitive schema measured by the PAIRS is independent of the cognitive set measured by the CEQ. Moreover, PAIRS scores were not related significantly to either the low back (r = 0.07; P 7 0.30) or general (r = - 0.07; P 7 0.15) cognitive error subscales of the CEQ. Given that the CEQ was developed to measure cognitive distortions associated with depression, a correlational analyses between the CEQ and the BDI was conducted. Results revealed a significant positive relationship between the two tests (r = 0.54; P < 0.001). Even after controlling for impairment (SIP) through regression analysis, the CEQ accounted for a significant amount of variance in the BDI (R2 change = 0.11;
54
P < 0.015) thus indicating that the CEQ is a strong predictor of depression. Analyses of the relations~p between the PAIRS and the BDI also revealed that the PAIRS was correlated positively with the BDI (Y = 0.42; P < 0,015). A closer look at the relationship between impairment (SIP) and depression (BDI), however, also revealed a strong, significant relations~p (r = 0.75; P < 0.0001). Therefore, further analyses were calculated to clarify the relationship between the PAIRS and the BDJ, since both depression and pain beliefs were significantly correlated with impairment (r = 0.75 and r = 0.54, respectively). A regression analysis of the PAIRS and BDI relationship was conducted to assess whether the PAIRS was related to emotional distress after level of impairment was taken into account. Results of the regression analysis revealed that the PAIRS was no longer related to emotional distress ( R2 change = 0.0003; P > 0.90) once level of impairment (SIP) was controlled. Fig. 1 grap~~ally summarizes the interrelationships among the PAIRS, CEQ, SIP and BDI. There is a strong relationship between impairment (SIP) and depression (BDI), but not between the cognitive schema assessed by the PAIRS and depression (BDI). This cognitive schema (PAIRS) is specifically related to impairment (SIP), whife the cognitive set assessed by the CEQ is specifically related to depression (BDI). These analyses support the divergent validity of the PAIRS by demonstrating that the instrument is more strongly related to levels of impairment (SIP) than to measures of psychological distress (BDI and CEQ). Thus, these re-
suits indicate that the PAIRS is specific to its intended area of measurement, rather then a8general measure of distorted thinking or emotional distress. Test-retest
reliability:
is the PAIRS
reliable over time?
Test-retest results were obtained through a mail-out follow-up re-administration, with 29 of 31 CLBP subjects returning completed retests. Retesting did not occur at a fixed point in time for each subject. Rather, retests were obtained after at least 4 weeks had elapsed from the original testing to obtain a range of intervals of administration. The number of weeks between administrations ranged from 4 to 39, with a mean of 21.8. Despite this long interval, test-retest results reveal a strong correlation between PAIRS scores at time 1 and time 2 (r = 0.66; P < 0.05). A paired t test was calculated to determine whether there was a significant difference between the mean scores at the two testings. Results were significant {mean score at time 1 = 71.4 (S.D. = 12.1) and at time 2 = 75.6 (SD. = 13.4); t (28) = 2.1; P < O.OS), indicating that the retest scores were significantly higher. Thus, in summary, PAIRS scores were relatively consistent (reliable) over an extended interval, with overall test scores increasing over time, suggesting that impairment maintaining beliefs become more prominent with continued chronicity of pain. Vulnerability to self-report response bias: to what degree is the PAIRS independent of favorable self-report response bias? S&e response bias is often an issue with self-report,
group performance on the Self-Perception Scale (SPS) was evaluated. The groups did not differ significantly with respect to either of the response biases tested by the SPS (i.e., the propensity to deny negative attitudes or to falsely endorse positive attributes, all Ps > 0.60). Further, correlational analyses between the SPS and the PAIRS revealed no significant correlations between the PAIRS and either type of response bias (all Ps > 0.10). These results suggest that the PAIRS is not related to favorable self-report bias, at least as measured by the SPS. These results also suggest that the obtained group differences on the PAIRS cannot be attributed to a differentially biased presentation by either the CLBP group or the healthy control group.
Discussion Pig. 1. Summary of regression analyses on the PAIRS, SIP, BDI, and CEQ. Area “A” illustrates that cognitive factors assessed on the PAIRS are specifically related to impairment (SIP). Area “B” shows that impairment (SIP) and depression (BDI) are strongly related. Area “c” represents the relationship between cognitive distortion (CEQ) and depression (BDI). See text for further discussion of the relationships among these variables and the implications for the convergent and divergent validity of the PAIRS.
These findings confirm and extend previous findings from other investigators. Riley et al. [19] demonstrated that the PAIRS has adequate internal consistency and construct validity when used with a specialty pain clinic sample. The study presented here extends this work to a sample that was more homogeneous in terms of gender (all male), and pain syndrome (CLBP). It is also im-
55
portant to note that the present sample was not collected from a specialty pain clinic, but rather from an outpatient orthopedic clinic. This may be an important difference given that psychological distress may be over represented in pain clinic samples [1,14,16]. In addition, we conducted several important tests of reliability and validity for the first time on the PAIRS. This study has demonstrated that the PAIRS reliably discriminates between pain and non-pain groups, and that the PAIRS is correlated in the expected direction with functional impairment. Moreover, it should be noted that the PAIRS was significantly related to impairment in function even after pain intensity, duration, and severity of spine dysfunction were controlled. A comparison of effect size estimates (R*) indicates that the beliefs tapped by the PAIRS account for more variance in impairment than do those traditional “disease” variables combined. These results, in combination with those from Riley et al. [19], indicate that the PAIRS is a valid instrument in both a general and a highly selected group of CLBP patients. Present results also indicate that beliefs and attitudes held by the CLBP group, and reflected by the PAIRS scores, were not related to cognitive distortions associated with depression or with depression itself. Additionally, scores on the PAIRS were not related to favorable self-report bias. These findings further substantiate the proposal that the cognitive schema measured by the PAIRS are independent of cognitive distortion, emotional distress or response bias, supporting the specificity of the instrument. Finally, the PAIRS was found to be reliable over extended time intervals. The PAIRS operationalizes the frequently described cognitive schema that characterizes the stereotypical clinical presentation of chronic pain patients [68,10,11,23,24]. This cognitive schema asserts that impairment is a necessary aspect of pain, that activity despite pain is dangerous, and that pain is an acceptable excuse for neglecting responsibilities [10,22]. These beliefs frequently may be appropriate in the case of acute pain, but often may be maladaptive in cases of chronic nonmalignant pain [23]. The cognitive schema measured by the PAIRS have major implications for the rehabilitation of the CLBP patient. For example, Follick and colleagues [lo] describe the importance of confronting maladaptive beliefs and attitudes regarding pain and disability. The PAIRS directly measures these important beliefs and attitudes. Other studies support the importance of maladaptive beliefs in rehabilitation [13,21] and have suggested that altering cognitive schemas to incorporate the possibility of living a meaningful life despite pain is a prominent factor in the success of rehabilitation [13]. In this respect, the PAIRS may be useful in providing targets for cognitive treatments, to improve rehabilitation potential and in predicting response to treatment.
Further studies are under way to determine the utility of the PAIRS in these direct clinical applications. In the interim, the current .data suggest that the PAIRS may be a particularly useful addition to pretreatment evaluation batteries. Elevated scores on the PAIRS identify patients with strong beliefs about their pain that may impede functioning to capacity, increase the likelihood of deconditioning, augment pain through fear, and decrease compliance with rehabilitation efforts, Impairment beliefs elicited by the PAIRS should be addressed directly through education, behavioral demonstration [ 111 and/or cognitive restructuring techniques [24] as a part of a comprehensive pain management program. The PAIRS may also be useful as a manipulation check to determine the impact of cognitive interventions. Repeated administration during or after pain treatment can be helpful in identifying whether treatment has been effective in altering critical pain/impairment beliefs, and in identifying cognitive factors that may lead to relapse following treatment. In conclusion, the initial validation of the PAIRS is encouraging. The PAIRS appears to contribute a useful new perspective to the assessment armamentarium of researchers and clinicians interested in chronic pain. Further validation studies on larger samples of various pain syndromes and settings are needed to enlarge the validation base of this measure. Additionally, investigation into the limitations of this measure are also warranted. Although further studies using the PAIRS have been initiated in our laboratories, the present results are sufficiently encouraging to suggest that the measure be employed by other investigators interested in assessing the beliefs and attitudes patients have about their functional abilities as measured by the PAIRS.
Acknowledgements
Support for this research was provided by grants from the Veterans Administration Medical Research Program. The authors gratefully appreciate JoAnn Grant, R.N., for her contributions to data collection on this projecct.
References Atkinson, J.H., Slater, M.A., Grant, I., Patterson, T.L. and Garfin, S.R., Depressed mood in chronic low back pain: relationship with stressful life events, Pain, 35 (1988) 47-55. Beck, A.T., Cognitive Therapy and Emotional Disorders, Intemational University Press, New York, 1976. Beck, A.T., Ward, C., Mendelson, M., Mock, J. and Erbaugh, J., An inventory for measuring depression, Arch. Gen. Psychiat., 4 (1961) 53-63. Bergner, M., Bobbitt, R.A., Carter, W.B. and Gilson, B.S., Sickness Impact Profile: development and final revision of health status measure, Med. Care, 19 (1981) 787-805.
5 Bergner, M., Bobbitt, R.A., Pollard, W.E., Martin, D.P. and Gilson, B.S., The sickness impact profile: validation of a health status measure, Med. Care, 14 (1976) 57-67. 6 Black, R.G., Chronic pain syndrome, Clin. Med., 82 (1975) 17-20. 7 Follick, M.J., Aberger. E.W., Ahern, D.K. and McCartney. J.R., Chronic low back pain syndrome: identification and management, RI Med. J., 67 (1984) 219-224. 8 Follick. M.J., Ahem, D.K. and Aberger, E.W., Development of an audiovisual taxonomy of pain behavior: reliability and discriminant validity, Hlth Psychol., 4 (1985) 555-568. 9 Follick. M.J., Smith, T.W. and Ahern, D.K.. The sickness impact profile: a global measure of disability in chronic low back pain, Pain, 21 (1985) 67-76. 10 Follick, M.J., Zitter, R.E. and Ahem, D.K., Failures in operant treatment of chronic pain. In: E.B. Foa and P. Emmelkamp (Eds.), Failures in Behavior Therapy, Wiley, New York, 1983, pp. 311334. 11 Fordyce, W.E., Behavioral Methods for Chronic Pain and Illness, Mosby, St. Louis, MO, 1976. 12 Gilson, B.S., Gilson, J., Bergner, M., Bobbitt, R., Kressel, L., Pollard. W.E. and Vesselagg, M., The sickness impact profile: development of an outcome measure of health care, Am. J. Publ. Hlth, 65 (1975) 1304-1310. 13 Herman, E. and Baptiste, S., Pain control: mastery through group experience, Pain, 10 (1981) 79-86. 14 Katon, W., Egan, K. and Miller, D., Chronic pain: lifetime psychiatric diagnosis and family history. Am. J. Psychiat., 142 (1985) 115661160. 15 Kendall. P.C., Hollen. S.D.. Beck, A.T., Hummen, C.L. and In-
16 17
18
19
20 21
22
23 24
25
gram, R.E.. Issues and recommendations regarding use of the Beck Depression Inventory, Cogn. Ther. Res., 11 (1987) 289-299. Kramlinger, D.G., Swanson, P.W. and Manna, T., Are patients with chronic pain depressed? Am. J. Psychiat., 140 (1983) 747-749. Lefebvre, M.F.. Cognitive distortion and cognitive errors in depressed psychiatric and low back pain patients, J. Consult. Clin. Paychol.. 49 (1981) 517-525. Pollard, W.E., Bobbitt, R.A., Bergner, M., Martin, C.P. and Gilson. B.S., The sickness impact profile: reliability of a health status measure, Med. Care, 14 (1976) 1466155. Riley, J.F., Ahem, D.K. and Follick, M.J.. Chronic pain and functional impairment: assessing beliefs about their relationship, Arch. Phys. Med. Rehab., 69 (1988) 579-582. Roth. D.L., Snyder, C.R. and Pace, L.M., Dimensions of favorable self-presentation, J. Pers. Sot. Psychol.. 51. (1986) 867-874. Schwartz, D.P., DeGood, D.E. and Shutty, M.S., Direct asssessment of beliefs and attitudes of chronic pain patients, Arch. Phys. Med. Rehab., 66 (1985) 806-809. Smith, T.W., Follick, M.J., Ahern, D.K. and Adams, A.E.. Cognitive distortion and disability in chronic low back pain, Cogn. Ther. Res., 10 (1986) 201-210. Sternbach, R.A., Chronic pain as a disease entity, Triangle, 20 (1981) 27-32. Turk, D.C., Meichenbaum, D. and Genest, M.. Pain and Behavioral Medicine: Cognitive-Behavioral Perspective, Guilford Press, New York, 1983. Waddell, G. and Main, C.J., Assessment of severity in low-back disorders, Spine, 9 (1984) 204-208.
