177

Pain, 46 (1991) 177-184

0 1991 Elsevier Science Publishers B.V. 0304-3959/91/$03.50 ADONIS 030439599100172LJ

PAIN 01822

Depression and the chronic pain experience Jennifer A. Haythornthwaite

a, William J. Sieber b and Robert D. Kerns ’

a National Institute on Aging, Baltimore, MD 21224 (U.S.A.), ’ Yale Urkersity, New Hawn, CT 06520 (U.S.A.i und (’ Yale Unkersity School of Medicine and West HaL,en Veterans Affairs Medical Center, Psychology Sewice, West Harem CT 06516 (U.S.A.)

(Received 26 February 1990, revision received 14 November 1990, accepted 20 November 19901

The present study examined the relationship between depression and a constellation of pain-reSummary lated variables that describe the experience of chronic pain patients. Thirty-seven depressed and 32 non-depressed heterogeneous chronic pain patients were identified through structured interviews, use of standardized criteria and scores on the Beck Depression Inventory (BDI). The 2 groups were compared on demographic variables and scores on the Marlowe-Crowne Social Desirability scale (MC), as well as measures of disability and medication use, pain severity, interference due to pain and reported pain behaviors. The depressed group was found to be younger and to score lower on the MC than the non-depressed group. Multivariate analyses of covariance (MANCOVA), using age and MC as covariates, reveaIed that depressed chronic pain patients, relative to their non-depressed counterparts, reported greater pain intensity, greater interference due to pain and more pain behaviors. There were no group differences on the measures of disability and use of medications. The results provide further support for the importance of incorporating depression into clinical and theoretical formulations of chronic pain. Future use of structured interviews and standardized criteria for diagnosing depression may clarify some of the inconsistencies found in the literature. Key words: Chronic pain; Depression;

Depression diagnosis; Pain intensity; Activity; Pain behaviors

Introduction An apparently high rate of depression among chronic pain patients has spurred increased attention to this problem in both clinical and theoretical formulations of chronic pain. Although the exact nature of the relationship between pain and depression remains somewhat controversial, depression has been demonstrated to have a substantial effect on both clinical presentation [14,21,25,401 and treatment response 13,111. Progress has recently been made in identifying aspects of the pain experience that differentiate depressed and non-depressed chronic pain patients; however, many inconsistencies remain when results are compared across studies. Demographic variables, such as sex and level of education, and medical history

Correspondence too: Jennifer A. Haythornthwaite, National Institute on Aging, 4940 Eastern Avenue, Baltimore, MD 21224. U.S.A.

variables, such as duration of pain, type of pain and number of operations, generahy have not been found to differentiate these 2 groups. Two studies have reported opposite findings when comparing the age of depressed and non-depressed pain patients [8,491, and 2 recent studies indicated that depressed pain patients were more likely than non-depressed pain patients to receive a variety of medications, including sedative/ hypnotics and narcotics [211 and antidepressants and anxiolitics [6]. Comparisons of depressed and non-depressed chronic pain patients on variables related to the experience of pain, such as pain intensity, activity impairment and pain behaviors, have yielded inconsistent findings. Some studies have reported a positive relationship between depression and pain intensity ratings E6,10,11,21,331 whereas other studies reported no reliable relationship [25,34,39,481. On the other hand, depressed chronic pain patients have been consistently found to be less active than their non-depressed counterparts [6,10,17,21,25]. Although Keefe and his col-

17s

leagues ]2 l] reported that depression reliably predicted a variety of pain behaviors. an earlier study did not demonstrate a relationship between depressed mood and observed pain behaviors [35]. The inconsistencies across studies may bo due to the wide range of methods used to identify depression, a practice which has hampered progress in this area [X,42]. There is clearly a need to incorporate structured assessment and criteria into the measurement of depression among chronic pain patients (421. After reviewing the use of self-report instruments that assess depression, such as the Beck Depression Inventory (BDIJ [21. a group of prominent investigators [23] recently expressed concern over the reliance upon selfreport measures when studying depression. These authors cautioned that careful diagnostic procedures arc necessary in order to identify depression that is not confounded by other psychopathology and emphasized the need to employ multiple methods for assessment. While the BDI has been demonstrated to be a useful index of depression among chronic pain patients [49]. a recent study denlonstr~tt~d high correlations between the BDI and measures of general psychological distress [lo]. This type of finding, also found among community samples [38], raises issues about the psychopathological specificity of the BDI [22]. At issue is the extent to which high scores on the BDI indicate clinical depression and not depressed mood of a transient nature or some other psychopathological state. The use of multiple methods for assessing dcpression, including structured interviews and diagnostic criteria, has generally been limited to studies of prevalence of depression among chronic pain patients [3 1,32] and biogenetic factors shared by these 2 disorders [8,15,16.50]. Studies examining cognitive and behavioral characteristics that discriminate depressed from non-depressed pain patients have rarely used standardized methods to diagnose depression. Structured interviews and diagnostic criteria have the benefit of identifying chronic pain patients whose complaints of depression are clinically significant and not of a transient nature and allow comparison of findings across studies. The impact of response style, for example social desirability, on self-reports of pain has also rarely been examined. Beck and Beamesderfer [l] observed that depressed patients described themselves in an unfavorable way, even in response to questions not addressing their depressive symptomatology. These authors suggested that a socially undesirable response style was a distinguishing characteristic of depression. Given this description of depression, chronic pain patients who are also depressed would be expected to portray themselves in an unfavorable way even in response to questions about areas other than their affective state, such as questions about their experience of pain. The present study was designed to delineate reliable

differences in the experience of chronic pain between groups of patients with and without coexisting depression. Three methodological strengths extend previous findings. First. diagnostic interviewing. standardized criteria and a self-report measure were used to identify depression. The patients who met diagnostic criteria for major depression and presented with at least ;I mildly depressed score on the BDI [22] were included in the depressed group: patients who did not meet criteria for major depression and presented with i-l normal score on the BDI were included in the non-depressed group. Second, multivariate statistical proccdures were used to examine multiple aspects of the pain experience, including disability and use of mcdica[ions, ratings of pain severity. frequency of pain behaviors and impairment of daily activities. And third. ;L frequently used mcasurc of social desirability, the Marlowe-Crowne Social Desirability scale. was used to examine the intluence of this response style on the experience of pain.

Method .Su&ect.s

Subjects were obtained from consecutive admissions to an interdisciplinary pain rehabilitation program and met the following criteria: (a) pain of at least 6 months duration, (b) pain not related to malignancies, and tc) no evidcncc of active psychosis or acute risk of suicide. Only subjects who completed structured interviewing and all questionnaire measures were included in the analyses. Two criteria were used to identify groups of depressed and non-depressed subjects. First, the Schcdule for Affective Disorders and Schizophrenia (SADS) [ 131. a structured psychiatric interview. was used to establish a diagnosis based on the Research Diagnostic Criteria (RDC) [l&46]. SADS interviews were conducted by 3 interviewers (advanced doctoral level psychology students1 who received extensive training in administering the SADS and using the RDC from II research psychiatrist. Training consisted of watching experienced interviewers conduct the SADS interview while trainees simultaneously rated patients’ responses. then of participating in the interviews conducted by trained interviewers and finally of conducting intcrviews with trained interviewers watching and supervising. Upon completion of each SADS interview, RDC were used to determine a diagnosis of major depression, minor depression or no depression. A sample of 12 interviews were selected for computing inter-rate] reliability. The agreement between raters for diagnosis of major depression, minor depression or no deprcssion was 100%. Since diagnoses were based on the patient’s descrip-

179

tion of the worst week of the identified episode, a second criterion was used to reflect the severity of s~ptoms of depression at the time of evaluation. A BDI score of 11 was used to identify at least a mild level of current depressive symptomatology [22]. Using both criteria, 37 patients were classified as depressed; these pain patients were diagnosed with major depression using the RDC and scored 11 or higher on the BDI. Thirty-two patients were classified as non-depressed; these pain patients were diagnosed as having either minor depression or no depression based on the RDC and scored 10 or less on the BDI. Twenty-two patients were excluded using the double criteria. The sample, including the depressed and the nondepressed groups, ranged in age from 26 to 78 years (M = 49.3, S.D. = 13.6) and had attained a mean educational level of 12.1 years (S.D. = 2.2). Eighty-four percent were males and 59% were married. The most common pain complaint was low back pain (49%), with neck and shoulder musculoskeletal pain (14%) and other musculoskeletal pain (16%) comprising the majority of the remaining complaints. The average duration of pain was just over 8 years (S.D. = 7.2 years), with 48% of the patients reporting at least one surgery for their pain prior to the evaIuation. Only 20% of the subjects were actively employed, and 5170 of all patients were receiving some form of compensation for their pain.

Questionnaire and interview data were grouped into 4 theoretically derived constructs relevant to a comprehensive pain assessment. These constructs induded disability and use of medications, pain severity, interference due to pain and pain behaviors. Empirical support for these constructs is based in part on the published factorial validity of the West Haven-Yale Multidimensional Pain Inventory (WHYMPII [291. Each construct was measured by multiple scales which are described below. Disability and medication use. Six variables derived from a semi-structured pain assessment interview were used to index pain-related disability and patients’ use of medications. Use of medications for pain, including narcotic and non-narcotic analgesics and use of antidepressant medications were each coded as present or absent, The chronic@ of the pain problem was coded in months. Previous surgery for the pain complaint, compensation for pain and current employment were also coded as present or absent for each patient. Pain secerity. Two measures were used to measure pain severity: the Pain Rating Index (PRI) [36] from the McGill Pain Questionnaire and the Pain Severity scale from the WHYMPI [29]. Both scales have been demonstrated to have good psychometric properties. Interference. This construct assessed the degree of

interference caused by pain and the frequency with which patients engaged in a variety of common activities that may be affected by chronic pain. Five subscales from the WHYMPI were included to measure this construct, including Interference, Social Activities, Outdoor Activities, Activities Away From Home and Household Activities. These scales have shown good internal consistency and test-retest reliability 1291. Pain behaviors. The Pain Behavior Checklist (PBCL) [26] was used as a measure of self-reported pain behavior. Patients rated how often they engaged in 17 different behaviors on 7-point scales (0 = not at all; 6 = almost all the time). These items have been previously factor analyzed into 4 reliable scales: Distorted Ambulation (e.g., walking in a protective fashion), Facial/ Audible Expressions (e.g., grimacing, clenching teeth), Affective Distress (e.g., becoming irritable or angry), and Seeking Help (asking someone to do something to help the pain). The 4 scales have been shown to have good internal consistency and stability. The scales have been validated against observed pain behaviors and other self-report measures of the pain experience [26]. Social desirability. The Marlowe-Crowne Social Desirability scale (MC) [9] consisted of 33 true-false items. Work done on the MC since its original development indicates that the scale measures affect inhibition, defensiveness and presenting oneself in a positive light. The scale has been demonstrated to have good internal consistency and stability [9]. Procedure

Subjects completed the SADS and pain interviews and questionnaires as part of a comprehensive assessment protocol that preceded the implementation of the rehabilitation program. The assessment protocol has been described elsewhere [24,28].

Results As a first step in the analyses, demographic variables and MC scores were compared between the groups of depressed and non-depressed patients. Next, simple correlations between variables which differed between the 2 groups and the measures of pain intensity, interference and pain behaviors were examined. Since 5 of the 6 measures of disability and medication use were dichotomous scores, these measures were not included in the correlational analysis. Since univariate analyses (chi-square and analysis of covariance) were used to examine group differences on disability and use of medications, the alpha for these tests was adjusted (0.008) using the Bonferroni method (alpha/number of univariate tests) to maintain the familywise error rate. Multivariate analyses of covariance (MANCOVA) with age and social desirability as the covariates were used

IS0

to examine group differences on the constructs of pain intensity, interference and pain behaviors. Eta-squared was computed as an estimate of the strength of association in the population between depression and the pain experience constructs [5]. When multivariate analyses were significant, univariate analyses of covariance (ANCOVA) were conducted on individual measures. A l-way muItivariate analysis of variance was performed comparing the 2 groups on age and education. A multivariate main effect of group (F (2, 65) = 6.63, P < 0.0 1) was found with univariate analyses to be due to the younger mean age of the depressed patients compared to the non-depressed patients t F (1, 66)= 12.71. P < 0.001; see Table I). There was no significant difference on education. Chi-square analyses showed no significant differences between groups on type of pain, gender or marital status (Ps > 0.10). A l-way ANOVA between groups (F (1. 66) = 9.61. P < 0.01) revealed that non-depressed patients scored significantly higher than did depressed patients on the Marlowe-Crowne (M = 22.0 (S.D. = 5.4) and M = 17.4 (SD. = 6.7). respectively). The 2 variables found to be significantly different between the 2 groups, age and Marlowe-Crowne, were used as covariates for the analyses of pain intensity, interference and pain behaviors. Simple correlations between age and MC and the measures of pain intensity, interference and pain behaviors were examined next. Significant relationships were found between age and the McGilI Pain Questionnaire (f. = -0.31, P < 0.01; see Table II), the Interference scale from the WHYMPI (I = --(1.29, t” c( 0.05). the Outdoor Activity scale from the WHYMPI (r = 0.26, P < 0.05) and the Affective Distress scale from the PBCL (r = -0.28. P < 0.05). Marlowe-

TABLE

I

TABLE

II

CORRELATION BETWEEN AND PAIN MEASURES

AGE.

Pain measure

Age

-

Pain severity McGill pain questionnaire WHYMPI-I’S

(,..j *

SOCIAL

DESIRABILITY

Social desirability ._.~ --

.i *

O.Oh

- 0.07 0.13

Interference ~I~YMPI-lnterferencc Activities away from home Social activilies Outdoor activities Ilousehold activities Pain behavior> Distorted arnbul~itjon Facial expressions Affective d&treks Seeking help

0.70 * 0.12 0. I5 0.x + 0.03

-- 0.15 ‘^I0. I I O.Ijf, 0.23 - 0.01~

-~.I),iP -~0. 19 -. ().‘X :**

0.0x ci..w *-* _ 0.43 %*

-- 0. I fl

-- I). I2 --. _-

c I’ c 0.05: * ,b P c O.()l,

Crowne scores correlated significantly with 2 of the PBCL subscales, Facial Expressions (P = - 0.30, P < 0.01) and Affective Distress (r = -0.43, 1’ < 0.01). The first pain construct examined was disability and medication use. A l-way ANCOVA, using age and MC as covariates, showed no difference between the groups of depressed and non-depressed pain patients on duration of the pain. Chi-square analyses showed no group differences in the likelihood of receiving pain medications, anti-depressant medications, whether the patient was receiving compensation for pain, whether the patient had received surgery for the pain complaint or employment status iail Ps < 0.01). These results arc presented in Table III. A l-way MANCOVA on the pain intensity measures yielded a significant effect of group (F (2. 04) = 0.23. P < 0.01: see Table IV). Depression diagnosis accounted for 14% of the variance in pain intensity

DESCRIPTIVE STATISTICS FOR DEPRESSED AND NON-DEPRESSED PATIENTS ON BACKGROUND VARIABLES TABLE Vnriahle

Age * Mean S.D. Education Mean S.W. Gender (52 male) Marital status (c: married) Type of pain (r’r 1 Low back pain Neck/shoulder Other muscul~sekeltai Other pain

III

Depressed

Non-depressed

N = 37

N = 32

44.3 11.4

55.0 13.x

12.3 ‘.O

11.x 2.4 8X

Duration (years) Mean SD. Using pain medication

hfl

Using anti-depressant

37

Surgery for pain

81

DESCRIPTIVE STATISTICS FOR DEPRESSED AND NON-DEPRESSED PATIENTS ON MEASURES OF DISABILITY AND PREDICATION USE --Measure

N = 31

medication 0 IY

75

Depressed

(3)

complaint(S)

Receiving compensation for pain (%) E,mployment

status (Q)

)

N = 31 _~

9.5

x2 7.6 75.0

13.3

21.0

59.5

34.3

61.1 if>.?

4O.h

?,‘) 7.11 ic?

______ Non-depressed

75 .o

181 TABLE IV MEANS AND STANDARD DEVIATIONS FOR DEPRESSED AND NON-DEPRESSED SITY, INTERFERENCE AND PAIN BEHAVIORS construct

Depressed

Non-depressed

N = 37

N = 32

Intensity

McGill Pain Questionnaire Mean S.D, WHYMPI-PS Mean S.D.

0.47 0.20

0.27 0.18

4.41 0.86

4.29 1.09

hrerference

WHYMPI-Interference Mean SD. Activities away from home Mean S.D. Social actities Mean

S.D. Outdoor activities Mean SD. Household activities Mean S.D.

4.80 0.90

3.64

1.91 1.10

2.62 1.11

1.97

2.60

1.11

1.27

0.95 1.31

1.86 1.63

2.65 1.20

3.26 1.46

F

df

Eta ’

6.23 ** 7.04 ***

2,64 365

0.14

1.84

365

4.25 ** 6.96 **

5.61 3,65

3.96 *

3,65

1.65

3.65

3.02 *

3,65

1.42

3.65

5.86 ***

4,62 3,65

4.44 **

3.65

13.42 ***

3,65

* P < 0.05; **p the cvidencc support a relationship?. Psychol. Bull., Y7 (IYX5) 1x-34. 43 Rucker, L.. Frye, B. and Cygan, R., Feasibility and usefulh~ess ot depression screening in medical outpatients, Arch. Intern. Med.. 136 (IYXh) 729-731. 44 Rudy, T.. Kerns, R. and Turk. D., Chronic pain and depression: toward a cognitive-behavioral mediation model. Pain. 3.S (1088) 129-130. 45 Schulberg, II.. Saul. M., McClelland, M., Ganguli, M.. Christy. W. and Frank, R.. Assessing depression in primary medical and psychiatric practices. Arch. Gen. Psychiat.. 42 (1085) 1164-I 170. 46 Spitzer, R.. Endicott, J. and Robins, E., Research diagnostic criteria: rationale and reliability. Arch. Cen. Psychiat., 35 (1978) 773-782. 47 Sternbach, R.A., Pain Patients: Traits and Treatment. Academic Press, New York, lY74. 4X Timmermans. G. and Sternbach, R.A., Human chronic pain and personality: a canonical correlation analysis. In: J.J. Bonica and D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy. Raven Press, New York. 1976, pp. 307-310. 3Y Turner, J. and Romano. J., Self-report screening measures for depression in chronic pain patients, J. Clin. Psychol.. 40 (IYX4) YOO-Y13. 50 Ward, N., Bloom, V.. Dworkin. S., Fawcett. J.. Narisimhachari. N. and Friedel, R.. Psychobiological markers in coexisting pain and depression: toward a unified theory, J. C‘lin. Psychiat.. 43 t 1982) 32-j’).

Depression and the chronic pain experience.

The present study examined the relationship between depression and a constellation of pain-related variables that describe the experience of chronic p...
1MB Sizes 0 Downloads 0 Views