Personality disorders and response to outpatient ... - Springer Link

8 downloads 0 Views 1MB Size Report
Timothy R. Elliott, 1 Warren T. Jackson, 1,3 Molly Layfield, 2 and. Debra Kendall 2 ... Elliott, Jackson, Layfield, and Kendall ... (1993) found that 60% of their sam-.
Journal of Clinical Psychology in Medical Settings, VoL 3, No. 3, 1996

Personality Disorders and Response to Outpatient Treatment of Chronic Pain Timothy R. Elliott, 1 Warren T. Jackson, 1,3 Molly Layfield, 2 and Debra Kendall 2

As part of a comprehensive interdisciplinary evaluation conducted prior to participation in an outpatient chronic pain treatment program, the psychological status of 101 persons was assessed. The majority of participants was found to have a form of personality disorder, determined by conservative cutoff scores applied to their Millon Clinical Multiaxial Inventory (MCMI) profiles. DSM-III-R Cluster C disorders (Le., Avoidan~ Dependent, Obsessive-Compulsive, and Passive-Aggressive) were overrepresented in this sample. Subsequent ana~ses revealed that personality disorders were related to higher levels of self-reported distress and pain at both the beginning and the end of outpatient treatment. Differential responses to treatment were observed on self-report measures; however, few relations were found between personality disorder and physical therapist ratings of impairment and improvement. Implications for the assessment of personality disorders in outpatient pain treatment programs are discussed and appropriate intervention strategies are considered. KEY WORDS: chronic pain; health psychology; Millon Clinical Multiaxial Inventory; personality disorder; ambulatory care.

INTRODUCTION

Personality mechanisms have been recognized for some time as important factors in the development and maintenance of chronic pain 1Department of Physical Medicine and Rehabilitation, University of Alabama at Birmingham, Birmingham, Alabama 35233. 2Richmond, Virginia. 3presently with the Department of Psychiatry and Behavioral Neurobiology, University of

Alabama at Birmingham,Birmingham,Alabama. 219 I068-9583/96/09~-0219509~0~©

1996Plenum PublishingCorporation

220

Elliott, Jackson, Layfield, and Kendall

syndromes. Generally, research has focused on the correlates of commonplace individual difference variables either from a trait perspective [e.g., neuroticism, extroversion (Harkins, Price, & Braith, 1989; Wade, Dougherty, Hart, & Rafii, 1992)] or from contemporary social learning theories [e.g., locus of control (Wallston, Stein, & Smith, 1994)]. Clinicians have been inclined to infer personality tendencies from instruments that lack linkage with theoretical models of personality [e.g., MMPI (Armentrout, Moore, Parker, Hewitt, & Feltz, 1982; Fordyce, 1976)]. Unfortunately, few studies have examined personality disorders among persons seeking treatment for chronic pain syndromes. In the updated DSM-IV (1994), a personality disorder is defined as "an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individual's culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time, and leads to distress or impairment" (p. 629). This definition is consistent with the definition found in DSM-III-R (1987, p. 335); however, the concept of deviation from cultural expectations is more strongly emphasized in the later version. Overall, the definition of personality disorder translates quite well from DSM-III-R to DSM-IV. For the purposes of this paper, personality disorder is discussed according to the DSM-III-R framework that groups 11 personality disorders into three clusters. Cluster A includes Paranoid, Schizoid, and Schizotypal Personality Disorders. People with Cluster A disorders often appear "odd or eccentric." Cluster B personality disorders are characterized by "dramatic, emotional, or erratic" behavior. Antisocial, Borderline, Histrionic, and Narcissistic Personality Disorders comprise Cluster B. Finally, Cluster C consists of Avoidant, Dependent, Obsessive-Compulsive, and Passive-Aggressive Personality Disorders. People with Cluster C disorders often appear "anxious or fearful." Available research suggests that a significant number of persons seeking treatment for chronic pain syndromes may in fact meet DSM diagnostic criteria for a personality disorder. In one of the few descriptive studies utilizing rigorous criteria, Kinney et al. (1993) found that 60% of their sample of persons with chronic pain met the diagnostic criteria for a personality disorder. The most prevalent characterological conditions were Paranoid, Passive-Aggressive, Avoidant, and Borderline personality disorders. In contrast, only 21% of patients with acute low back pain were found to have a personality disorder. Typically, personality disorders are considered to predate the onset of injury and complicate the course of a pain syndrome (Katon, Egan, & Miller, 1985; Lustman, Velozo, Eubanks, Montag, & Cole, 1991; Polatin, Kinney, Gatchel, Lillo, & Mayer, 1993). Personality disorders can potentially influence patients' responses to chronic pain rehabilitation. For example, Large (1986) noted that a signifi-

Personality Disorders and Pain

221

cant number of persons with chronic pain syndromes met criteria for Dependent and Obsessive-Compulsive personality disorders. These persons likely have a more fearful, anxious style in coping with their pain condition, and thus may be more likely to develop debilitating pain behaviors than those with other conditions, perhaps due to the cycle of avoidance-inactivity-increased pain. Prospective study has found that persons with anxious, fearful personality characteristics are more likely to develop pain-related disability than those without these tendencies (Klenerman et al., 1995). In contrast, persons with Antisocial features may be more impulsive, erratic, and aggressive. These individuals may demonstrate problems adhering to therapeutic regimens and treatment protocols. Personality disorders have been linked with poor vocational outcomes among persons with chronic back pain (Gatchel, Polatin, & Kinney, 1995), but this relationship has not been confirmed in other research (Gatchel, Polatin, Mayer, & Garcy, 1994). It should also be noted that other evidence suggests that personality disorders may be unrelated to surgical outcomes for chronic low back pain (Herron, Turner, & Weiner, 1986). The present study was conducted to assess the presence of personality disorders among persons under consideration for admission to an outpatient chronic pain treatment program. Personality disorders were assessed using the Millon Clinical Multiaxial Inventory (MCMI; Millon, 1983), a popular self-report questionnaire that has clear theoretical linkage with a contemporary model of personality (Millon & Klerman, 1986) and with the DSM diagnostic nomenclature for defining personality disorders. MCMI responses were compared with measures of adjustment and functional status at the admission evaluation and with available measures of treatment response among those who participated in the outpatient program.

METHOD Participants The sample consisted of 101 people (50 women, 51 men) referred for psychological evaluation prior to admission to a CARF-accredited outpatient chronic pain treatment program. All participants completed a series of self-report questionnaires as a part of their evaluation. Of this number, 81 were also evaluated by a physical therapist (licensed by the state board of medicine) at the clinic. The sample was predominately Caucasian (72.6%) and the average age was 40.2 years (SD = 8.9 years). A boardcertified physiatrist diagnosed 45 persons with chronic low back pain, 21 with chronic cervical pain (e.g., cervical disc damage), 8 with chronic pain

222

Elliott, Jackson, Layfield, and Kendall

in an upper extremity (e.g., reflex sympathetic dystrophy), 8 with chronic lower extremity pain (e.g., chronic knee pain), and 14 persons with a chronic pain condition that involved both upper and lower extremity (e.g., fibromyalgia). Five participants had unspecified pain conditions, based on chart records. Each patient participated in a comprehensive medical examination that included diagnostic x-rays, EMGs, MRIs, assessment of current physical capabilities for activities of daily living, enumeration of prescribed medications, and a complete medical history. Individuals admitted for treatment participated in a structured, 5-dayper-week outpatient program. The outpatients participated in a daily regimen of physical therapy, group therapy, biofeedback, exercise, and instruction in behavior modification and stress management. The average length of stay in the program was 17.8 days. At the time of discharge, participants rated their levels of emotional distress and pain and the physical therapist reevaluated their functional mobility. Materials

Measurement of Personality Disorders. The Millon Clinical Multiaxial Inventory (MCMI; Millon, 1983) was used to assess personality disorders among the participants. This instrument was available to the clinic at the time data collection was initiated (March 1990) and thus was selected as a matter of convenience. The MCMI has 175 items that are answered in a true-false format. It was developed to assess personality disorders according to the DSM-III/-III-R diagnostic framework and it is considered appropriate for use in rehabilitation with certain caveats (Elliott & Umlauf, 1995). The diagnostic scales on the MCMI are interpreted according to "base rates" that parallel clinical prevalence rates for each disorder. The personality scales are also grounded in Millon's explicit personality theory, which delineates characterological styles in terms of biological predispositions and learning experiences (MiUon & Klerman, 1986). The MCMI has 11 scales that coincide with DSM-III-R personality disorder categories: Schizoid, Avoidant, Dependent, Histrionic, Narcissistic, Antisocial, Compulsive, Passive-Aggressive, Schizotypal, Borderline, and Paranoid. These scales assess both long-standing features and current presentation of each dimension of the respondent's personality. An additional nine scales on the MCMI assess clinical syndromes that parallel DSM Axis I disorders: Anxiety, Somatoform, Hypomania, Dysthymia, Alcohol Abuse, Drug Abuse, Psychotic Thinking, Psychotic Depression, and Psychotic Delusion. The Personality Disorder scales on the MCMI have received considerable attention. The reliability and validity of the Avoidant, Dependent,

Personality Disorders and Pain

223

Schizotypal, Histrionic, Borderline, Narcissistic, and Paranoid Scales have been particularly well-supported (Torgensen & Alnaes, 1990). The MCMI may be especially adept at classifying Cluster C disorders [Avoidant, Dependent, Obsessive-Compulsive, and Passive-Aggressive (Wetzler, 1990)]. However, individuals with no known history of aberrant behavioral patterns may evidence clinical elevations on scales assessing Cluster C (Reich & Troughton, 1988), and the MCMI may "overdiagnose" individuals with no known behavioral disturbance when the recommended cut-off score of 74 is applied (Piersma, 1987). Critics of the MCMI have noted a low test/retest reliability for some of the personality disorder scales (Piersma, 1987), lack of consistent connection with DSM criteria for characterological disorders (Widiger, 1985), and a relatively high false positive rate for the MCMI compared to structured interview systems and other measures (Reich, 1987; Repko & Cooper, 1985; Wetzler, 1990). Following the more conservative recommendations of Hsu and Maruish (1992, p. 15), we used a cutoff base rate of 84 to determine a person's "most prominent" characterological style and to classify participants into separate personality clusters. Thus, the highest scale score determined the personality disorder and diagnostic cluster of each participant. Individuals who had no score greater than 84 were classified as "normal." Finally, MCMI base rate scores for each personality disorder scale were used in correlational analyses to maximize interpretation of our data via comparison with other continuous variables. Beck Depression Inventory. The Beck Depression Inventory (BDI, Beck & Steer, 1987) is a frequently used measure of depressive behaviors. It consists of 21 items that are rated on a 0 (no symptom) to 3 (extreme form of symptom) scale. Substantive data indicate that the BDI has an adequate internal consistency (.84) and its correlations with other depression measures range from .54 to .68 (Tanaka-Matsumi & Kameoka, 1986). The BDI is a sensitive index of depression severity and it appears to load heavily on cognitive and affective symptoms among patients in primary care settings (Brown, Schulberg, & Madonia, 1995). Total scores on the BDI at the time of admission evaluation (BDI 1) and upon program completion (BDI 2) were used as criterion variables. A depressive symptom change score (BDI 1-BDI 2 = BDIA) was also computed to serve as an indicator of response to rehabilitation for use as a criterion variable. Thus, a higher BDIA value reflects greater improvement (i.e., reduction) in distress over the course of the treatment program. Pain Ratings. Participants were asked to rate their lowest, average, and worst pain levels at admission and upon discharge using separate 7point Likert-type scales (0 = lowest intensity, 6 = highest intensity). The average score across these three ratings was computed to obtain an admis-

224

Elliott, Jackson, Layfield, and Kendall

sion pain score (Pain 1) and a discharge pain score (Pain 2). Higher scores indicate higher pain levels. A pain level change score (PainA) was completed by subtracting the discharge score from the admission score. Thus, a higher score on the PainA variable indicates greater improvement (i.e., reduction) in pain over the course of the rehabilitation program. Functional Evaluations. The staff physical therapist rated participants' range of motion (ROM) for their cervical, trunk, upper extremity, and lower extremity regions both at the preadmission evaluation and upon program completion. Higher scores for each region indicate a higher percentage of functional ROM. Days in Treatment. The total number of days during which each participant attended the outpatient program was determined from clinic records.

RESULTS Table I presents means and standard deviations of the demographic data, self-report variables, physical therapists ratings, and outcome variables. No significant differences were found between men and women on the MCMI subscale scores. Table II displays high scale categories and the distribution of men and women throughout the personality clusters. Correlations used in subsequent analyses are presented in Table III. Participant attrition, unavailable records, and admission denials resulted in an unequal distribution of participants across variables. Of the 99 persons who completed the MCMI, 93 persons also completed the BDI during the admission evaluation and 58 completed the BDI at discharge. Pain ratings were available for 81 participants at the time of admission; 53 completed pain ratings at discharge. Data on total days in treatment were available for 90 participants. The number of participants who completed the physical therapy admission evaluation ranged from 85 (trunk) to 91 (cervical); upon discharge assessment, the number ranged from 49 (trunk) to 57 (upper body/lower body).

Personality Disorders Of the 101 persons referred for evaluation, 99 had valid profiles suitable for analysis. Using a conservative cutoff base rate of 84, 34 persons evidenced a profile indicative of normal, nonpathological personality functioning. However, the remainder of the sample (66%) evidenced a high-point scale above the cutoff, indicative of a personality disorder. As

Personality Disorders and Pain

225

Table I. Means and Standard Deviations of Participant Data by Gender

Male

Female

Mean

SD

Mean

SD

Age (yr) Education level (yr)

38.50 10.20

7.99 2.39

41.84 12.I0

9.35 1.97

BDI 1 Pain 1

18.02 4.23

8.41 0.62

1920 4.21

8.90 0.72

PT ratings (admission) Cervical Trunk Upper body Lower body

79.24 20.02 97.52 94.84

15.88 23.03 5.42 7.41

80.39 31.74 94.96 96.69

22.37 32.11 8.41 6.12

BDI 2 Pain 2

14.39 3.74

10.78 0.74

14.32 3.48

9.87 0.84

PT rating (discharge) Cervical Trunk Upper body Lower body

89.96 24.00 99.90 99.45

11.74 27.50 0.41 1.59

90.96 29.71 99,43 100.00

12.18 30.59 2.50 0.00

Total days in treatment

17.77

13.09

17.79

12.51

MCMI Personality Disorder scales Schizoid Avoidant Dependent Histrionic Narcissistic Antisocial Compulsive Passive-Aggressive Schizotypal Borderline Paranoid

53.09 54.09 64.42 54.76 59.64 60.56 58.11 51.62 52.76 61.16 66.11

28.49 28.11 25.50 21.85 20.98 20.63 17.65 30.48 17,44 i6.16 14.05

53.72 53.16 63.70 53.20 60.76 58.30 65.38 52.06 51.74 64.96 70.36

21.88 23.75 26.72 26.13 21.31 17.40 17.1 t 30.07 12.48 18.32 15.26

MCMI Axis I scales Anxiety Somatoform Hypomania Dysthymia Alcohol Abuse Drug Abuse Psychotic Thinking Psychotic Depression Psychotic Delusions

81.18 68.07 39,11 73.07 51.82 59.69 56.33 53.71 58.64

19.96 14.63 29.56 20.96 19.91 22.08 14.90 16.40 18.22

88.28 75.74 40.58 78.72 51.08 50.52 55.06 54.56 64.38

22.63 15.42 28.78 26.07 18.39 21.25 16.65 18.48 14.09

depicted in Table II, 35% of these cases were classified as Cluster C disorders: 21 people had pronounced elevations on the MCMI Dependent

Elliott, Jackson, Layfield, and Kendall

226

Table II. Distribution of Personality Disorders by Cluster and Gender Men

Women

17

17

DSM Cluster A: Odd, Eccentric Paranoid Schizoid Schizotypal Total Cluster A

1 4 0 5

5 1 0 6

DSM Cluster B: Dramatic, Emotional, Erratic Antisocial Borderline Histrionic Narcissistic Total Cluster B

2 1 3 3 9

0 5 2 3 10

1 9 1 4 15

1 12 6 1 20

Normal profiles

DSM Cluster C: Anxious, Fearful Avoidant Dependent Obsessive-Compulsive Passive-Aggressive Total Cluster C

scale and 7 had elevations on the Compulsive scale. This distribution was significantly different from that expected by chance [22(3) = 19.82, p < .001]. Cluster B disorders had the second-highest frequency (19%). Six of these persons had peak elevations on the Narcissistic scale. Another six persons had peak scores on the Borderline scale. Relations Between Personality Disorders and Admission Variables A one-way A N O V A revealed a significant difference between personality clusters for depression scores at the initial evaluation [F(3, 89) = 3.71, p < .05]. Cluster C had a higher average BDI score (M = 22.49) than the Cluster B (M = 15.33) and nonpathological (M = 17.03) groups. Regression analysis using the base rate scores for each Cluster C variable revealed that the Passive-Aggressive scale was the best single predictor of admission BDI scores [F(1, 91) = 95.67, 13 = .72, R 2 = .51]. Higher scores on the PassiveAggressive scale were significantly associated with higher BDI scores at the admission evaluation. The stepwise forward entry procedure selected the Avoidant scale as the next best predictor of BDI scores at admission [Fi, e(1,90) = 6.21, 13 = .23, R2mc = .03]. Higher Avoidant scale scores were

.38** .27" -.05

-.04

-.03 .06 -.12 -.02

-.14 -.02 -.01 .02

BDI 1 BDI 2 BDIA

Days in treatment

PT ratings (admission) Cerv ROM LBody ROM UBody ROM Trunk ROM

PT ratings (discharge) Cerv ROM LBody ROM UBody ROM Trunk ROM

-.13 .07 .24 .27

-.06 .00 -.12 .14

-.07

.60** .42** .05

.01 -.12 .24

M2

-.24 -.01 .12 .17

-.11 -.08 -.14 .12

-.05

.40** .21 .09

-.05 -.07 .09

M3

.18 -.02 .02 -.18

-.04 -.04 .01 -.08

.02

-.24* -.25 .00

-.07 -.04 -.08

M4

.16 .03 -.17 -.08

-.01 .06 .10 .00

-.06

-.39** -.29* -.12

-.04 .30* -.41"*

M5

.14 .01 -.08 .07

.01 .01 .00 .03

-.10

-.16 -.04 -.18

.14 .11 -.07

M6

-.05 -.09 -.35* -.07

.00 .12 -.03 .12

.02

-.44** -.25 -.18

-,09 .25 -.28*

M7

-.16 .09 .24 .28

-.13 -.08 -.06 .10

-.04

.72** .32* .33**

.05 -.18 .19

M8

-.10 .06 .08 .17

.IX) -.02 -.12 .08

-.13

.34** .40** -.09

.00 -.14 .15

Ms

-.04 .09 .05 .03

-.09 -.02 -.01 .01

-.28**

-.13 .01 -.t9

.00 .06 -.12

Mp

.01 .08 .35* .11

-.06 -.11 -.10 .10

-.26*

.57** .37** .17

.05 -.17 .21

Mc

aROM, range of motion; Cerv, cervical; Lbody, lower body; Ubody, upper body; M1, Schizoid; M E, Avoidant; M3, Dependent; M4, Histrionic; Ms, Narcissistic; M6, Antisocial; MT, Compulsive; Ms, Passive-aggressive; M s, Schizotypal; Mr, Borderline; Mp, Paranoid; *p < .05. **p < .01.

-.02 .03 -.03

Pain 1 Pain 2 PainA

M1

Table Ill. Correlations Among MCMI Personality Disorder Scales and Criterion Variablesa

Criterion variable

ItO It~ ",.,I

g.

an

O

228

Elliott, Jackson, Layfield, and Kendall

associated with higher BDI scores after taking the Passive-Aggressive scale into account. A one-way A N O V A revealed an association between the personality disorder clusters and the average pain ratings at the initial evaluation, [F(3, 77) = 2.22, p -- .09]. Post hoc tests revealed that persons with Cluster B personality disorders had lower pain ratings (M = 3.79) than Cluster C (M = 4.28) and the group with nonpathological profiles (M = 4.32). A series of one-way ANOVAs revealed no significant relations between personality clusters and the physical therapist R O M ratings (all F's < 1.1). Similarly, regression equations found no MCMI personality disorder scale to be significantly predictive of any admission R O M variable. Relations Between Personality Disorders and Discharge Variables A one-way A N O V A revealed no relation between the personality clusters and BDI scores at discharge IF(3, 54) < 1]. A stepwise regression equation using a forward entry technique found that the Avoidant scale was significantly predictive of BDI scores at discharge [F(1, 56) = 12.11, [3 = .42, R e = 18]. Higher Avoidant scale scores at the initial evaluation were significantly associated with higher BDI scores at the completion of the program. Another one-way A N O V A revealed no effect for personality disorder cluster on pain ratings at discharge IF(3, 49) = 1.55, ns]. However, regression analysis-employing a stepwise procedure with a forward entry technique--revealed that the Narcissism scale was significantly predictive of discharge pain ratings [F(1, 51) = 5.08, 13 = .30, R z = .09]. Higher Narcissism scale scores were associated with higher pain ratings at discharge. The Histrionic scale significantly augmented the equation at the second step [Fine(l, 50) = 4.94, 13 = -.36, RZinc = .08]. Unlike the Narcissism scale, higher scores on the Histrionic scale were significantly associated with lower pain ratings at discharge. One-way A N O V A indicated that personality disorder cluster did not have a significant effect on the total number of days in treatment IF(3, 86) < 1, ns]. However, inspection of the means in Fig. 1 revealed that individuals with Cluster B personality disorders averaged approximately 5 fewer days in outpatient treatment than those with no characterological problems. Correlational analyses revealed that the Borderline and Paranoid scale scores had the highest correlations with days in treatment (-.26 and -.28, respectively); therefore, these two variables were entered as a block in a regression equation to predict days in treatment. These variables accounted for 12% of the variance in the total number of days treated by the clinic [F(2, 87) = 6.17, p < .01]. The beta coefficients for the Borderline and Paranoid variables indicated that higher scores on these scales were significantly associated with fewer days in treatment (13 = -.22 and -.24, respectively).

229

Personality Disorders and Pain

Additional one-way ANOVAs found no significant relations between the personality disorder clusters and the discharge ROM for trunk, lower body, or upper body. A relation was found between cluster and the discharge cervical ROM variable IF(3, 46) = 2.47, p = .07]. Post hoc analysis indicated that the group with no personality disorder elevation had a significantly greater cervical ROM at discharge than the Cluster C group (M = 93.9 and 85.95, respectively). Those individuals with fearful, anxious personality features displayed less cervical ROM at discharge. Regression analysis using the forward-entry stepwise procedure, however, found no single personality disorder scale to be predictive of cervical, lower body, or trunk ROM at discharge. Compulsive scale scores were selected as the best single predictor of upper body ROM at discharge [F(1, 51) = 7.08, 13 = -.35, R 2 = .12]. Higher Compulsive scale scores were associated with less upper body ROM at discharge.

Personality Disorders and Response to Rehabilitation One-way A N O V A revealed no significant differences between personality disorder clusters for the BDIA variable [F(3, 53) < 1, ns]. Regression analysis using a forward entry technique indicated that the Passive-Aggressive scale was the best single predictor of improvement on the

2O 19 i .IIII.Z"

ili:i ~

15

c~

14 :i~iiiiiiii:ii::iii::!:Z::!iiii!:!i:

~2 11

i iiiiiiiiii

lO 9 No Personality Disorder N:31

Cluster A N=10

Cluster B N=14

Cluster C N=35

Fig. 1. Average days in outpatient pain treatment by personality disorder cluster.

230

Elliott, Jackson, Layfleld, and Kendall

BDI IF(l, 55) = 6.61,p < .05, 13 = 0.33, R 2 = .11]. Higher Passive-Aggressive scores were significantly associated with greater improvement on the BDI between initial evaluation and discharge from the program. As reported above, however, persons with higher Passive-Aggressive scale scores had higher BDI scores (more depressive symptoms) upon admission. For the Pain~ variable, one-way ANOVA showed a significant effect for personality disorder cluster IF(3, 46) = 3.14, p < .05]. Post hoc analyses using the least square means procedure revealed that Cluster B scores (M = -0.33) were significantly lower than Cluster A (M = 1.17), Cluster C (M = 0.59), and the group with normal profiles (M = 1.17). Persons in Cluster B, then, exhibited less improvement on the pain ratings than other participants. The means of the PainA variable for these groups suggest that those in Cluster B had a slight increase in pain over the course of the rehabilitation program. Stepwise regression analysis using the forward entry technique indicated that the Narcissism scale was the single best predictor of PainA [F(1, 48) = 9.63, 13 = -.41, R 2 = .17]. Higher Narcissism scale scores were significantly associated with less improvement in pain ratings after treatment.

DISCUSSION These data provide intriguing evidence that personality disorders may be overrepresented among persons seeking treatment in outpatient chronic pain clinics and these patterns of behavior may have ramifications for clinical assessment and treatment strategies. Notably, the rate of personality disorders in the present sample converges with previous research. Sixty-six percent of the persons in our sample had peak scores indicative of a personality disorder of some type. Prior work using diagnostic interview systems have found these rates to vary from 40% to 60% (Fishbain, Goldberg, Meagher, Steele, & Rosomoff, 1986; Gatchel, Polatin, Mayer, & Garcy, 1994; Kinney et al., 1993; Large, 1986; Polatin et al., 1993). Consistent with the Fishbain et al. and Large studies, we observed a higher rate of persons exhibiting features of a fearful, anxious personality disorder congruent with Cluster C (Avoidant, Dependent, Obsessive-Compulsive, and Passive-Aggressive Personality Disorders). Accumulating evidence from prospective studies suggests that persons who have a greater proclivity for apprehension, fear, and anxiety may be more susceptible to developing pain avoidance behaviors in reaction to a painful condition (Klenerman et al., 1995; Polatin et al., 1993). These individuals may likely experience intense fear and anxiety in response to painful stimuli and respond by emitting withdrawal and other disabling be-

Personality Disorders and Pain

231

haviors as they try to avoid activities that might elicit or exacerbate pain (Slade, Troup, & Lethem, 1983). Furthermore, persons with dependency characteristics may be susceptible to seeking secondary gains following onset of a pain syndrome. These persons may find the attention given to them by medical, insurance, and health-related professions to be reinforcing; similarly, they may also attract more solicitous and reinforcing behavior from caring family members, friends, and associates. As the chronic pain syndrome develops, these persons may then find successful treatment and the prospect of losing support threatening. Consequently, our data suggest that these persons may display more psychological distress upon admission for chronic pain treatment, as reflected in the BDI scores at the initial evaluation. Other research indicates that persons with Cluster C disorders may be prone to depression, generally, and higher rates of these disorders have appeared in samples of depressed psychiatric inpatients (Libb et aL, 1990; Mezzich, Fabrega, & Coffman, 1987). Heightened distress does not necessarily connote a greater subjective pain experience or physical dysfunction. We found no effects for these personality dimensions on pain ratings and physical therapist ratings at the initial evaluation. However, we did find that an element of the Cluster C domain--Passive-Aggressive features--was significantly associated with a greater reduction in distress over the course of treatment. It may be possible that the pain management program was most effective in alleviating the psychological distress experienced by these individuals. We do not know if these gains were maintained over time. Interestingly, we found significant correlates with Cluster B dramatic, emotional, or erratic behavioral tendencies (e.g., Antisocial, Borderline, Histrionic, and Narcissistic Personality Disorders) and indices of adjustment. Higher Narcissism scores were associated with greater distress and pain at discharge, and with less pain reduction at discharge. The Cluster B group also had the fewest days in treatment, although this difference was not statistically significant. Elevations on the Narcissism scale are indicative of an arrogant, exploitative, presumptuous and entitled style that might complicate response to a treatment regimen that emphasizes personal responsibility, education, and active coping. Thus, individuals who display marked elevations on this scale may report fewer subjective gains over the course of treatment, possibly in an attempt to preserve their sense of uniqueness or status. This might also reflect an attempt to sabotage or discount the treatment program or clinic personnel for presumably refusing to cater to their sense of entitlement to special favors. Few associations were found between Cluster A odd or eccentric tendencies (e.g., Paranoid, Schizoid, and Schizotypal Personality Disorders)

232

EUiott, Jackson, Layfleld, and Kendall

and criterion variables. Few persons in our sample had personality disorders of this nature. This group had the second fewest days in treatment, and the Paranoid scale was the best overall predictor of days in treatment. Elevations on this scale are suggestive of an outstanding mistrust of others, a fear of losing self-determination, and a readiness to perceive influence, conspiracy, and betrayal. Thus, persons with higher scores on this scale may be irascible, disparaging, and troubled in their interpersonal relationships over time (Millon, 1984). Individuals with elevations on this scale may have difficulty interacting with others in the course of the outpatient program, particularly if group activities are expected and the structure of the program allows few opportunities for self-directed activities. These persons probably harbor suspicions about clinic personnel and eventually provoke conflict and reasons for opting out of the program earlier than anticipated. These dynamics are consistent with Cluster A behavioral patterns and may occasionally accompany other clinical syndromes (e.g., Borderline, Antisocial disorders). Elevations on this scale, then, may warrant particular attention from the treatment planning team. In summary, health care providers working in the area of chronic pain treatment must develop an understanding of the maladaptive personality variables that they will frequently encounter. Personality disorders among chronic pain patients can be addressed through careful admission evaluation and treatment planning that includes cognitive-behavioral approaches to pain management. The MCMI appears to be a useful measure in chronic pain assessment and treatment planning. Further work is needed to determine to what extent the MCMI is "reactive" to external variables such as mode of treatment or changes in psychosocial stress levels. MCMI administration at discharge in addition to admission may lend valuable insights into the complex relation between maladaptive personality variables and chronic pain. Certain caveats are advised in the interpretation of these data. The MCMI has been criticized for "overdiagnosing" personality disorders (e.g., Wetzler, 1990); although we employed a more conservative cutoff score, this may not have been a sufficient safeguard. Individuals using the updated version--either the MCMI-II or the MCCMI-III--may find a different distribution of characterological problems in this population. Furthermore, there is evidence to suggest that responses to the MCMI may be adversely affected by depressed states and other moods (e.g., Overholser, Kabakoff & Norman, 1990). This overlap may have contributed to significant relation between several of the MCMI scales and the BDI. Future research might employ other measures of personality disorders to study their correlates among patients with chronic pain conditions.

Personality Disorders and Pain

233

REFERENCES American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Armentrout, D. P., Moore, J. E., Parker, J. C., Hewitt, J. E., & Feltz, C. (1982). Pain-patient MMPI subgroups: The psychological dimensions of pain. Journal of Behavioral Medicine, 5, 201-211. Beck, A. T., & Steer, R. A. (1987). Beck Depression Inventory manual. San Antonio, TX: The Psychological Corporation. Brown, C., Schulberg, H. C., & Madonia, M. J. (1995). Assessing depression in primary care practice with the Beck Depression Inventory and the Hamilton Rating Scale for depression. Psychological Assessment, 7, 59-65. Elliott, T., & Umlauf, R. (1995). Measurement of personality and psychopathology in acquired disability. In L. Cushman & M. Scherer (Eds.), Psychological assessment in medical rehabilitation settings (pp. 325-358). Washington, DC: American Psychological Association. Fishbain, D. A., Goldberg, M., Meagher, B. R., Steele, R., & Rosomoff, H. (1986). Male and female chronic pain patients categorized by DSM-III psychiatric diagnostic criteria. Pain, 26, 181-197. Fordyce, W. E. (1976). Behavioral methods for chronic pain and illness. St. Louis: C. V. Mosby. Gatchel, R. J., Polatin, P. B., Mayer, T. G., & Garcy, P. D. (1994). Psychopathology and the rehabilitation of patients with chronic low back pain disability. Archives of Physical Medicine and Rehabilitation, 75, 666-670. Gatchel, R. J., Polatin, P. B., & Kinney, R. K. (1995). Predicting outcome of chronic back pain using clinical predictors of psychopathology: A prospective analysis. Health Psychology, 14, 415-420. Harkins, S. W., Price, D. D., & Braith, J. (1989). Effects of extroversion and neuroticism on experimental pain, clinical pain, and illness behavior. Pain, 36, 209-218. Herron, L., Turner, J. & Weiner, P. (1986). A comparison of the Million Clinical Multiaxial Inventory and the Minnesota Multiphasic Personality Inventory as predictors of successful treatment by lumbar laminectomy. Clinical Orthopaedics and Related Research, Feb. (203), 232-238. Hsu, L. M., & Maruish, M. E. (1992). Conducting publishable research with the MCMI-II: Psychometric and statistical issues. Minneapolis, MN: National Computer Systems. Katon, W., Egan, K., & Miller, D. (1985). Chronic pain: Lifetime psychiatric diagnoses and family history. American Journal of Psychiatry, 142, 1156-1160. Kinney, R. K., Gatchel, R. J., Polatin, P. B., Fogarty, W. T., & Mayer, J. G. (1993). Prevalence of psychopathology in acute and chronic low back pain patients. Journal of Occupational Rehabilitation, 3, 95-103. Klenerman, L., Slade, P. D., Stanley, I. M., Pennie, B., Reilly, J. P., Atchison, L. E., Troup, J. D. G., & Rose, M. J. (1995). The prediction of chronicity in patients with an acute attack of low back pain in a general practice setting. Spine, 20, 478-484. Large, R. G. (1986). DSM-III diagnoses of chronic pain: Confusion or clarity? Journal of Nervous and Mental Disease, 174, 295-303. Libb, J. W., Stankovic, S., Freeman, A., Sokol, R., Switzer, P., & Houck, C. (1990). Personality disorders among depressed outpatients as identified by the MCMI. Journal of Clinical Psycholog}; 46, 277-284. Lustman, P. J., Velozo, C. A., Eubanks, B., Montag, J. A., & Cole, D. M. (1991). Prior psychiatric problems in rehabilitation clients with work-related injuries. Journal of Occupational Rehabilitation, 1, 227-233. Mezzich, J. E., Fabrega, H., & Coffman, G. A. (1987). Multiaxial characterization of depressive patients. Journal of Nervous and Mental Disease, 175, 339-346. Millon, T. (1983). Millon Clinical Multiaxial Inventory Manual, Minneapolis, MN: National Computer Systems.

234

Eiliott, Jackson, Layfieid, and Kendall

Millon, T. (1984). Interpretive guide to the Millon Clinical Multiaxial Inventory. In P. McReynolds & G. J. Chelune (Eds.), Advances in Psychological Assessment, 11"ol.6 (pp. 1-40). San Francisco: Jossey-Bass. Millon, T., & Klerman, G. L. (Eds.) (1986). Contemporary directions in psychopathology: Toward the DSM-IV. New York: Guilford. Overholser, J.C., Kabakoff, R., & Norman, W. (1989). The assessment of personality characteristics in depressed and dependent psychiatric inpatients. Journal of Personality Assessment, 53, 40-50. Piersma, H. (1987). The MCMI as a measure of DSM-III Axis II diagnoses: An empirical comparison. Journal of Clinical Psychology, 43, 478-483. Polatin, B. P., Kinney, R. K., Gatchel, R. J., Lillo, E., & Mayer, T. G. (1993). Psychiatric illness and chronic low back pain: The mind and the spine--Which goes first? Spine, 18, 66-71. Reich, J. (1987). Instruments measuring DSM-III and DSM-III-R personality disorders. Journal of Personality Disorders, 1, 220-240. Reich, J., & Troughton, E. (1988). Comparison of DSM-III personality disorders in recovered depressed and panic disorder patients. Journal of Nervous and Mental Disease, 176, 300-304. Repko, G. R., & Cooper, R. (1985). The diagnoses of personality disorders: A comparison of MMPI profile, Million inventory, and clinical judgment in a worker's compensation population. Journal of Clinical Psychology, 41, 867-881. Slade, P. D., Troup, J. D. G., & Lethem, J. (1983). The fear-avoidance model of exaggerated pain perception. 2. Preliminary studies of coping strategies for pain. Behavioral Research and Therapy, 2I, 409-416. Tanaka-Matsumi, J., & Kameoka, V. A. (1986). Reliabilities and concurrent validates of popular measures of depression, anxiety, and social desirability. Journal of Consulting and Clinical Psychology, 54, 328-333. Torgenson, S., & Atnaes, R. (1990). The relationship between the MCMI personality scales and DSM-III, Axis II. Journal of Personality Assessment, 55, 698-707. Wade, J. B., Dougherty, L. M., Hart, R., & Rafii, A. (1992). A canonical correlation analysis of the influence of neuroticism and extroversion on chronic pain, suffering, and pain behavior. Pain, M, 67-73. Wallston, K. A., Stein, M. J., & Smith, C. (1994). Form C of the MHLOC Scales: A condition specific measure of locus of control. Journal of Personality Assessment, 63, 534-553. Wetzler, S. (1990). The Million Clinical Multiaxial Inventory (MCMI): A review. Journal of

Personality Assessment, 55, 445-464. Widiger, T. (1985). The MCMI as a measure of DSM-III. Journal of Personality Assessment, 49, 336-378.