- Email: [email protected]
Physician judgments of chronic pain patients
Pergamon
Soc. Sci. Med. Vol.45, No. 8, pp. 1199-1205, 1997
PII: S0277-9536(97)00033-6
© 1997 ElsevierScienceLtd. All rights reserved Printed in Great Britain 0277-9536/97 $17.00 + 0.00
PHYSICIAN JUDGMENTS OF CHRONIC PAIN PATIENTS R A Y M O N D C. TAIT* and JOHN T. CHIBNALL Department of Psychiatry and Human Behavior, Saint Louis University School of Medicine, St. Louis, MO 63104, U.S.A. Abstract--Recent evidence has indicated that physician judgments of patients can be influenced by contextual factors. This study examined three contextual factors relevant to hypothetical patients with low back pain, using vignettes that were varied in a 2 x 2 x 2 factorial design: level of reported pain (high vs low), level of supporting medical evidence (high vs low), and the valence of the physician-patient interaction (positive vs negative). Perceived levels of pain, disability, emotional distress, and somatic preoccupation were rated by internists after reading a vignette. Ratings of pain and disability were lower for patients without supporting medical evidence: ratings of distress, somatic preoccupation, and disability were greater for patients who exhibited negative rather than positive affect; internist ratings of pain were lower than patient ratings among patients reporting high levels of pain, while ratings were inflated for patients with low levels of pain. The results suggest that characteristics of both the patient and the situation may influence medical judgments. ~ 1997 Elsevier Science Ltd Key words--chronic pain, assessment, physician judgment, medical decision-making
INTRODUCTION Because pain is a symptom that confounds standard medical diagnostic procedures, the patient-plhysician interaction is of great importance as a source of information for physicians making medical judgments about patients in pain (Rudy et al., 1988). While it is well-known that physician judgments are influenced by social context (Bass et al., 1986; Borkan et al., 1995; Clark et al., 1991; Ong et al., 1995; Schwartz and Griffin, 1986) and that pain is more susceptible to biasing influences than most other medical conditions (Goldman, 1991), little empirical attention has addressed the effects of social context on physician judgments of pati~:nts with intractable pain. The available research, while limited, indicates that judgments may be blase6 by several elements common to social interactiLons (Tait and Chibnall, in press): (1) characteristic:~ of the target (patient); (2) characteristics of the judge (physician); and (3) characteristics of the context in which the interaction occurs. One important patient characteristic that has been shown to affect judgments of pain is the level of pain reported by the patient. For example, several studies of lay judges (university students) have documented a tendency to discount pain intensity when patients report high pain levels (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994). Similar findings were reported among health professionals by Grossman et al. *Author for correspondence. SSM Rehabilitation Institute, 350 Village Square Drive, St. Louis, MO 63042, U.S.A.
(1991), who compared pain levels reported by cancer patients with those ascribed to them by nurses, house officers, and oncology fellows. While there was a high rate of patient-provider agreement when patients reported relatively low levels of pain (79% agreement), the rate of agreement dropped substantially as patient ratings of pain increased to moderate (37% agreement) and high levels (13% agreement). A tendency by physicians to underestimate patient pain levels also has been found with irritable bowel syndrome patients (Van Dulmen et al., 1994) and patients with acute extremity trauma presenting to an emergency department (Todd et al., 1994). Although medical diagnostics have been shown to correlate poorly with patient reports of pain intensity (Hadler, 1984; Rudy et al., 1988), diagnostic evidence represents an important characteristic of the situation in which the patient-judge interaction occurs. Several studies of university students suggest that judges rely heavily on such evidence when forming judgments about pain patients (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994): in the absence of clear diagnostic evidence, judge ratings of pain intensity, disability, and emotional distress were found to be significantly lower than when medical evidence was present. While not a direct study of medical judgments, Carey et al. (1988) found evidence that physicians also weigh medical evidence heavily: patients with strong medical evidence supporting their symptoms were given higher disability ratings than were patients with weaker medical evidence.
1199
1200
Raymond C. Tait and John T. Chibnall
Factors affecting the judge also appear to influence symptom judgments. For example, there is evidence that training can occasion either more accurate judgments (Cleeland et al., 1986) or be associated with a tendency to discount the severity of symptoms (Lenburg et al., 1970). It has been suggested that these discrepant findings may be mediated by the valence of the patient-physician relationship. Consistent with this premise, Weissman and Haddox (1989) have found physicians to undermedicate cancer patients whom they viewed negatively. While not involving physicians, several studies cited previously (Chibnall and Tait, 1995; Tait and Chibnall, 1994) also found that hypothetical patients who were viewed positively by judges received significantly higher symptom ratings than did patients who were viewed negatively. The present study sought to extend the experimental research on moderators of pain-related symptom judgments by using a factorial design to test the effects of several variables simultaneously in a physician sample. The obvious advantage of a study with a physician sample relative to studies with university students (Chibnall and Tait, 1995; Chibnall and Tait, in press; Swartzman and McDermid, 1993; Tait and Chibnall, 1994) is its increased external validity. In addition, given the evidence that training may impact symptom judgments, it is important to determine whether moderating variables have the same impact on physicians as on university students. Clearly, other features of the judge, patient, and situation also could be studied as potential influences on symptom judgments, including attractiveness (Hadjistavropoulis et al., 1990), gender (McDonald and Bridge, 1991), litigation status (Tait et al., 1988; Weighill, 1983), patient age (Ross et al., 1989), ethnicity (Todd et al., 1993), and social desirability sensitivity (Deshields et al., 1995). The independent variables chosen for the present study were selected for several reasons: (1) there was reasonable experimental evidence implicating them as potent influences on judgments; (2) they clearly were relevant to physicians; and (3) they were amenable to paper-and-pencil manipulation. Based on these criteria, we examined the effects of patient self-reported pain intensity, the level of medical evidence, and affective valence of the patient-physician interaction on physician ratings of pain intensity, emotional distress, somatic preoccupation, and pain-related disability for hypothetical chronic low back pain patients. It was hypothesized that each of these variables would have significant effects on symptom judgments.
METHOD
Subjects A total of 80 internal medicine physicians participated in the study. The physicians in the sample were primarily male (80%), had been practicing medicine for an average of 15.2 years (SD = 12.2), and had an average age of 45.2 years (SD = 12.4). Design and materials
The study was a 2 (pain intensity)x 2 (medical evidence)x2 (affective valence) between-subjects factorial design. The independent variables were manipulated across eight written case summaries, each describing a hypothetical chronic low back pain patient. In the summaries, patient pain report was either low (3 on an 1 l-point scale ranging from 0 "no pain" to 10 "unbearable pain") or high (7 on the same scale); objective medical evidence consistent with the patient's pain was either present or absent; and the valence of the patient-physician interaction was either positive or negative. A sample summary appears below showing all levels of all independent variables, with the information that was varied appearing in brackets. Case summary History and presenting problem. JES is a 43-yearold white male. He has been referred for evaluation of low back pain with radiation to the right hip and buttock. The pain began about 1.5 years ago. JES was the driver of a vehicle that was stationary when it was struck from the rear at high speed by another vehicle. JES has reported low back pain since the accident and has missed work intermittently because of the pain. On a scale ranging from 0 (no pain) to 10 (unbearable pain), JES rates his average pain as a [3]/[7]. Medical findings. JES has undergone extensive diagnostic testing. [X-rays show mild degenerative changes in the lumbar spine. CT scan shows mild bulging at L5-S1, but with no evidence of nerve root compression. Electrodiagnostic studies are negative for nerve damage. Physical examination shows no evidence of radiculopathy or sensory loss.]/[X-rays show grade III spondylolisthesis at L5-S1. CT scan indicates a pars interarticularis defect at that level. Eiectrodiagnostic studies show evidence of denervation at L5-S1. Physical examination reveals sensory loss consistent with denervation.] Medication use. JES is using ibuprofen or aspirin daily for relief of pain (three to six tablets). He is also using an anti-inflammatory medication (DayPro, two tablets each morning). Mood and Behavior. Initial impressions after one appointment are [negative; JES appears demanding and hostile]/[positive; JES appears cooperative and friendly].
Physician judgments of chronic pain patients
Dependent measures Subjects were asked to indicate the amount of pain, emotional distress, somatic preoccupation, and pain-related disability that they believed the patient was experiencing. Pain intensity was rated on an l 1-point modified VAS scale (Jensen e~ al., 1994) that ranged from 0 (no pain) to 10 (unbearable pain). Emotional distress ("e.g. depression, anxiety, anger, emotional upset") and somatization ("level of preoccupation with somatic s y m p t o m s " ) w e r e rated on scales that ranged from 0 (none) to 10
(extreme). Disability was measured with three items from the Pain Disability Index (PDI) (Gronblad et al., 1993; Jerome and Gross, 1991; Tait et al., 1990). The PDI assesses the degree to which pain interferes with or prohibits usual behavior or role functioning in seven life areas (family/home responsibiliLties, recreation, social activity, occupation, sex, self-care, and life-support). Three items were selected in an effort to limit the number of ratings required of the subjects. The items pertaining to family/home responsibilities, social activity, and occupation were used in the present study. These three items were judged to be the best proxy for the total PDI through an internal consistency reliability analysis of PDI data from 1059 chronic pain patients (Chibnall and Tait, 1994). Disability ratings were made on l l-point Likert-type scales that ranged from 0 (not at all) to 10 (totally).
Procedure The study sample was obtained from the physician database of the marketing department of a large university medical center. Internal medicine physicians were included because evidence indicates that they see a significant number of patients with chronic pain (Howie et al., 1994). A physician was randomly selected from the database and one of the eight case summaries was mailed to him or her. Each subject received a cover letter that described the study as an examination of physician perceptions of chronic pain patients. Case summaries and dependent measures appeared on a separate sheet included with the letter. To minimize the likelihood of subjects discerning the intent of the experimental manipulations, each physician received only one vignette, and vignettes were quite brief. Also, the vignettes contained experimentally irrelevant (i.e. constant) information on the patient's history, personal data (e.g. age, gender, ethnicity) and medication use. Successive mailings were done until a minimum of 10 subjects had responded to each case summary. In all, 92 physicians responded to the mailings. Because 12 of the respondents did not complete the materials as directed, they were eliminated from the sample. This resulted in 10 subjects in each of the eight cells of the experimental design (n = 80). The final sample of 80 physicians consti-
1201
tuted a convenience sample and, therefore, cannot be considered representative of the general population of internists. RESULTS
Relations among dependent measures As expected, the three disability items taken from the PDI were highly intercorrelated (0.64 < r < 0.83, P < 0.001). Cronbach's coefficient alpha for these items was 0.89. Based on these data, the disability items were averaged to form a single composite disability score that could range from 0 to 10. Correlations then were computed among all of the dependent measures, most of which also were intercorrelated (see Table 1). Particularly strong correlations were found between measures of somatization and emotional distress (r = 0.69, P < 0.001) and disability and pain intensity (r = 0.66, P < 0.001), while other associations were relatively weaker. To further clarify relations among the measures, a principal components analysis was done, using a factor extraction criterion of an eigenvalue greater than or equal to one. Two factors emerged that accounted for 86.5% of the total variance in the original variable set (see Table 2). An orthogonal varimax rotation revealed high loadings ( > 0.85) for the emotional distress and somatization items on Factor l, which accounted for 54.4% of the variance. The pain intensity and disability items loaded highly (>0.85) on Factor 2, which accounted for 32.1% of the variance. Because of these high loadings and low cross loadings (<0.35) for the items, subsequent analyses examined the factors separately.
Symptom judgments Based on the relations among the dependent measures, separate 2 x 2 x 2 MANOVAs were done, first for the emotional distress and somatization variables, and then for the disability and pain intensity variables. For the first MANOVA, a multivariate main effect emerged only for affective valence, F(2,71) = 18.4, P < 0.001. Subsequent univariate analyses of variance showed that internists rated emotional distress and somatization higher for patients whose affective valence with the physician was negative rather than positive (see Table 3). Table 1. Zero-order correlations among dependent variables
Somatization Emotional distress Pain intensity Disability ap < 0.05; bp < 0.00|.
Emotional distress
Pain intensity
Disability
0.69b
-0.05 0.26" 0.66b
0.24" 0.48 b
Raymond C. Tait and John T. Chibnall
1202
nificantly lower (mean = 3.9, SD = 1.6) than the average level of 5 reported in the summaries, t(39) = - 4 . 3 , P < 0.001.
Table 2. Rotated factor loadings from principal components analysis Somatization Emotional distress Pain intensity Disability
Factor 1
Factor 2
0.94 0.87 -0.05 0.31
-0.05 0.32 0.94 0.86
DISCUSSION
The second MANOVA, examining the pain/disability factor, yielded multivariate main effects for all three independent variables: affective valence, F(2,71) = 6.2, P < 0.01; medical evidence, F(2,71) = 3.3, P < 0.05; and pain report, F(2,71) = 10.3, P < 0.001. The multivariate interaction effects were not significant (all Ps>0.05). Subsequent univariate analyses (see Table 4) showed that internists ascribed higher levels of disability (but not pain) to patients for whom affective valence was negative. They ascribed higher levels of both pain and disability to patients when medical evidence was present rather than absent and when the patient reported high as opposed to low pain.
Comparisons of patient and physician pain judgments Since pain level functioned as both an independent and dependent variable, it was possible to compare the mean pain intensity ratings given by the internists with those provided in the case summaries. These comparisons were computed only for pain and medical evidence variables, the independent variables where prior analyses showed significant univariate effects. Internist ratings of pain for patients reporting low pain (mean = 3.6, SD = 1.5) were significantly higher than the intensity rating (3) given in the summaries, t(39) = 2.5, P < 0.05. Ratings for patients with high pain, on the other hand, were significantly lower (mean = 5.1, SD = 1.6) than the rating (7) provided in the case summaries, t(39) = - 7 . 5 , P < 0.001. When medical evidence was available to support the patient's claims of pain, pain intensity levels were comparable (mean = 4.8, SD = 1.8) to the average rating of 5 given in the summaries, t(39) = - 0 . 7 , ns. When medical evidence was absent, however, internist ratings of pain were sig-
The results of this study support the general hypotheses that physician judgments of pain-related symptoms are influenced by properties of the patient, the situation, and the valence of the physician-patient interaction. Medical evidence, considered to be a situational variable in this study because of research evidence indicating that it correlates poorly with pain complaints among patients with intractable pain, had a significant influence on internist judgments of pain and disability. Reported level of pain, a patient characteristic, occasioned discounting of symptoms when reported pain was high. Finally, the affective valence of the physicianpatient interaction influenced ratings of somatic preoccupation, emotional distress, and, interestingly, levels of disability associated with pain. This pattern of results is of interest as it relates to prior research on symptom judgments, especially judgments rendered by lay samples (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994). In some respects, the judgments of physicians and university students follow similar patterns: both groups weight medical evidence heavily and both groups discount pain severity when it is reported at high levels. On the other hand, their judgments also differ in some respects. Complex interaction effects were found among university students (Tait and Chibnall, 1994), who were inclined to weight pain level and medical evidence differentially, depending to a large degree on the valence of the relationship between the target and the judge. By contrast, interactions among independent variables were not found among physicians, suggesting that they did not weight the independent variables differentially. Instead, physicians may have utilized stereotypic models to facilitate judgment, models that may not have been available to the students (Forgas, 1995). Another way in which lay and physician judges differed is the effect on judgments of the affective
Table 3. Univariate main effects on physician judgments of patient emotional distress and somatization Emotional distress Effect ~
Somatization
Mean (SD)
F
eta 2
Mean (SD)
F
eta 2
5.0 (2.1) 7.1 (1.2)
30.5 b
0.30
4.5 (2.4) 6.9 (1.9)
27.2 b
0.27
Affective state
Positive Negative Medical evidence ¢
Present Absent
6.0 (1.7) 6.1 (2.3)
5.4 (2.3) 6.0 (2.6)
5.6 (2.1) 6.4 (1.9)
5.3 (2.6) 6.1 (2.3)
Pain r e p o r f
Low High
an = 40 per group for all main effects; bp < 0.001; Cmultivariate effect not significant, univariate statistics not reported.
Physician judgments of chronic pain patients
1203
Table 4. Univariate main effects on physician judgments of patient pain intensity and disability Effect~ Affective state Positive Negative Medical evidence Present Absent Pain report Low High
Mean (SD) 4.2 (1.7) 4.5 (1.7)
Pain intensity F 0.4
Disability F
eta2
3.9 (1.7) 5.1 (1.9)
9.8c
0.12
eta2
Mean (SD)
0.01
4.8 (1.8) 3.9 (1.6)
6.0 b
0.08
4.9 (1.7) 4.0 (1.9)
4.7 b
0.06
3.6 (1.5) 5.1 (1.6)
20.4d
0.22
4.0 (2.0) 4.9 (1.7)
5.3 b
0.07
an = 40 per group for all main effects; bp < 0.05; ~P < 0.01; dp < 0.001.
valence of the judge-target interaction. Among lay judges, symptom judgments were positively influenced when affective valence was positive: symptom ratings for distress, disability, and pain were higher for positively valenced targets. Among physicians, however, the opposite effect occurred: patients presented in a negative light were seen as more distressed, disabled, and somatically preoccupied than patients presented positively. Because of differences in experimental inductions across studies, it is impossible to compare these differences directly. It may be that the induction failed in the cur:rent study: physicians may not have interpreted the patient descriptors as relevant to the patient-physician relationship. Alternatively, physicians, by virtue of their experience in the clinical assessment of patients with pain, may have used the information regarding affective valence more diagnostically. For example, they may have seen hostility, demandingness, and somatic preoccupation as direct consequences of high levels of pain and/or as contributors to painrelated disability. In either case, the augmentar~ion of symptom judgments of distress, disability, and somatization in response to a negative patient interaction makes clinical sense. It would be instruclive, however, to determine whether these judgments would extrapolate to other contexts. For example, disability assessments made in a context with significant consequences and high standard for accuracy, such as a recommendation to the Social Security Administration regarding a disability rating, may not show this pattern of effects. Comparisons of patient reports of pain, manipulated in this study as low (3/10) and high (7/10), with physician ratings also are noteworthy. In the presence of low pain, physicians tended to overestimate the pain actually reported by the pati,~nt. With high levels of pain, however, physicians discounted pain levels. This finding is compatible with other studies reported in the literature (Chibnall and Tait, in press; Grossman et al., 1991), where high levels of pain have been consistently discounted by health professionals and medical students. Such a finding is unfortunate in light of the
fact that patients most in need of medical attention are also those most vulnerable to underestimation of their symptom reports and, potentially, undertreatment (Weissman and Haddox, 1989). Predictably, medical evidence influenced concordance between patient and physician pain estimates. Confirmatory medical findings led to physician judgments that were comparable to patient reports, while non-confirming findings occasioned physician underestimates (regardless of the absolute level of pain intensity reported by the patient). While this is not surprising in light of the ample evidence that physicians depend on medical benchmarks to anchor their judgments (Carey et al., 1988; Hadler, 1984), it provides further evidence of the predicament facing many patients with intractable pain whose objective findings offer little diagnostic value (Hadler, 1984; Rudy et al., 1988). Of course, the results of this study must be considered in light of its methodologic limitations. Most importantly, the study relied on short, written case summaries to convey information about the patient and Likert-type paper-and-pencil measures to assess symptom judgments. Such methods lack the fidelity of actual clinical interactions and limit the generalizability of the findings reported here to clinic settings. Future research should incorporate stimuli and measures of a higher fidelity. Second, the study suffers from what Cook and Campbell (1979) have called "mono-operation bias." This term applies to all studies where the independent variables underrepresent the causal constructs. Here, it was hypothesized that variance in the amount of medical evidence, patient pain levels, and affective valence would cause systematic differences in symptom judgments. However, there are dozens of ways in which "medical evidence," "patient pain level," and "affective valence" could be operationalized other than those selected for the study. It is impossible to determine whether this study adequately represented the causal constructs in the manipulations. Further complicating the experimental induction, it is possible that physicians discerned the purpose of the study, and consciously modified their responses accordingly. Third, the
Raymond C. Tait and John T. Chibnall
1204
results lack external validity. Although the experimental methodology allows us to conclude that the manipulations caused the systematic differences observed on the dependent variables, the non-random nature of the sample prevents generalization of the findings to the larger population of internal medicine physicians. Enhancing the external validity of the effects is the task of future replications. Despite the methodologic caveats, the present research is an important step in the empirical study of the physician judgment process as it applies to patients with chronic pain. The results suggest that physician judgments about pain patients are substantially influenced by characteristics of the patient and the situation in which judgments are made. In conjunction with previous studies (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994), the present study suggests that judgments are affected by characteristics of the judge (i.e. medical professional vs lay person). The identification of sources of bias in physician judgments is important to the health and well-being of the patient in pain. If future research should be consistent with that reported here in identifying systematic sources of variance in symptom judgments, steps could be taken (e.g. education) to reduce such variance. Future research, however, is clearly needed before such decisions can be made.
REFERENCES
Bass, M. J., Buck, C., Turner, L., Dickie, G., Pratt, G. Robinson, C. (1986) The physician's actions in the outcome of illness in family practice. Journal of Family Practice 23, 43-47. Borkan, J., Reis, S., Hermoni, D. and Biderman, A. (1995) Talking about pain: a patient-centered study of low back pain in primary care. Social Science & Medicine 40, 977-988. Carey, T. S., Hadler, N. M., Gillings, D., Stinnett, S. and Wallsten, T. (1988) Medical disability assessment of the back pain patient for the Social Security Administration: the weighting of presenting clinical features. Journal of Clinical Epidemiology 41, 691-697. Chibnall, J. T. and Tait, R. C. (1994) The Pain Disability lndex: factor structure and normative data. Archives of Physical Medicine and Rehabilitation 75, 1082-1086. Chibnall, J. T. and Tait, R. C. (1995) Observer perceptions of low back pain: effects of pain report and other contextual factors. Journal of Applied Social Psychology 25, 418-439. Chibnall, J. T. and Tait, R. C. (in press) The effects of medical evidence and pain intensity on medical student judgments of chronic pain patients. Journal of Behavioral Medicine. Clark, J. A., Potter, D. A. and McKinlay, J. B. (1991) Bringing social structure back into clinical decision making. Social Science & Medicine 32, 853-866. Cleeland, C. S., Cleeland, L. M., Dar, R. and Rinehardt, L. C. (1986) Factors influencing physician management of cancer pain. Cancer 58, 796-800. Cook, T. D. and Campbell, D. T. (1979) QuasiExperimentation. Houghton Mifflin, Boston. Deshields, T. L., Tait, R. C., Gfeller, J. D. and Chibnall, J. T. (1995) The relationship between social desirability
and self-report in chronic pain patients. Clinical Journal of Pain 11, 189-193. Forgas, J. P. (1995) Mood and judgment: the Affect Infusion Model (AIM). Psychological Bulletin 117, 3966. Goldman, B. (1991) Chronic-pain patients must cope with chronic lack of physician understanding. Journal of the Canadian Medical Association 144, 1492-1497. Gronblad, M., Hupli, M., Wennerstrandk, P., Jarvinen, E., Lukinmaa, A., Kouri, J. P. and Karaharju, E. O. (1993) Intercorrelation and test-retest reliability of the Pain Disability Index (PDI) and Oswestry Disability Questionnaire (ODQ) and their correlation with pain intensity in low back pain patients. Clinical Journal of Pain 9, 189-195. Grossman, S. A., Sheidler, V. R., Swedeen, K., Mucenski, J. and Piantadosi, S. (1991) Correlation of patient and caregiver ratings of cancer pain. Journal of Pain and Symptom Management 6, 53-57. Hadjistavropoulis, H. D., Ross, M. A. and von Baeyer, C. L. (1990) Are physicians' ratings of pain affected by patients' physical attractiveness?. Social Science & Medicine 31, 69-72. Hadler, N. M. (1984) Medical Management of the Regional Musculoskeletal Diseases: Backache, Neck Pain, and Disorders of the Upper and Lower Extremities. Grune and Stratton, New York. Howie, J. G.R., Heaney, D. J. and Maxwell, M. (1994) Evaluating care of patients reporting pain in fundholding practices. British Medical Journal 3119, 705-712. Jensen, M. P., Turner, J. A. and Romano, J. M. (1994) What is the maximum number of levels needed in pain intensity measurement?. Pain 58, 387-392. Jerome, A. and Gross, R. T. (1991) Pain Disability Index: construct and discriminant validity. Archives of Physical Medicine and Rehabilitation 72, 920-922. Lenburg, C. B., Glass, H. P. and Davitz, L. J. (1970) Inferences of physical pain and psychological distress. 1I. In relation to the stage of the patient's illness and occupation of the perceiver. Nursing Research 19, 392398. McDonald, D. D. and Bridge, R. G. (1991) Gender stereotyping and nursing care. Research in Nursing and Health 14, 373-378. Ong, L. M. L., De Haes, J. C. J. M., Hoos, A. M. and Lammes, F. B. (1995) Doctor-patient communication: a review of the literature. Social Science & Medicine 40, 903-918. Ross, M. J., Tait, R. C., Grossberg, G. T., Handal, P. J., Brandeberry, L. and Nakra, R. (1989) Age differences in body consciousness. Journal of Gerontology: PsTchological Sciences 44, 23 24. Rudy, T. E., Turk, D. C. and Brena, S. F. (1988) Differential utility of medical procedures in the assessment of chronic pain patients. Pain 34, 53-60. Schwartz, S. and Griffin, T. (1986) Medical Thinking: The Psychology of Medical Judgment and Decision Making. Springer, New York. Swartzman, L. C. and McDermid, A. J. (1993) The impact of contextual cues on the interpretation of and response to physical symptoms: a vignette approach. Journal of Behavioral Medicine 16, 183-198. Tait, R. C. and Chibnall, J. T. (1994) Observer perceptions of chronic low back pain. Journal of Applied Social Psychology 24, 415-431. Tait, R. C. and Chibnall, J. T. (in press) Symptom judgments in patients with pain: a review. Advances in Medical Psychotherapy. Tait, R. C., Margolis, R. B., Krause, S. and Leibowitz, E. (1988) Compensation status and symptoms reported by patients with chronic pain. Archives of Physical Medicine and Rehabilitation 69, 1027-1029.
Physician judgments of chronic pain patients Tait, R. C., Chibnall, J. T. and Krause, S. J. (1990) The Pain Disability Index: psychometric properties. Pahr 411, 171-182. Todd, K. H., Samaroo, N. and Hoffman, J. R. (it993) Ethnicity as a risk factor for inadequate emergency department analgesia. Journal of the American Medical Association 269, 1537-1539. Todd, K. H., Lee, T. and Hoffman, J. R. (1994) The effect of ethnicity on physician estimates of pain severity in patients with isolated extremity trauma. Journal of the American Medical Association 271, 925-928.
1205
Van Dulmen, A. M., Fennis, J. F. M., Mokkink, H. G. A., van der Velden, H. G. M. and Bleijenberg, G. (1994) Doctors' perception of patients' cognitions and complaints in irritable bowel syndrome at an out-patient clinic. Journal of Psychosomatic Research 38, 581-590. Weighill, V. E. (1983) Compensation neurosis: a review of the literature. Journal of Psychosomatic Research 27, 97-104. Weissman, D. E. and Haddox, J. D. (1989) Opioid pseudo-addiction: an iatrogenic syndrome. Pain 36, 363-367.
Soc. Sci. Med. Vol.45, No. 8, pp. 1199-1205, 1997
PII: S0277-9536(97)00033-6
© 1997 ElsevierScienceLtd. All rights reserved Printed in Great Britain 0277-9536/97 $17.00 + 0.00
PHYSICIAN JUDGMENTS OF CHRONIC PAIN PATIENTS R A Y M O N D C. TAIT* and JOHN T. CHIBNALL Department of Psychiatry and Human Behavior, Saint Louis University School of Medicine, St. Louis, MO 63104, U.S.A. Abstract--Recent evidence has indicated that physician judgments of patients can be influenced by contextual factors. This study examined three contextual factors relevant to hypothetical patients with low back pain, using vignettes that were varied in a 2 x 2 x 2 factorial design: level of reported pain (high vs low), level of supporting medical evidence (high vs low), and the valence of the physician-patient interaction (positive vs negative). Perceived levels of pain, disability, emotional distress, and somatic preoccupation were rated by internists after reading a vignette. Ratings of pain and disability were lower for patients without supporting medical evidence: ratings of distress, somatic preoccupation, and disability were greater for patients who exhibited negative rather than positive affect; internist ratings of pain were lower than patient ratings among patients reporting high levels of pain, while ratings were inflated for patients with low levels of pain. The results suggest that characteristics of both the patient and the situation may influence medical judgments. ~ 1997 Elsevier Science Ltd Key words--chronic pain, assessment, physician judgment, medical decision-making
INTRODUCTION Because pain is a symptom that confounds standard medical diagnostic procedures, the patient-plhysician interaction is of great importance as a source of information for physicians making medical judgments about patients in pain (Rudy et al., 1988). While it is well-known that physician judgments are influenced by social context (Bass et al., 1986; Borkan et al., 1995; Clark et al., 1991; Ong et al., 1995; Schwartz and Griffin, 1986) and that pain is more susceptible to biasing influences than most other medical conditions (Goldman, 1991), little empirical attention has addressed the effects of social context on physician judgments of pati~:nts with intractable pain. The available research, while limited, indicates that judgments may be blase6 by several elements common to social interactiLons (Tait and Chibnall, in press): (1) characteristic:~ of the target (patient); (2) characteristics of the judge (physician); and (3) characteristics of the context in which the interaction occurs. One important patient characteristic that has been shown to affect judgments of pain is the level of pain reported by the patient. For example, several studies of lay judges (university students) have documented a tendency to discount pain intensity when patients report high pain levels (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994). Similar findings were reported among health professionals by Grossman et al. *Author for correspondence. SSM Rehabilitation Institute, 350 Village Square Drive, St. Louis, MO 63042, U.S.A.
(1991), who compared pain levels reported by cancer patients with those ascribed to them by nurses, house officers, and oncology fellows. While there was a high rate of patient-provider agreement when patients reported relatively low levels of pain (79% agreement), the rate of agreement dropped substantially as patient ratings of pain increased to moderate (37% agreement) and high levels (13% agreement). A tendency by physicians to underestimate patient pain levels also has been found with irritable bowel syndrome patients (Van Dulmen et al., 1994) and patients with acute extremity trauma presenting to an emergency department (Todd et al., 1994). Although medical diagnostics have been shown to correlate poorly with patient reports of pain intensity (Hadler, 1984; Rudy et al., 1988), diagnostic evidence represents an important characteristic of the situation in which the patient-judge interaction occurs. Several studies of university students suggest that judges rely heavily on such evidence when forming judgments about pain patients (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994): in the absence of clear diagnostic evidence, judge ratings of pain intensity, disability, and emotional distress were found to be significantly lower than when medical evidence was present. While not a direct study of medical judgments, Carey et al. (1988) found evidence that physicians also weigh medical evidence heavily: patients with strong medical evidence supporting their symptoms were given higher disability ratings than were patients with weaker medical evidence.
1199
1200
Raymond C. Tait and John T. Chibnall
Factors affecting the judge also appear to influence symptom judgments. For example, there is evidence that training can occasion either more accurate judgments (Cleeland et al., 1986) or be associated with a tendency to discount the severity of symptoms (Lenburg et al., 1970). It has been suggested that these discrepant findings may be mediated by the valence of the patient-physician relationship. Consistent with this premise, Weissman and Haddox (1989) have found physicians to undermedicate cancer patients whom they viewed negatively. While not involving physicians, several studies cited previously (Chibnall and Tait, 1995; Tait and Chibnall, 1994) also found that hypothetical patients who were viewed positively by judges received significantly higher symptom ratings than did patients who were viewed negatively. The present study sought to extend the experimental research on moderators of pain-related symptom judgments by using a factorial design to test the effects of several variables simultaneously in a physician sample. The obvious advantage of a study with a physician sample relative to studies with university students (Chibnall and Tait, 1995; Chibnall and Tait, in press; Swartzman and McDermid, 1993; Tait and Chibnall, 1994) is its increased external validity. In addition, given the evidence that training may impact symptom judgments, it is important to determine whether moderating variables have the same impact on physicians as on university students. Clearly, other features of the judge, patient, and situation also could be studied as potential influences on symptom judgments, including attractiveness (Hadjistavropoulis et al., 1990), gender (McDonald and Bridge, 1991), litigation status (Tait et al., 1988; Weighill, 1983), patient age (Ross et al., 1989), ethnicity (Todd et al., 1993), and social desirability sensitivity (Deshields et al., 1995). The independent variables chosen for the present study were selected for several reasons: (1) there was reasonable experimental evidence implicating them as potent influences on judgments; (2) they clearly were relevant to physicians; and (3) they were amenable to paper-and-pencil manipulation. Based on these criteria, we examined the effects of patient self-reported pain intensity, the level of medical evidence, and affective valence of the patient-physician interaction on physician ratings of pain intensity, emotional distress, somatic preoccupation, and pain-related disability for hypothetical chronic low back pain patients. It was hypothesized that each of these variables would have significant effects on symptom judgments.
METHOD
Subjects A total of 80 internal medicine physicians participated in the study. The physicians in the sample were primarily male (80%), had been practicing medicine for an average of 15.2 years (SD = 12.2), and had an average age of 45.2 years (SD = 12.4). Design and materials
The study was a 2 (pain intensity)x 2 (medical evidence)x2 (affective valence) between-subjects factorial design. The independent variables were manipulated across eight written case summaries, each describing a hypothetical chronic low back pain patient. In the summaries, patient pain report was either low (3 on an 1 l-point scale ranging from 0 "no pain" to 10 "unbearable pain") or high (7 on the same scale); objective medical evidence consistent with the patient's pain was either present or absent; and the valence of the patient-physician interaction was either positive or negative. A sample summary appears below showing all levels of all independent variables, with the information that was varied appearing in brackets. Case summary History and presenting problem. JES is a 43-yearold white male. He has been referred for evaluation of low back pain with radiation to the right hip and buttock. The pain began about 1.5 years ago. JES was the driver of a vehicle that was stationary when it was struck from the rear at high speed by another vehicle. JES has reported low back pain since the accident and has missed work intermittently because of the pain. On a scale ranging from 0 (no pain) to 10 (unbearable pain), JES rates his average pain as a [3]/[7]. Medical findings. JES has undergone extensive diagnostic testing. [X-rays show mild degenerative changes in the lumbar spine. CT scan shows mild bulging at L5-S1, but with no evidence of nerve root compression. Electrodiagnostic studies are negative for nerve damage. Physical examination shows no evidence of radiculopathy or sensory loss.]/[X-rays show grade III spondylolisthesis at L5-S1. CT scan indicates a pars interarticularis defect at that level. Eiectrodiagnostic studies show evidence of denervation at L5-S1. Physical examination reveals sensory loss consistent with denervation.] Medication use. JES is using ibuprofen or aspirin daily for relief of pain (three to six tablets). He is also using an anti-inflammatory medication (DayPro, two tablets each morning). Mood and Behavior. Initial impressions after one appointment are [negative; JES appears demanding and hostile]/[positive; JES appears cooperative and friendly].
Physician judgments of chronic pain patients
Dependent measures Subjects were asked to indicate the amount of pain, emotional distress, somatic preoccupation, and pain-related disability that they believed the patient was experiencing. Pain intensity was rated on an l 1-point modified VAS scale (Jensen e~ al., 1994) that ranged from 0 (no pain) to 10 (unbearable pain). Emotional distress ("e.g. depression, anxiety, anger, emotional upset") and somatization ("level of preoccupation with somatic s y m p t o m s " ) w e r e rated on scales that ranged from 0 (none) to 10
(extreme). Disability was measured with three items from the Pain Disability Index (PDI) (Gronblad et al., 1993; Jerome and Gross, 1991; Tait et al., 1990). The PDI assesses the degree to which pain interferes with or prohibits usual behavior or role functioning in seven life areas (family/home responsibiliLties, recreation, social activity, occupation, sex, self-care, and life-support). Three items were selected in an effort to limit the number of ratings required of the subjects. The items pertaining to family/home responsibilities, social activity, and occupation were used in the present study. These three items were judged to be the best proxy for the total PDI through an internal consistency reliability analysis of PDI data from 1059 chronic pain patients (Chibnall and Tait, 1994). Disability ratings were made on l l-point Likert-type scales that ranged from 0 (not at all) to 10 (totally).
Procedure The study sample was obtained from the physician database of the marketing department of a large university medical center. Internal medicine physicians were included because evidence indicates that they see a significant number of patients with chronic pain (Howie et al., 1994). A physician was randomly selected from the database and one of the eight case summaries was mailed to him or her. Each subject received a cover letter that described the study as an examination of physician perceptions of chronic pain patients. Case summaries and dependent measures appeared on a separate sheet included with the letter. To minimize the likelihood of subjects discerning the intent of the experimental manipulations, each physician received only one vignette, and vignettes were quite brief. Also, the vignettes contained experimentally irrelevant (i.e. constant) information on the patient's history, personal data (e.g. age, gender, ethnicity) and medication use. Successive mailings were done until a minimum of 10 subjects had responded to each case summary. In all, 92 physicians responded to the mailings. Because 12 of the respondents did not complete the materials as directed, they were eliminated from the sample. This resulted in 10 subjects in each of the eight cells of the experimental design (n = 80). The final sample of 80 physicians consti-
1201
tuted a convenience sample and, therefore, cannot be considered representative of the general population of internists. RESULTS
Relations among dependent measures As expected, the three disability items taken from the PDI were highly intercorrelated (0.64 < r < 0.83, P < 0.001). Cronbach's coefficient alpha for these items was 0.89. Based on these data, the disability items were averaged to form a single composite disability score that could range from 0 to 10. Correlations then were computed among all of the dependent measures, most of which also were intercorrelated (see Table 1). Particularly strong correlations were found between measures of somatization and emotional distress (r = 0.69, P < 0.001) and disability and pain intensity (r = 0.66, P < 0.001), while other associations were relatively weaker. To further clarify relations among the measures, a principal components analysis was done, using a factor extraction criterion of an eigenvalue greater than or equal to one. Two factors emerged that accounted for 86.5% of the total variance in the original variable set (see Table 2). An orthogonal varimax rotation revealed high loadings ( > 0.85) for the emotional distress and somatization items on Factor l, which accounted for 54.4% of the variance. The pain intensity and disability items loaded highly (>0.85) on Factor 2, which accounted for 32.1% of the variance. Because of these high loadings and low cross loadings (<0.35) for the items, subsequent analyses examined the factors separately.
Symptom judgments Based on the relations among the dependent measures, separate 2 x 2 x 2 MANOVAs were done, first for the emotional distress and somatization variables, and then for the disability and pain intensity variables. For the first MANOVA, a multivariate main effect emerged only for affective valence, F(2,71) = 18.4, P < 0.001. Subsequent univariate analyses of variance showed that internists rated emotional distress and somatization higher for patients whose affective valence with the physician was negative rather than positive (see Table 3). Table 1. Zero-order correlations among dependent variables
Somatization Emotional distress Pain intensity Disability ap < 0.05; bp < 0.00|.
Emotional distress
Pain intensity
Disability
0.69b
-0.05 0.26" 0.66b
0.24" 0.48 b
Raymond C. Tait and John T. Chibnall
1202
nificantly lower (mean = 3.9, SD = 1.6) than the average level of 5 reported in the summaries, t(39) = - 4 . 3 , P < 0.001.
Table 2. Rotated factor loadings from principal components analysis Somatization Emotional distress Pain intensity Disability
Factor 1
Factor 2
0.94 0.87 -0.05 0.31
-0.05 0.32 0.94 0.86
DISCUSSION
The second MANOVA, examining the pain/disability factor, yielded multivariate main effects for all three independent variables: affective valence, F(2,71) = 6.2, P < 0.01; medical evidence, F(2,71) = 3.3, P < 0.05; and pain report, F(2,71) = 10.3, P < 0.001. The multivariate interaction effects were not significant (all Ps>0.05). Subsequent univariate analyses (see Table 4) showed that internists ascribed higher levels of disability (but not pain) to patients for whom affective valence was negative. They ascribed higher levels of both pain and disability to patients when medical evidence was present rather than absent and when the patient reported high as opposed to low pain.
Comparisons of patient and physician pain judgments Since pain level functioned as both an independent and dependent variable, it was possible to compare the mean pain intensity ratings given by the internists with those provided in the case summaries. These comparisons were computed only for pain and medical evidence variables, the independent variables where prior analyses showed significant univariate effects. Internist ratings of pain for patients reporting low pain (mean = 3.6, SD = 1.5) were significantly higher than the intensity rating (3) given in the summaries, t(39) = 2.5, P < 0.05. Ratings for patients with high pain, on the other hand, were significantly lower (mean = 5.1, SD = 1.6) than the rating (7) provided in the case summaries, t(39) = - 7 . 5 , P < 0.001. When medical evidence was available to support the patient's claims of pain, pain intensity levels were comparable (mean = 4.8, SD = 1.8) to the average rating of 5 given in the summaries, t(39) = - 0 . 7 , ns. When medical evidence was absent, however, internist ratings of pain were sig-
The results of this study support the general hypotheses that physician judgments of pain-related symptoms are influenced by properties of the patient, the situation, and the valence of the physician-patient interaction. Medical evidence, considered to be a situational variable in this study because of research evidence indicating that it correlates poorly with pain complaints among patients with intractable pain, had a significant influence on internist judgments of pain and disability. Reported level of pain, a patient characteristic, occasioned discounting of symptoms when reported pain was high. Finally, the affective valence of the physicianpatient interaction influenced ratings of somatic preoccupation, emotional distress, and, interestingly, levels of disability associated with pain. This pattern of results is of interest as it relates to prior research on symptom judgments, especially judgments rendered by lay samples (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994). In some respects, the judgments of physicians and university students follow similar patterns: both groups weight medical evidence heavily and both groups discount pain severity when it is reported at high levels. On the other hand, their judgments also differ in some respects. Complex interaction effects were found among university students (Tait and Chibnall, 1994), who were inclined to weight pain level and medical evidence differentially, depending to a large degree on the valence of the relationship between the target and the judge. By contrast, interactions among independent variables were not found among physicians, suggesting that they did not weight the independent variables differentially. Instead, physicians may have utilized stereotypic models to facilitate judgment, models that may not have been available to the students (Forgas, 1995). Another way in which lay and physician judges differed is the effect on judgments of the affective
Table 3. Univariate main effects on physician judgments of patient emotional distress and somatization Emotional distress Effect ~
Somatization
Mean (SD)
F
eta 2
Mean (SD)
F
eta 2
5.0 (2.1) 7.1 (1.2)
30.5 b
0.30
4.5 (2.4) 6.9 (1.9)
27.2 b
0.27
Affective state
Positive Negative Medical evidence ¢
Present Absent
6.0 (1.7) 6.1 (2.3)
5.4 (2.3) 6.0 (2.6)
5.6 (2.1) 6.4 (1.9)
5.3 (2.6) 6.1 (2.3)
Pain r e p o r f
Low High
an = 40 per group for all main effects; bp < 0.001; Cmultivariate effect not significant, univariate statistics not reported.
Physician judgments of chronic pain patients
1203
Table 4. Univariate main effects on physician judgments of patient pain intensity and disability Effect~ Affective state Positive Negative Medical evidence Present Absent Pain report Low High
Mean (SD) 4.2 (1.7) 4.5 (1.7)
Pain intensity F 0.4
Disability F
eta2
3.9 (1.7) 5.1 (1.9)
9.8c
0.12
eta2
Mean (SD)
0.01
4.8 (1.8) 3.9 (1.6)
6.0 b
0.08
4.9 (1.7) 4.0 (1.9)
4.7 b
0.06
3.6 (1.5) 5.1 (1.6)
20.4d
0.22
4.0 (2.0) 4.9 (1.7)
5.3 b
0.07
an = 40 per group for all main effects; bp < 0.05; ~P < 0.01; dp < 0.001.
valence of the judge-target interaction. Among lay judges, symptom judgments were positively influenced when affective valence was positive: symptom ratings for distress, disability, and pain were higher for positively valenced targets. Among physicians, however, the opposite effect occurred: patients presented in a negative light were seen as more distressed, disabled, and somatically preoccupied than patients presented positively. Because of differences in experimental inductions across studies, it is impossible to compare these differences directly. It may be that the induction failed in the cur:rent study: physicians may not have interpreted the patient descriptors as relevant to the patient-physician relationship. Alternatively, physicians, by virtue of their experience in the clinical assessment of patients with pain, may have used the information regarding affective valence more diagnostically. For example, they may have seen hostility, demandingness, and somatic preoccupation as direct consequences of high levels of pain and/or as contributors to painrelated disability. In either case, the augmentar~ion of symptom judgments of distress, disability, and somatization in response to a negative patient interaction makes clinical sense. It would be instruclive, however, to determine whether these judgments would extrapolate to other contexts. For example, disability assessments made in a context with significant consequences and high standard for accuracy, such as a recommendation to the Social Security Administration regarding a disability rating, may not show this pattern of effects. Comparisons of patient reports of pain, manipulated in this study as low (3/10) and high (7/10), with physician ratings also are noteworthy. In the presence of low pain, physicians tended to overestimate the pain actually reported by the pati,~nt. With high levels of pain, however, physicians discounted pain levels. This finding is compatible with other studies reported in the literature (Chibnall and Tait, in press; Grossman et al., 1991), where high levels of pain have been consistently discounted by health professionals and medical students. Such a finding is unfortunate in light of the
fact that patients most in need of medical attention are also those most vulnerable to underestimation of their symptom reports and, potentially, undertreatment (Weissman and Haddox, 1989). Predictably, medical evidence influenced concordance between patient and physician pain estimates. Confirmatory medical findings led to physician judgments that were comparable to patient reports, while non-confirming findings occasioned physician underestimates (regardless of the absolute level of pain intensity reported by the patient). While this is not surprising in light of the ample evidence that physicians depend on medical benchmarks to anchor their judgments (Carey et al., 1988; Hadler, 1984), it provides further evidence of the predicament facing many patients with intractable pain whose objective findings offer little diagnostic value (Hadler, 1984; Rudy et al., 1988). Of course, the results of this study must be considered in light of its methodologic limitations. Most importantly, the study relied on short, written case summaries to convey information about the patient and Likert-type paper-and-pencil measures to assess symptom judgments. Such methods lack the fidelity of actual clinical interactions and limit the generalizability of the findings reported here to clinic settings. Future research should incorporate stimuli and measures of a higher fidelity. Second, the study suffers from what Cook and Campbell (1979) have called "mono-operation bias." This term applies to all studies where the independent variables underrepresent the causal constructs. Here, it was hypothesized that variance in the amount of medical evidence, patient pain levels, and affective valence would cause systematic differences in symptom judgments. However, there are dozens of ways in which "medical evidence," "patient pain level," and "affective valence" could be operationalized other than those selected for the study. It is impossible to determine whether this study adequately represented the causal constructs in the manipulations. Further complicating the experimental induction, it is possible that physicians discerned the purpose of the study, and consciously modified their responses accordingly. Third, the
Raymond C. Tait and John T. Chibnall
1204
results lack external validity. Although the experimental methodology allows us to conclude that the manipulations caused the systematic differences observed on the dependent variables, the non-random nature of the sample prevents generalization of the findings to the larger population of internal medicine physicians. Enhancing the external validity of the effects is the task of future replications. Despite the methodologic caveats, the present research is an important step in the empirical study of the physician judgment process as it applies to patients with chronic pain. The results suggest that physician judgments about pain patients are substantially influenced by characteristics of the patient and the situation in which judgments are made. In conjunction with previous studies (Chibnall and Tait, 1995; Chibnall and Tait, in press; Tait and Chibnall, 1994), the present study suggests that judgments are affected by characteristics of the judge (i.e. medical professional vs lay person). The identification of sources of bias in physician judgments is important to the health and well-being of the patient in pain. If future research should be consistent with that reported here in identifying systematic sources of variance in symptom judgments, steps could be taken (e.g. education) to reduce such variance. Future research, however, is clearly needed before such decisions can be made.
REFERENCES
Bass, M. J., Buck, C., Turner, L., Dickie, G., Pratt, G. Robinson, C. (1986) The physician's actions in the outcome of illness in family practice. Journal of Family Practice 23, 43-47. Borkan, J., Reis, S., Hermoni, D. and Biderman, A. (1995) Talking about pain: a patient-centered study of low back pain in primary care. Social Science & Medicine 40, 977-988. Carey, T. S., Hadler, N. M., Gillings, D., Stinnett, S. and Wallsten, T. (1988) Medical disability assessment of the back pain patient for the Social Security Administration: the weighting of presenting clinical features. Journal of Clinical Epidemiology 41, 691-697. Chibnall, J. T. and Tait, R. C. (1994) The Pain Disability lndex: factor structure and normative data. Archives of Physical Medicine and Rehabilitation 75, 1082-1086. Chibnall, J. T. and Tait, R. C. (1995) Observer perceptions of low back pain: effects of pain report and other contextual factors. Journal of Applied Social Psychology 25, 418-439. Chibnall, J. T. and Tait, R. C. (in press) The effects of medical evidence and pain intensity on medical student judgments of chronic pain patients. Journal of Behavioral Medicine. Clark, J. A., Potter, D. A. and McKinlay, J. B. (1991) Bringing social structure back into clinical decision making. Social Science & Medicine 32, 853-866. Cleeland, C. S., Cleeland, L. M., Dar, R. and Rinehardt, L. C. (1986) Factors influencing physician management of cancer pain. Cancer 58, 796-800. Cook, T. D. and Campbell, D. T. (1979) QuasiExperimentation. Houghton Mifflin, Boston. Deshields, T. L., Tait, R. C., Gfeller, J. D. and Chibnall, J. T. (1995) The relationship between social desirability
and self-report in chronic pain patients. Clinical Journal of Pain 11, 189-193. Forgas, J. P. (1995) Mood and judgment: the Affect Infusion Model (AIM). Psychological Bulletin 117, 3966. Goldman, B. (1991) Chronic-pain patients must cope with chronic lack of physician understanding. Journal of the Canadian Medical Association 144, 1492-1497. Gronblad, M., Hupli, M., Wennerstrandk, P., Jarvinen, E., Lukinmaa, A., Kouri, J. P. and Karaharju, E. O. (1993) Intercorrelation and test-retest reliability of the Pain Disability Index (PDI) and Oswestry Disability Questionnaire (ODQ) and their correlation with pain intensity in low back pain patients. Clinical Journal of Pain 9, 189-195. Grossman, S. A., Sheidler, V. R., Swedeen, K., Mucenski, J. and Piantadosi, S. (1991) Correlation of patient and caregiver ratings of cancer pain. Journal of Pain and Symptom Management 6, 53-57. Hadjistavropoulis, H. D., Ross, M. A. and von Baeyer, C. L. (1990) Are physicians' ratings of pain affected by patients' physical attractiveness?. Social Science & Medicine 31, 69-72. Hadler, N. M. (1984) Medical Management of the Regional Musculoskeletal Diseases: Backache, Neck Pain, and Disorders of the Upper and Lower Extremities. Grune and Stratton, New York. Howie, J. G.R., Heaney, D. J. and Maxwell, M. (1994) Evaluating care of patients reporting pain in fundholding practices. British Medical Journal 3119, 705-712. Jensen, M. P., Turner, J. A. and Romano, J. M. (1994) What is the maximum number of levels needed in pain intensity measurement?. Pain 58, 387-392. Jerome, A. and Gross, R. T. (1991) Pain Disability Index: construct and discriminant validity. Archives of Physical Medicine and Rehabilitation 72, 920-922. Lenburg, C. B., Glass, H. P. and Davitz, L. J. (1970) Inferences of physical pain and psychological distress. 1I. In relation to the stage of the patient's illness and occupation of the perceiver. Nursing Research 19, 392398. McDonald, D. D. and Bridge, R. G. (1991) Gender stereotyping and nursing care. Research in Nursing and Health 14, 373-378. Ong, L. M. L., De Haes, J. C. J. M., Hoos, A. M. and Lammes, F. B. (1995) Doctor-patient communication: a review of the literature. Social Science & Medicine 40, 903-918. Ross, M. J., Tait, R. C., Grossberg, G. T., Handal, P. J., Brandeberry, L. and Nakra, R. (1989) Age differences in body consciousness. Journal of Gerontology: PsTchological Sciences 44, 23 24. Rudy, T. E., Turk, D. C. and Brena, S. F. (1988) Differential utility of medical procedures in the assessment of chronic pain patients. Pain 34, 53-60. Schwartz, S. and Griffin, T. (1986) Medical Thinking: The Psychology of Medical Judgment and Decision Making. Springer, New York. Swartzman, L. C. and McDermid, A. J. (1993) The impact of contextual cues on the interpretation of and response to physical symptoms: a vignette approach. Journal of Behavioral Medicine 16, 183-198. Tait, R. C. and Chibnall, J. T. (1994) Observer perceptions of chronic low back pain. Journal of Applied Social Psychology 24, 415-431. Tait, R. C. and Chibnall, J. T. (in press) Symptom judgments in patients with pain: a review. Advances in Medical Psychotherapy. Tait, R. C., Margolis, R. B., Krause, S. and Leibowitz, E. (1988) Compensation status and symptoms reported by patients with chronic pain. Archives of Physical Medicine and Rehabilitation 69, 1027-1029.
Physician judgments of chronic pain patients Tait, R. C., Chibnall, J. T. and Krause, S. J. (1990) The Pain Disability Index: psychometric properties. Pahr 411, 171-182. Todd, K. H., Samaroo, N. and Hoffman, J. R. (it993) Ethnicity as a risk factor for inadequate emergency department analgesia. Journal of the American Medical Association 269, 1537-1539. Todd, K. H., Lee, T. and Hoffman, J. R. (1994) The effect of ethnicity on physician estimates of pain severity in patients with isolated extremity trauma. Journal of the American Medical Association 271, 925-928.
1205
Van Dulmen, A. M., Fennis, J. F. M., Mokkink, H. G. A., van der Velden, H. G. M. and Bleijenberg, G. (1994) Doctors' perception of patients' cognitions and complaints in irritable bowel syndrome at an out-patient clinic. Journal of Psychosomatic Research 38, 581-590. Weighill, V. E. (1983) Compensation neurosis: a review of the literature. Journal of Psychosomatic Research 27, 97-104. Weissman, D. E. and Haddox, J. D. (1989) Opioid pseudo-addiction: an iatrogenic syndrome. Pain 36, 363-367.