Pain, 48 (1992) 227-236 0 1992 Elsevier Science
B.V. All rights
of attitudes towards and beliefs about pain
Jenny Strong ‘, Roderick Ashton a and David Chant b ’ Department
of Psychology and ’ Social Sciences Group, The lJnit,ersity of Queensland, Queensland 4072 (Australia) (Received
5 April 1991, revision
3 June 1991, accepted
This study compared the psychometric properties of two scales designed to measure attitudes Summary towards and beliefs about pain. The Survey of Pain Attitudes (Revised) SOPA (Jensen and Karoly 1987) and the Pain Beliefs and Perceptions Inventory (PBPI) (Williams and Thorn 1989) were examined in terms of internal consistency, discriminant validity, factor structure, construct validity and sensitivity to age and gender effects. Results provided strong support for the SOPA as a useful measurement tool for use with patients with chronic low back pain. Further work is suggested for the PBPI, as the reported factor structure was not replicated. Discussion centred around the possible reasons for this finding, with issues such as the possible orientation of different treatment facilities, the possible differences in attitudes between patients with different types of pain, and the possible influence of length of years in pain or the receipt of workers compensation payments being considered. Key words: Assessment; Survey of Pain Attitudes (Revised); Pain Beliefs and Perceptions
Introduction While the concept of attitude-behaviour relations is far from new (see Ajzen and Fishbein 19771, its application to the area of chronic pain management is a relatively recent, and increasingly popular, development. Since the mid-1980s a number of pain researchers have argued for the routine assessment of a patient’s attitudes towards, beliefs about, and expectations about his/her pain and its subsequent treatment (Schwartz et al. 1985; Jensen and Karoly 1987; Jensen et al. 1987; Riley et al: 1988; Williams and Thorn 1989; Shutty et al. 1990; Shutty and DeGood 1991). Thus it is argued that an assessment of attitudes should be included as part of a multidimensional assessment prior to the commencement of treatment programmes (Strong et al. 1990; Slater et al. 1991). Such information is valuable to treatment planning, as a patient’s attitudes/beliefs about his/her pain condition may influence compliance to treatment, ability to cope with pain, and resultant treatment outcome (Schwartz et al.
Correspondence to: Jenny Strong, Dept. of Psychology, versity of Queensland, Queensland 4072, Australia.
1985; Jensen et al. 1987; Riley et al. 1988; Williams and Thorn 1989; Shutty and DeGood 1991). It has been suggested that attitudes are underlying variables which influence behaviour (Fishbein and Ajzen 1975). Attitudes are conceptualised as the degree of feeling or affect held towards an object (Fishbein and Ajzen 1975). Pain beliefs have been defined as ‘patients own conceptualizations of what pain is and what pain means for them’ (Williams and Thorn 1989). Beliefs have been distinguished from attitudes to the extent that a belief is the information known about the object (Fishbein and Ajzen 1975). While pain attitudes and beliefs may be informally observed by staff in the clinical setting, it is important for more formal, empirically sound methods to be used (Schwartz et al. 1985). There are now four dedicated published measures available which assess attitudes towards beliefs about pain and its treatment. The first published scale was the Pain Information and Beliefs Questionnaire (PIBQ) (Schwartz et al. 1985; Shutty et al. 1990; Shutty and DeGood 1991), which was designed to assess factual information about conservative pain management and the extent to which patients agreed with such a treatment approach (Shutty et al. 1990). In its original form it consisted of a 2-part
questionnaire containing 19 true-false questions on factual information and 13 questions about beliefs using a b-point rating scale (Schwartz et al. 1985). The questionnaire was used with a 23 min psychoeducational videotape about chronic pain. Later work used a modified PIBQ, consisting of 9 attitude items rated on a S-point scale, the same factual questions, and a 15 min videotape (Shutty and DeGood 1991). The 4 factors reported to emerge from the PIBQ were: (1) admission of emotionality; (2) perceived relevance of videotape; (3) acknowledgement of personal responsibility in treatment; and (4) discrimination of non-invasive treatment (Schwartz et al. 1985). Two problems have been identified with the use of the PIBQ methodology for pain attitude assessment. Firstly, only 2 of the factors (admission of emotionality and personal responsibility) can be considered as ‘attitude’ factors (Jensen et al. 1987). Secondly, the method relies on the USC of videotape equipment, with such hardware not always available in the clinic setting (Jensen et al. 1987). An additional drawback with the PIBQ is that its authors recommend the development and use of a locally produced videotape at each pain clinic. The use of different videotapes would, however, introduce variability into the procedure. The Pain and Impairment Relationship Scale (PAIRS) was devised to measure the extent to which patients with chronic pain believe that pain interferes with their functioning (Riley et al. 1988). It consists of 15 items scored on a 7-point Likert scale. A recent study has provided support for the discriminant, convergent and divergent validity and reliability over time of the PAIRS @later et al. 1991). A limitation of the PAIRS is its consideration of one attitude alone - that of the pain-impairment link. The Survey of Pain Attitudes (SOPA) instrument consists of 24 true-false items arranged into 5 subscales: medical cure, pain control, solicitude, disability
and medication (Jensen ct al. 1987). After carIF promising results on the SOPA, a reviscd instrument (SOPA( was developed (Jensen and Karoly 19X7). The revised instrument (SOPA( contains the 5 subscales of the earlier version. plus a 6th suhscalc which taps the attitude that pain may be intlucnccd hy an emotional link. The most recent development is the Pain Beliefs and Perceptions Inventory (PBPI), which asscsscs a patient’s beliefs about the stability of pain over time. pain as a mystery, and self-blame (Williams and Thorn 1989). Given the increased interest in the assessment of pain attitudes/beliefs, and the availability of 4 dedicated measurement tools, which assessment should the researcher or clinician incorporate into their assessment procedures? The aim of this study was to investigate and compare the psychometric properties of scales which attempt to measure attitudes towards and beliefs about pain. The PIBQ was not used in this study because the non-comparability of results clearly reduces the usefulness of this method, nor was the PAIRS included in this study, as it reportedly only measures one dimension of attitudes. A comparative analysis was therefore made of the SOPA and the PBPI. Each scale was examined for internal consistency, discriminant validity, factor structure, construct validity, and sensitivity to age and gender effects.
Subjects One hundred patients with written, informed consent served were drawn from the pain clinics major metropolitan hospitals and
chronic low back as subjects in this and neurosurgical the pain clinic of a
pain who gave study. Patients services of two third metropoli-
Age Years in pain Number of operations Employment (%) Invalid pension Home duties Sickness benefits Other pension Compensation Paid employment Unemployed Unknown ” Skewness;
96 90 90
46.1 9.8 1.3
13.3 10.1 2.5
0.3 2.0 6.4
-0.8 3.9 s1.2
43 6 1
22-74 o-47 o-22
22.9 19.x 16.7 8.3 6.3 7.3 4.2 14.6
Gender cc% Male Female
Litigation Yes No Unknown
6.25 91.67 2.08
tan hospital. Given that 4 subjects did not provide data on some items, the sample number was reduced to 96 throughout, thereby ensuring integrity of data across the whole analysis. Descriptive information of the subject sample is illustrated in Table I.
Procedure The attitude/belief questionnaires were administered individually to each subject by the first investigator as part of a larger, multidimensional assessment battery. The investigator remained with the subjects throughout the assessment to answer queries. All measures in the assessment battery were presented in random order.
could not be more severe’ as the endpoints (Scott and Huskisson 1979). Subjects are asked to put a mark on the line at the point which best represents their present pain intensity.
Data preparation and analysis Prior to analysis of the data items on both the SOPA and the PBPI with reverse scoring were recalculated. Results on the PBPI were then modified to ensure that equal intervals existed between all responses (in its current form, the numbers -2, - 1, + 1, +2 are used with an unequal interval between - 1 and + 1). All data analyses were performed using SAS statistical software.
Instruments Sunbey of Pain Attitudes (Reked) (SOPA(R The SOPA is a 35-item questionnaire in which subjects indicate their agreement using a 5-point Likert scale, where 0 = very untrue for me and 4 = very true for me (Jensen and Karoly 1987). The attitudes tapped by the SOPA are (1) that there is a medical cure for pain; (2) that one can control one’s pain; (3) that others should assist people in pain; (4) that one is disabled because of the pain; (5) that medication is the best treatment for pain; and (6) that pain, may be influenced by emotional states. Unfortunately, when developed, its authors did not provide information on whether subjects were receiving workers compensation payments, whether they had litigation pending, whether they were employed or on invalid pensions. Such matters may well have some bearing upon a patient’s attitudes towards his/her pain. Also of note is the variety of pain complaints of the subjects in the sample used by Jensen and Karoly (1987). Pain Beliefs and Perception lncentory (PBPI). This scale measures the extent of agreement or disagreement with 16 pain belief items (Williams and Keefe 1989; Williams and Thorn 1989). A 4-point Likert scale is used, with -2 indicating strongly disagree and + 2 indicating strongly agree. The beliefs measured by the scale are (1) time: the belief that pain is and will be an enduring part of the patient’s life; (2) mystery: the belief that pain is a mysterious and poorly understood phenomenon in the patient’s life; and (3) selfblame: the belief that pain is caused or perpetuated by the patient. The study in which the psychometric properties of the PBPl were reported used as subjects industrially injured workers attending a rehabilitation centre; all were receiving workers compensation benefits (Williams and Thorn 1989). The question as to how much the subjects’ attitudes/beliefs were influenced by the compensation payments being received needs to be considered. Furthermore, the pain in these subjects was in a variety of sites. In addition to the 2 attitudes/beliefs measures, the following instruments were used in the validity checks of the above measures. Coping Strategy Questionnaire (CSQ). This questionnaire consists of 50 items which measure the frequency with which 7 coping strategies are used, plus the perceived effectiveness of such strategies in terms of ability to control and decrease pain (Rosensteil and Keefe 1983). The coping strategies are (1) diverting attention; (2) reinterpretation of pain sensations; (3) ignoring pain sensation; (4) coping self-statements; (5) praying or hoping; (6) increased behavioural activities; and (7) catastrophising. Pain Disability Index (PDI). This is a self-report questionnaire in which subjects are asked to indicate the degree to which pain interferes with seven areas of daily functioning (Tait et al. 1990). Each item is scored on a O-10 scale, where 0 indicates no disability and 10 indicates total disability. Box Scale CBS). This pain intensity measure consists of the numbers from 0 to 10 written side by side and enclosed in boxes (Jensen et al. 1986). Subjects are asked to put a cross through the box with the number which best corresponds to their present pain intensity. Visual Analogue Scale (Horizontal) (VASH). This scale consists of a 100 mm horizontal line with the words ‘no pain’ and ‘pain which
Results Internal consistency
Cronbach’s alpha reliability coefficients for the SOPA and the PBPI subscales are contained in Table II. With the exception of the ‘medication’ subscale of the SOPA which had an alpha value of 0.49, all scales demonstrated satisfactory reliability estimates. A much improved alpha value was found for the ‘disabled’ scale here than in the original study (Jensen and Karoly 1987). The finding of poor internal consistency on the ‘medication’ scale was also found by Jensen and Karoly (1987). Inspection of the individual items of the ‘medication’ scale suggested that 2 items (‘pain medication is the best treatment for pain’ and ‘stop giving me medication’), if deleted, would increase the overall alpha value. However, this would result in a scale with only 2 items, which clearly does not constitute an adequate scale. Discriminant validity
As a screening procedure to check that each of the subscales of the 2 instruments measured different constructs the intercorrelations between the subscales should be substantially lower than Cronbach’s alpha coefficients. The intercorrelations between the SOPA subscales and the PBPI subscales are illustrated in Table III. In the current study, interscale correlations on the SOPA ranged from -0.15 to 0.28. All correlations were considerably smaller than the alpha values. For the PBPI, the interscale correlations ranged from 0.09 to -0.19, which no correlations reaching significance. All were clearly much lower than the corresponding alpha values. Factor structure
The factor structure of the SOPA was examined using a principal components analysis with varimax rotation. While 12 eigenvalues were greater than 1.0, the first 6 factors alone accounted for 48.27% of the total variance. All further factors after the 6th factor each accounted for less than 5% of the variance, and
MedCure = medical cure; CtrlP = pain Disah = disabled; Medic = medications time; Myst = mystery; SfBI = self-blame. Subscale
control; Solic = solicitude: Emot = emotion; Time =
Williams and Thorn (1989)
SOPA MedCure CtrtP Solic Disah Medic Emot
0.63 0.76 0.7 I 0.5 1 0.32 0.75
PBPI Time Myst SfBi
0.80 0.80 0.65
0.7X 0.74 0.67
thus added little to the solution. This, in addition to an inspection of the scree plot pointed to a &factor solution as the obvious choice for factor extraction. A direct comparison with the Csubscale structure of the SOPA was thus possible. Table IV contains the items of the SOPA and their loadings on the factors. ‘Solicitude’, ‘control pain’, ‘medical cure’, ‘disabled’ and ‘emotion’ emerged as clear factors, as did a weak 3-item ‘medication’ factor. Two items from the original ‘medication’ subscale failed to load on any factor. This is not surprising given that
Study Jensen Karoly
II) IS IX 75 34 I Y 13 16 21 24 2X 33 35 5 7 14 19 26 2’) 2 8 22 30 4 II 23 31 3 12 17 20 27 32
0.08 0.04 0.1x - 0.20 0.16 - 0.08 0.1 I -0.14 - 0.30 -0.05 -0.10 - 0.3s 0. I1 - 0.04 0.15 0.6 I 0.73 0.62 0.73 0.78 0.73 0.09 0. I I - 0.05 0.14 0.18 0.0’) - 0.00 ~ 0.2Y 0.30 0.05 0.24 0.02 - 0.04 -0.10
- 0.0 I
IU 0.6 1 0.57 0.64 0.46 0.53 0.73 0.10 - 0.18
- 0.02 - 0.04 --0.16 0. I4 - 0.20 0.5 1 0.56 0.52 0.53 0.60 0.27 0.63 0.70 0.52 - 0.06 -0.10 - 0.07 0.1 I -0.11 0.01 - 0.26 0.07 - 0. IO - 0.23 -- 0.00 - 0.02 -c).lO - 0.26 - 0.02 0.04 - 0.08 - 0.02 0.14 - 0.08
0.05 0.12 - 0.03 -0.17 - 0.20 - 0.05 - 0.07 - 0.08 t1.12 0.25 0.15 -0.14 0.117 0.06 0.02 0.15 -0.lh 0.00 - 0.04 - 0.07 0.08 0.04 - 0.0s 0.08 0.07 - 0.02 - 0.07