PAIN IS AN UNPLEASANT sensory and emotional

Transcription

1 782 Test-Retest Reliability and Internal Consistency of the Quebec-French Version of the Survey of Pain Attitudes Josée Duquette, MSc, Patricia A. McKinley, PhD, June Litowski, BSc ABSTRACT. Duquette J, McKinley PA, Litowski J. Testretest reliability and internal consistency of the Quebec-French version of the Survey of Pain Attitudes. Arch Phys Med Rehabil 2005;86: Objective: To determine the test-retest reliability and internal consistency of the Quebec-French version of the Survey of Pain Attitudes (QAD/F-SOPA). Design: Measurement of test-retest reliability (2-wk interval) and internal consistency. Setting: Five rehabilitation settings that offer services to chronic pain patients. Participants: Convenience sample of 69 Francophone adults (having either French as the mother tongue or a good mastery of French) with musculoskeletal pain for a minimum of 6 months and stable pain condition during the test-retest interval. Interventions: Completion of the QAD/F-SOPA twice within a 2-week interval. Main Outcome Measures: Test-retest reliability (Pearson r, 2-tail paired t test, P.001) and internal consistency (Cronbach at time 1). Results: Fifty-six subjects completed the QAD/F-SOPA on both occasions. Except for the disability subscale, the r values fell between 0.7 and 0.9 (high correlation). For the paired t test, all subscales (except for control and medication) had a P value greater than.05, confirming their test-retest stability. All subscales showed satisfactory internal consistency estimates ( ) except for the harm (.67) and disability (.64) subscales. Conclusions: Globally, the QAD/F-SOPA has good reliability and validity properties and meets the prerequisites for use for clinical and research purposes. The disability subscale shows weaker properties; further studies would help determine how it could be improved. Key Words: Chronic disease; Pain; Pain measurement; Rehabilitation; Reliability and validity; Translations by American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From CRIR (Duquette, McKinley); Constance-Lethbridge Rehabilitation Centre (Duquette, McKinley, Litowski); and School of Physical and Occupational Therapy, McGill University (McKinley), Montreal, QC, Canada. Supported by the Réseau de Recherche en Réadaptation du Montréal Métropolitain et de l Ouest du Québec. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Josée Duquette, MSc, CRIR, Constance-Lethbridge Rehabilitation Centre, 7005 de Maisonneuve Blvd W, Montreal, QC H4B 1T3, Canada, josee_duquette@ssss.gouv.qc.ca /05/ $30.00/0 doi: /j.apmr PAIN IS AN UNPLEASANT sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage. 1(p210) It is a highly personal and complex perceptual phenomenon. Indeed, many people report pain in the absence of tissue damage or any likely pathophysiologic cause, and this usually happens for psychologic reasons. Many psychosocial and behavioral factors may worsen and perpetuate pain and disability 2 ; for example, the subjective sensation of pain may be influenced by many factors such as cultural mores, the meaning of the context in which pain is experienced, the degree of attention, anxiety, distraction given to the pain, and the impression of control over the pain. 3 From a cognitive-behavioral perspective, a person s beliefs with respect to pain represent his/her understanding of pain and what it means personally. 4 Its numerous dimensions (eg intensity, quality, time course, effect, personal meaning) are uniquely experienced by each person and, accordingly, can only be assessed indirectly. 2 Because of this idiosyncratic nature, individual attitudes and beliefs toward pain become important factors to be considered in assessing and treating chronic pain and have a primary role in the adjustments that one makes to chronic pain. 4,5 It is important for rehabilitation specialists in chronic pain management to evaluate pain attitudes and beliefs of people who present with chronic pain because these factors can influence their receptivity to interventions and affect their capacity to adapt to the pain. 4-8 In fact, certain people may not be receptive to therapeutic interventions if they are incompatible with their beliefs and attitudes. Moreover, early detection of abnormal or inappropriate beliefs and attitudes toward his/her pain is important in determining whether a person is at risk to develop chronic disabilities. For example, someone who avoids exercise and other physical activities because of concerns about increased tissue damage or pain intensity will be predisposed to inactivity. This concern places that person into a downward spiral of inactivity and increased perception of pain because avoidance of physical activity leads to physical detraining, fatigue, and muscle microtraumas. Those conditions will, in turn, further increase the perception of pain, reinforcing fears and inappropriate pain behaviors. It is important that such avoidance and anticipation behaviors be detected by therapists, who could then educate patients to the fact that pain does not automatically signify damage. Although attitudes toward pain can be observed informally by a clinician, it is important that these attitudes be evaluated with a more formal and empiric methodology that has measurable psychometric properties. 9 The Survey of Pain Attitudes (SOPA) was developed to address this need. The SOPA is a self-administered questionnaire with 57 statements that permit the evaluation of 7 categories of beliefs and attitudes about pain that are considered critical for the adjustment to long-term chronic pain. 10 Each category is comprised of 6 to 10 statements. Respondents must indicate their agreement with each statement on a 5-point Likert scale with the following anchors at each end, This is very untrue for me and This is very true for me. The SOPA assesses to what extent the patients believe (1) they can control their pain (control subscale); (2) they are disabled by their pain (disability subscale); (3) that pain means they are damaging themselves and that they should avoid

2 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette 783 exercise (harm subscale); (4) that their emotions affect their experience of pain (emotion subscale); (5) that medications are an appropriate treatment for chronic pain (medication subscale); (6) that others, especially family members, should be solicitous in response to their experience of pain (solicitude subscale); and (7) in a medical cure for their pain problem (medical cure subscale). During its development, the SOPA has been through multiple validation studies and revisions and has shown good internal consistency 10,13 and test-retest reliability. 10 Further, it can be administered in person, by telephone interview, either partially 5,14,15 or fully, and takes about 20 minutes to complete. Thus, it is an extremely flexible instrument that is easily administered a low cost and takes little time to complete. For the therapist, it can be a gauge with which to monitor change in attitudes and beliefs during and after an intervention to evaluate the efficacy of a treatment plan. Or, it can be used to tailor an intervention to address a particular attitude or belief that may be problematic, thus derailing the downward spiral of inactivity and increased pain. 10,16-18 Researchers have used it to measure the relation between certain beliefs toward pain and the perception of pain intensity, adherence to treatment, adjustment to chronic pain, and the function of people with chronic pain. 5,8,10,14-16,18-21 The SOPA has been translated into French by a research team from Belgium, and initial validation studies have been conducted. 22 However, it is not adapted for the French Canadian syntax. Two items have been eliminated and the composition of the subscales modified to increase the internal consistency of some subscales. Thus, it cannot be used for research purposes in Quebec, where the tests are administered bilingually and colloquialisms differ. For example, in the Belgium version, the control category, renamed pain modulation, contains 18 items compared with 10 in the original SOPA. Thus, the English and French questionnaires are not comparable and cannot be used in a research study that includes both Francophone and Anglophone subjects. Currently in Quebec, both therapists and researchers need to use measurement tools that have comparable versions in French and English and similar psychometric properties in both languages. To this end, we recently translated and adapted the SOPA for the Francophone community in Quebec by using the process of forward-backward translation by committee. 23 This Francophone version is entitled Questionnaire sur les Attitudes envers la Douleur (QAD/F-SOPA) and was judged to be clinically useful as a first version, although psychometric properties were not yet measured. That being said, there were difficulties with the translation of items 3 (I consider my pain to be a disability) and 35 (I consider myself to be disabled), because the word disability could be translated either as incapacité or as handicap. 23 These 2 statements are found in the disability category of the SOPA; therefore, we decided to standardize the word disability in both. The word handicap was retained in the first version. However, we decided that in a subsequent validation of the questionnaire, it would be important to measure the test-retest reliability and internal consistency of these 2 statements by using the word incapacité as compared with the word handicap. Thus, the main objectives of this study were (1) to determine the test-retest reliability and the internal consistency of the 7 subscales of the QAD/F-SOPA, (2) to compare the results with those of the original SOPA, and (3) to determine if a different translation of items 3 and 35 would affect the test-retest reliability and internal consistency of the disability subscale of the QAD/F-SOPA. METHODS Participants Subjects (N 69) were recruited from among patients who were treated in 1 of 5 rehabilitation programs in the greater Montreal area: Constance-Lethbridge Rehabilitation Centre (CLRC), Centre de Réadaptation Lucie-Bruneau (CRLB), Jewish Rehabilitation Hospital (JRH), Montreal Pain Clinic (MPC), and St. Mary s Hospital Center (SMHC). The ethics committees of each institution approved the project. To ensure that the pain condition was stable during the test-retest interval, subjects were recruited at the time of their final discharge from the rehabilitation program or, exceptionally, while they were on the waiting list for admission. They did not have any treatment during that 2-week period or changes in their medication. Inclusion criteria were: age of at least 18 years; musculoskeletal pain for a minimum of 6 months; sufficient physical, perceptual, and cognitive capacities to answer the questionnaire independently; being Francophone or having a good mastery of French; and stable pain condition. Exclusion criteria included psychiatric problems that could affect the validity of the responses and an event experienced during the test-retest interval that could have significantly increased the level of pain (eg, car collision, fall). Procedure The project was explained to eligible subjects by a researcher or by a collaborating clinician. A detailed explanation was given to those who agreed to participate and they were assigned a subject number and asked to complete a consent form and a general information questionnaire. They also indicated their level of general pain on a pain visual analog scale (pain VAS). They then completed the QAD/F-SOPA. The questionnaires were returned in sealed envelopes to ensure confidentiality. The few patients who could not complete the questionnaires at the clinic were asked to complete them at home later the same day and mail it back immediately in a prestamped envelope. A few patients were recruited by telephone contact. If they agreed to participate, they were sent a package including a letter of explanation, the consent form, the pain VAS, and the QAD/F-SOPA. This first phase corresponds to time 1 (t1). All subjects were also given a second copy of the QAD/F-SOPA and the pain VAS, which had to be completed 2 weeks later and returned in a prestamped envelope (time 2 [t2]). Subjects had to complete the QAD/F-SOPA and the pain VAS on the same day for both t1 and t2, and indicate the date on the subscales. Although the target test-retest interval was 2 weeks (long enough to forget the answers but short enough to ensure condition stability), a range between 1 to 4 weeks was deemed acceptable. 24,25 People who forgot to send back the second questionnaires were reminded through a telephone call. On their return, the dates on which they were completed were verified to ensure that they were within our acceptable time range. Outcome Measures Demographics and pain. The general information questionnaire included questions on demographics, working and family status, identification of the painful area(s) (head, neck, back, lower limbs, upper limbs, other), and duration of the pain. The pain VAS consisted of a 100-mm horizontal line with the end-point anchors at no pain and very intense pain ; subjects marked on the line the point that best represented their current pain intensity. In the last few decades, VASs have been widely used in clinical and research milieus; their particular

3 784 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette advantages are their sensitivity, simplicity, reproducibility, and universality. 26 QAD/F-SOPA. The original QAD/F-SOPA was used. 23 Because this first study indicated difficulty with the translation of the word handicap with the items 3 and 35, those items were posed twice once with the word handicap and the second time with the word incapacités. Statistics Descriptive statistics were used to describe the demographic and clinical profiles of the subjects and the distribution of the QAD/F-SOPA results at both t1 and t2. Test-retest reliability. We used 2 measures; the Pearson correlation coefficient (r) and the paired t test (2-tail P.001). The Pearson r was used for comparisons with the results reported by Jensen et al. 10 However, this correlation is based on regression analysis and measures to what extent the relation between 2 variables can be described by a straight (regression) line; therefore, a perfect fit can be obtained, resulting in a r of 1.0, despite the fact that the intercept is not a zero and the slope is not equal to Thus, the Pearson correlation coefficient may be higher than the true test-retest reliability, and therefore a high r score (ie, a linear function) does not mean necessarily that the results are stable over time. For example, a subject might have for a QAD/F-SOPA subscale a score of 0 for all the items at t1 but a score of 4 for all of them at t2. In that situation r would be 1.0 because a perfect correlation would be obtained because of a linear transformation, but the results would show absolutely no test-retest stability. For that reason, we also used a paired t test to verify whether the means at t1 and t2 differed. Internal consistency. The homogeneity of each subscale of attitudes and behaviors toward the pain (internal consistency) was verified with the Cronbach coefficient, because the items should tap different aspects of the subscale but not different parts of different traits. 24 The lower limit of acceptability for was set at 0.7 and the upper limit at 0.9 to ensure moderate correlation and to avoid a high level of item redundancy. 24 In addition to calculating values, that coefficient was also recalculated for each subscale, each time omitting 1 item. A significant increase in the value when a specific item is left out would indicate that its exclusion would increase the homogeneity of the subscale. 24 Translation of disability. The word disability was translated once by incapacité and another time by handicap(é) for items 3 and 35. This provided 4 different combinations for the disability subscale. For each, the Pearson r and the Cronbach were calculated for comparison purposes. Also, a chi-square analysis was used (1) to compare the answers distribution of items 3 and 35 (considered separately) when using incapacité versus handicap(é), at t1 and at t2, and (2) the stability of the answers distribution over time (t1 vs t2) for both translation possibilities of items 3 and 35. RESULTS Demographics Of the 69 patients who agreed to participate, 7 (10.1%) did not participate at all, 6 (8.7%) answered only at t1, and 56 (81.2%) completed the QAD/F-SOPA on both occasions. For the latter sample, the mean test-retest interval was days. At t1, 56 subjects completed the questionnaires in the clinical setting and 6 at home. The 56 subjects who completed the questionnaires both times were compared with the 13 people who did not through a series of paired t tests (for age, duration of pain, and number of painful sites) and chi-square Table 1: Demographics Variable n % Sex Women Men Age (y) Marital status Married or living with a partner Single/divorced/separated/widow Unknown Education Elementary High school College University Missing 1 (1.8) Work status Working Not working Unknown analyses (for sex and diagnosis). No significant differences were found for any of these variables between the 2 groups. Thus, the 56 subjects who completed the QAD/F-SOPA on both occasions were retained for the analysis and no further consideration was given to those who did not. Of these 56 subjects, 25 (44.6%) were from the CLRC, 11 (19.6%) from the CRLB, 11 (19.6%) from the JRH, 5 (8.9%) from the SMHC, and 4 (7.1%) from the MPC. The demographics are listed in table 1. Most subjects were women (76.8%) and were evenly distributed across ages (between 29 79y in 10-y increments). More than 50% were married or living with a partner. At least 50% of the sample had completed high school, and nearly 50% had completed college or university. Of the subjects, 18 (30%) were working either full or part time while the remaining 38 (68%) were not working. Pain and Diagnosis Subjects reported having pain for an average of years. Fourteen subjects (25%) had had pain for less than 2 years, 17 (30.4%) between 2 and 5 years, 14 (25%) between 5 and 10 years, and 10 (17.9%) for 10 years or more. The data were missing for 1 subject. The most frequently encountered diagnosis was fibromyalgia (n 25 [32.1%]), followed by arthritic disease (n 9 [16.1%]). The mean VAS pain scores at t1 and t2 were, respectively, and ; these values did not differ significantly (P.943). Test-Retest Reliability Table 2 shows for each subscale the P values for paired t tests between the t1 and t2 scores, the Pearson r of our study (P.001), and of the original SOPA study of Jensen. 10 Our results show that, except for the disability subscale, all the r values of the QAD/F-SOPA were much higher than those calculated for the original SOPA; they fell between 0.7 and 0.9, which is interpreted as having a high correlation and a marked relationship. 27 For the disability subscale, an r value of.56 can still be considered as a moderate correlation and a substantial

4 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette 785 Subscale Table 2: Test-Retest Reliability Measures t1 t2 t2 t1 Mean SD Mean SD Mean SD P Pearson r Current Study (SOPA*) Control (.68) Harm (.66) Emotion (.68) Medication (.63) Solicitude (.68) Medical cure (.68) Disability (.63) Abbreviation: SD, standard deviation. *Pearson r of the SOPA as presented in Jensen et al. 10 P.05. relationship. 27 A chi-square independence analysis of the individual items composing the disability subscale showed a significant difference in scores between t1 and t2 for items 19 (P.048), 42 (P.037), 49 (P.155), and 52 (P.253). For the paired t test, P values for 5 of the 7 the subscales were greater than.05, indicating that the null hypothesis was accepted and there were no significant statistical differences between the mean scores at t1 and t2. This confirms the test-retest stability of the subscales harm, emotion, solicitude, medical cure, and disability, but not for the control (P.006) and medication subscales (P.013), in which significant statistical differences were found. Internal Consistency The Cronbach values are shown in table 3 along with those reported by Jensen et al 10 and Tait and Chibnall. 13 The latter sample was comprised of 395 consecutive chronic pain patients evaluated at a comprehensive pain service. In our study, all subscales showed satisfactory internal consistency estimates (between 0.7 and 0.9), except for the harm and disability subscales (.67,.64, respectively), which were close to the satisfactory level. The values of 4 subscales of our study (medication, solicitude, medical cure, control) were equal to or higher than those in the 2 other studies. On the other hand, the of the disability subscale was lower. Inspection of the individual items of the various subscales suggests that only item 4, if deleted, would give an increase of at least.05 to the overall value. Deleting this item would add.087 to the value of the emotion subscale. Moreover, it is the only subscale with a negative item-test correlation (.146). The item-test correlation was also verified for items 3 and 35 because they were particularly difficult to translate and both belonged to the disability subscale. Results show that deleting either of those items would greatly decrease the value of the disability subscale (.60 when deleting item 3,.54 for Table 3: Cronbach Coefficients Subscales Current Study Jensen et al 10 Tait and Chibnall 13 Control Harm Emotion Medication Solicitude Medical cure Disability item 35). This suggests that both items contributed significantly to the global value and that their deletion would weaken it. Translation of Disability The words disability and disabled, for the items 3 and 35 of the SOPA, were difficult to translate but finally the words handicap and handicapé were kept in the published version of the QAD/F-SOPA. 23 One goal in this study was to verify whether the use of incapacité instead of handicap, in 1 or both items, would affect the test-retest reliability and internal consistency values of the disability subscale. Items 3 and 35 were then placed twice into the questionnaire, once with handicap(é) and another time with incapacité. Using both translations for the 2 items provided 4 different combinations that could be used to calculate the Pearson r and the Cronbach for the disability subscale. Results shown in table 4 show that the r and values of the various combinations were comparable. No statistical difference was found for either items 3a versus 3b or items 35a versus 35b. To complete the analysis of items 3 and 35 (considered separately), we used a chi-square analysis to compare the answers distribution when using incapacité versus handicap, once at t1 and another time at t2, and the stability of the answers distribution between t1 and t2 when using incapacité versus handicap. Results are presented in table 5. No significant difference was found for the answer distributions between items 3a and 3b, either at t1 or at t2. Regarding the test-retest stability of the answers, item 3a showed a good and significant stability (t1 vs t2) in contrast with 3b. For items 35a and 35b, no significant statistical difference was found between them at t1, but there was a slight difference at t2. Item 35a showed the best and the more highly significant test-retest stability. Table 4: Pearson r and Cronbach Coefficients of the QAD/F-SOPA Disability Subscale, for Each Possible Combination of Translation of Items 3 and 35 Items Combination* r 3a 35b b 35b a 35a b 35a *Item 3 (I consider my pain to be a disability) translated as Je considère ma douleur comme étant un handicap (3a) or Je considère que ma douleur me cause des incapacités (3b). Item 35 (I consider myself to be disabled) translated as Je me considère handicapé(e) (35a) or Je me considère comme ayant des incapacités (35b).

5 786 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette Table 5: Chi-Square Independence Test Results for Translation Alternatives of Items 3 and 35 Item No. Time 2 (independent) Comparison of answers distribution 3a vs 3b t a vs 3b t a vs 35b t a vs 35b t * Time stability of answers 3a t1 vs t b t1 vs t a t1 vs t b t1 vs t NOTE. 3a Je considère ma douleur comme étant un handicap; 3b Je considère que ma douleur me cause des incapacités; 35a Je me considère handicapé(e); 35b Je me considère comme ayant des incapacités. *P.05. P.01. P.001. Discriminant Validity For each QAD/F-SOPA subscale to measure different constructs, the intercorrelations between the subscales should be substantially lower than the Cronbach coefficients. 12 The intersubscales correlation values of the QAD/F-SOPA subscales (at t1) are shown in table 6. The interscale correlations ranged from.363 to.550 and all correlations were considerably smaller than the values. DISCUSSION Despite the fact that the physical and psychologic conditions of the subjects might have changed over the 2-week interval, the test-retest reliability coefficients were high for all the subscales except for that of disability, for which reliability is considered as moderate. Interestingly, except for this latter subscale, the Pearson r values of the QAD/F-SOPA were higher than those found for the original SOPA. 10 There are several reasons for the differences between both studies. First, in our study, the mean test-retest interval was much shorter (2.5wk vs 6wk) than in the original study. 10 This shorter test-retest interval might have ensured a better stability of the pain condition over time and thus could account for greater test-retest reliability coefficients. Second, our population was seen in outpatient rehabilitation centers as contrasted with the subjects in Jensen et al 10 who were hospitalized. Thus, the pain conditions of the population in that study might have been more acute and not as stable as in our population, which could also affect the test-retest reliability. Third, in the QAD/F- SOPA, the wording of certain items was modified to generate positive-meaning sentences, rather than negative ones. These modifications were made after personal communication (Jan 1, 1999) from Mark Jensen, principal coauthor of the SOPA, who sent us a modified version of the SOPA in which the items 3, 11, 12, 13, 16, 19, 28, 43, and 47 were written in a positive way. For example, the original item 12, I do not expect a medical cure for my pain, was changed to I expect a medical cure for my pain. Jensen (personal communication) believes that these modifications increase the test-retest reliability of the relevant subscales because they give more consistency to the values of the Likert scales (affirmation being highest and negation being lowest). Therefore, we followed his suggestion and included the modifications in the QAD/F-SOPA, 23 which might have contributed to its improved test-retest reliability. Fourth, there may be cultural differences between our subjects and those in the original study. 10 Our sample was comprised of people of Quebec-French origin, whereas subjects in the original study were American. Indeed, various studies have shown that attitudes and beliefs may differ among various cultures, 28,29 and this may have been the case in our study. The cultural particularity of our sample may have also had a negative impact on test-retest results of the disability subscale, for which the answers to items 19, 42, 49, and 52 were not stable over time. Those items are item 19 My pain problem will always interfere with my activity level; item 42 My pain does not stop me from leading a physically active life; item 49 Pain will never stop me from doing I really want to do; and item 52 Whether or not a person is disabled by pain depends more on your attitude than the pain itself. A closer examination of these 4 sentences show they share a common feature of being written in a very affirmative way, without nuance. Indeed, extreme affirmative words are found in most of these sentences, such as never, always, not, whereas the 6 other items composing that subscale have less extreme connotations. In this study, these 4 items may show less test-retest stability because our respondents may have found it more difficult to clearly define themselves with such strong affirmations; their opinions may have been more nuanced than extreme. The use of nuance in describing feelings may be more common in French than in English; our participants may have felt more uncertain of their responses to these items and, consequently, were more prone to change their mind over time. Because disability is a culturally defined construct, its perceptions may vary within culturally diverse backgrounds. 30 Considering that the items of the SOPA disability subscale were chosen and validated to obtain an optimal reliability with an Anglo-American culture, it is possible that these items may be perceived somehow differently and less clearly by Quebec-French people, which, in turn, may lower their test-retest stability. Interestingly, although the disability subscale had low testretest reliability coefficients (see table 2, fig 1), the mean scores Table 6: Subscales Intercorrelations at t1 Subscales Control Harm Emotion Medication Solicitude Medical Cure Disability Control Harm Emotion Medication Solicitude Medical cure Disability 0.265* *P.01. P.001.

6 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette 787 Fig 1. Regression graphs of (A) medication and (B) disability categories, t1 vs t2. at t1 and t2 did not differ significantly (paired t test). However, a closer study of the results revealed a large standard deviation in this measure. Caution should be used when assessing this analysis as evidence of test-retest stability. By contrast, the medication and control subscales had high test-retest reliability coefficients, but their mean scores at t1 and t2 differed significantly. These results could seem contradictory, but in fact they are not. For example, as shown in figure 1, the medication showed good test-retest reliability because the slope of the regression line was close to 1 (.792). However, to show good test-retest stability, the y intercept should be close to 0, which at.656, is not the case here. So, in that particular case, the Pearson r did not indicate that the scores between t1 and t2 had been stable but rather that they had shown a constant change. Indeed, the paired t test indicated the means at t1 and t2 differed significantly. This example clearly shows the importance of using more than 1 type of measure when verifying the stability of results over time. Based on our results, we suggest that, in a test-retest situation, the scores of the disability, control, and medication subscales be interpreted with more caution than for the other subscales and countered-checked with a paired t test. However, for clinical use, a mean difference of 0.2 is often seen and could be acceptable. Globally, 5 subscales of the QAD/F-SOPA showed satisfactory internal consistency. Four (control, emotion, solicitude, medical cure) had an equal to or higher than the original study 16 and that of Tait and Chibnall. 13 This is rarely seen in translated tools; usually, the psychometric properties of translated tools are lower. 25 The harm and disability subscales were the least homogeneous but remained close to the acceptability limit; the elimination of an item would not enhance the homogeneity because the exclusion of any of the items would not significantly improve the values. The value for the harm subscale was comparable to that of Tait and Chibnall 13 but lower than that of Jensen et al. 10 In contrast to the 2 American studies, in which the disability subscale showed the highest or second highest internal consistency, in our study the disability subscale had the lowest internal consistency. We might ask then, if the original wording of items 3 and 19 had been retained in their original negative sense, would the internal consistency be similar to that reported in the English version? Also, could it be possible that the weak internal consistency of this subscale is a result in part of cultural differences? As previously noted, 23 Quebec Francophones seem to have more difficulty in accepting themselves as being handicapped or disabled; the same is observed in other Latin cultures. For example, in a pilot survey by Santiago et al 31 of American Latinos with disabilities, 56% were classified, according to specific criteria, as being severely disabled, yet only 37% considered themselves as such. Perhaps, for our Francophone subjects, referring to themselves as having a disability may have a more negative connotation or a greater emotional significance than it does for Americans. Therefore, it might have been more difficult for our subjects to respond consistently to these items. Again with respect to internal consistency, item 4 had a negative item-test correlation; its exclusion would substantially increase the internal consistency of the emotion subscale. However, we have decided to retain that item because that subscale shows a good value. Also, keeping that item ensures that the QAD/F-SOPA is identical and comparable to the SOPA, which is important because in the province of Quebec, research is often conducted simultaneously with Francophone and Anglophone populations. During the adaptation and translation process of the SOPA, typical difficulties arose with the translation of items 3 and 35 because the word disability could be translated either by incapacité or by handicap. 23 These 2 items are in the disability category of the SOPA and therefore we decided to retain 1 word for both items; the term we chose was handicap(é). One of our aims in this study was to determine if using the incapacité in rather than handicap, in 1 or both items, would improve the test-retest reliability and internal consistency of the disability subscale of the QAD/F-SOPA. Our results revealed marginal acceptance r and values of that specific subscale when using the word handicap. However, use of the term incapacité did not increase or decrease the acceptance values. Moreover, and unexpectedly, whether the word disability was translated as handicap or as incapacité, both items 3 and 35 contributed significantly to the value of the disability subscale. By contrast, a better test-retest stability was obtained for both items when the word handicap is used. Taken together, these results do not justify a modification of the translation of the original QAD/F-SOPA, and the word handicap(é) is kept for both items. Our study had some weaknesses. First, we used convenience sample that might not have been representative of all the chronic pain conditions for which the QAD/F-SOPA can be

7 788 TEST-RETEST RELIABILITY OF THE QAD/F-SOPA, Duquette used. Moreover, the number of subjects was not high and the distribution within the categories of diagnosis was unequal; results might have differed with a larger sample and a better distribution of diagnoses. More research would be useful to further verify the psychometric properties we measured and to explore whether the reliability of the subscales was influenced by independent variables such as demographics, diagnosis, and pain variables. CONCLUSIONS The QAD/F-SOPA is among the rare tools rigorously translated and validated in Quebec-French for assessing and understanding attitudes and beliefs about chronic pain. Globally, the QAD/F-SOPA shows good test-retest and internal consistency properties and meets the prerequisites for use as a periodical evaluation tool, both for clinical and research purposes. For most of its subscales, the tested reliability properties are comparable or equal to those in the original SOPA study 16 and that of Tait and Chibnall. 13 The disability subscale shows weaker reliability properties; further studies would help verify how it could be improved. Given that the QAD/F-SOPA can be easily integrated into routine assessment of patients attitudes toward and beliefs about their pain, it now provides therapists working with Francophone patients a new evaluation tool for evidence-based practice. Acknowledgments: Our appreciation to Mario Côté (CRLB), Martha Visintin (JHR), Aurelio Sita (MPC), and Myra Siminovitch (SMHC) for their clinical collaboration, and to Manon Lemelin, for data entry. References 1. Merskey H, Lindblom U, Mumford JM, Nathan PW, Sunderland S. Pain terms. In: Merskey H, Bogduk M, editors. Classification of chronic pain. 2nd ed. Seattle: IASP Pr; p Turk DC. Assess the person, not just the pain. Pain Clin Updates 1993;I(3). Available at: Accessed August 4, Melzack R, Wall PD. The challenge of pain. New York: Basic Books; Williams DA, Thorn BE. An empirical assessment of pain beliefs. Pain 1989;36: Jensen MP, Karoly P. Control beliefs, coping efforts, and adjustment to chronic pain. J Consult Clin Psychol 1991;59: Feuerstein M, Beattie P. Biobehavioral factors affecting pain and disability in low back pain: mechanisms and assessment. Phys Ther 1995;75: Shutty MS, DeGood DE, Tuttle DH. Chronic pain patients beliefs about their pain and treatment outcomes. Arch Phys Med Rehabil 1990;71: Strong J, Ashton R, Cramond T, Chant D. Pain intensity, attitude and function in back pain patients. Aust Occup Ther J 1990;37: Schwartz DP, DeGood DE, Shutty MS. Direct assessment of beliefs and attitudes of chronic pain patients. Arch Phys Med Rehabil 1985;66: Jensen MP, Turner JA, Romano JM, Lawler B. Relationship of pain-specific beliefs to chronic pain adjustment. Pain 1994;57: Jensen MP, Karoly P, Chant D. The development and preliminary validation of an instrument to assess patient s attitudes towards pain. J Psychosom Res 1987;31: Strong J, Ashton R, Chant R. The measurement of attitudes towards and beliefs about pain. Pain 1992;48: Tait R, Chibnall JT. Development of a brief version of the Survey of Pain Attitudes. Pain 1997;70: Lai YH, Keefe FJ, Sun WZ, T et al. Relationship between painspecific beliefs and adherence to analgesic regimens in Taiwanese cancer patients: a preliminary study. J Pain Symptom Manage 2002;24: Haythornthwaite JA, Menefee LA, Heinberg LJ, Clark M. Pain coping strategies predict control over pain. Pain 1998;7: Jensen MP, Turner JA, Romano JM. Correlates of improvement in multidisciplinary treatment of chronic pain. J Consult Clin Psychol 1994;62: Duquette J, Litowski J. [Traduction francophone du Survey of Pain Attitudes : processus scientifique et réflexions sur la signification socioculturelle des termes employés dans la traduction] [French]. In: Résumés du 5ième Congrès Québécois de la Réadaptation en Déficience Physique; 2000 Oct 19-20; Trois-Rivières (QC). p Coughlin AM, Badura AS, Fleischer TD, Guck TP. Multidisciplinary treatment of chronic pain patients: its efficacy in changing patient locus of control. Arch Phys Med Rehabil 2000;81: Jensen MP, Karoly P. Pain-specific beliefs, perceived symptom severity, and adjustment to chronic pain. Clin J Pain 1992;8: Jensen MP, Romano JM, Turner JA, Good AB, Wald LH. Patient beliefs predict patient functioning: further support for a cognitivebehavioral model of chronic pain. Pain 1999;81: Lefort SM. Test of Braden s self-help model in adults with chronic pain. J Nurs Scholarsh 2000;32: Grisart J, Masquelier E, Ophoven E. [Adaptation et validation en français d un questionnaire d attitudes vis-à-vis de la douleur chronique: étude préliminaire] [French]. Douleur Analg 1999;12: Duquette J, McKinley PA, Litowski J. Traduction et pré-test du Questionnaire sur les attitudes envers la douleur (Survey of Pain Attitudes). Rev Québécoise d Ergothér 2001;10: Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 2nd ed. New York: Oxford Medical; Vallerand RJ. [Vers une méthodologie de validation transculturelle de questionnaires psychologiques: implications pour la recherche en langue française] [French]. Can J Psychol 1989;30: Huskisson EC. Visual analogue scales. In: Melzack R, editor. Pain measurement and Assessment. New York: Raven Pr; p Martin P, Bateson P. Measuring behaviour. An introductory guide. 2nd ed. Cambridge: Cambridge Univ Pr; Bates MS, Rankin-Hill L, Sanchez-Ayendez M. The effects of the cultural context of health care on treatment of and response to chronic pain and illness. Soc Sci Med 1997;45: Sanders S, Brena S, Spier C, Beltrutti D, McConnell H, Quintero O. Chronic low back pain patients around the world: cross-cultural similarities and differences. Clin J Pain 1992;8: Schaller J, Parker R, Garcia SB. Moving toward culturally competent rehabilitation counseling services: issues and practices. J Appl Rehabil Counsel 1998;29: Santiago AM, Villarruel FA, Leahy MJ. Latino access to rehabilitation services: evidence from Michigan. Am Rehabil 1996;22: 10-7.