Increased use of safety-seeking behaviors in chronic back pain patients with high health anxiety

ARTICLE IN PRESS

Behaviour Research and Therapy 45 (2007) 2821–2835 www.elsevier.com/locate/brat

Increased use of safety-seeking behaviors in chronic back pain patients with high health anxiety Nicole K.Y. Tanga,, Paul M. Salkovskisa, Elena Poplavskayaa, Kelly J. Wrighta, Magdi Hannab, Joan Hesterb a

Department of Psychology (PO 77), Institute of Psychiatry, King’s College London, de Crespigny Park, Denmark Hill, London SE5 8AF, UK b Pain Relief Research Unit, King’s College Hospital, UK Received 29 January 2007; received in revised form 19 April 2007; accepted 3 May 2007

Abstract Many patients with chronic pain also exhibit elevated levels of health anxiety. This study examined the effect of health anxiety on the use of safety-seeking behaviors (SSBs) in pain-provoking situations. Participants were 20 chronic back pain patients with high health anxiety (Group H), 20 with low health anxiety (Group L) and 20 pain-free controls (Group C). Two physical tasks were video recorded, and compared both for overt pain behavior (identified by blind observers following a standardized procedure) and for the occurrence of SSB (identified by showing the participants video playback and asking them to specify motivation for all actions/behaviors displayed during the tasks). While there were no differences in the display of overt pain behaviors, Group H deployed a greater number of SSBs than Groups L and C. This finding held true for both tasks and remained significant when concurrent pain and mood ratings were statistically controlled for. SSB was correlated with catastrophizing thoughts but not pain intensity; pain intensity was correlated with overt pain behavior but not catastrophizing. Taken together, these findings suggest that SSB is distinct from overt pain behavior and may be a defining characteristic of chronic pain patients reporting high levels of health anxiety. r 2007 Elsevier Ltd. All rights reserved. Keywords: Chronic pain; Health anxiety; Safety-seeking behavior; Pain behavior; Catastrophizing; Fear avoidance

Introduction Health anxiety is characterized by severe and persistent concerns about health. In diagnostic terms, people who are disabled by their health anxiety can receive the diagnosis of ‘hypochondriasis’ (American Psychiatric Association, 2000; World Health Organization, 1994). Although cases meeting the criteria of hypochondriasis are rare in the community (.2%; Looper & Kirmayer, 2001), lower levels of health anxiety are extremely common in individuals who are physically ill or where there is reason for people to regard their longer-term health status as ambiguous. Elevated health anxiety has been observed among patients undergoing treatment Corresponding author. Tel.: +44 207 848 5010; fax: +44 207 848 5037.

E-mail address: [email protected] (N.K.Y. Tang). 0005-7967/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2007.05.004

ARTICLE IN PRESS 2822

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

for chronic pain (Hadjistavropoulos, Owens, Hadjistavropoulos, & Asmundson, 2001). In a recent study, Rode, Salkovskis, Dowd, and Hanna (2006) examined the occurrence of severe and persistent health anxiety in patients consecutively recruited from a specialist pain clinic. On the basis of responses to the Short Health Anxiety Inventory (SHAI; Salkovskis, Rimes, Warwick, & Clark, 2002), these authors conservatively estimated the prevalence of hypochondriasis (i.e., SHAIX18) to be at least 37% and of severe health anxiety (i.e., SHAIX15) to be at least 51% among patients with chronic pain. These figures were much higher than the rates observed in both controls with pain (hypochondriasis: 3%; severe health anxiety: 18%) and controls without pain (hypochondriasis: 1%; severe health anxiety: 10%), consistent with the possible role of health anxiety as a causal and/or maintaining factor in chronic pain. The value of linking chronic pain to health anxiety lies in the possibility of applying theoretical and empirical work across these currently distinct areas. The cognitive theory of health anxiety (Salkovskis & Bass, 1997; Salkovskis & Warwick, 1986; Warwick & Salkovskis, 1990) suggests that people with persistent and severe health anxiety have a relatively enduring tendency to misinterpret bodily symptoms and health-relevant information as evidence that they are suffering from (or are at high risk of developing) a serious physical illness. Such negative/catastrophic interpretation of health-relevant information is characteristic of people experiencing high levels of health anxiety, and results in a range of reactions, which can serve to maintain and increase their negative beliefs. These reactions can manifest in multiple domains including the cognitive (e.g., selective and enhanced attention to illness-related information, negative appraisal of symptoms), affective (e.g., anxiety, depression, anger), physiological (e.g., increased bodily arousal, sleep disturbance) and behavioral (e.g., increased checking/reassurance seeking) systems. Though not intended, such reactions increase the sufferer’s preoccupation with health and may worsen hypochondriacal symptoms in the longer term (Clark, 1999; Salkovskis, 1996b; Salkovskis, Clark, & Gelder, 1996). It has been proposed that the cognitive–behavioral model of health anxiety can be applied to chronic pain patients in whom health anxiety is a major feature of clinical presentation (Hadjistavropoulos et al., 2001; Rode et al., 2006). The enduring tendency of health anxious patients to misinterpret pain sensations either as an indication of an underlying disease process (e.g., cancer) or as a sign that they may be vulnerable to or have already sustained serious physical damage (e.g., a crumbling spinal column) is likely to intensify the pain experience through similar cognitive, affective, physiological and behavioral mechanisms found in hypochondriasis. If this is so, then it should be possible to adapt the treatment strategies which have proven so effective in the treatment of health anxiety (Salkovskis, Warwick, & Deale, 2003) for use in people suffering from chronic pain syndromes and co-existing health anxiety. Such an approach is consistent with the recognition of chronic pain as a heterogeneous disorder, and the suggestion that matching treatment to patients’ psychological characteristics may improve the efficacy of cognitive behavior therapies for chronic pain (Vlaeyen & Morley, 2005). In a well-designed series of experimental studies, Hadistavropoulos and her colleagues (see Hadjistavropoulos & Hadjistavropoulos, 2003 for a review) have been able to demonstrate that, compared to non-health anxious individuals, health anxious pain patients are more likely to show selective attention to bodily sensations, detect heightened physical symptoms, report more intense pain, exhibit lower pain tolerance, report greater anxiety, engage in catastrophizing thinking and seek medical reassurance. These findings not only illustrate the interaction between health anxiety and pain but also demonstrate the clinical benefits of psychological subtyping according to patients’ levels of health anxiety. However, full application of the health anxiety model to the study of chronic pain relies on research validating the specific mechanisms hypothesized to be involved in maintaining the syndrome. One major prediction of the model that awaits empirical evaluation is the use of ‘‘safety-seeking behavior (SSB)’’ among chronic pain patients with high health anxiety. The concept of ‘‘SSB’’ has its origin in the anxiety disorder literature. It is defined as a behavior performed as a strategy intended to prevent or minimize a feared catastrophe (Salkovskis, 1996b; Salkovskis et al., 1996, 1986). SSB refers not only to total avoidance of a feared situation (e.g., ‘‘I won’t lift anything heavy because that would damage my back’’) but also to escape behaviors which are performed to terminate exposure to a feared situation (e.g., ‘‘I’ll stop doing these household chores at the first sign of pain in case I damage my back’’) and subtle avoidance behaviors which are performed to prevent the feared catastrophe while remaining

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2823

in the feared situation (e.g., ‘‘I’d make sure that I don’t lean forward when I’m holding something heavy because that might break my spine’’). Thus, while there is some degree of overlap, SSB is theoretically distinguished from ‘‘overt pain behavior’’ described in the operant theory (e.g., Fordyce, 1982) and ‘‘pain avoidance’’ outlined in the existing cognitive–behavioral models of chronic pain (e.g., Lethem, Slade, Troup, & Bentley, 1983; Philips, 1987; Vlaeyen, Kole-Snijders, Boeren, & van, 1995; Vlaeyen & Linton, 2000). The introduction of the concept of SSB has led to significant advances in the theoretical understanding and clinical management of a range of anxiety disorders (e.g., Salkovskis, Clark, Hackmann, Wells, & Gelder, 1999; Wells, Clark, Salkovskis, & Ludgate, 1995); its specific application to the study of chronic pain under the health anxiety framework therefore needs further development (Sharp, 2001). This is the purpose of the present study. An observational study was designed to examine the use of SSB in chronic pain. SSB was elicited by asking the participants to perform two physical tasks that were potentially pain-provoking. Video playback was used to elicit details of the pain behaviors exhibited by the participants during the tasks, including the thoughts and emotions, which the participants understood as being involved in the motivation of these behaviors. In order to separate the effects of pain and health anxiety on the deployment of SSB, participants of this study included a group of pain clinic patients with high health anxiety, a group of pain clinic patients with low health anxiety and a group of pain-free volunteers from the community. Given the key role of SSB in the cognitive model of the maintenance of health anxiety, it was predicted that pain patients with high health anxiety would employ more SSB when they were asked to perform a task that they believe is likely to provoke pain relative to both pain patients with low health anxiety and pain-free controls. In this study, although the participants were asked to perform two bag-carrying tasks, they were given the opportunity to decide whether they wanted to progress to the second task after completing the first one. Based on previous theoretical work on fear avoidance in chronic pain (Asmundson, Norton, & Vlaeyen, 2004), it was predicted that patients with high health anxiety would be more inclined to avoid physical activity given a chance than patients with low health anxiety and pain-free controls. Finally, while SSB is thought to be a construct distinguished from the pain behavior as described in the behavioral theory of pain, this assumption has not been evaluated empirically. The current study therefore also aimed to investigate to what extent SSB and overt pain behavior were distinct. Method Participants Participants of this study were 40 chronic back pain patients from a pain clinic and 20 pain-free volunteers recruited from the local community. Chronic back pain patients were recruited from an outpatient pain clinic within an inner city teaching hospital. Patients were included in the study if they (i) were aged between 18 and 65, (ii) were English speaking, (iii) had been having back pain for at least 6 months and (iv) had no severe learning disability. Patients were excluded from the study if they (v) had a terminal illness (e.g., cancers, HIV/AIDS), (vi) had previously identified severe psychopathological co-morbidity (e.g., schizophrenia, bipolar disorder or substance dependence disorders), (vii) suffered from traumatic brain injury, (viii) were wheelchair bound/on crutches or (ix) were enrolled in a drug trial at the time of the study. A total of 44 back pain patients met the criteria and took part in the study. However, four participants’ data could not be used for analysis due to video file corruption. Pain-free volunteers were recruited by approaching adult passers-by on the street and by placing recruitment posters in multiple locations within the catchment area of the hospital. Individuals who expressed interest in the research were recruited to the study if they did not have a history of pain problems for 41 month and if they reported only negligible pain or a total absence of pain in the past week. Of the 27 volunteers who responded to the recruitment drive, six could not be contacted and one volunteer with migraine and sickle cell disease was not included in the study. Thus, the final sample comprised 20 pain-free volunteers (Group C), 20 chronic pain patients with high health anxiety (Group H) and 20 chronic pain patients with low health anxiety (Group L) (see the section below for details of the group assignment/definition).

ARTICLE IN PRESS 2824

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

Measures Standardized questionnaires All participants were asked to provide basic demographic information and complete the measures described below: SHAI (Salkovskiset al., 2002): To measure baseline health anxiety, we asked the participants to complete the SHAI with reference to their general concern about health over the past 6 months. The SHAI consists of 14 groups of four statements, each of which is ranked on a 4-point scale (0–3) and summed to give a total score that ranges from 0 to 42. A cut-off point of 18 or higher in the SHAI can reliably identify people meeting the criteria for hypochondriasis as defined by the DSM-IV (American Psychiatric Association, 2000), while a score between 15 and 17 represents a mixture of both hypochondriacal patients and people who are very health anxious but just missed the criteria for clinical diagnosis (Rode et al., 2006). The SHAI has good internal consistency (a ¼ .89) comparable to that of its long version and is sensitive to treatment for hypochondriasis if referenced ‘‘over the past week’’. In this study, the cut-off score of 15 was used to assign the chronic pain patients to either the high (scoringX15; Group H) or low (scoring o15; Group L) health anxiety group. Short-Form McGill Pain Questionnaire (SF-MPQ; Melzack, 1987). To assess baseline pain levels, we asked the participants to complete the SF-MPQ with reference to their pain experience over the past week. The SF-MPQ comprises 15 representative words from the Sensory (11 items) and Affective (4 items) categories of the standard MPQ (Melzack, 1975). Each pain descriptor is ranked on a 4-point intensity scale (0 ¼ ‘‘None’’, 1 ¼ ‘‘Mild’’, 2 ¼ ‘‘Moderate’’, 3 ¼ ‘‘Severe’’). The sum of these rank values generates a sensory pain score (range ¼ 0–33) and an affective pain score (range ¼ 0–12). The SF-MPQ also includes a 10-cm horizontal visual analogue scale (VAS) as an assessment of the overall pain intensity. The SF-MPQ has well-established validity and reliability (Melzack & Katz, 2001) and correlates highly with the major painrating indices of the standard MPQ (Melzack, 1987). In this study, those participants who reported no persistent pain problems and scored o2 on the VAS of the SF-MPQ were assigned to the control group (Group C). Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983). To index baseline moods, we asked the participants to complete the HADS based on how they had been feeling over the past week. The HADS is a 14-item questionnaire that measures the severity of anxiety and depression (7 items each). Each item is rated on a 4-point scale (0–3), generating a total anxiety/depression score that ranges from 0 to 21. The HAD has good internal consistency (mean Cronbach’s a for HADS-A ¼ .83 and mean Cronbach’s a for HADS-D ¼ .82; Bjelland, Dahl, Haug, & Neckelmann, 2002). Scores of X8 identify anxiety/depression cases (as defined by ICD-9) with around 80% sensitivity and specificity (Bjelland et al., 2002). Catastrophising in Pain Scale (CIPS; Rode, 2003). Based on the cognitive model of health anxiety, it is predicted that pain patients with high health anxiety should exhibit greater levels of pain-relevant fear and catastrophizing compared with those with low health anxiety and pain-free controls. We incorporated the CIPS, a 20-item scale, to measure catastrophic thinking in the current sample. The items describe physical (e.g., ‘‘The pain is caused by a serious disease my doctors have missed’’) or mental (e.g., ‘‘The pain will drive me insane’’) pain-related catastrophes, and the participants were asked to indicate to what extent these catastrophic concerns occurred in their mind ‘‘the last time the pain was really bad’’. Five of the 20 items were taken from the fear subscale of the Pain Anxiety Symptoms Scale (PASS; No. 1, 2, 3, 19 and 20; McCracken, Zayfert, & Gross, 1992). Another 3 PASS items (No. 4, 13 and 16) were adapted so that the items reflected more catastrophic outcomes. The remaining 12 items were constructed by clinical psychologists, doctors and researchers who specialized in chronic pain, on the basis of their understanding of catastrophizing as an extreme type of negative appraisal process. Each item is rated on a 5-point scale (0–4; 0 ¼ Thought never occurs; 4 ¼ Thought always occurs), generating a total score that ranges from 0 to 80. The CIPS has been validated in a sample of 342 chronic pain patients and demonstrated excellent internal consistency (Cronbach’s a ¼ .92), good test–retest reliability over a 1-day interval (r ¼ .93; n ¼ 38) and a significant difference in scores between chronic pain patients (M ¼ 21.4, SD ¼ 15.1) and non-clinical controls (M ¼ 11.0, SD ¼ 9.3).

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2825

Bag-carrying task A bag-carrying task was adapted from the bag-holding task developed by Vlaeyen et al. (1995) to elicit SSB. The bag-carrying task was designed as a circuit training exercise to be performed within a laboratory. Each ‘‘task’’ is made up of ‘‘cycles’’, each of which involves (i) walking from A to B (distance ¼ 3 m), (ii) lifting a moderately heavy shopping bag for 1 min at B, (iii) carrying the bag from B to A, then from A back to B, (iv) unloading the bag to the floor at B and (v) walking from B to A, the starting point. The variations were added to the original bag-holding task so as to allow the study of SSB under both static and dynamic conditions. The weight of the bag was 5 kg for female and 6 kg for male chronic back pain patients, 10 kg for female and 12 kg for male pain-free volunteers. Although this weight arrangement did not allow for individual differences in muscle strength, it adjusted for general differences relating to gender and patient status (Rode, Salkovskis, & Jack, 2001). The bag-carrying task tended to be mildly pain-provoking, enabling us to observe and study SSB. The participants were asked to do the bag-carrying tasks twice in succession. For each task, the participants were asked to do ‘‘as many cycles as possible until [they] no longer [felt] safe/comfortable if [they were] to continue’’. Tasks 1 and 2 were separated by a break, during which the participants decided whether they wanted to attempt Task 2 based on their assessment of their own physical state. This design permitted the measure of the proportion of participants who would completely avoid physical activities in the acute presence of pain. Pain, Mood and Thoughts Record Sheet A Pain, Mood and Thoughts Record Sheet was administered to the participants immediately after each bagcarrying task. On this record sheet, the participants were asked to give ratings of the levels of pain (0 ¼ ‘‘No pain at all’’, 10 ¼ ‘‘Worst possible pain’’), general anxiety (0 ¼ ‘‘Not at all’’, 4 ¼ ‘‘Extremely’’) and depression (0 ¼ ‘‘Not at all’’, 4 ¼ ‘‘Extremely’’) experienced during the performance of the task. In addition, they were asked to note down thoughts that ran through their mind when they were performing the task [‘‘What were you thinking just now during Task 1 (or 2)?’’]. The thoughts were later subjected to blind rating by two independent raters for their level of catastrophizing (0 ¼ ‘‘Not at all catastrophizing’’, 5 ¼ ‘‘Extremely catastrophizing’’). The intraclass correlations between the catastrophizing ratings provided by the two raters for both tasks were high (Task 1: r ¼ .87, po.001; Task 2: r ¼ .78, po.001). The average of these ratings was taken for subsequent analyses. Safety-behavior Record Sheet A Safety-behavior Record Sheet was administered to the participants in the final part of the experiment, during which the participants were shown a video of their session and asked to systematically note down any actions/behaviors they had displayed ‘‘because they were experiencing pain or expecting pain’’. For each entry they were asked to answer two follow-up questions to (a) indicate their recollection of the reasons behind their actions (‘‘What was this action/behavior for? What did you think would have happened had you not performed this action/behavior?’’) and (b) rate how anxious they would have been had they not acted the way they did (response scale: 0 ¼ ‘‘Not at all anxious’’, 10 ¼ ‘‘Very anxious’’; referred to below as ‘‘Would-have Anxiety Rating’’). Based on previous definitions of SSB (e.g., Salkovskis, 1991, 1996b), those actions that were carried out (1) to pre-emptively minimize discomfort/pain or prevent (re)injury/accidents and (2) to reduce anxiety (i.e., ‘‘Would-have Anxiety Rating’’X5) were considered as SSBs. The total number of SSBs, for both tasks, was derived using these two criteria by two independent blind raters. The reliability of this method was excellent; the degree of agreement between the two raters was high for both Task 1 (k ¼ 1, po.001) and Task 2 (k ¼ .94, po. 001). The average number of SSBs across the two observers was taken for subsequent analyses. Design and procedure The protocol of this study received full ethical approval from the local Research Ethics Committee. All participants gave informed consent before participating. The experimental session was approximately 90 min long and took place in a closed circuit television (CCTV)-covered experimental room. Given that pain

ARTICLE IN PRESS 2826

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

behaviors may be reinforced by a supportive social environment (Flor, Kerns, & Turk, 1987; Fordyce, Shelton, & Dundore, 1982), all participants took part individually in the absence of a spouse/partner. The experimental session was structured into three parts. Part one—questionnaire and instructions After obtaining informed consent, participants were asked to complete a set of questionnaires that includes the SHAI, SF-MPQ, HADS and CIPS to assess their pain and psychological characteristics. They were introduced to the lab and were explained the bag-carrying task. Before commencing the task, the participants were given 5 min to warm up and familiarize themselves with the environment. This is to minimize performance anxiety, enabling the participants to perform the experimental task in a natural manner. Part two—bag-carrying tasks The participants began the first bag-carrying task soon after the experimenter had exited the experimental room. The experimenter monitored and video-recorded their performance via the CCTV in an adjacent control room. Although the participants were asked to complete as many cycles as they could, for ethical reasons the task was terminated once the participants had reached the pre-determined upper performance limit, 4 cycles. After Task 1, the experimenter re-entered the room to administer the first Pain, Mood and Thoughts Record Sheet. Conversation between the experimenter and the participants was restricted at this point. The experimenter gave instructions following a standardized protocol and was not allowed to offer any verbal or non-verbal encouragement to the participants. There was then a 5-min break, during which the participants were asked whether they would like to attempt Task 2. Those who chose to do Task 2 went through a procedure identical to the one described above, i.e., to perform the second bag-carrying task and then complete the second Pain, Mood and Thoughts Record Sheet. Those who decided to quit at this point skipped Task 2 and proceeded straight to video reviewing. Part three—video reviewing The video reviewing took place in the control room. Before watching the video,1 the participants were given the Safety-behavior Record Sheet. They were instructed to look out for and note down any actions/behaviors displayed because they were experiencing/expecting pain while they were reviewing their performance. This is to minimize potential memory failure and reporting bias associated with self-report. The video was paused at the end of each task, giving the participants sufficient time to specify the reasons behind their actions and give a ‘‘Would-have Anxiety Rating’’. Finally, the participants were thoroughly debriefed and paid before leaving the session. Post-experiment—overt pain behavior rating As part of the effort to clarify the relationship between SSB and overt pain behaviors, two independent blind observers were asked to (1) view the videos and (2) record any verbal/non-verbal concomitant pain behaviors (e.g., guarding, bracing, rubbing, grimacing, sighing), following the observation and coding methodology validated by Keefe and Block (1982). However, instead of scoring the occurrence/nonoccurrence of pain behaviors, the blind observers were instructed to note the frequency of the five pain behaviors. Interrater reliability calculated for the total number of pain behavior was moderate (Task 1: r ¼ .48, po.001; Task 2: r ¼ .26, po.05). The average numbers of overt pain behavior across the two observers were taken for subsequent analyses. Treatment of data As the three groups were set up according to the levels of pain and health anxiety experienced by the participants, one-way ANOVA (and Tukey/Dunnett’s T3 post-hoc comparisons) was carried out for their scores on the SHAI, SF-MPQ, HADS and CIPS to establish the criterion validity of the groups. For the rest 1

The video contained the participants’ performance in both tasks. Length of the video varied (range: 1–15 min) depending on the number of tasks and cycles the participants completed.

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2827

of the demographic variables, w2 tests were performed to detect whether there were any significant differences between the three groups, with the exception of age (analyzed with a one-way ANOVA) and pain duration (analyzed with an independent sample t-test). Identical statistical treatments were given to data obtained from both Tasks 1 and 2. Three sets of analysis were conducted for each task: First, a series of one-way ANOVAs were conducted to check whether the three groups differ on the pain and mood ratings obtained during the task. Second, controlling for the differences in the levels of pain, anxiety and depression, a univariate ANCOVA was performed to detect whether there were any significant between-group differences in the utilization of SSB during the task. Third, a univariate ANOVA plus an ANCOVA (controlling for the levels of pain, anxiety and depression) were performed to detect whether there were any significant between-group differences in their display of overt pain behavior during the task. Data for the two tasks were analyzed separately in order to avoid potential sampling bias introduced by selective attrition (Ahern & Le Brocque, 2005), given there was an option after Task 1 to not participate in Task 2. To examine whether chronic pain patients could be distinguished from pain-free volunteers on the basis of total physical activity avoidance, the proportion of participants totally avoided in Task 2 was compared across the three groups using a w2 test. Correlational analyses were carried out to examine the relationship between (1) the utilization of SSB and the display of overt pain behavior and (2) self-reported pain, anxiety, depression and catastrophizing thoughts that occurred during the bag-carrying tasks. This was pre-planned and was done only with chronic pain patients. Results Participant characteristics Consistent with other studies conducted in the UK, the current sample consisted of largely white (72%), middle-aged (M age ¼ 44.7 years) participants with a 2–1 female to male ratio. Thirty-five percent of the sample were educated up to tertiary level, 38% were married/cohabiting and 47% were on sick leave/ unemployed at the time of testing. There were no significant differences between the three groups on these demographic variables (see Table 1). However, as expected, participants with chronic pain (Groups H and L) had significantly higher levels of pain, health anxiety, anxiety, depression and pain-related catastrophizing compared to participants without chronic pain (Group C). The pain duration did not differ between the two chronic pain patient groups, who on average had been suffering from back pain for 10 years. The validity of the group assignment of chronic pain patients was supported by the finding that Group H had significantly higher mean scores on the SHAI and the CIPS compared to Group L. While the two groups of chronic pain patients did not differ in terms of their scores on the SF-MPQ sensory scale and the pain intensity VAS, Group H scored significantly higher on the SF-MPQ affective subscale and on both the anxiety and depression subscales of the HADS compared to Group L. Sample SSB Given that SSB has never been specifically studied in the context of chronic pain, Table 2 presents several examples of SSB identified in the current study, along with the strategy rationale and the ‘‘Would-have Anxiety Rating’’ provided by the participants. SSBs were defined as actions that were carried out (1) to preemptively minimize discomfort/pain or prevent (re)injury/accidents and (2) to reduce anxiety (i.e., ‘‘Wouldhave Anxiety Rating’’X5). Task 1 Pain intensity, mood and thoughts ratings Table 3 presents the ratings of pain, anxiety and depression by tasks and group. There were significant between-group differences for the pain, anxiety, depression and catastrophizing thought ratings obtained in Task 1. Follow-up tests indicated that pain ratings were higher in Group H (po.001) and Group L

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2828 Table 1 Participant characteristics

Demographic characteristics Age

Group H (n ¼ 20)

Group L (n ¼ 20)

Group C (n ¼ 20)

45.4 (11.8)

46.6 (9.4)

42.1 (14.0)

Post-hoc comparisons F (2, 57) .76 w2 (2, n ¼ 60) .15 .66 3.08

p .47 p .93 .72 .22 .75 .15 p .60

H¼L

p o.001 o.001 o.001 o.001 o.001 o.001 o.001

H4L4C H ¼ L4C H4L4C H ¼ L4C H4L4C H4L4C H4L4C

Gender (% female) Ethnicity (% white) Education (% with university education) Marital status (% married/cohabiting) Employment (% unemployed)

65 65 30

65 75 25

70 75 50

35 50

45 60

35 30

Pain duration (years)

10.5 (10.9)

8.9 (7.7)

N/A

.56 3.75 t (37) .53

Pain and psychological characteristics SHAI SF-MPQ—sensory SF-MPQ—affective Pain intensity VAS HADS—anxiety HADS—depression CIPS

20.4 (6.0) 15.1 (7.5) 5.9 (3.1) 6.8 (2.6) 11.3 (3.6) 9.8 (3.2) 35.0 (13.8)

9.7 (3.2) 11.5 (6.9) 3.1 (2.9) 5.1 (2.6) 8.6 (3.0) 6.2 (4.0) 16.5 (9.8)

6.0 (2.6) .4 (.9) .4 (1.2) .1 (.3) 3.3 (2.6) 1.2 (2.0) 6.0 (7.1)

F (2, 57) 63.12 33.73 24.11 51.56 34.16 36.61 38.56

H¼L¼C H¼L¼C H¼L¼C H¼L¼C H¼L¼C H¼L¼C

Note: Means are presented with standard deviations in parentheses. Group H ¼ chronic back pain patients with high health anxiety; Group L ¼ chronic back pain patients with low health anxiety; Group C ¼ pain-free volunteers. SHAI ¼ Short Health Anxiety Inventory. SF-MPQ—sensory ¼ Short Form McGill Pain Questionnaire—sensory subscale. SF-MPQ—affective ¼ Short Form McGill Pain Questionnaire—affective subscale. VAS ¼ Visual Analogue Scale. HADS—anxiety ¼ Hospital Anxiety and Depression Scale— anxiety subscale. HADS—depression ¼ Hospital Anxiety and Depression Scale—depression subscale. CIPS ¼ Catastrophising in Pain Scale. N/A ¼ not applicable. Table 2 Examples of safety-seeking strategies employed by chronic pain patients during the bag-carrying tasks

1 2 3

Description of the strategy

Rationale for the strategy

Anxietya

‘‘Lifting and loading the bag cautiously with back kept straight’’ ‘‘Standing with weight on one side’’

‘‘To prevent pain and sudden jolt. Lifting could have put me out of circulation and my back would have seized up.’’ ‘‘I can’t stand still with both feet on the floor otherwise pain would have increased and I’d be in agony.’’ ‘‘So that no weight will be on my ‘bad’ side. This is to avoid something drastic from happeningy I’ll panic. I’ll die.’’ ‘‘To keep muscles in back moving otherwise tension will build up in my neck leading to severe headache.’’ ‘‘To protect my spine and avoid pain.’’

10

6 7

‘‘Holding and pushing my back to the ‘good’ side’’ ‘‘Constantly rocking. Shifting weight between my legs’’ ‘‘Using the strength of the arm to hold bag instead of the strength of my body’’ ‘‘Holding bag high up at chest level’’ ‘‘Taking deep breath’’

8

‘‘Continually tensing my stomach muscle’’

9

‘‘Concentrating hard on the task’’

4 5

a

‘‘To prevent pain from getting worse and slipping a disc.’’ ‘‘To increase control otherwise I would have been overwhelmed by pain.’’ ‘‘To protect my lower back. If not, I would have suffered more pain later.’’ ‘‘To block out pain and to prevent myself from doing something silly and hurt myself.’’

8 10 7 7 10 9 8 7

Would-have Anxiety Rating (0 ¼ ‘‘Not at all anxious’’, 10 ¼ ‘‘Very anxious’’).

(po.001) compared to Group C, and anxiety ratings were significantly higher in Group H compared to Group L (po.05) and Group C (po.001). Though there was a main effect for group on the depression ratings, post-hoc tests indicated no significant individual between-group differences. There was also a trend

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2829

Table 3 Pain, mood and thought ratings by tasks and groups Task 1

Pain intensity (0–10)a Anxiety (0–4)b Depression (0–4)c Catastrophizing (0–5)d

Task 2

Group H (n ¼ 20)

Group L (n ¼ 20)

Group C (n ¼ 20)

F (2,57)

Group H (n ¼ 17)

Group L (n ¼ 18)

Group C (n ¼ 20)

F (2,57)

6.4 (2.8)

4.4 (2.8)

.4 (.9)

33.2*** H ¼ L4C

5.9 (2.6)

4.9(2.8)

.6 (.8)

30.8*** H ¼ L4C

1.4 (1.5)

.4 (.5)

.1 (.2)

11.0*** H4L ¼ C

1.3 (1.3)

.5 (1.0)

.0 (.0)

8.8**

H4C

.7 (1.4)

.1 (.3)

.1 (.2)

3.8*

H¼L¼C

.9 (1.4)

.3 (.8)

.1 (.2)

4.2*

H¼L¼C

.8 (1.4)

.5 (.9)

.0 (.1)

3.3*

H¼L¼C

.4 (.8)

.2 (.3)

.2 (.8)

.6

H¼L¼C

Between-group comparisons

Between-group comparisons

Note: Means are presented with standard deviations in parentheses. Group H ¼ chronic back pain patients with high health anxiety; Group L ¼ chronic back pain patients with low health anxiety; Group C ¼ pain-free volunteers. a Response scale: 0 ¼ No pain at all, 10 ¼ Worst possible pain. b Response scale: 0 ¼ ‘‘Not at all’’, 1 ¼ ‘‘A little’’, 2 ¼ ‘‘Moderately’’, 3 ¼ ‘‘Quite a bit’’, 4 ¼ ‘‘Extremely’’. c Response scale: 0 ¼ ‘‘Not at all’’, 1 ¼ ‘‘A little’’, 2 ¼ ‘‘Moderately’’, 3 ¼ ‘‘Quite a bit’’, 4 ¼ ‘‘Extremely’’. d Response scale: 0 ¼ ‘‘Not at all catastrophizing’’, 5 ¼ ‘‘Extremely catastrophizing’’. *po.05. **po.01. ***po.001.

Group H Mean number of SSB utilized

3

Group L

2.7

Group C 2.82

2 * 1.25

* 1.28

1 * 0.3

* 0.3 0 Task 1

Task 2

Fig. 1. Mean number of SSBs utilized during the bag-carrying tasks by groups. Two analyses of covariance were performed to detect between-group differences, controlling for the effects of pain, anxiety and depression. Asterisks indicate a difference with Group H at the po.01 level.

that thoughts recorded by Group H were rated to be more catastrophizing than thoughts recorded by Group C (p ¼ .067). Utilization of SSB A univariate ANCOVA was performed to detect any significant differences between groups in their utilization of SSB controlling for the effects of concurrent pain, anxiety and depression. For Task 1, there were significant differences between groups [F (2, 54) ¼ 7.8, po.01] in their utilization of SSB (see Fig. 1). Post-hoc comparisons indicated that Group H employed a greater number of SSB than both Group L (po.01) and Group C (po.01). Display of overt pain behavior Overt pain behaviors were identified using the observation and coding methodology validated by Keefe and Block (1982). Fig. 2 depicts the total number of overt pain behaviors displayed across groups by tasks. An ANOVA indicated that there were significant between-group differences in the display of overt pain behavior

ARTICLE IN PRESS 2830

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

Mean number of overt pain behavior displayed

20

Group H

Group L

Group C

18.45 16.88

15.85 15

12.81 10

9.28

10

5

0

Task 1

Task 2

Fig. 2. Mean number of overt pain behaviors displayed during the bag-carrying tasks by groups. Two analyses of covariance were performed to detect between-group differences, controlling for the effects of pain, anxiety and depression. No significant differences were observed.

during Task 1 [F (2, 57) ¼ 5.2, po.01]. Follow-up tests indicated that the numbers of overt pain behavior exhibited by Group H (po.01) and Group L (po.05) were significantly greater relative to Group C. However, these differences between groups became non-significant once the effects of pain intensity, anxiety and depression were controlled for in an ANCOVA. Total avoidance of Task 2 Following the completion of Task 1, while none of the participants in Group C quit doing Task 2, three (15%) participants in Group H and two (10%) participants in Group L chose not to attempt Task 2. A w2 analysis indicated that the proportion of participants who totally avoided Task 2 did not differ between the three groups [w2(2, N ¼ 60) ¼ 3.1, p ¼ .217]. Task 2 Pain intensity, mood and thoughts ratings As is shown in Table 3, there were significant between-group differences for the pain, anxiety and depression ratings obtained in Task 2. Follow-up tests indicated that pain ratings were higher in Group H (po.001) and Group L (po.001) compared to Group C. The anxiety rating was higher in Group H than in Group C (po.001). There was a non-significant trend for the depression rating to be higher in Group H relative to Group C (p ¼ .062). Utilization of SSB As in Task 1, a univariate ANCOVA was performed to detect differences between groups in the utilization of SSB in Task 2 (see Fig. 1), controlling for the effects of pain, anxiety and depression. There were significant differences observed between the three groups [F(2, 49) ¼ 5.8, po.01]. Post-hoc comparisons indicated that Group H employed a greater number of SSB than Group L (po.01) and Group C (po.01). Display of overt pain behavior An ANOVA indicated that there were marginally significant differences between the three groups in their display of overt pain behavior during Task 2 [F(2, 50) ¼ 3.1, p ¼ .05; see Fig. 2]. Follow-up tests indicated that the number of overt pain behaviors exhibited by Group H was significantly greater than that by Group C (po.05). However, as in Task 1, these differences between groups lost significance once the effects of pain intensity, anxiety and depression were controlled for using an ANCOVA.

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2831

Table 4 Intercorrelation between SSB, overt pain behavior, pain, moods and catastrophizing thoughts among chronic pain participants Task 1

Part 1 SSB—Task 1 SSB—Task 2 Part 2 Overt pain behavior—Task 1 Overt pain behavior—Task 2

Task 2

Pain intensity

Anxiety

Depression

Catastrophizing thoughts

Pain intensity

Anxiety

Depression

Catastrophizing thoughts

.27 –

.03 –

.02 –

.42** –

– .21

– .26

– .28

– .46**

.18

.10

.31y

–

–

–

–

–

–

.12

.02

.39* –

–

.34y

.21

Note: Task 1: n ¼ 40 chronic pain participants. Task 2: n ¼ 35. ypo.06. *po.05. **po.01. SSB ¼ safety-seeking behavior.

Relationship of SSB and overt pain behavior with pain intensity, mood and catastrophizing thoughts among chronic pain participants As can be seen from Table 4, SSB and overt pain behavior had different patterns of correlation with pain, mood and catastrophizing thoughts. While the use of SSB in chronic pain patients was positively correlated with the level of catastrophizing in both Task 1 (r ¼ .42, po.01) and Task 2 (r ¼ .46, po.01), the display of overt pain behavior in chronic pain patients was correlated with the level of pain intensity (r ¼ .39, po.05) in Task 1, and marginally correlated with the level of pain intensity (r ¼ .34, p ¼ .052) in Task 2. There was also a marginal correlation between overt pain behavior and catastrophizing thoughts in Task 1 (r ¼ .31, p ¼ .052). The correlation between SSBs and catastrophizing remained significant when the effect of pain intensity was partialled out [Task 1: r (37) ¼ .37, po.05; Task 2: r (32) ¼ .47, po.01]. Likewise, the correlation between overt pain behaviors and pain intensity remained more or less the same when the effect of catastrophizing was partialled out [Task 1: r (37) ¼ .33, po.05; Task 2: r (30) ¼ .34, p ¼ .06]. Discussion The current experiment represents the first empirical investigation of behavior in chronic pain from the perspective of the cognitive construct of SSB. Participants were observed on video performing two bagcarrying tasks. A novel feature of the study was the strategy of showing participants their performance in the task using immediate video playback in order to identify the motivation for the actions/behaviors occurring during the tasks. This strategy was used because pilot work indicated that ‘‘online’’ questioning interfered with task performance and vice versa. A wide range of SSBs were identified, particularly in pain patients with high health anxiety. These SSBs had, by definition, the common goal of protecting from further pain, injury or danger. Some of the SSBs were similar to an overt pain behavior (e.g., ‘‘holding the back’’), while some others were very subtle (e.g., ‘‘continually tensing my stomach muscles’’) and creative (e.g., ‘‘constantly rocking, shifting the weight between legs’’). Most SSBs continued throughout the time the person remained in a feared situation. As can be seen from the examples listed in Table 2, the use of SSBs could have two unintended effects: firstly, to exacerbate pain (e.g., ‘‘using the strength of the arm instead of strength of my body’’, ‘‘continually tensing my stomach muscles’’) and, secondly, to prevent the disconfirmation of catastrophizing beliefs (e.g., ‘‘something drastic will happen unless I do this’’, ‘‘if I don’t, I’ll panic and die’’), thus potentially maintaining the anxiety/disability in the long term, and maintaining pain itself. As predicted, it was found that pain patients with high health anxiety employed significantly more SSBs relative to both pain patients with low health anxiety and pain-free volunteers; this was noted in both Task 1 and Task 2. The greater use of SSB in patients with high health anxiety did not seem to be accounted for by the levels of pain, anxiety and depression occurring during the tasks, because the differences remained

ARTICLE IN PRESS 2832

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

significant even when these variables were statistically controlled for. By contrast, the number of overt pain behaviors did not significantly differentiate pain patients with high health anxiety from those with low health anxiety. Although pain patients with high health anxiety did exhibit a greater number of overt pain behaviors relative to pain-free volunteers, this difference disappeared once the effects of pain, anxiety and depression experienced during the tasks were statistically controlled for. Taken together, the data are consistent with the hypothesis that SSB may be a defining characteristic of chronic pain patients with high health anxiety. It was also hypothesized that when pain was activated health anxious pain patients would be more inclined to avoid physical activity compared to non-health anxious pain patients and pain-free volunteers. There was no significant difference between the three groups in the proportion of people opting not to do Task 2 following the completion of Task 1. The majority of chronic pain patients (85–90%) in this study chose to carry on with the task even though they understood that they were not obliged to complete Task 2. Future studies could also vary the task by increasing the effort or duration required to avoid ceiling effects. This study sought to strike a balance between experimental control and clinical/ecological relevance. This inevitably involves compromises in methodology. To maximize ecological validity, we adopted a bag-carrying task (closely simulating real-life activities), but participants were observed doing this in a laboratory using a carefully standardized procedure. Future research should consider replicating the experiment in the ‘‘real world’’ (e.g., in the participant’s home). The laboratory setting also highlights the issue of reactivity, given that participants knew that the session involved observation and measurement. Previous research has identified reactivity as a potential limitation (Keefe, 1982; Keefe & Smith, 2002). Reactivity was minimized by (i) giving participants time to warm up and familiarize themselves with the setting before commencing the study, (ii) using instructions that did not identify the specific type of pain-related behaviors that were being observed, (iii) using standardized instructions to limit interaction between the experimenter and the participants, (iv) starting the task only when the experimenter had left the room, (v) videotaping being inconspicuous, and (vi) eliciting self-reported motivation for behaviors observed after the task (during the video replay). The goal of refining the concept of pain behavior is shared by other research groups. For example, an attempt was made in a recent experimental study to classify observable pain behaviors based on their presumed functions (Sullivan et al., 2006). Facial and (para)verbal pain expressions were categorized as ‘‘communicative pain behaviors’’, whereas guarding, holding, touching or rubbing were categorized as ‘‘protective pain behaviors’’. The notion of protective pain behavior overlaps with the idea of pain-related SSB as described and tested in the current study. However, the two concepts are distinct in their scope. Unlike protective pain behavior, SSB refers to not only overt but also covert pain behaviors both within and outside a feared situation. The two concepts also differ in the way they are operationalized and in their theoretical drivers. While the presence of ‘‘protective pain behavior’’ is determined by observers based on their interpretation of the motive behind a behavior, the identification of SSBs is informed by the subject’s own appraisal of the function of a behavior and requires the use of subjective report of motivation, as is clear in the use of the term ‘‘safety-seeking behaviour’’ as opposed to the more theoretically neutral term ‘‘safety behavior’’. This can be regarded as a disadvantage (Nisbett & Wilson, 1977) or a strength (Turk & Flor, 1987). We sought to maximize the reliability of the self-report data (e.g., the participants were helped to accurately report safety strategies used by showing them videos of themselves performing the tasks) and the validity of the SSBidentification process (e.g., judgment was made based on clearly pre-specified criteria). The reliability of the procedure was checked by two independent blind raters and found to be good. The present study adapted the concept of SSB from cognitive models of anxiety disorders. Other concepts derived from research on anxiety disorders have already been used in chronic pain. However, these have focused on conventional exposure-based approaches emphasizing the role of ‘‘avoidance behavior’’ (Vlaeyen et al., 1995). Over the last decade, there has been continuing refinement of the fear-avoidance model of chronic pain (Norton & Asmundson, 2003; Vlaeyen et al., 1995; Vlaeyen & Linton, 2000), but this has remained substantially behavioral in content, continuing to emphasize avoidance rather than safety seeking. The present analysis therefore examines the cognitive–motivational element as has recently been done in anxiety disorders from a cognitive perspective, particularly in panic disorder (Salkovskis, 1991), obsessive–compulsive disorder and severe and persistent health anxiety (Salkovskis, 1996b). It is hypothesized that pain patients who believe that physical activity will cause pain/(re)injury will do things to make themselves safe (SSBs), thereby

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2833

counteracting the potential benefits of remaining in the feared situation. Thus, SSBs prevent the pain patients from disconfirming their beliefs about danger from physical activities, in the same way as agoraphobic patients who use SSB during exposure do not improve as much as those who drop or reverse these (Salkovskis, Hackmann, Wells, Gelder, & Clark, 2007). ‘‘Pain-related SSB’’ is theoretically and functionally distinct from ‘‘overt pain behaviors’’, which refer to observable actions/behaviors that communicate the presence of pain (Fordyce, 1982). The present results are consistent with this distinction, for example, in the specificity of the link between SSBs (but not pain behaviors) and health anxiety. Frequent use of motivated SSBs differentiated health anxious pain patients from those who are non-health anxious. Correlational analyses also indicated that the use of SSB was more strongly correlated with the presence of catastrophizing thoughts during the tasks, whereas the display of overt pain behavior was more strongly correlated with the level of pain experienced during the tasks. Clinically, findings of the current study may have important implications to the treatment of chronic pain patients in whom health anxiety is a major feature of clinical presentation. The observation that few pain patients actually chose to totally avoid physical activity and the identification of subtle within-situation SSB when pain patients were engaged in feared activities reveal the need to look beyond communicative pain behavior and activity total avoidance for targets of intervention. Based on our experience treating patients with anxiety disorders, it is vital to eliminate any SSB, overt or covert, that serves to maintain catastrophic cognitions, and clinicians should seek to assist patients in dropping their SSB. Cognitive–behavioral treatments for a range of anxiety disorders emphasize the use of ‘‘behavioral experiments’’ (rather than exposure) to challenge maladaptative thoughts and beliefs (Salkovskis, 1996a). A behavioral experiment is a planned experiential activity that is undertaken by patients during or between therapy sessions. Behavioral experiments are individually tailored and developed collaboratively with the patient, with the ultimate aim of giving patients the opportunity to experimentally test out their beliefs/behaviors and, by doing so, provide the patients with clear and memorable demonstrations that their beliefs/behaviors are unhelpful. The ‘‘loosening up’’ of negative beliefs prior to behavioral challenge will allow patients to engage with the process of change and to evaluate the impact of engaging in (and/or reversing) SSB as part of treatment (Salkovskis et al., 2003). The behavioral experiment then follows, with the therapist helping the patient to understand the results of the behavioral experiment once completed. Just as the observation of pain behavior proves to be useful for the assessment of pain (Keefe, 2000), the study of SSB may be a particularly useful way of understanding catastrophizing thoughts and beliefs and their role in maintaining chronic pain. By the same token, helping patients to find ways of dropping or reversing SSB may help the patients not only to reduce behaviors which interfere with daily functioning but also to challenge catastrophizing appraisals, replacing such cognitive and behavioral responses with more helpful reactions. Acknowledgments This research was supported by grants from the Croucher Foundation Hong Kong awarded to Nicole Tang as part of her post-doctoral research fellowship. The authors are grateful to Amy Hodges and Natasha Cole for providing assistance to the blind rating. Thanks go to staff at the King’s College Hospital Pain Relief Research Unit for their help in patient recruitment. References Ahern, K., & Le Brocque, R. (2005). Methodological issues in the effects of attrition: Simple solutions for social scientists. Field Methods, 17, 53–69. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., Text Revision ed.). Washington, DC: American Psychiatric Association. Asmundson, G. J. G., Norton, P. J., & Vlaeyen, J. W. S. (2004). Fear-avoidance models of chronic pain: An overview. In G. J. G. Asmundson, J. W. S. Vlaeyen, & G. Crombez (Eds.), Understanding and treating fear of pain (pp. 3–24). Oxford: Oxford University Press. Bjelland, I., Dahl, A. A., Haug, T. T., & Neckelmann, D. (2002). The validity of the hospital anxiety and depression scale. An updated literature review. Journal of Psychosomatic Research, 52(2) 69–77, 52, 69–77.

ARTICLE IN PRESS 2834

N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

Clark, D. M. (1999). Anxiety disorders: Why they persist and how to treat them. Behaviour Research and Therapy, 37, S5–S27. Flor, H., Kerns, R. D., & Turk, D. C. (1987). The role of spouse reinforcement, perceived pain, and activity levels of chronic pain patients. Journal of Psychosomatic Research, 31, 251–259. Fordyce, W. E. (1982). A behavioural perspective on chronic pain. British Journal of Clinical Psychology, 313–320. Fordyce, W. E., Shelton, J. L., & Dundore, D. E. (1982). The modification of avoidance learning pain behaviors. Journal of Behavioral Medicine, 5, 405–414. Hadjistavropoulos, H. D., & Hadjistavropoulos, T. (2003). The relevance of health anxiety to chronic pain: Research findings and recommendations for assessment and treatment. Current Pain and Headache Reports, 7, 98–104. Hadjistavropoulos, H. D., Owens, K. M. B., Hadjistavropoulos, T., & Asmundson, G. J. G. (2001). Hypochondriasis and health anxiety among pain patients. In G. J. G. Asmundson, S. Taylor, & B. J. Cox (Eds.), Health anxiety: Clinical and research perspectives on hypochrondriasis and related conditions (pp. 298–323). Chichester, UK: Wiley. Keefe, F. J. (1982). Behavioral assessment and treatment of chronic pain: Current status and future directions. Journal of Consulting & Clinical Psychology, 50(6), 896–911. Keefe, F. J. (2000). Pain behavior observation: Current status and future directions. Current Review of Pain, 4(1), 12–17. Keefe, F. J., & Block, A. R. (1982). Development of an observation method for assessing pain behavior in chronic low back pain patients. Behavior Therapy, 13, 363–375. Keefe, F. J., & Smith, S. (2002). The assessment of pain behavior: Implications for applied psychophysiology and future research directions. Applied Psychophysiology & Biofeedback, 27, 117–127. Lethem, J., Slade, P. D., Troup, J. D., & Bentley, G. (1983). Outline of a fear-avoidance model of exaggerated pain perception—I. Behaviour Research & Therapy, 21, 401–408. Looper, K. J., & Kirmayer, L. J. (2001). Hypochondriacal concerns in a community population. Psychological Medicine, 31, 577–584. McCracken, L. M., Zayfert, C., & Gross, R. T. (1992). The pain anxiety symptoms scale: Development and validation of a scale to measure fear of pain. Pain, 50, 67–73. Melzack, R. (1975). The McGill Pain Questionnaire: Major properties and scoring methods. Pain, 1, 277–299. Melzack, R. (1987). The short-form McGill Pain Questionnaire. Pain, 30, 191–197. Melzack, R., & Katz, J. (2001). The McGill Pain Questionnaire: Appraisal and current status. In D. C. Turk, & R. Melzack (Eds.), Handbook of pain assessment (2nd ed., pp. 35–52). New York: Guilford. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231–259. Norton, P. J., & Asmundson, G. J. G. (2003). Amending the fear-avoidance model of chronic pain: What is the role of physiological arousal? Behavior Therapy, 34, 17–30. Philips, H. C. (1987). Avoidance behaviour and its role in sustaining chronic pain. Behaviour Research & Therapy, 25, 273–279. Rode, S. (2003). Cognitive– behavioural factors in pain and health anxiety. Unpublished doctoral dissertation, University of London, UK. Rode, S., Salkovskis, P. M., Dowd, H., & Hanna, M. (2006). Health anxiety levels in chronic pain clinic attenders. Journal of Psychosomatic Research, 60, 155–161. Rode, S., Salkovskis, P. M., & Jack, T. (2001). An experimental study of attention, labelling and memory in people suffering from chronic pain. Pain, 94, 193–203. Salkovskis, P. M., & Bass, C. (1997). Hypochondriasis. In D. M. Clark, & C. G. Fairburn (Eds.), Science and practice of cognitive behaviour therapy (pp. 313–339). Oxford, UK: Oxford University Press. Salkovskis, P. M., Warwick, H. M. C., & Deale, A. C. (2003). Cognitive–behavioural treatment for severe and persistent health anxiety (hypochondriasis). Brief Treatment and Crisis Intervention, 3, 353–368. Salkovskis, P. M. (1991). The importance of behaviour in the maintenance of anxiety and panic: A cognitive account. Behavioural Psychotherapy, 19, 6–19. Salkovskis, P. M. (1996a). Avoidance behaviour is motivated by threat beliefs: A possible resolution of the cognitive–behaviour debate. In P. M. Salkovskis (Ed.), Trends in cognitive and behavioural therapies (pp. 25–41). Chichester: Wiley. Salkovskis, P. M. (1996b). The cognitive approach to anxiety: Threat beliefs, safety-seeking behavior, and the special case of health anxiety and obsessions (pp. 48–74). New York: Guilford. Salkovskis, P. M., & Warwick, H. M. (1986). Morbid preoccupations, health anxiety and reassurance: A cognitive–behavioural approach to hypochondriasis. Behaviour Research and Therapy, 24, 597–602. Salkovskis, P. M., Clark, D. M., & Gelder, M. G. (1996). Cognition–behaviour links in the persistence of panic. Behaviour Research and Therapy, 34, 453–458. Salkovskis, P. M., Clark, D. M., Hackmann, A., Wells, A., & Gelder, M. G. (1999). An experimental investigation of the role of safetyseeking behaviours in the maintenance of panic disorder with agoraphobia. Behaviour Research and Therapy, 37, 559–574. Salkovskis, P. M., Hackmann, A., Wells, A., Gelder, M. G., & Clark, D. M. (2007). Belief disconfirmation versus habituation approaches to situational exposure in panic disorder with agoraphobia: A pilot study. Behaviour Research & Therapy, 45, 877–885. Salkovskis, P. M., Rimes, K. A., Warwick, H. M. C., & Clark, D. M. (2002). The health anxiety inventory: Development and validation of scales for the measurement of health anxiety and hypochondriasis. Psychological Medicine, 32, 843–853. Sharp, T. J. (2001). The ‘‘safety seeking behaviours’’ construct and its application to chronic pain. Behavioural and Cognitive Psychotherapy, 29, 241–244. Sullivan, M. J., Thibault, P., Savard, A., Catchlove, R., Kozey, J., & Stanish, W. D. (2006). The influence of communication goals and physical demands on different dimensions of pain behaviour. Pain, 125, 270–277.

ARTICLE IN PRESS N.K.Y. Tang et al. / Behaviour Research and Therapy 45 (2007) 2821–2835

2835

Turk, D. C., & Flor, H. (1987). Pain greater than pain behaviors: The utility and limitations of the pain behavior construct. Pain, 31, 277–295. Vlaeyen, J. W., Kole-Snijders, A. M., Boeren, R. G., & van, E. H. (1995). Fear of movement/(re)injury in chronic low back pain and its relation to behavioral performance. Pain, 62, 363–372. Vlaeyen, J. W., & Linton, S. J. (2000). Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain, 85, 317–332. Vlaeyen, J. W., & Morley, S. (2005). Cognitive–behavioral treatments for chronic pain: What works for whom? Clinical Journal of Pain, 21, 1–8. Warwick, H. M., & Salkovskis, P. M. (1990). Hypochondriasis. Behaviour Research and Therapy, 28, 105–117. Wells, A., Clark, D. M., Salkovskis, P. M., & Ludgate, J. (1995). Social phobia: The role of in-situation safety behaviors in maintaining anxiety and negative beliefs. Behavior Therapy, 26, 153–161. World Health Organization. (1994). International classification of diseases, 10th revision (ICD-10). Geneva: World Health Organization. Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavia, 67(6) 361–370, 67, 361–370.