Attentional Biases Toward Sensory Pain Words in Acute and Chronic Pain Patients

The Journal of Pain, Vol 11, No 11 (November), 2010: pp 1136-1145 Available online at www.sciencedirect.com

Attentional Biases Toward Sensory Pain Words in Acute and Chronic Pain Patients Sonia P. Haggman,* Louise A. Sharpe,* Michael K. Nicholas,y,* and Kathryn M. Refshaugez * School of Psychology, The University of Sydney, NSW 2006 Australia. y ADAPT Pain Management Centre, Royal North Shore Hospital, Sydney, Australia. z School of Physiotherapy, The University of Sydney, NSW 2006 Australia.

Abstract: Attentional biases towards pain-related words of chronic and acute low back pain (LBP) patients were compared with healthy pain-free controls. Specifically, the aims were to determine: 1) whether chronic LBP patients demonstrate attentional biases compared to pain-free controls; 2) whether observed biases are also present in those with acute LBP; and 3) whether observed biases are associated with pain-related fear among the pain groups. Four groups were recruited: 1) acute LBP patients; 2) chronic LBP patients from physiotherapy practices; 3) chronic LBP patients from a tertiary referral pain-management center; and 4) healthy pain-free controls. Participants were assessed on the dot-probe computer task for attentional bias to pain-related words. All 3 pain groups demonstrated biases compared to controls on sensory but not on affective, disability, or threat words. Among the pain groups, those with low and moderate levels of fear of (re)injury demonstrated biases towards sensory pain words that were absent in those with high levels of fear, which is counterintuitive to what the fear of (re)injury model suggests. These results suggest that the experience of pain, rather than duration, is the primary indicator of the presence of pain-related biases. Perspective: Attentional biases are present in chronic and acute pain. Biases towards sensory-pain stimuli were demonstrated regardless of pain duration; however, they were present in those with low and moderate levels of fear of (re)injury only and not those high in fear. These findings are not consistent with the fear of (re)injury model. Crown Copyright ª 2010 Published by Elsevier Inc on behalf of the American Pain Society Key words: Attentional biases, acute pain, chronic pain, fear of (re)injury, dot-probe.

T

he existence and nature of attentional biases in chronic pain has received interest due to the potentially causal role of hypervigilance in the development of chronicity. The fear of (re)injury model proposes that pain-fearful individuals become involved in a vicious cycle of avoidance and disability that culminates in and maintains chronic pain. According to Vlaeyen and Linton,37 this cycle develops when painfearful individuals interpret pain as threatening, become hypervigilant to painful sensations, and avoid activity as

Received November 11, 2008; Revised October 28, 2009; Accepted February 12, 2010. Supported by a grant from the National Health and Medical Research Council (NHMRC) of Australia to L.A.S, M.K.N, and K.M.R. S.P.H. was supported by an NHMRC Scholarship. L.A.S. is also supported by a Senior NHMRC Research Fellowship. Address reprint requests to Louise Sharpe, School of Psychology, Brennan McCallum (A18), The University of Sydney, NSW Australia 2006. E-mail: [email protected]. 1526-5900/$36.00 Crown Copyright ª 2010 Published by Elsevier Inc on behalf of the American Pain Society doi:10.1016/j.jpain.2010.02.017

1136

a result of the fear.9,36 This ultimately contributes to ongoing pain-related disability and perpetuates the cycle of pain. Attentional bias is defined as the consistent disruption or facilitation of the response of an individual to a defined group of stimuli.27 There is robust evidence for attentional biases across various anxiety disorders.4 However, findings for attentional biases in pain patients are mixed.3,27 Pincus and Morley27 concluded that there was evidence of a small effect size for attentional biases towards sensory-related but not affective pain words, based on their review of the literature involving predominantly research using the modified Stroop task.27 When mood disturbance co-exists with pain, however, biases towards affective pain words have been demonstrated.27 More recent research using the dot-probe paradigm (thought to be a less ambiguous measure of attentional bias) has also been mixed. Three studies have failed to find an effect.1,3,29 These studies are marred by small sample sizes (combined n = 107) (eg, Asmundson and Hadjivropoulos1). Only 2 studies have found attentional

Haggman et al

The Journal of Pain 10,33

biases to sensory words using the dot-probe task. Both used large samples sizes (combined n = 268), and both found biases towards sensory but not other categories of pain words. However, neither compared the biases of pain patients to a control group. Therefore, where biases have been found in heterogeneous groups of chronic pain patients, it has been for sensory pain words. Comparison between those studies identifying differences and those who have failed to do so are made difficult by the use of different words, and parameters of the task. Hence, there is a pressing need for more large-scale research, using the precise paradigms tested in the literature. Further, theories suggest that the biases should be associated with the degree to which participants fear (re)injury. However, the little empirical attention within pain samples that this has received1 and the findings in healthy samples are mixed.15 The present study aimed to address these gaps in knowledge. That is, firstly we aimed to identify whether a representative sample of both private and tertiary chronic LBP patients demonstrate attentional biases compared to healthy, pain-free controls, using the same paradigm that has successfully shown the presence of biases towards sensory pain words in previous studies.10,33 Secondly, this study aimed to determine whether biases observed in chronic LBP are also present in those experiencing an acute injury. The final aim was to determine whether attentional biases varied as a function of level of fear of (re)injury in the pain samples. The following hypotheses were made: 1. The 3 groups of pain patients would demonstrate attentional biases towards sensory pain-related words, but not other categories of pain-related stimuli, compared with the healthy control group. 2. Within the pain groups, those with high levels of fear of (re)injury would demonstrate biases towards sensory pain words not observed in those with moderate levels of fear, whereas those with low levels of fear of (re)injury would demonstrate a bias away from sensory pain words.

Methods Participants The participants in this study were recruited between May 2005 and December 2007, and comprised 4 groups: 1) 51 acute LBP patients; 2) 50 chronic LBP patients from physiotherapy practices; 3) 57 chronic pain patients from a tertiary referral pain clinic; 4) and 50 healthy pain-free controls. All participants were aged between 18 and 75 years and the mean age of the entire sample was 43.6 years (SD = 14.5). Fifty-six percent of the sample were male and 62% were married or living in a de facto relationship. Thirty-one percent of the sample had less than 10 years of education while 43% had completed a university degree. Fifty-five percent were working full-time, with a further 17% in part-time employment. Forty-one percent of the pain sample had used alcohol for pain relief while 49% were using analgesics at the

1137

time of assessment and 73% had used analgesics in the past 3 months. Participants in the pain-free control group were matched against the acute pain group for age (±3 years), sex and education. Participants in the control group were aged between 18 and 73 years (mean = 40.52 years, SD = 14.8), and 64% of the sample were male. Sixty-six percent of the control sample was married or living in a de facto relationship, 28% had never married, and 6% had been divorced. Thirty-four percent had less than 10 years education, whereas 18% had completed a trade certificate and 48% had completed a university degree. Seventy-four percent of the control sample was working full-time and 16% were employed part-time. The remainder of the sample comprised participants who were retired, students, or involved in home duties. For the participants in the acute LBP group, patients were to be experiencing a single acute episode of LBP of less than 3 month’s duration. Sixty-five percent of the sample had experienced pain for less than 2 weeks, and only 10% had experienced pain between 2 and 3 months. Hence, this group was recruited very early following their injury. Participants in the chronic groups had to have experienced constant pain for more than 3 month’s duration. Participants were excluded if they suffered comorbid medical illness, psychosis, or if their level of English language was insufficient to complete the assessment. The chronic patients were recruited from 2 different sources. One group of chronic LBP patients (chronic physiotherapy group) were recruited from 24 private physiotherapy clinics throughout the Sydney metropolitan area, from which the acute LBP pain patients were also recruited. The physiotherapy clinics were located in 5 regions of Sydney and were selected from the 2005 Sydney telephone directory. Forty-five clinics agreed in principal to participate; however, only 24 recruited participants for the study. Hence, the method of recruitment was controlled for the acute pain and chronic physiotherapy pain groups. In the chronic physiotherapy group, nearly a quarter of the sample (23%) had experienced pain for between 3 and 6 months, with another 23% having pain for 6 to 12 months. The remainder (41%) had experienced pain in excess of a year. Another set of participants formed the tertiary referral chronic pain group (chronic tertiary group) and were recruited from a tertiary pain management center in Sydney. Consecutive patients were recruited prior to taking part in the pain management program and from new referrals to the center. Patients were invited to participate at the commencement of the 3-week program for chronic pain management or when they attended their assessment appointment. Most of the published studies in the attentional bias literature have recruited patients from similar clinics and hence these patients had a range of pain sites, whereas the chronic physiotherapy group was solely LBP patients. The vast majority of patients from the chronic tertiary group had experienced pain for in excess of 1 year (81%). The pain-free control group was recruited from the community and via advertisements placed around the campus of The University of Sydney. Participants

1138

The Journal of Pain

Attentional Biases in Acute and Chronic Pain

were matched individually against the participants in the acute pain group for gender, age (±3 years) and education (primary, secondary, technical certificate, university degree). Participants in the pain-free control group were excluded if they had a history of recurrent or chronic pain, had recently been injured, were currently in pain, were regularly using pain medication (including headache tablets), and/or if they had used pain medication in the past 24 hours.

Roland-Morris Disability Questionnaire31

Measures

Fear of Pain Questionnaire III23

Participants from each of the 3 pain categories completed all questionnaires. Participants from the control group provided demographic information and those questionnaires that did not refer specifically to pain or injury (DASS and FPQ). The complete battery of questionnaires is outlined below.

The Fear of Pain Questionnaire III consists of 30 selfreport items that assess pain-fearfulness. Internal consistency is excellent, with Cronbach’s a for the total scale being .92 and good test-retest reliability (r = .74).23

Demographic Information Participants reported on demographic information including sex, age, marital status, current work status, educational level, onset of LBP (sudden or gradual), and use of health care providers since onset of the LBP episode.

Visual Analogue Scale6 A visual analogue scale (VAS) of pain intensity comprised a 100-mm horizontal line anchored with ‘‘no pain’’ at left and ‘‘extreme pain’’ at the right.

Orebro Musculoskeletal Pain Screening Questionnaire20 The Orebro Musculoskeletal Pain Screening Questionnaire (OMPSQ) is a self-administered measure of psychosocial factors that have been found to reliably predict chronicity and disability in patients with LBP. It consists of 24 items divided into 5 domains (function, pain, psychological factors, fear avoidance, and miscellaneous, eg, sick leave, age, gender). The overall test-retest reliability for the questionnaire was r = .83.19

Depression Anxiety Stress Scale 2121 The Depression Anxiety Stress Scale (DASS-21) is a 21-item measure with 3, 7-item subscales assessing depression, anxiety, and stress, respectively. The psychometric properties are good with reliability consistently greater than r = .9.35 Psychometric properties are similar in clinical and nonclinical samples.8 The DASS has been compared to several other measures of depression in chronic pain patients and was found to have the best psychometric properties.35 In chronic pain patients, Cronbach a coefficients for the 3 DASS scales were reported as Depression = .96, Anxiety = .90 and Stress = .94.35 DASS scores are interpreted for each scale according to severity ranging from normal, mild, moderate, severe and extremely severe.

The Roland-Morris Disability Questionnaire is a 24-item checklist to determine the degree to which a person is limited by pain in completing daily activities. This questionnaire has good psychometric properties,30 with Cronbach’s a for the scale estimated as .93. The modified versions of the scale also show similar levels of internal consistency a = .91.34

Tampa Scale for Kinesiophobia15 The Tampa Scale for Kinesiophobia (TSK) is a 17-item measure assessing fear of movement using a 4-point Likert-type scale. Psychometric properties have shown the TSK to have good validity and reliability with internal consistency ranging from a = .68 to .80.36

Stimulus Words Stimulus words were adopted from Dehghani et al10 and comprised 4 pain-related word groups: sensory, affective, disability, and threat. These words were originally carefully selected from words on the McGill Pain Questionnaire24 and those that have been used in previous studies.5,11 During the initial development of the task, each category was developed to comprise 10 pain words which were matched for length and frequency with a neutral word.18 The final set of stimuli consisted of 40 pairs of cue/neutral words.10 (Table 1)

Dot-Probe Task22 Attentional bias was assessed using a modified version of the dot-probe task.10 The dot-probe paradigm was presented on a laptop computer (screen 30  24 cm) with word pairs displayed in yellow (Arial font, size 14) against a royal blue background. Word pairs were presented 2 cm above and 2 cm below the yellow fixation point presented in the center of the screen. Participants were presented with a fixation point ‘‘.’’ in the middle of a computer monitor for 500 ms. The point was then replaced by a word pair; 1 word above the location of the fixation point and 1 below. The words appeared for 500 ms, and were subsequently deleted and replaced by a probe of either the letter ‘‘p’’ or ‘‘q.’’ Participants were instructed to indicate as quickly as possible using the computer keyboard, whether a ‘‘p’’ or ‘‘q’’ had appeared. A probe identification task was chosen because it had been suggested that the appearance of a probe could be detected with peripheral vision and hence probe detection may give a less sensitive measure of attentional bias than probe identification.25 The computer program recorded the reaction time. The probe was deleted as soon as a response was recorded or after

Haggman et al Table 1.

The Journal of Pain

1139

Word Pairs Used in the Dot-Probe Task

SENSORY

DISABILITY

THREAT

AFFECTIVE

Flickering / Waterfalls Throbbing / Sailboat Shooting / Drinking Boring / Swivel Drilling / Whirling Sharp / Items Burning / Moment Stiff / Skirt Tugging / Refresh Pinching / Postmark

Vulnerable / Afterwards Suffering / Expensive Disable / Semester Paralyzed / Detectors Unhealthy / Timetable Ill / Joy Sick / Rear Inactive / Flooring Chronic / Grasped Injury / Mirror

Crushing / Elephant Fearful / Tribune Frightful / Flowering Terrifying / Innovative Killing / Crystal Suffocating / Advertisers Scared / Garage Danger / Waited Harmful / Airmail Threat / Golden

Vicious / Lessons Annoying / Chivalry Miserable / Undertake Troublesome / Restraining Unbearable / Metabolite Cruel / Drums Tiring / Cotton Exhausting / Blackberry Punishing / Advocates Discouraging / Subcommittee

1500 ms. The next trial commenced after 500 ms. (see Fig 1 for a diagrammatic presentation of the task). Four different presentation combinations were possible using the cue and probe appearing in the upper or lower locations of the computer screen. Two combinations were comprised of the cue and probe appearing in the same location of the computer screen (ie, once in the upper and once in the lower location of the screen). These pairings are congruent trials. The remaining 2 combinations were made up of the cue and the probe appearing in different locations of the computer screen (ie, one in the upper and the other in the lower location of the screen). These pairings are incongruent trials. A positive bias index score denotes attentional shifts toward the cue word indicating speeded processing when the cue is replaced by the probe, which represents attentional bias towards the pain-related stimuli. A negative bias index score indicates faster processing when the probe replaces the neutral word and attentional bias away from pain-related stimuli. Each word/probe combination appeared in a random order and each pair of words was presented 4 times, once in each of the 4 possible combinations described above. Participants were given a 1-minute break after each set of 40 words. Five pairs of practice words were presented at the beginning of the task and 2 more pairs of practice words were presented at the start of each set. The entire task took approximately 12 to 15 minutes to complete.

Procedure Participants in the acute pain and chronic physiotherapy pain groups were invited to participate in the study by their treating physiotherapist on the first or second visit for treatment of LBP. All participants received an information sheet, written consent form, and battery of questionnaires to complete at home and bring to the following treatment session, in accordance with the protocol that was approved by the University of Sydney’s Human Research Ethics Committee. Physiotherapists were provided with a checklist of inclusion and exclusion criteria and an outline of the recruitment procedure. When a suitable patient was referred, the physiotherapist asked the participant to attend the practice early and contacted the researchers who attended the practice half an hour prior to the scheduled appointment. The physiotherapists were encouraged to contact the researcher if they were uncertain of a patient’s suitability

and to discuss any concerns regarding the inclusion and exclusion criteria. In addition, the researcher made regular contact with the physiotherapists to maintain standardization of the procedure and to clarify any questions or problems that arose. Participants in the chronic tertiary pain group who were to commence the hospital-based pain management program were invited to participate in the study by the staff on their initial visit to the Unit. Those who agreed to participate were given the information sheet and written consent form to complete before commencing the program. All participants completed the questionnaires prior to completing the dot-probe task. Participants in the pain-free control group were recruited from the community through advertisements. Control group participants were tested on the dotprobe task at their convenience in their home or workplace. Each control group participant was provided the questionnaires at the time of testing and completed them prior to completing the dot-probe task. On arrival, participants were seated in front of a laptop computer and given instructions for the dot-probe task. Instructions were given verbally and presented on the monitor, as follows: ‘‘This is an attention task. First you will see a ‘‘.’’ in the center of the screen. Focus your gaze on the dot when it appears. After the dot you’ll be shown a pair of words, one above the other. Read the words silently. The words will be followed by the letters ‘‘p’’ or ‘‘q.’’ Press the ‘‘p’’ key as fast as you can when you see ‘‘p.’’ Press the ‘‘q’’ key as fast as you can when you see ‘‘q.’’ The computer will record your responses.’’

Analyses All analyses were conducted using SPSS 15.0. Differences between the groups were assessed using analyses of variance (ANOVAs) to identify potential covariates. To determine differences between groups in the allocation of attention, we performed a 4 (words: sensory, affective, disability, threat)  2 (congruence: congruent vs incongruent)  4 (group: acute, chronic physiotherapy, chronic tertiary, control) mixed model ANCOVAs for each set of pain words (ie, sensory, affective, disability, threat). To determine the nature of interaction effects, we also conducted 2 (congruence)  4 (group) ANOVAs separately for each word. We also conducted one-way t-tests for those indices that were significantly

1140

Attentional Biases in Acute and Chronic Pain

The Journal of Pain Congruent Trials

•

500ms Burning 500ms

Moment p

1500ms

•

500ms Moment

Burning

500ms

p 1500ms

Incongruent Trials

•

500ms Burning

Moment

500ms

p

1500ms

•

500ms Moment 500ms

Burning p

1500ms

Figure 1. Dot-Probe diagram demonstrating congruent and incongruent trials.

different from one another to determine whether those indices differed from zero. The data was checked for normality prior to analysis and in each analysis reaction time was the dependent variable. To determine level of fear of (re)injury, we divided participants into groups (high, medium, and low) according to their scores on the TSK questionnaire. We decided to use a tertile split procedure, rather than a multiple regression analysis, because prior research using fear of pain has found that participants high in levels of fear show a bias towards pain words, whereas those with a low level of fear demonstrate a bias away from pain words, and those with moderate levels of fear show no bias.1,12,13,28 The direction of any bias as a function of fear is an interesting issue that is not available through the use of correlational analyses. Hence, to establish the differences between the groups we performed 2, 3 (tertile: high vs medium vs low)  4 (word: sensory, affective, disability, threat) ANOVAs for the TSK scores for the combined pain groups. In order to aid interpretation of interaction effects, we also calculated indices of attentional bias for each word set (sensory, affective, disability, threat) using the following formula:12

Index ¼ ððcupl  clplÞ 1 ðclpu  cupuÞÞ = 2; c = cue; p = probe; u = upper location; and l = lower location. That is, ‘‘cupl’’ indicates a trial where the cue appeared in the upper part of the screen followed by a probe in the lower location. When the formula is applied, the resulting number gives the relative savings in terms of reaction time when the probe and cue appear in the same location, compared with when they appear in different locations. Hence a positive score indicates selective attention towards a cue, whereas a negative score indicates a bias away from the cue. Based on attaining a moderate effect size of .2728 for the analysis of 4 groups to show changes at 5% significance with 80% power, 45 subjects were required for each group.6

Results Participant Characteristics Table 2 presents the participants’ characteristics across the 4 groups. No significant differences were found

Haggman et al Table 2.

The Journal of Pain

1141

Participant Characteristics – Mean (SD)

VARIABLE

CONTROLa

ACUTEb

CHRONIC PHYSIOTHERAPYc

CHRONIC TERTIARYd

F

P VALUE

Age (yrs) Gender (males) Employment status (F/T) Education level (>12 yrs) Marital status (m/defac) Days off work Fear of pain DASS (Depression) DASS (Anxiety) DASS (Stress) RMDQ TSK OMPSQ VAS

40.52 (14.81) 64% (N = 32) 74% (N = 37) 80% (N = 40) 66% (N = 33) – 71.62 (21.33) 5.68 (4.60) 4.28 (4.85) 11.00 (7.08) – – – –

41.23 (13.03) 59% (N = 30) 67% (N = 34) 86% (N = 44) 61% (N = 31) 5.13 (10.87) 68.96 (15.98) 5.79 (6.79) 5.46 (6.67) 12.25 (8.50) 8.85 (5.62) 37.10 (6.81) 83.54 (28.04) 2.74 (2.22)

46.88 (15.07) 3% (N = 27) 53% (N = 27) 77% (N = 39) 55% (N = 28) 168.7 (416.51) 72.67 (21.34) 8.65 (9.99) 7.35 99.63) 14.61 (10.25) 9.42 (5.52) 40.71 (7.85) 101.50 (26.27) 2.61 (2.25)

45.68 (14.41) 40% (N = 23) 16% (N = 9) 68% (N = 39) 48% (N = 27) 95.00 (110.0) 75.15 (20.07) 13.27 (11.34) 9.49 (9.25) 15.67 (10.41) 11.75 (5.26) 41.36 (7.35) 122.74 (25.26) 4.85 (2.81)

2.394 – – – – 3.489 .842 8.013 3.939 2.586 3.558 4.523 22.844 1.811

.070 – – – – .035 .472 .000 .009 .055 .031 .013 .000 .169

NOTE. Participants: an = 50; bn = 51; cn = 50; dn = 57.

between the 4 groups (acute, chronic physiotherapy, chronic tertiary, control) for age, gender, education level, marital status, weekly income, stress, or fear of pain. Significant differences were found between the groups for employment status (F(3,192) = 18.903, P < .001) such that the chronic tertiary group differed from all other groups (P < .001). No significant differences were found between the 3 pain groups (acute, chronic physiotherapy, chronic) for fear of pain or pain ratings. However, there were significant differences between the 3 pain groups for OMPSQ (F(2,113) = 22.844, P < .001), such that the chronic tertiary group scored higher than the acute (P < .001) and chronic physiotherapy pain groups (P < .006). Significant differences were also found between the 3 pain groups for the TSK (F(2,138) = 4.523, P <.013) in that the chronic tertiary (P < .021) and chronic physiotherapy pain groups (P < .051) scored higher than the acute group on fear of movement. Similarly, significant differences were found between the chronic tertiary and acute pain groups on the RMDQ (F(2,140) = 3.558, P = .031) where the chronic tertiary pain group (P < .037) scored higher on disability than the acute pain group, but not the chronic physiotherapy group. There were significant differences between the groups for depression (F(3,190) = 10.467, P < .001) and anxiety (F(3,189) = 3.939, P < .009). The significant differences in depression were found between the control group and the chronic tertiary group (P < .001), the acute pain group and the chronic tertiary group (P < .001), and the chronic physiotherapy group and chronic tertiary group (P = .021). In contrast, for anxiety, the only significant difference was found between the control group and the chronic tertiary group (P = .009). Anxiety and depression were highly correlated (r = .699, P < .001) and therefore there was no need to control for both. Since significant differences emerged between all 3 pain groups and the control group for depression, but only between the chronic tertiary and control group for anxiety, depression was chosen as the most appropriate covariate. Although there was no significant effect of covariance on sensory (F(3,203) = .195 P = .659, h2 = .001), af-

fective (F(3,203) = .059, P = .809, h2 = .000) or threat words (F(3,203) = .898, P = .344, h2 = .004) there was a significant effect for depression on disability words (F(3,204) = 10.807, P = .001, h2 = .050). This indicates that those who demonstrated symptoms of depression responded more slowly to disability words than those who did not.

Main Analyses The following reported results are based on analyses performed when controlling for the covariate depression. In order to exclude the fact that both depression and attentional processes were related and obscured significant findings, analyses were also conducted without controlling for the covariate. The results of the 2 sets of analyses did not differ substantially from each other and hence we report only the results of the ANCOVAs. Initially, we performed the 4 (word type: sensory, affective, disability, threat)  2 (congruence)  2 (pain group: acute, chronic physiotherapy, chronic tertiary, control) ANCOVA. The analysis revealed a main effect for word (F(3,203) = 12.037, P = .000), a word  congruence interaction (F(3,203) = 10.858, P = .000), and a three-way word  congruence  group interaction (F(3,203) = 2.396, P = .011). Follow-up ANCOVAs were performed separately to determine the nature of the differences between groups for different pain-related words. For affective words, there was no main effect for congruence (F(3,203) = .008, P = .927, h2 = .000), or group (F(3,203) = 1.794, P = .149, h2 = .026), and no interaction effect (F(3,203) = .840, P = .474, h2 = .012). Similarly for disability words, there was no main effect for congruence (F(3,204) = 2.931, P = .088, h2 = .014), and although there was a trend towards a main effect for group, suggesting that pain patients responded more slowly than controls, this failed to reach significance (F(3,204) = 2.543, P = .057, h2 = .036). There was no significant interaction effect between group and congruence (F(3,204) = 1.109, P = .346, h2 = .016). For threat words there was a significant main effect for congruence (F(3,203) = 9.572, P = .002, h2 = .045), but both group (F(2,203) = 1.991, P = .117, h2 = .029) and interaction effects failed to reach significance

1142

Attentional Biases in Acute and Chronic Pain

The Journal of Pain

Table 3.

Group Mean (SD) Reaction Times (ms) for Congruence

PAIN WORDS Sensory Congruent Incongruent Bias Affective Congruent Incongruent Bias Disability Congruent Incongruent Bias Threat Congruent Incongruent Bias

CONTROLa

ACUTEb

CHRONIC PHYSIOTHERAPYc

CHRONIC TERTIARYd

553.43 (112.29) 552.43 (96.19) .99 (36.90)

566.88 (104.37) 592.21 (109.24) 25.33 (37.54)

589.92 (97.93) 609.18 (115.67) 19.25 (44.76)

596.65 (115.67) 615.59 (124.24) 18.94 (37.59)

551.9 (95.47) 550.09 (100.02) 1.81 (32.14)

580.51 (110.67) 578.09 (108.65) 2.42 (45.55)

590.75 (109.03) 595.23 (102.32) 4.48 (44.38)

592.49 (109.00) 604.81 (117.44) 12.31 (37.77)

540.96 (98.47) 550.37 (93.94) 9.41 (41.08)

582.31 (104.64) 579.25 (104.77) 3.05 (36.27)

597.58 (118.97) 597.02 (108.48) 5.60 (37.19)

602.99 (115.09) 608.59 (126.16) 5.60 (48.02)

544.63 (101.39) 536.46 (97.36) 8.16 (31.24)

570.76 (114.09) 555.17 (95.12) 15.59 (45.53)

595.22 (106.43) 568.98 (118.20) 26.24 (44.60)

601.64 (116.90) 589.55 (118.20) 12.09 (50.20)

NOTE. Participants: an = 50; bn = 51; cn = 50; dn = 57.

(F(3,203) = 1.595, P = .192, h2 = .023). This indicates that all groups responded more quickly to trials when the cue replaced the neutral word indicating a bias away from threat words. For sensory words, there was a significant main effect for congruence (F(3,203) = 21.002, P < .001, h2 = .094) but no main effect for group (F(3,303) = 2.100, P = .101, h2 = .030). However, there was a significant interaction effect for group  congruence (F(3,203) = 4.474, P = .005, h2 = .062). See Table 3 for mean reaction times for congruence.

Analyses of Biases Between Groups To aid interpretation of the significant interaction effect for sensory words, a 4 (group: acute, chronic physiotherapy, chronic tertiary, control)  4 (word: sensory, affective, disability, threat) ANOVA was conducted on bias scores. This revealed a main effect for word category (F(3,203) = 10.858, P < .001, h2 =.051) but not a main effect for group (F(2, 203) = 1.175, P = .320, h2 = .017). Importantly, a significant word  group interaction was found (F(3,203) = 2.396, P = .011, h2 = .034). This interaction effect is due to a difference between sensory pain words in the 3 pain groups compared with the pain-free control group. That is, participants in the 3 pain groups demonstrated biases towards sensory pain words that were not demonstrated in the pain-free control group. When comparing the word biases, significant differences were found between sensory and disability words (F(2,203) = 5.478, P = .001, h2 = .075) and sensory and threat words (F(2,203) = 3.702, P = .013, h2 = .052). However, the difference between sensory and affective words was not significant (F(2,230) = 1.924, P = .127, h2 = .028). We also compared the word biases to zero in the pain groups, using a series of t-tests. Significant attentional biases were found towards sensory words (t(1,207) =5.66, P < .001) and away from threat words (t(1,207) = –5.06, P < .001).

Effect of Fear of (Re)Injury The 3 pain groups were collapsed, since there were no differences in the pattern of biases between them and

they were re-categorized on the basis of their scores on the TSK (see Table 4). We conducted 2, 3 (tertile: high vs medium vs low)  4 (word: sensory, affective, disability, threat) ANOVAS for the combined pain groups according to scores on the TSK. There was an effect of words (F(1,135) = 20.34, P < .0005, h2 = .131) but no main effect of fear (F(2,135) = .023, P = .977, h2 = .000). There was an interaction effect between words and TSK (F(2,135) = 3.831, P = .024, h2 = .054). This interaction effect showed that the difference was between the medium and high TSK groups (P = .021) for the bias towards sensory pain words. For both the low and medium pain groups, the bias towards sensory pain words was significantly greater than zero (t(1,47) = 4.930, P = .000 and t(1,38) = 4.537, P = .000, respectively), however, this was not the case for the high fear group (t(1,50) = 1.557, P = .126). (see Table 5)

Discussion We investigated whether attentional biases are present in individuals experiencing pain compared with pain-free controls. To do so, we used a group of acute pain patients and 2 groups of chronic pain patients, one recruited from physiotherapy practices and the other from tertiary care. As hypothesized, all 3 pain groups showed biases compared to controls on sensory but not other pain words. Indeed, the 3 pain groups were indistinguishable from each other in the pattern of biases, despite the fact that more than half of the acute pain group had been in pain for less than 2 weeks and there was evidence that the chronic tertiary pain group was more severe than the other groups on a number of outcomes. Our results, however, did not support our second hypothesis. Participants with low or moderate levels of fear of (re)injury demonstrated significant biases towards sensory pain words, which were not present in participants high in fear of (re)injury as one might expect. These results are counterintuitive to what the fear of (re)injury model predicts37 and suggest that not only does duration of pain not account for attentional

Haggman et al

The Journal of Pain

Group Distribution for Fear of (Re)Injury (TSK) Tertiles

Table 5. Group Means (SD) for Fear of (Re)Injury (TSK) Tertiles

Table 4.

FEAR OF (RE) INJURY Low Moderate High Total

CHRONICb ACUTE 18 17 13 48

a

1143

PHYSIOTHERAPY

CHRONICc TERTIARY

TOTAL

17 10 22 49

13 12 17 42

48 39 52 139

BIASES

LOWa

MODERATEb

HIGHc

Sensory Affective Disability Threat

24.88 (34.96) 5.52 (37.70) 8.99 (38.16) 15.26 (36.61)

31.49 (43.35) 3.12 (48.77) 1.37 (42.29) 27.50 (53.55)

8.61 (39.52) 2.63 (43.90) 11.82 (53.32) 13.56 (47.61)

NOTE. Participants: na = 48; nb = 39; nc = 52. NOTE. Participants: an = 48; bn = 49; cn = 42.

biases towards pain but that biases towards sensory pain words may represent a normal response to the experience of pain. Despite careful attention to methodology, limitations remain. First, the dot-probe was conducted in a variety of settings. Participants in the pain groups were tested at different physiotherapy clinics or in the consulting rooms in a hospital whereas participants in the control group were tested in their home or workplace. It is possible extraneous factors related to the different testing environments may have influenced participants’ attention to the task. While this is a possibility, it seems unlikely since considerable effort was made to standardize the administration of the dot-probe task, including administration by the same researchers using the same laptop computers. The 3 pain groups were assessed in a pain-related setting (either prior to physiotherapy or an assessment or pain management group); hence, it is possible that they were primed to respond to painrelated words. However, we did ensure that the control group knew that the study was about pain and completed the questionnaires, which mention pain among other issues, prior to the task in an attempt to minimize the impact of any possible priming. Second, due to their obvious lack of pain status, the control group did not complete measures relating to disability, fear of movement, and the effect of LBP on lifestyle and work. As a result, such comparisons between the groups were limited to the 3 pain groups. It was therefore not possible to control for these differences in analyses comparing the 4 groups. This is not problematic since there were no differences in attentional biases between the 3 pain groups despite differences on each of these measures. Hence, it is unlikely that differences between these 3 groups in other variables contributed to attentional biases. Third, the chronic tertiary group comprised pain patients with numerous pain sites. While chronic pain patients have demonstrated different recall biases across different pain syndromes,38 chronic pain patients from tertiary referral pain centers are the group most represented in the pain literature. Since we found no differences in the pattern of biases between the 3 pain groups, the likelihood of possible confounding effects is low. Fourth, the modified version of the dot-probe task utilized in this study did not have neutral-neutral trials. There have been arguments in the literature that including neutral-neutral trials allows the nature of attentional

biases that are observed to be disentangled.16,17 More recently, however, authors have questioned whether or not this is the case.26 We opted to use a task that had been used to identify biases in other groups of chronic pain patients.10,11,33 There has been considerable debate about what particular parameters of the dotprobe task are necessary to find an effect, as there is considerable variation in the use of identification versus probe detection tasks, presentation times, stimuli, etc.32 Equally, stimuli such as pictures or faces may have been more salient in representing the threat posed by pain.14 Nonetheless, using a paradigm where biases have previously been found in chronic pain samples allowed the best opportunity to identify whether these biases are seen in different groups of pain patients and whether they are absent in healthy controls. The use of the same procedure allows the results to be directly compared. Finally, this is a cross-sectional study where attentional biases of pain patients were assessed on 1 occasion only. Therefore, no causal inferences can be drawn from these findings. It is possible that the biases found are a consequence of being in pain or indeed a normal and adaptive response to pain. These results, however, suggest that acute pain patients develop attentional biases to sensory pain words early in their experience of pain and/or injury since the majority had experienced pain for only a few weeks. However, whether the biases of acute pain patients are related to their subsequent prognosis is unknown. These limitations notwithstanding, this study has some important findings. This is the first study using the dotprobe to compare acute and chronic pain patients with healthy controls in the same experiment that has found chronic pain patients show biases not evident in healthy controls. Previous studies have failed to find an effect in chronic pain patients compared to controls.1-3,27 However, the number of pain patients tested in these 4 papers is only 107. In contrast, attentional biases have been found towards sensory pain words in comparison to other categories of pain words in chronic pain patients10,11 and rheumatoid arthritis patients.33 The combined number of patients in these three trials is 268. However, none of these trials included a control group. Therefore, the fact that we were able to replicate the findings of a bias towards sensory pain words and not other categories of pain words, but demonstrate that this bias is not present in healthy controls, is significant.

1144

The Journal of Pain

Attentional Biases in Acute and Chronic Pain

Our results are strengthened by the fact that we used 3 diverse groups of pain patients. Not only did we utilize a consecutive group of chronic pain patients recruited from a tertiary referral center similar to pain samples used in other studies, but we also used a group presenting to physiotherapy for their chronic LBP condition. Very few studies of chronic pain patients have been conducted outside of tertiary referral centers. The fact that we included 2 chronic pain samples, recruited from different settings, and found evidence of the same biases in both groups is also significant. Furthermore, the fact that the same biases were found for both groups of chronic pain patients and acute pain patients suggests that attentional biases have considerable relevance to the study of pain. The results of this study are therefore likely to be generalizable to treatment-seeking chronic pain patients at different levels of service utilization. The fear of (re)injury model hypothesizes that those who fear (re)injury are most likely to demonstrate biases and that these biases have a causal relationship with chronic pain. Our results on the contrary indicated that biases towards sensory pain words were greater in groups with low-to-moderate levels of fear of (re)injury rather than high levels of fear as one would expect. The groups that had low or moderate levels of fear of (re)injury both demonstrated biases towards sensory pain words that were identified in the heterogeneous pain sample, but not in those who were high in fear of (re)injury. Although these findings are opposite to that predicted, there are only 2 studies that have examined biases in fear of pain in healthy participants, and one similarly found effects in the opposite direction to that predicted.13 Nonetheless, the results challenge the fear of (re)injury model and question the role of fear of (re)injury as an important correlate of attentional biases in pain patients. The findings suggest that biases that do occur early in the acute phase of pain, however, are stronger for those with low-to-moderate levels of pain related fear. If attentional biases have a causal role in the development of chronicity, then one would expect to observe biases prior to the development of chronicity, as found in this study. However, those biases would be

expected to be associated with other variables known to confer risk of future chronification, such as fear of (re)injury. One explanation for these findings is that attending to the sensory aspects of pain is adaptive to the presence of pain (particularly in the acute pain state). Indeed, Pincus and Morley27 suggested that this may be the case. Another explanation is that only a small proportion of the acute group (27%) were in the high fear of (re)injury category, indicating an over-representation of those who demonstrated a bias in the moderate (44%) to low groups (38%). Clearly, prospective studies are needed to assess the validity of this claim. In conclusion, the present study provides strong support for an attentional bias towards sensory pain words in a large sample of pain patients with different characteristics that is not present in pain-free controls. This bias is specific to sensory pain words and corresponds to a larger effect size than is reported in the literature (Cohen’s d = .53).7 The fact that when the pain groups were divided into 3 groups on the basis of fear of (re)injury, biases were observed for the low and moderate pain-fearful groups, but not the high fear groups does not support the basic tenets from the fear of (re)injury model of chronic pain. Rather, the results raise the possibility that biases towards sensory aspects of pain are a ‘‘normal’’ response to pain. Future research should investigate whether these biases are associated with future functioning to determine whether they might have an adaptive rather than maladaptive function at least in acute pain patients.

References

5. Boston A, Sharpe L: The role of threat-expectancy in acute pain: Effects on attentional bias, coping strategy effectiveness and response to pain. Pain 119:168-175, 2005

1. Asmundson GJG, Hadjivropoulos HD: Is high fear of pain associated with attentional biases for pain-related or general threat? A categorical reanalysis. J Pain 8:11-18, 2007 2. Asmundson GJG, Kuperos JL, Norton GN: Do patients with chronic pain selectively attend to pain-related information? Preliminary evidence for the mediating role of fear. Pain 72:27-32, 1997 3. Asmundson GJG, Wright KD, Hadjistravopoulos HD: Hypervigilance and attentional fixedness in chronic musculoskeletal pain: Consistence of findings across modified stroop and dot-probe tasks. J Pain 6:497-506, 2005 4. Bar-Haim Y, Lamy D, Pergamin L, BakermansKranenburg MJ, van IJzendoorn MH: Threat-related attentional bias in anxious and non-anxious individuals: A meta-analytic study. Psychol Bull 133:1-24, 2007

Acknowledgments We extend our thanks to the participants who took part in this study and the physiotherapy practices that assisted us with recruitment of participants. In particular we would like to thank Sports, Spinal and Rehab Solutions Hornsby NSW, Burns Bay Physiotherapy Lane Cove NSW, Camperdown Physiotherapy Camperdown NSW and Manly Physiotherapy and Sports Injury Centre Manly NSW. There are no conflicts of interest. We kindly thank the reviewers for their helpful contribution to this paper.

6. Carlsson AM: Assessment of chronic pain: I Aspects of the reliability and validity of the visual analogue scale. Pain 89: 127-134, 1983 7. Cohen J: A power primer. Psychol Bull 112:155-159, 1992 8. Crawford JR, Henry JD: The Depression Anxiety Stress Scales (DASS): Normative data and latent structure in a large non-clinical sample. Br J Clin Psychol 42:111-131, 2003 9. Crombez G, Van Damme S, Eccleston C: Hypervigilance to pain: an experimental and clinical analysis. Pain 116: 4-1, 2005 10. Dehghani M, Sharpe L, Nicholas MK: Selective attention to pain-related information in chronic musculoskeletal pain patients. Pain 105:37-46, 2003

Haggman et al 11. Dehghani M, Sharpe L, Nicholas MK: Modification of attentional bias in chronic pain patients: A preliminary study. Eur J Pain 8:585-594, 2004

The Journal of Pain

1145

A replication study using a modified version of the probe detection task. Behav Res Ther 37:595-604, 1999

12. Keogh E, Ellery D, Hunt C, Hannent I: Selective attentional bias for pain-related stimuli amongst pain fearful individuals. Pain 91:91-100, 2001

26. Mogg K, Holmes A, Garner M, Bradley BP: Effects of threat cues on attentional shifting, disengagement and response slowing in anxious individuals. Behav Res Ther 46:656-667, 2008

13. Keogh E, Thompson T, Hannent I: Selective attentional bias, conscious awareness and the fear of pain. Pain 104: 85-91, 2003

27. Pincus T, Morley S: Cognitive-processing bias in chronic pain: A review and integration. Psychol Bull 127:599-617, 2001

14. Khatibi A, Dehghani M, Sharpe L, Asmundson GJG, Pouretemad H: Selective attention towards painful faces among chronic pain patients: Evidence from a modified version of the dot-probe. Pain 142:42-47, 2009

28. Roelofs J, Peters ML, Fassaert T, Vlaeyen JWS: The role of fear of movement and injury in selective attentional processing in chronic low back pain patients: A dot-probe evaluation. J Pain 6:294-300, 2005

15. Kori SH, Miller RP, Todd DD: Kinesiophobia: A new view of chronic pain behavior. Pain Manag 3:35-43, 1990

29. Roelofs J, Peters ML, van der Zijden M, Thielen FGIM, Vlaeyen JWS: Selective attention and avoidance of painrelated stimuli: A dot-probe evaluation in a pain-free population. J Pain 4:322-328, 2003

16. Koster EHW, Crombez G, Verschuere B, De Houwer J: Selective attention to threat in the dot probe paradigm: Differentiating vigilance and difficulty to disengage. Behav Res Ther 42:1183-1192, 2004 17. Koster EHW, Crombez G, Verschuere B, De Houwer J: Attention to threat in anxiety prone individuals: Mechanisms underlying attentional bias. Cogn Ther Res 30: 635-645, 2006 18. Kucera H, Francis WN: Computational Analysis of Present-Day American English. Providence, RI, Brown University Press, 1967 19. Linton SJ, Boersma K: Early identification of patients at risk of developing a persistent back problem: The predictive validity of the orebro musculoskeletal pain questionnaire. Clin J Pain 19:80-86, 2003 20. Linton SJ, Halden K: Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. Clin J Pain 14: 209-215, 1998 21. Lovibond SH, Lovibond PF: Manual for the Depression Anxiety Stress Scales. Sydney, AU, Psychology Foundation, 1995 22. McLeod C, Matthews A, Tata P: Attentional bias in emotional disorders. J Abnorm Psychol 95:15-20, 1986 23. McNeil DW, Rainwater AJ: Development of the fear of pain questionnaire III. J Behav Med 21:389-410, 1998 24. Melzack R: The short-form McGill Pain Questionnaire. Pain 30:191-197, 1987 25. Mogg K, Bradley BP: Some methodological issues in assessing attentional biases for threatening faces in anxiety:

30. Roland M, Fairbank J: The Roland-Morris disability questionnaire and the oswestry disability questionnaire. Spine 25:3115-3124, 2000 31. Roland M, Morris R: A study of the natural history of back pain: Part 1. Development of a reliable and sensitive measure of disability in low back pain. Spine 8:145-150, 1983 32. Salemink E, van den Hout M, Kindt M: Selective attention and threat: Quick orienting versus slow disengagement and two versions of the dot probe task. Behav Res Ther 45: 607-615, 2007 33. Sharpe L, Dear BF, Schrieber L: Attentional biases on chronic pain associated with rheumatoid arthritis: Hypervigilance or difficulties disengaging? J Pain 10:329-335, 2009 34. Stratford PW, Binkley JM, Riddle DL, Guyatt GH: Sensitivity to change of the Roland-Morris back pain questionnaire: Part 1. Phys Ther 78:1186-1196, 1998 35. Taylor R, Lovibond PF, Nicholas MK, Cayley C, Wilson PH: The utility of somatic items in the assessment of depression in patients with chronic pain. Clin J Pain 21:191-100 36. Vlaeyen JWS, Kole-Sniders AM, Boeren RG, van Eek H: Fear of movement/(re)injury in chronic low back pain and its relation to behavioural performance. Pain 62:363-372, 1995 37. Vlaeyen JWS, Linton SJ: Fear avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain 85:317-332, 2000 38. Wells H, Pincus T, McWilliams E: Information processing biases among chronic pain patients and ankylosing spondylitis patients: The impact of diagnosis. Eur J Pain 7:105-111, 2003