ORIGINAL ARTICLE

Exploring Attentional Bias for Real-World, Pain-related Information in Chronic Musculoskeletal Pain Using a Novel Change Detection Paradigm Daniel E. Schoth, PhD, Yizhu Ma, MSc, and Christina Liossi, DPsych

Objectives: Attentional biases for pain-related words and images have commonly been reported in individuals with chronic pain. In former studies, however, pain-related stimuli have been presented without context, for example, facial expressions of pain with no accompanying information regarding the location, severity, or cause of pain or injury. The present study investigated attentional biases for pain-related information using complex, real-world scenes in an ecologically valid experimental paradigm. Methods: Participants with chronic musculoskeletal pain (n = 20) and healthy, pain-free controls (n = 23) completed a version of the change detection paradigm, the flicker task, which requires participants to detect a single difference between 2 otherwise identical versions of the same scene. These change-scenes were presented in a continuous cycle for approximately 3 minutes, with an unrelated distractor-scene interspersed between. Both pain-related and neutral scenes were used in 4 experimental conditions: change-pain/ distractor-pain, change-pain/distractor-neutral, change-neutral/ distractor-pain, and change-neutral/distractor-neutral. Results: Individuals with chronic musculoskeletal pain, relative to healthy control participants, took significantly longer to detect changes when the change-scene was pain-related. Within-group analysis showed healthy control participants to take significantly longer to detect changes in neutral change-scenes compared with pain-related change-scenes. Discussion: This study is the first to show individuals with chronic pain possess attentional biases for pain-related information presented as part of complex, real-world scenes. Possible future research includes the use of real-world scenes in visual-search paradigms modifying attentional biases, and exploration into the relations and effects of combined cognitive biases (eg, attention, memory, and interpretation) in chronic pain. Key Words: chronic pain, attentional bias, change detection, flicker task, real-world scenes

(Clin J Pain 2015;31:680–688)

A

ttentional bias for pain-related information has been established in individuals with chronic pain,1,2 providing support for theoretical models of attention and pain Received for publication April 24, 2014; revised September 29, 2014; accepted August 15, 2014. From the Academic Unit of Psychology, University of Southampton, Southampton, UK. The authors declare no conflict of interest. Reprints: Daniel E. Schoth, PhD, Academic Unit of Psychology, University of Southampton, Highfield, Southampton SO17 1BJ, UK (e-mail: [email protected]). Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Website, www.clinicalpain.com. Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/AJP.0000000000000149

that predict such biases. The Schema Enmeshment Model of Pain,3 for example, proposes that the enmeshment of 3 cognitive schemata (pain, illness, and self) is responsible for information processing biases in chronic pain, and predicts all patients with pain to demonstrate biases for sensorypain information. Van Damme et al’s4 motivational account of pain proposes that, when pain management becomes the patient’s focal goal, attentional biases for pain and pain-related information arise. Previous research exploring attentional bias has commonly used the visualprobe task with linguistic5–8 and pictorial9,10 stimuli. Recently, eye-tracking research has used more ecologically valid experimental paradigms, providing evidence for painrelated biases when multiple emotionally salient images (ie, facial expressions) competing for attention are presented concurrently in visual-scanning11 and visual-search tasks.12 A limitation of former investigations, however, is that stimuli are presented without context. For example, facial expressions of pain are presented with no contextual information regarding the location, severity, or cause of pain or injury. In real life, however, such information rarely occurs in isolation, and observers will likely be presented with a wealth of additional information, either from the individual in pain (eg, additional pain behaviors such as guarding, rubbing, or touching the painful area) or the surrounding environment (eg, signs of accident or assault).13 Considering this, the present study investigated whether individuals with chronic musculoskeletal pain show attentional biases for pain-related information that is presented as part of complex, real-world scenes. A version of the change detection paradigm, the flicker task,14 was used, which allows for exploration of visual search performance. During this task participants are presented with 2 versions of the same scene (ie, the change-scene), which differ only in the removal of a single object. A third distractor-scene is interspersed between them, the presence of which removes local motion signals caused by the image change, preventing attention being drawn to the change in a bottom-up manner.14,15 In the flicker task, the perception of change to an object requires focused attention, but is also dependent upon visual working memory as the individual must perform a serial search of objects in the scene.14,16 Research shows individual and group differences, such as anxiety,15 alcohol craving,17 and domain expertise,18 also influence attention and change detection performance when stimuli of relevance to the individual are used. Pain-related information in change-scenes and distractor-scenes may therefore engage top-down attentional processes to varying degrees in individuals with chronic pain and healthy controls, according to the relevance of such information to the individual. In individuals with chronic pain, visual-probe

680 | www.clinicalpain.com Clin J Pain  Volume 31, Number 8, August 2015 Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Clin J Pain



Volume 31, Number 8, August 2015

task studies have provided evidence of more pronounced pain-related bias during longer stimuli presentation times associated with maintained attention and rumination,1,2 and eye-tracking studies have provided evidence of bias in initial orienting of attention.11,12,19 The present study investigated the time course of attentional bias using real-world, complex scenes. To achieve this, change-scenes and distractor-scenes controlled for low-level features were presented at exposure durations associated with processes of maintained attention and initial orienting of attention, respectively. On the basis of relevant theories of pain and attention, and former attentional bias research, it was hypothesized that individuals with chronic musculoskeletal pain, relative to healthy controls, would demonstrate significant attentional bias for pain-related information presented as part of change-scenes and distractor-scenes, as evidenced by significantly slower change detection response times.

MATERIALS AND METHODS Participants Participants were recruited from the University of Southampton and the South of England through press announcements, posters, and Psychobook, an online recruitment system. Inclusion criteria for the chronic musculoskeletal pain group were: (1) experiencing chronic musculoskeletal pain, that is, pain lasting for Z3 months affecting the muscles, ligaments, tendons, or joints,20 (2) 18 to 60 years of age (inclusive), (3) fluency in English language, and (4) normal or corrected-to-normal vision. Exclusion criteria were: (1) experiencing any other form of chronic pain, and (2) having a diagnosis or receiving treatment for a psychiatric disorder, either currently or within the past 5 years. Inclusion criteria for the healthy control group were: (1) 18 to 60 years of age (inclusive), (2) fluency in English language, and (3) normal or corrected-tonormal vision. Exclusion criteria were: (1) having a diagnosis or receiving treatment for a psychiatric disorder, either currently or within the past 5 years, (2) suffering from any form of chronic or recurrent pain, and (3) taking any psychotropic or analgesic medication regularly. Eligibility criteria were determined in a preliminary telephone interview with the first author. On the basis of these criteria, 1 individual who contacted the researchers was excluded due to the presence of chronic headache in addition to musculoskeletal pain. One individual with chronic musculoskeletal pain was excluded after recruitment due to a failure to understand the experimental instructions. An a priori power analysis indicated >85% power to detect differences of magnitude 0.5 between groups for a sample size of 38 (effect size = 0.50, critical F1,36 = 4.11, l = 9.50).21 Forty-three participants (mean age = 29.63; SD = 12.46; range, 18 to 60 y) were recruited, including 20 participants with chronic musculoskeletal pain (mean age = 31.75; SD = 13.97; range, 19 to 56 y) and 23 healthy control participants (mean age = 27.78; SD = 10.96; range, 18 to 60 y). Twenty-seven (63%) participants were female. In the chronic musculoskeletal pain group, 7 (35%) participants reported low back pain, 4 (20%) lower limb pain, 2 (10%) upper back pain, and 7 (35%) pain in >1 location (ie, back pain and either lower limb or hip pain). Participants reported living with pain for a mean duration of 8.11 years (SD = 8.83; range, 2 to 40 y), and 6 (30%) participants reported at least 1 first-degree relative to also Copyright

r

Chronic Pain Change Detection

experience recurrent pain. Nine (45%) participants reported taking medication regularly in response to their pain, with an average of 28.82% (SD = 33.89) pain relief from treatments and medications over the previous week, as indexed by the Brief Pain Inventory.22

Measures The following questionnaires were used to characterize the sample and assess participants’ pain experience: Brief Pain Inventory-Short Form (BPI)22 is a commonly used measure of pain intensity and interference. Pain intensity is assessed by 4 items asking individuals to rate their worst pain, least pain, and average pain over the past week, along with their present pain, on an 11-point numeric rating scale (0 = no pain, 10 = pain as bad as you can imagine). The average of these 4 items is calculated to form the pain intensity scale. Pain interference is assessed by 7 items asking the degree to which pain has interfered with general activity, mood, walking ability, normal work, relations with other people, sleep, and enjoyment of life over the past week, which are assessed on an 11-point numeric rating scale (0 = does not interfere, 10 = completely interferes). The average of these 7 items is calculated to form the pain interference scale. The BPI also includes a single item assessing how much relief pain treatments and medications have provided over the past week, and a body map so individuals can graphically indicate the location of their pain. Cronbach a in the current investigation for the pain intensity and pain interference scales were 0.84 and 0.91, respectively. Hospital Anxiety and Depression Scale23 is a 14-item measure of the severity of anxiety and depression symptoms experienced over the past week (7 items for each). All items are rated on a 4-point scale. Possible scores for both subscales range from 0 to 21, with higher scores indicating a greater severity of symptoms. Cronbach a in the current investigation for the anxiety and depression subscales were 0.75 and 0.84, respectively. State-Trait Anxiety Inventory24 is 40-item measure of state (ie, how the respondent currently feels) and trait (ie, how the respondent generally feels) anxiety. All items are rated on a 4-point scale. Possible scores range between 20 and 80 for both state and trait subscales, with higher scores indicating more intense or more frequent feelings of anxiety. Cronbach a in the current investigation for the state and trait subscales were 0.95 and 0.94, respectively. Fear of Pain Questionnaire III25 is a 30-item measure of pain-related fear. Respondents rate how fearful they are of pain associated with specific situations and events (eg, breaking your arm) on a 5-point scale. Possible scores range between 30 and 150, with higher scores representing a more intense fear of pain. Cronbach a in the present study was 0.92.

Experimental Stimuli Stimuli Description A total of 56 unique images were used in the present investigation, which were either collected from the Internet or photographed by the researchers, each comprising of a real-world scene. Images corresponded to pain-related and neutral categories. Pain-related images featured either people with injuries and experiencing pain (eg, a man sat on the ground holding his leg in pain, a woman sitting on a chair holding her shoulder in pain, a man with an injured ankle, an injured woman being lifted into an ambulance by

2014 Wolters Kluwer Health, Inc. All rights reserved.

www.clinicalpain.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

681

Clin J Pain

Schoth et al

paramedics) or situations and objects associated with the experience of pain and injury (eg, an ambulance, a man in a wheelchair). Neutral images featured either people without visible injury and pain in everyday situations (eg, a woman sitting at a desk in an office working, a man playing cricket, a man and woman sitting eating at a table, a woman speaking on her mobile phone) or neutral situations and objects (eg, a van, an image of a kitchen). Pain-related and neutral images were matched in terms of number of people in the scene and their sex, and their distance and orientation from the camera (eg, an image of a man lying in a hospital bed in pain was matched with an image of a man lying on a sofa), and the 2 categories overall featured a similar number of foreground people (pain-related = 24 male, 20 female; neutral = 26 male, 17 female). Pain-related and neutral images were divided into change-scenes and distractor-scenes. Change-scenes were edited using Adobe Photoshop CC to remove 1 object from the scene, which was replaced with the surrounding background content in a naturalistic manner (eg, as shown in Fig. 1, the notepad is removed and replaced with a continuation of the underlying kitchen work surface). Two versions of each change-scene were therefore used in each trial; the unedited original scene (change-scene A) and the edited scene (change-scene B). Pain-related and neutral image categories were matched on the location of change26 within change-scenes (pain-related = 3 top left, 6 top right, 2 bottom left, 3 bottom right; neutral = 2 top left, 4 top right, 4 bottom left, 4 bottom right), with no significant differences between the 2 conditions in change location, w2(3) = 1.41, P = 0.703. All images were in full color and measured 1024 pixels wide 768 pixels high. Once collected, and following analysis of image properties and low-level features as described below, the images were divided into 4 image conditions for use in the flicker task: (1) change-pain/distractor-pain (CpDp), (2) change-pain/distractor-neutral (CpDn), (3) change-neutral/ distractor-pain (CnDp), and (4) change-neutral/distractorneutral (CnDn). Thus CpDn represents the condition where the change-scene is pain-related, and the distractor-scene is neutral. Image conditions were matched based upon the overall number of foreground people (CpDp = 14 male, 13



Volume 31, Number 8, August 2015

female; CpDn = 8 male, 10 female; CnDp = 11 male, 8 female; and CnDn = 17 male, 6 female). Image conditions were also matched on the location of change in changescenes, under the criteria that each condition must have at least 1 change, and no more than 4 changes, in each quadrant of the image (CpDp 1 top left, 4 top right, 1 bottom left, 1 bottom right; CpDn = 2 top left, 2 top right, 1 bottom left, 2 bottom right; CnDp = 1 top left, 1 top right, 3 bottom left, 2 bottom right; and CnDn = 1 top left, 3 top right, 1 bottom left, 2 bottom right). No significant difference in change location was found between the image conditions, w2(9) = 5.02, P = 0.832.

Properties of Pain-related Versus Neutral Images Low-level features of real-world scenes influence attention and gaze,27,28 therefore an approach similar to that adopted by Nummenmaa et al29 was used to control for the images’ complexity, luminance, mean contrast level, and color saturation levels in red, green, and blue channels. Aside from complexity, all parameters were computed through Adobe Photoshop CC, and their data are available online as supplementary material (see Supplemental Digital Content 1, Tables S1 to S7, http://links.lww.com/CJP/ A128). Image complexity was assessed as the size of each compressed JPEG image in kilobytes, based on the assumption that the more complex the image the larger the file in kilobytes.30 Luminance refers to the intensity of light emitted from a surface; contrast refers to the difference in areas of lightness and darkness in an image; color saturation channel refers to the vividness of a color’s hue (http://www.dictionary.reference.com). Analysis of the lowlevel features of all pain-related and neutral images used in the present investigation (ie, from all 4 image conditions) was conducted by independent t tests. Pain-related and neutral images did not differ significantly in terms of complexity, mean contrast level, luminance, or red, green, and blue color saturation (see Supplemental Digital Content 1, Table S1, http://links.lww.com/CJP/A128). Further to low-level features, images were rated on valence and arousal (see Supplemental Digital Content 1, Table S2, http://links.lww.com/CJP/A128). An independent sample of 10 participants (n = 10; 5 female; mean

Presentation continues for 80 cycles (approximately 3 minutes) or until participant responds.

Distractor-scene 150 ms Change-scene B 1000 ms

+

Distractor-scene 150 ms Change-scene A 1000 ms

Fixation cross

FIGURE 1. Schematic illustration in black and white of a presentation cycle of a change-neutral/distractor-neutral trial in the flicker task.

682 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Clin J Pain



Volume 31, Number 8, August 2015

age = 27.00; SD = 4.11) provided ratings on the valence and arousal of each image by a 9-point computerized SelfAssessment Manikin.31 All images were randomly presented to participants for 3 seconds each. Following this, two 9-point Self-Assessment Manikin scales were presented, 1 for valence (unpleasant = 1 to pleasant = 9) and 1 for arousal (calm = 1 to excitement = 9). Participants were instructed to indicate how happy and aroused they felt while viewing each image, using the computer mouse to provide their responses. Paired sample t tests were used to compare participant responses to pain-related and neutral images. Pain-related images were rated as significantly less pleasant (P < 0.001), than neutral images, although no statistical difference in arousal was found. Change-scenes were also rated for difficulty of change detection (see Supplemental Digital Content 1, Table S3, http://links.lww.com/CJP/A128). An independent sample of 12 participants (n = 12; 5 female; mean age = 34.08; SD = 15.83) unfamiliar with the images completed a paper spot the difference task. During this task, both changescenes (ie, change-scene A and change-scene B) were presented one above the other on A4 paper, in randomized order. Both images were presented in full color and measured 18 cm wide 13 cm high on paper. Participants were instructed to spot the single difference between the image pairs, and for each image rate on an 11-point numerical rating scale how difficult it was to spot the difference (0 = not difficult at all, 10 = very difficult). To avoid potential familiarity effects, a separate sample performed this task to the sample that performed the valence and arousal rating task described above. A paired sample t test on difficulty ratings revealed no significant difference between neutral and pain-related change-scenes.

Properties of Experimental Image Conditions The same low-level feature analyses as those described above (ie, image complexity, mean contrast level, luminance, and red, green, and blue color saturation) were conducted between the 4 image conditions (ie, CpDp, CpDn, CnDp, and CnDn) by a series of 1-way analysis of variances (ANOVAs) (see Supplemental Digital Content 1, Table S4, http://links.lww.com/CJP/A128), with no significant differences found. One-way ANOVAs were conducted on valence and arousal ratings for pain-related images used in CpDp, CpDn, and CnDp conditions (see Supplemental Digital Content 1, Table S5, http://links. lww.com/CJP/A128), with no significant differences found. One-way ANOVAs were also conducted on valence and arousal ratings for neutral images used in CpDn, CnDp, and CnDn image conditions (see Supplemental Digital Content 1, Table S6, http://links.lww.com/CJP/A128), and again no significant differences were found. Similarly, a 1way ANOVA was conducted on difficulty of change detection across CpDp, CpDn, CnDp, and CnDn image conditions (see Supplemental Digital Content 1, Table S7, http://links.lww.com/CJP/A128), with no significant differences found.

The Flicker Task The flicker task was developed in Presentation (version 12.2),32 and run on a personal computer with a 15-inch color monitor. The task included 28 experimental trials (ie, 7 trials per image condition) that were presented in a new randomized order for each participant. Two practice trials (CnDn) were initially completed to familiarize participants Copyright

r

Chronic Pain Change Detection

with the requirements of the task. Participants then completed the experimental trials, which were divided into 2 blocks of 14 trials, with a short break between. The average completion time was approximately 20 minutes. Each trial began with the display of a fixation cross in the center of the screen. The researcher started the presentation cycle for each trial once the participant verbally indicated they were ready to proceed. Each presentation cycle consisted of an alternating sequence of (1) changescene A, (2) distractor-scene, (3) change-scene B, and (4) distractor-scene (Fig. 1). Within this cycle, both changescenes were presented for 1000 ms, and the distractor-scene for 150 ms. Presentation cycles followed one another continuously and seamlessly until either the participant detected the change or 80 cycles (lasting approximately 3 min) had elapsed. Participants were instructed to press a button on a response box as soon as they detected the change, and then to describe the change to the researcher. Images were presented full screen, and instruction text and fixation crosses were presented in white font against a black background.

Procedure Ethical approval for this investigation was obtained from the Research Ethics Committee of the University of Southampton. Informed consent was obtained before beginning the investigation. Participants first completed the flicker task, which similar to recent research15 was described as a computerized “spot the difference” task, with each trial featuring a single change that always consisted of the removal of 1 object from the scene (eg, a car, a camera, a hanging painting). Participants were informed that the removed object could be located anywhere in the scene, although were not told before each trial what this object was, which would likely guide their search strategy.28 Participants completed the task in a dimly lit room, seated approximately 60 cm from the computer monitor. The researcher was present in the laboratory during this task, although seated out of sight of the participant. Participants completed the questionnaire measures following the flicker task to avoid potential priming on the task. The total experimental duration was approximately 45 minutes.

Data Reduction and Analytic Plan Data were analyzed in IBM SPSS Statistics 21 for Windows. Practice trials were excluded from final analysis, as were trials with incorrect responses. Mean response times of correct trials were calculated for each participant, as was the percentage of change blindness (ie., percentage of trials in which a change was not detected). Between-group differences for demographic characteristics and emotional functioning were explored by the t test and w2 test for continuous and categorical variables, respectively. Mixeddesigns ANOVA was used to compare chronic pain and healthy control groups on mean change detection response times. The t tests were used in post hoc analyses to clarify significant effects. Effect sizes for ANOVA and t tests were quantified using partial Z2p and Cohen d, respectively. Cohen d and associated 95% confidence intervals (CI) were calculated by ECSI.33 For ANOVA analyses, the a-level was set at 0.05, 2-tailed. Pearson correlation coefficients were used to assess the relationship between questionnaire measures and mean change detection response times, with the a-level set at 0.01 due to the large number of calculations conducted.

2014 Wolters Kluwer Health, Inc. All rights reserved.

www.clinicalpain.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

683

Clin J Pain

Schoth et al

RESULTS Group Characteristics Chronic musculoskeletal pain and healthy control groups did not differ significantly in age (chronic musculoskeletal pain group mean age = 31.75 y [SD = 13.97]; range, 19 to 56; healthy control group mean age = 27.78 y [SD = 10.96]; range, 18 to 60) or sex (chronic musculoskeletal pain group = 10 female [50%]; healthy control group = 17 female [74%]). Mean self-report data for questionnaire measures are presented in Table 1. The chronic musculoskeletal pain group reported significantly higher depression (P = 0.024) compared with the healthy control group. Depression was not included as a covariate in the ANOVA analysis as no linear relationship was found between this variable and mean change detection response times.34

Change Detection Response Time Analysis Chronic musculoskeletal pain and healthy control groups did not differ significantly in overall response time (M = 21.34 s, SD = 7.12), percentage of change blindness trials (M = 8.31%, SD = 8.31), or percentage of incorrect responses (M = 0.33%, SD = 1.05). A 2 (group; chronic musculoskeletal pain, healthy control) 2 (changescene; pain-related, neutral) 2 (distractor-scene; painrelated, neutral) mixed-designs ANOVA was calculated on mean change detection response times (Table 2 and Fig. 2). A significant main effect was found for distractor-scene, F1,41 = 26.56, P < 0.001, Z2p = 0.393; participants took significantly longer to detect changes when the distractorscene was pain-related than when it was neutral (mean difference [MD] = 7.54). Significant main effects were not found for change-scene, F1,41 = 0.17, P = 0.687, Z2p = 0.004, or group, F1,41 = 1.99, P = 0.166, Z2p = 0.046. Significant 2-way interactions were found for change-scene by group, F1,41 = 11.55, P = 0.002, Z2p = 0.220, changescene by distractor-scene, F1,41 = 5.01, P = 0.031, Z2p = 0.109, but not for distractor-scene by group, F1,41 = 0.29, P = 0.595, Z2p = 0.007. The 3-way interaction for change-scene by distractor-scene by group was not significant, F1,41 = 0.98, P = 0.328, Z2p = 0.023. To explore the change-scene by group interaction, post hoc independent samples t tests were conducted, with a Bonferonni correction applied and the a-level adjusted to 0.01. The chronic musculoskeletal pain group, compared



Volume 31, Number 8, August 2015

TABLE 2. Mean (SD) Change Detection Response Times in Seconds for Chronic Musculoskeletal Pain and Healthy Control Groups

Image Condition

Chronic Musculoskeletal Pain Group (n = 20)

Healthy Control Group (n = 23)

28.07 (12.37)

19.55 (9.46)

21.07 (7.21)

16.23 (7.73)

26.19 (13.07)

27.06 (10.71)

16.54 (7.13)

16.88 (9.47)

Change-pain/ distractor-pain Change-pain/ distractorneutral Change-neutral/ distractor-pain Change-neutral/ distractorneutral

with the healthy control group, took significantly longer to detect changes when the change-scene was pain-related, t41 = 3.02, P = 0.004, d = 0.93, CI of d (0.29, 1.55). No between-group differences were found when the changescene was neutral, t41 = 0.24, P = 0.816, d = 0.07, CI of d ( 0.53, 0.67). Paired sample t tests were also used to compare performance within each group. The healthy control group took significantly longer to detect changes when the change-scene was neutral than when it was pain-related, t22 = 3.27, P = 0.003, d = 0.51, CI of d (0.17, 0.85). No significant difference in change detection time between pain-related and neutral change-scenes was found for the chronic musculoskeletal pain group, t19 = 1.78, P = 0.091, d = 0.41, CI of d ( 0.06, 0.88). To explore the change-scene by distractor-scene interaction, post hoc paired sample t tests were conducted. For pain-related change-scenes, participants across both groups took significantly longer to detect changes when the distractor-scene was pain-related compared with when it was neutral, t42 = 2.78, P = 0.008, d = 0.54, CI of d (0.16, 0.91). For neutral change-scenes, participants also took significantly longer to detect changes when the distractor-scene was pain-related compared with when it was neutral, t42 = 5.60, P < 0.001, d = 0.81, CI of d (0.42, 1.20). To clarify this, MD scores were calculated between pain-related distractor and neutral-distractor response times. The MD score was significantly greater when the change-scene was neutral (MD = 9.94, SD = 11.64)

TABLE 1. Means (SD) for Questionnaire Measures Completed by Chronic Musculoskeletal Pain and Healthy Control Groups

Measure HADS anxiety HADS depression STAI state anxiety STAI trait anxiety FOP-III total BPI pain intensity scale BPI pain interference scale BPI present pain

Chronic Musculoskeletal Pain Group (n = 20) 6.65 5.35 35.50 42.35 73.85 4.98 4.60 4.00

(2.98) (4.52) (12.80) (11.09) (23.56) (2.02) (2.60) (2.72)

Healthy Control Group (n = 23) 5.78 2.70 32.83 38.48 82.30

(3.37) (2.20) (10.56) (10.57) (16.56)

Mean Difference 0.87 2.65 2.67 3.87 8.45

t

df

P

0.89 2.39 0.75 1.17 1.38

41 27 41 41 41

0.380 0.024 0.457 0.249 0.177

95% Cohen Confidence d Interval of d 0.27 0.76 0.23 0.36 0.42

0.33, 0.87 0.14, 1.38 0.38, 0.83 0.25, 0.96 0.19, 1.02

BPI indicates Brief Pain Inventory; FOP-III, Fear of Pain Questionnaire III; HADS, Hospital Anxiety and Depression Scale; STAI, State-Trait Anxiety Inventory.

684 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Clin J Pain



Volume 31, Number 8, August 2015

Chronic Pain Change Detection

FIGURE 2. Ninety-five percent confidence interval error bars for mean change-detection response times across all conditions for chronic musculoskeletal pain and healthy control groups.

compared with when the change-scene was pain-related (MD = 5.03, SD = 11.89), t42 = 2.31, P = 0.026, d = 0.42, CI of d (0.05, 0.78).

Correlation Analysis Correlational analyses were conducted between questionnaire measures and mean change detection response times for chronic musculoskeletal pain and healthy control groups separately, the results of which are available online as supplementary material (see Supplemental Digital Content 2, Table S8, http://links.lww.com/CJP/A129). No significant correlations were found at the a-level 0.01 for either group.

DISCUSSION The aim of this investigation was to explore whether individuals with chronic musculoskeletal pain show attentional biases for real-world, pain-related information while performing an experimental task with high ecological validity. The results, in line with theoretical models of pain and attention3,4 and previous studies,7,10 provide support for the adopted hypothesis; individuals with chronic musculoskeletal pain, compared with healthy controls, took significantly longer to detect changes when the flicker task change-scene was pain-related. Pain-related information presented within changescenes captures the attention of individuals with chronic pain. As per task requirements, participants had to inspect all areas of the change-scene to locate the change efficiently. From these results, individuals with chronic pain appear less able to ignore the pain-related information contained within these scenes compared with healthy controls. Individuals with chronic pain ruminate over pain, and such worries are often difficult to dismiss, capture attention, and maintain vigilance to threat.35 As change-scenes were Copyright

r

presented for 1000 ms, this provided enough time on each presentation for participants to locate pain-related information and ruminate over its meaning. The present results support those obtained with the visual-probe task, suggesting biases for pain-related information are particularly prominent at stimuli presentation times associated with rumination and maintained attention.1,2 Furthermore, it is notable that the between-group effect size in the present study (ie, d = 0.93) is similar to that reported in a recent visual-scanning task (ie, d = 0.79),11 and also those reported in pictorial visual-probe tasks (eg, d = 0.719; d = 0.7010). This indicates that attentional biases are robust features of chronic pain, and that their strength does not diminish despite increased stimuli richness and complexity or task-related cognitive requirements. No significant between-group differences were found when the change-scene was neutral. Although pain demands attention and interrupts performance of ongoing activities,36 the present results demonstrate that in the flicker task, it is the presence of pain-related information within the change-scene that notably disrupts task performance for individuals with chronic pain. In contrast to this finding, no significant differences were found between groups in change detection performance when the distractor-scene was pain-related. Distractor-scenes were presented for 150 ms, and as such any evidence of bias toward pain-related distractor-scenes would likely result from processes of initial orienting of attention. Although previous studies have found evidence of initial orienting biases in individuals with chronic pain, the flicker task differs to the paradigms adopted in these former studies in terms of task requirements and instructions. Specifically, the visualprobe task19 required participants to locate a single dotprobe in 1 of 2 possible locations, the visual-scanning task11 allowed participants to view 4 facial expressions freely, and the visual-search task12 instructed participants to search for

2014 Wolters Kluwer Health, Inc. All rights reserved.

www.clinicalpain.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

685

Schoth et al

a specific facial expression embedded among 8 faces. The flicker task is the only paradigm which presents participants with complex, real-world scenes and asks them to search for an unspecified disappearing object. Such differences between paradigms therefore preclude any direct comparison of results. Across all participants change detection response times were significantly longer when the distractor-scene was pain-related compared with when it was neutral. Scene perception research suggests that while at least 150 ms is required to encode stimulus properties and process a scene normally during each gaze fixation, only 40 to 100 ms is needed to determine the gist of a scene.37 The present results suggest that participants were able to at least determine the gist of the 150 ms distractor-scenes which, when featuring pain-related information, had detrimental effects upon change detection performance. Interestingly, this detriment was significantly more pronounced when the change-scene was neutral compared with when it was pain-related. The saliency of task-irrelevant stimuli has been shown to influence visual-search performance.38 The present results suggest that irrelevant pain-related information is particularly salient and distracting when taskrelevant information is neutral. It is possible that, when paired with neutral change-scenes, pain-related distractorscenes are more unexpected and require greater elaborative processing. Change detection in the flicker task requires both focused visual attention and visual working memory.14,16 During each trial, participants must allocate visual attention to an image, encode a representation of that image, retain the representation for a time interval, retrieve the representation, and compare this with a subsequent version of the image.39 This raises 2 important points. First, although there is evidence of working memory deficits in individuals with chronic pain,40 the chronic musculoskeletal pain and healthy control groups in the present study did not differ in overall response time, percentage of change blindness trials, or percentage of incorrect responses. Deficits in working memory in the chronic musculoskeletal pain group therefore cannot account for the results obtained. Second, the requirement of both attention and memory processes in the flicker task points to the interplay of the various stages of information processing (ie, attention, interpretation, memory, cognitive control), and highlights the relevance of the combined cognitive bias hypothesis41 to this line of research. According to this hypothesis, rather than operating in isolation, cognitive biases (eg, attention, memory, interpretation) influence and interact with one another and cognitive control impairments, and collectively influence the etiology and maintenance of conditions such as anxiety and depression and, as we propose, possibly chronic pain. The clinical implications of attentional bias have been debated,42 with evidence supporting the therapeutic benefits of attentional bias modification (ABM) in individuals with chronic pain.43–45 Future research could begin examining the effects of combined cognitive biases in chronic pain, whether these have clinical implications pertaining to the onset and/or maintenance of pain, and whether their collective modification enhances therapeutic outcomes. Within-group analysis showed healthy control participants to take significantly longer detecting changes when the change-scene was neutral compared with when it was pain-related. This result may reflect an overall avoidance of

Clin J Pain



Volume 31, Number 8, August 2015

pain-related information. That is, rather than ruminating on pain-related information, healthy control participants may have preferred to engage with change detection and thus end the trials sooner. Avoidance of pain-related information in healthy participants has also been reported by Khatibi et al.46 As neutral and pain-related changescenes did not differ significantly in change detection difficulty, the alternative explanation that healthy controls simply found the former easier than the latter can be ruled out. The recording of participant eye movements in future research would likely provide further information on how healthy participants attend to information across neutral and pain-related change-scenes. A notable lack of significant correlations was found between change detection response times and questionnaire measures. This finding largely matches those reported in former chronic pain attentional bias studies, which has provided little consistent evidence for underlying correlates of bias. This suggests that in individuals with chronic pain, attentional biases are likely to be driven by the presence of pain itself, rather than associated increases in emotional distress such as depression. Visual search is a ubiquitous, everyday behavior made necessary by the complexity and richness of information existing in everyday scenes.47 A strength of the flicker task that involves visual search of complex, real-world scenes, is therefore its ecological validity. Low-level features of an image such as complexity, luminance, and color,27,28 along with properties of valence and arousal,48,49 guide visual attention. An additional strength of the present study was therefore the matching and preliminary analyses of these parameters across image conditions. Differences in change detection response times between pain-related and neutral change-scenes cannot be accounted for by variations in such parameters therefore. Limitations of the present study should also be noted. First, the relatively small sample size precludes the detection of small effect sizes. Second, image ratings of valence and arousal, and change detection difficulty, were collected from samples of separate participants to those completing the flicker task. Although this eliminates the possibility of familiarity effects influencing ratings, obtaining ratings from the same sample would allow for correlational analysis between valence, arousal, and change difficulty with mean change detection response times. Third, participants were recruited from the community and therefore may not be representative of a clinical sample. The contextual features of a visual-scene guides eye movements and attention.50 Although great effort was made to match neutral and pain-related images as closely as possible on numerous dimensions, it should be acknowledged that these categories likely differed in terms of familiarity and/or novelty for participants (although it should be noted participants had not seen before the specific images used in the flicker task). That is, in everyday life it is less common to witness others in pain than it is to witness them going about their daily behaviors and performing routine activities. For example, it is perhaps more common to view a person lying on a sofa than lying in a hospital bed. Future research using the flicker task could therefore use the same visual scenes to explore attentional biases for pain-related versus neutral information. For each visual scene, 1 version would be created which features a neutral object (eg, a pen), and another which features a pain-related object (eg, a syringe). In this design, change

686 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Clin J Pain



Volume 31, Number 8, August 2015

detection response times would provide a measure of attentional bias for the disappearing/reappearing object itself, while controlling for the effects of scene familiarity and novelty across conditions. In the real-world, people frequently search the same environment for multiple objects (eg, when working at a desk, a person may search for a pen, a stapler, a calculator). It has also been shown in experimental tasks that searching the same visual-scene for different objects across trials does not improve performance.51 Numerous other possibilities for future research building upon the present results exist. First, real-world images offer an ecologically valid method of exploring the combined cognitive bias hypothesis,41 which in addition to exploration of attentional bias in the flicker task may also be used to explore interpretation (eg, asking participants to provide a narrative of what is happening in each scene) and memory (eg, assessing recognition or recall for scenes at a later timepoint) biases. Second, ABM using a visual-search paradigm has shown beneficial effects in individuals with social phobia52 and stress,53 although to date no study has used real-world scenes. Assuming the results from the present study are replicated, future research could use real-world scenes to train attention away from pain-related information (eg, by asking participants to search for specified neutral objects). The complexity of such scenes offers a variety of competing content to attract attention, which is therefore more analogous to real-world situations. By using real-world images, the generalizability of ABM training to real-world situations may be increased, with potential improvements in training effectiveness. Third, recording of participant eye movements during the flicker task would establish whether pain-related information within complex scenes is subject to initial orienting of attention, and/ or more frequent or longer gaze fixations. In conclusion, the present study is the first to show that individuals with chronic pain demonstrate attentional biases for pain-related information presented as part of complex, real-world scenes, while performing a task that has high ecological validity (ie, visual search). Such biases are prominent during stimuli presentation times associated with rumination and maintained attention. ACKNOWLEDGMENT

Chronic Pain Change Detection

7.

8.

9.

10.

11.

12.

13.

14.

15. 16. 17.

18.

19.

20.

21.

22.

The authors thank Dr Jin Zhang, PhD (Senior Experimental Officer, Academic Unit of Psychology, University of Southampton, Southampton, Hampshire) for developing the flicker task used in this investigation, and for her continued technical support.

23.

REFERENCES

25.

1. Schoth D, Delgado Nunes V, Liossi C. Attentional bias towards pain-related information in chronic pain: a meta-analysis of visualprobe investigations. Clin Psychol Rev. 2012;32:13–25. 2. Crombez G, Van Ryckeghem DM, Eccleston C, et al. Attentional bias to pain-related information: a meta-analysis. Pain. 2013;154:497–510. 3. Pincus T, Morley S. Cognitive-processing bias in chronic pain: a review and integration. Psychol Bull. 2001;127:599–617. 4. Van Damme S, Legrain V, Vogt J, et al. Keeping pain in mind: a motivational account of attention to pain. Neurosci Biobehav Rev. 2010;34:204–213. 5. Haggman SP, Sharpe LA, Nicholas MK, et al. Attentional biases toward sensory pain words in acute and chronic pain patients. J Pain. 2010;11:1136–1145. 6. Liossi C, White P, Schoth DE. Time-course of attentional bias for threat-related cues in patients with chronic daily headache–

Copyright

r

24.

26. 27. 28.

29.

30.

2014 Wolters Kluwer Health, Inc. All rights reserved.

tension type: evidence for the role of anger. Eur J Pain. 2011; 15:92–98. Sharpe L, Dear BF, Schrieber L. Attentional biases in chronic pain associated with rhumatoid arthritis: hypervigilance or difficulties disengaging? J Pain. 2009;10:329–335. Liossi C, Schoth DE, Bradley BP, et al. Time course of attentional bias for pain-related cues in chronic daily headache sufferers. Eur J Pain. 2009;13:963–969. Schoth D, Liossi C. Attentional bias towards pictorial representations of pain in individuals with chronic headache. Clin J Pain. 2010;26:244–250. Schoth DE, Liossi C. Specificity and time-course of attentional bias in chronic headache: a visual-probe investigation. Clin J Pain. 2013;29:583–590. Liossi C, Schoth DE, Godwin HJ, et al. Using eye movements to investigate selective attention in chronic daily headache. Pain. 2014;155:503–510. Schoth DE, Godwin HJ, Liversedge SP, et al. Eye movements during visual search for emotional faces in individuals with chronic headache. Eur J Pain. [in press]. Hadjistavropoulos T, Craig KD. A theoretical framework for understanding self-report and observational measures of pain: a communications model. Behav Res Ther. 2002;40:551–570. Rensink RA, O’Regan JK, Clark JJ. To see or not to see: the need for attention to perceive changes in scenes. Psychol Sci. 1997;8:368–373. Gregory SM, Lambert A. Anxiety, conscious awareness and change detection. Conscious Cogn. 2012;21:69–79. Rensink RA. The dynamic representation of scenes. Vis Cogn. 2000;7:17–42. Hobson J, Bruce G, Butler SH. A flicker change blindness task employing eye tracking reveals an association with levels of craving not consumption. J Psychopharmaco. 2013;27:93–97. Werner S, Thies B. Is “change blindness” attenuated by domain-specific expertise? An expert-novices comparison of change detection in football images. Vis Cogn. 2000;7:163–173. Yang Z, Jackson T, Chen H. Effects of chronic pain and painrelated fear on orienting and maintenance of attention: an Eye Movement Study. J Pain. 2013;14:1148–1157. Merskey H, Bogduk N. Classification of Chronic Pain: Descriptions of Chronic Pain Syndromes and Definitions of Pain Terms. Seattle: IASP Press; 1994. Erdfelder E, Faul F, Buchner A. GPOWER: a general power analysis program. Behavior Research Methods, Instruments, and Computers. 1996;28:1–11. Cleeland C, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Annals of the Academy of Medicine, Singapore. 1994;23:129–138. Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand. 1983;67:361–370. Spielberger CD, Gorsuch RL, Lushene RE. State Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press; 1970. McNeil DW, Rainwater AJ. Development of the Fear of Pain questionnaire-III. J Behav Med. 1998;21:389–410. Mitroff SR, Simons DJ. Changes are not localized before they are explicitly detected. Vis Cogn. 2002;9:937–968. Henderson JM. Human gaze control during real-world scene perception. Trends Cogn Sci. 2003;7:498–504. Torralba A, Oliva A, Castelhano MS, et al. Contextual guidance of eye movements and attention in real-world scenes: the role of global features in object search. Psychol Rev. 2006;113:766–786. Nummenmaa L, Hyo¨na¨ J, Calvo MG. Eye movement assessment of selective attentional capture by emotional pictures. Emotion. 2006;6:257–268. Buodo G, Sarlo M, Palomba D. Attentional resources measured by reaction times highlight differences within pleasant and unpleasant, high arousing stimuli. Motiv Emotion. 2002;26:123–138.

www.clinicalpain.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

687

Schoth et al

31. Bradley M, Lang P. Measuring emotion: the Self-Assessment Manikin and the semantic differential. J Behav Ther Exp Psychiatry. 1994;25:49–59. 32. Presentations [Computer Program]. Version 12.2. Neurobehavioural Sciences Systems. http://www.neurobs.com. 33. Cumming G. Understanding the New Statistics: Effect Sizes, Confidence Intervals, and Meta-Analysis. New York: Routledge; 2012. 34. Tabachnick BG, Fidell LS. Using Multivariate Statistics. 6th ed. Boston: Pearson; 2012. 35. Eccleston C, Crombez G, Aldrich S, et al. Worry and chronic pain patients: a description and analysis of individual differences. Eur J Pain. 2001;5:309–318. 36. Eccleston C, Crombez G. Pain demands attention: a cognitiveaffective model of the interruptive function of pain. Psychol Bull. 1999;125:356–366. 37. Rayner K, Smith TJ, Malcolm GL, et al. Eye movements and visual encoding during scene perception. Psychol Sci. 2009;20:6–10. 38. Lamy D, Zoaris L. Task-irrelevant stimulus salience affects visual search. Vision Res. 2009;49:1472–1480. 39. Mitroff SR, Simons DJ, Levin DT. Nothing compares 2 views: change blindness can occur despite preserved access to the changed information. Percept Psychophys. 2004;66:1268–1281. 40. Berryman C, Stanton TR, Jane Bowering K, et al. Evidence for working memory deficits in chronic pain: a systematic review and meta-analysis. Pain. 2013;154:1181–1196. 41. Hirsch CR, Clark DM, Mathews A. Imagery and interpretations in social phobia: support for the combined cognitive biases hypothesis. Behav Ther. 2006;37:223–236. 42. Liossi C. Attentional biases in chronic pain: do they exist and does it really matter? Pain. 2012;153:9–10. 43. Carleton R, Richter AA, GJG Asmundson. Attention modification in persons with fibromyalgia: a double blind, randomized clinical trial. Cogn Behav Ther. 2011;40:279–290.

Clin J Pain



Volume 31, Number 8, August 2015

44. Schoth DE, Georgallis T, Liossi C. Attentional bias modification in people with chronic pain: a proof of concept study. Cogn Behav Ther. 2013;42:233–243. 45. Sharpe L, Ianiello M, Dear BF, et al. Is there a potential role for attention bias modification in pain patients? Results of 2 randomised, controlled trials. Pain. 2012;153:722–731. 46. Khatibi A, Dehghani M, Sharpe L, et al. Selective attention towards painful faces among chronic pain patients: evidence from a modified version of the dot-probe. Pain. 2009;147:42–47. 47. Chun MM. Scene perception and memory. In: Irwin DE, Ross BH, eds. Cognitive Vision. London: Elsevier Science; 2003: 79–109. 48. Fernandes MA, Koji S, Dixon MJ, et al. Changing the focus of attention: the interacting effect of valence and arousal. Vis Cogn. 2011;19:1191–1211. 49. Ni J, Jiang H, Jin Y, et al. Dissociable modulation of overt visual attention in valence and arousal revealed by topology of scan path. PLoS One. 2011;6:e18262. 50. Torralba A, Oliva A, Castelhano MS, et al. Contextual guidance of eye movements and attention in real-world scenes: the role of global features in object search. Psychol Rev. 2006;113:766–786. 51. Vo˜ ML, Wolfe JM. When does repeated search in scenes involve memory? Looking at versus looking for objects in scenes. J Exp Psychol Hum Percept Perform. 2012;38:23–41. 52. De Voogd E, Wiers R, Prins P, et al. Visual search attentional bias modification reduced social phobia in adolescents. J Behav Ther Exp Psychiatry. 2014;45:252–259. 53. Dandeneau SD, Baldwin MW, Baccus JR, et al. Cutting stress off at the pass: reducing vigilance and responsiveness to social threat by manipulating attention. J Pers Soc Psychol. 2007; 93:651–666.

688 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Exploring attentional bias for real-world, pain-related information in chronic musculoskeletal pain using a novel change detection paradigm.

Attentional biases for pain-related words and images have commonly been reported in individuals with chronic pain. In former studies, however, pain-re...
250KB Sizes 0 Downloads 6 Views