Journal Logo

Research Paper

Attentional biases in pediatric chronic pain: an eye-tracking study assessing the nature of the bias and its relation to attentional control

Soltani, Sabinea,*; van Ryckeghem, Dimitri M.L.b,c,d; Vervoort, Tineb; Heathcote, Lauren C.e; Yeates, Keitha,f,g; Sears, Christophera,g; Noel, Melaniea,f,g,h

Author Information
doi: 10.1097/j.pain.0000000000001916

1. Introduction

Pediatric chronic pain is highly prevalent, affecting 11% to 38% of youth.31 It is associated with significant distress, functional impairment, and high psychiatric comorbidity.11,53,55 Despite its prevalence and impact, a comprehensive understanding of the factors contributing to its development and maintenance remains elusive. Theoretical models of chronic pain posit that attentional biases contribute to its onset and maintenance,2,12,54 yet pain-related attentional biases remain poorly understood, particularly in youth. Given the significant increase in the onset of chronic pain in adolescence,29 and the fact that adolescent chronic pain often persists into adulthood,55 a better understanding of underlying mechanisms during this developmental period is critical for mitigating potentially lifelong impacts.32

Attentional bias is defined as preferential attention to emotionally relevant or salient information.10,13 Few studies have examined pain-related attentional biases in youth,24,52 and only a handful have examined clinical samples of youth with chronic pain.3,5,23 To date, the literature on attentional biases in youth echoes findings with adult samples: results are equivocal, with some studies reporting the presence of an attentional bias toward pain-related stimuli,3,24 and others reporting no evidence of attentional bias.23 A recent systematic review found weak evidence in favour of attentional bias for pain-related information in youth with chronic pain as compared to youth without chronic pain.6 Only one of the studies in this review assessed attentional bias using eye-tracking methodology.24 Eye-tracking is especially advantageous for such research because it provides a direct measure of attention to stimuli over an extended interval, unlike tasks that use response latencies to infer the focus of attention at a specific moment in time.44 In this study, healthy youth showed an initial orienting bias toward pain images, and this preferential attending was maintained over a 3500-ms presentation time.24 There was also evidence of a moderating effect of attentional control, such that for youth with lower attentional control, higher anxiety was associated with less dwell time on pain faces.24 Similar research with clinical samples is necessary to determine whether and how attentional biases in youth with chronic pain differ from youth without chronic pain because this remains unclear.

This is the first study to use eye-tracking to examine attentional biases to pain-related facial stimuli in a clinical sample of youth with chronic pain. The purpose of the investigation was to assess: (1) the nature of attentional bias (ie, initial orienting bias; total fixation bias) in a clinical sample of youth with chronic pain as compared to a pain-free control group; and (2) the moderating effect of attentional control on attentional bias. We hypothesized that, consistent with fear-avoidance models, youth with chronic pain would exhibit an attentional bias for pain faces that would differ from the bias of controls. We hypothesized that attentional control would moderate attentional biases, such that lower attentional control would be associated with longer total fixation times for pain faces.

2. Methods

2.1. Participants

All study procedures were approved by the University of Calgary Conjoint Health Research Ethics Board. Consistent with the Canadian research ethics policy,41 youth who were at least 14 years of age provided their consent to participate and signed a consent form. Youth who were younger than 14 years of age were asked to provide their assent and signed an assent form. In addition, one parent of each participant (regardless of age) attended the laboratory visit and provided their consent and, for youth younger than 14 years of age, consent for their child to participate in the study. The participants were youth aged 10 to 18 years with chronic pain (N = 102, 71% girls, Mage = 14.20 years; SD = 2.29) and youth without chronic pain (the control group; N = 53, 50% girls, Mage = 13.49 years; SD = 2.71). As a token of appreciation for their participation, youth and their parents were told at the outset that they would each receive $20 gift cards.

2.1.1. Chronic pain group

Youth were eligible if they were between 10 and 18 years of age and were referred to a chronic pain program for pain assessment and/or treatment. Youth who did not speak English or were diagnosed with a developmental disorder were not eligible for the study. Participants were recruited from 3 outpatient clinics (Headache, Abdominal Pain, and Complex Pain) housed within the pain and rehabilitation center of a children's hospital in Western Canada. Recruiting a mixed sample of youth with various pain conditions is consistent with previous research on pediatric chronic pain.37,39,56 To facilitate recruitment, clinical staff provided the study team with the contact information of new patients and patients who had received care in the program within the past year. Research staff also generated a list of participants who were participating in a clinical outcomes study and who had consented to be contacted about future studies. Research staff contacted prospective participants to provide information about the study, along with an option to decline participation. During recruitment, youth were confirmed to have experienced persistent or recurrent pain for at least 3 months, consistent with the current definition of chronic pain endorsed by the International Association for the Study of Pain.34 These youth were recruited as part of a larger longitudinal study that examined mechanisms underlying chronic pain and mental health issues.

2.1.2. Control group of youth without chronic pain

Youth were eligible if they were between 10 and 18 years of age and did not endorse the presence of chronic pain (ie, recurrent or persistent pain lasting 3 months or more). As with the chronic pain sample, youth who did not speak English or who were diagnosed with a developmental disorder were not eligible for the study. Youth who comprised the control group were recruited through a hospital-based registry of healthy families in Western Canada who were interested in participating in pediatric health-related research. Similar to the protocol used to recruit the chronic pain sample, research staff contacted prospective participants through email and telephone to provide information about the study.

2.2. Procedure

Either during or 1 to 4 weeks before their laboratory visit, youth and parents provided consent using an online consent form and completed questionnaires through REDCap, a secure online data collection tool.21 Parents completed questionnaires that collected sociodemographic information. Youth completed self-report measures to assess pain characteristics and attentional control. Youth and parents then visited the hospital-based research laboratory, located within the clinical milieu of a tertiary-level chronic pain and rehabilitation center. During this laboratory visit, participants completed an eye-tracking task and a behavioural measure of attentional control (the flanker visual filtering task).

2.3. Self-report measures

2.3.1. Demographic characteristics

Parents were asked to report sociodemographic information including youth age, sex, ethnicity, and annual household income.

2.3.2. Pain characteristics

Youth in the chronic pain group completed the Pain Questionnaire.40 Youth were asked to report their primary pain location using a body map.43 Youth reported how long their pain problem had been present (in years and months) and its frequency over the past 7 days (rated on a 5-point Likert scale ranging from “not at all” to “daily”). The Pain Questionnaire was developed by Palermo et al.40 to assess a variety of pain characteristics in youth with chronic pain. It comprised a series of single-item questions (eg, “How much do aches or pains bother or upset you?”). This measure has documented reliability and validity in youth with chronic pain40 and has been used in previous research to assess various pain characteristics in pediatric clinic samples, such as pain duration, location, unpleasantness, and interference.4,36

2.3.3. Pain outcomes: pain intensity and interference

The Patient-Reported Outcomes Measurement Information System (PROMIS-25) Pediatric Profile (version 1.0) was used to assess pain interference. Patient-Reported Outcomes Measurement Information System instruments are short forms developed by the National Institutes of Health to assess a variety of physical and mental health symptoms across the life span. The scales were created using item response theory, which produces more precise and informative measures27 using fewer items, thereby reducing respondent burden.30 The 4 items of the Pain Interference subscale are rated using a five-point Likert scale ranging from 0 (“never”) to 4 (“almost always”). Total scores range from 0 to 16, with higher scores indicating greater difficulty over the past 7 days. This scale is valid for use with youth with chronic pain.30 Total scores were transformed into standardized T-scores. For the current sample, internal consistency of pain interference subscales was good (α = 0.82). Pain intensity was rated on an 11-point numerical rating scale ranging from 0 (“no pain”) to 10 (“worst pain you can think of”), with the rating reflecting average pain intensity in the past 7 days. The numerical rating scale has been shown to be a valid and reliable measure for assessing pain intensity in youth with chronic pain.8

2.3.4. Attentional Control Scale

The Attentional Control Scale (ACS) is a 20-item measure that assesses an individual's self-reported level of attentional control.14 Items are rated on a 4-point Likert scale, with higher total scores indicating greater self-reported attentional control. The measure consists of 2 subscales that assess attention focusing and attention shifting. The ACS has good reliability.14 The ACS has been shown to have good construct validity and reliable factor structure in pediatric samples.49 Exploratory and confirmatory factor analyses of the ACS have consistently supported a two-factor structure (attention focusing and attention shifting), with some variations as to which items can be excluded from the calculation of total and subscale scores.28,38,41,49 For the analyses reported below, the ACS total score and corresponding scales were calculated excluding item 9, as recommended to enhance factor structure fit and increase measure precision.28,38,41,49 For the current sample, internal consistency of the ACS was good (α = 0.85).

2.3.5. Eye-tracking apparatus and task

Eye movements were recorded using an EyeLink 1000 eye-tracking system (SR Research Ltd, Ottawa, Ontario, Canada), which uses infrared video-based tracking technology. The system has a 1000-Hz sampling rate, allowing for a temporal resolution of 1 ms, with an average gaze error of less than 0.5 degrees of visual angle. Images were shown on a 21-inch BenQ XL2430T computer display positioned approximately 90 cm away from the participant. Images were 14 cm in height and 13 cm in width, and the centre of each image was located 8.5 cm from the fixation marker. Before the presentation of each pair of images, a fixation marker was displayed in the centre of the display for 500 ms to standardize the starting position of participants' gaze for each trial. Participants were instructed to focus on the fixation marker during the 500 ms and then to view the images subsequently presented. The 2 images were arranged horizontally in the display, with an image placed on the left and the right of the fixation marker location. Participants' eye gaze was measured continuously throughout the 3000 ms presentation time for each trial. The 2 images presented were defined as interest areas in the system programming, and fixations to these areas were automatically registered by the eye-tracking system. Fixation data were processed using the EyeLink Data Viewer software (SR Research) to filter for blinks, missing data, and other recording artifacts (using the default settings). To be included in analyses, a fixation had to be at least 100 ms in duration; sequential, adjacent fixations less than 100 ms in duration were merged into a single fixation.

2.3.6. Eye tracking stimuli

Stimuli consisted of 40 grayscaled images of 10 different children (5 boys and 5 girls, ranging in age from 9 to 16 years) depicting pain and neutral facial expressions. These images were used in previous research investigating attentional biases for pain-related stimuli in healthy children.52 The images were taken from video recordings of children experiencing an experimental pain task (the cold-pressor task51), and levels of pain expression assigned to the images correspond to observer ratings of pain intensity.50 Children and their parents provided permission for these images to be used for research purposes. For each of the 10 children in the images, 4 images represent 4 categories of facial expression: neutral face, low pain face, moderate pain face, and high pain face (Fig. 1). The stimulus set consisted of 30 pairs of images, each pair showing 2 images of the same child, one with a neutral face and the other one of the 3 pain expressions (low, moderate, and high; Fig. 2). Each pair was duplicated, with the neutral and pain face switching locations, thereby resulting in a total of 60 pairs (20 neutral-low pain pairs, 20 neutral-moderate pain pairs, and 20 neutral-high pain pairs).

Figure 1.
Figure 1.:
Example set of face images graded according to their pain expressiveness (ie, neutral, low pain, moderate, and high pain).
Figure 2.
Figure 2.:
Example trial for the eye-tracking task, depicting a neutral face paired with a pain face. Each pair of faces was presented for 3000 ms.

2.3.7. Flanker visual filtering task

Participants completed a flanker visual filtering task19 to provide a behavioural measure of attentional control. This task measures an individual's ability to ignore irrelevant or distracting stimuli while processing target stimuli.9 The flanker visual filtering task is the most frequently used task for assessing this aspect of executive inhibition. This study used the same version of the task used in previous studies,9,23 which is a simplified version of the visual filtering task with child-friendly stimuli. In this version of the task, a central target (an image of a fish facing either left or right) is presented on a computer display and is flanked by congruent (facing the same way) or incongruent (facing the opposite way) fish placed on either side of the target.

Participants were seated in front of a computer monitor and were shown a series of trials depicting a horizontal row of 5 fish. For each trial, participants were asked to respond as quickly as possible and indicate whether the middle fish was facing either left or right by selecting the left or right keyboard key. There were 2 types of trials: congruent and incongruent. Congruent trials were trials where all 5 fish pointed in the same direction (> > > > > or < < < < <). On incongruent trials, the 4 fish pointed in the opposite direction of the target fish (> > < > > or < < > < <). For each trial, stimuli were presented until participants selected a response or more than 3000 ms elapsed. If a response was not made within 3000 ms, an audio tone and the message “too slow” were presented. If a participant responded incorrectly, a tone and the message “wrong response” were presented. The intertrial interval was 1500 ms. Participants completed 2 practice blocks of 20 trials each to familiarize them with the task. After the practice trials, participants completed 120 randomly presented experimental trials (60 congruent and 60 incongruent trials). Three 30-second breaks were provided during the task (at 40-trial intervals).

2.4. Statistical analyses

Probability of first fixation (used to calculate first fixation bias) and mean total fixation time (used to calculate total fixation bias) were calculated as attention parameters to assess the initial orientation of attention and sustained attention, respectively.24 Probability of first fixation refers to whether a participant first fixated on the pain face or the neutral face when the pair of images was presented. A first fixation bias to pain is thus characterized as a higher probability that a participant first fixates on the pain face relative to the neutral face. The first fixation bias was calculated as the proportion of trials in which the pain face was first fixated divided by the total number of trials where the first fixation was made to either face. Proportion scores were calculated for each level of pain face expressiveness (high, moderate, and low) separately. The resulting proportion scores are interpreted as follows: a first fixation bias greater than 0.5 reflects an initial orienting bias toward pain faces; a first fixation bias equal to 0.5 indicates no bias; and a first fixation bias less than 0.5 reflects an initial orienting bias toward neutral faces and hence an initial attentional avoidance of pain faces.

Total fixation time was calculated by averaging the total time spent fixating on each face (pain face and neutral face) for each level of pain expressiveness separately (ie, high pain, moderate pain, and low pain). To quantify the bias, 3 total fixation bias scores (one for each level of pain face expressiveness) were calculated by subtracting the average of the total fixation time for the neutral faces from the averages of the total fixation time for the pain faces (high pain, moderate pain, and low pain). Positive values indicate that total fixation times for pain faces were longer than for the paired neutral faces, indicative of a total fixation bias (ie, greater attention) to pain faces. Negative values indicate that total fixation times for neutral faces were longer than for the paired pain faces, indicative of attentional avoidance of pain faces.

Statistical analyses were conducted using SPSS version 25 (IBM Corp, Armonk, NY). Statistical power was determined on the basis of previously published research.24,52 A power analysis for the planned mixed-model analysis of variance (ANOVA) with a medium effect size (ƒ2 = 0.25, α = 0.05, groups = 2; repeated measurements = 3) indicated that a total sample size of 82 participants (41 in each group) would provide 80% power to detect a two-way interaction. Thus, the current sample of 155 participants (102 chronic pain and 53 healthy control) provided more than adequate statistical power. An additional statistical power analysis was conducted to assess the power of the regression analyses with attentional control as a moderator. The results indicated that the sample size of 155 participants would provide 81% power to detect a 5% increase in variance accounted for (ie, increase in R-square) in the outcome variable by the interaction effect (a small effect size; ƒ2 = 0.05). Effect sizes for statistically significant t-tests (Cohen's d) are reported for key comparisons (where d values of 0.20, 0.50, and 0.80 correspond to small, medium, and large effects, respectively).

Data were examined before prorating and determined to be missing completely at random using Little's missing completely at random test, an essential assumption before proceeding with various techniques for the imputation of missing data.33 For participants with missing data constituting < 20% of responses within a scale, a prorated score was calculated for that scale.18 Descriptive, correlational, and ANOVA analyses were conducted using two-tailed hypothesis testing. Independent-samples t-tests and chi-square tests were used to test for group differences in key variables (ie, age, sex, and attentional control).

T-tests were used to determine whether the first fixation bias score for each pain face, for each group, differed significantly from chance (0.50), which would reflect an orienting bias toward or away from pain faces. Paired t-tests were used to determine whether the total fixation time for each pain face, for each group, differed significantly from the total fixation time for each of the corresponding neutral faces. To test for group differences in attentional biases, the first fixation bias and total fixation bias scores were analyzed using mixed-model ANOVA, with group (chronic pain, control) as the between-subjects factor and pain expressiveness (low pain face, moderate pain face, high pain face) as the within-subjects factor. Significant interactions were probed using t-tests. Pearson correlations were used to test for associations between attentional bias measures and measures of attentional control (ie, scores on the ACS and flanker task performance). Moderation models, testing whether attentional control moderated the magnitude of attentional bias within each group, were conducted using the Hayes' PROCESS macro for SPSS (Hayes, 2018).

3. Results

3.1. Processing of eye-tracking data

A total of 168 youth completed the eye-tracking task. The raw eye-tracking data were examined before analyses to screen for suboptimal data recording and potential outliers. Individual trials wherein no data were recorded (ie, total fixation time for the trial equal to 0 ms) were interpreted as reflecting a lack of attending or recording errors and were coded as missing (3.9% of all trials). No eye-tracking data were available for 7 participants due to inadequate calibration. The data from 4 participants were excluded from all analyses due to suboptimal eye-tracking data (ie, mean total trial fixation time less than 2000 ms). Finally, 2 participants were identified as statistical outliers on one or more of the attentional bias measures (using the SPSS Explore function) and were excluded from all analyses. The final sample consisted of 155 youth (102 in the chronic pain group and 53 in the control group).

3.2. Processing of flanker visual filtering task data

Consistent with previous research,9,23 trials with errors and trials with response times shorter than 200 ms or longer than 3000 ms were excluded from all analyses (2.7% of trials). In addition, response times 3.0 SDs above or below the mean were considered outliers and were excluded (0.61% of response times on congruent trials and 0.91% of response times on the incongruent trials). For each participant, mean response times for congruent trials were subtracted from mean response times for incongruent trials to produce a conflict score, with higher scores indicating greater interference in the presence of distracting information.42

3.3. Sociodemographic characteristics and descriptive statistics

Sociodemographic information (ie, sex, age, ethnicity, and household annual income) is listed in Table 1. For the youth in the chronic pain group, 61% were originally referred to a headache program and 39% were referred to a complex pain program. There were more females in the chronic pain group (70.6%) than in the control group (50.0%), X2(2) = 7.11, P = 0.029. This outcome is consistent with the sociodemographic characteristics of chronic pain samples in previous research,36,39 as well as the epidemiology of the condition in pediatric populations, which finds that a higher proportion of girls are affected by chronic pain as compared to boys.31 Over half of the youth in the chronic pain group (54%) reported pain in one location and just under half (43%) reported pain in multiple locations. Sixty-eight percent of youth reported headache, 26% reported musculoskeletal pain, 18% reported abdominal pain, 14% reported leg pain, 11% reported chest pain, and 25% reported pain in the “other” category. The average pain intensity level in the past week was 5.60 out of 10 (SD = 1.78). On average, youth reported a pain duration of 3.38 years (SD = 3.25). When asked about pain frequency over the past 7 days, 52.0% of youth in the chronic pain group endorsed daily pain, 15.7% endorsed pain “4 to 6 times per week,” 22.5% endorsed pain “2 to 3 times per week,” 6.9% endorsed pain “1 time per week,” and 2.9% endorsed no pain in the preceding week.

Table 1 - Sociodemographic and pain characteristics of the sample (by group).
Chronic pain group (N = 102) Control group (N = 53)
Mean age in y 14.20 (2.29) Mean age in y 13.49 (2.71)
Sex (% female) 70.6* Sex (% female) 50.0*
Ethnicity (%) Ethnicity (%)
 White 86.3  White 79.2
 Aboriginal 4.9  Aboriginal 9.4
 Black 2.0  Black 3.8
 Latin American 2.0  Latin American 3.8
 Arab/West Asian 2.0  Chinese 3.8
 South Asian 1.0  Other 0.0
 Other 4.9  Declined to answer 0.0
 Declined to answer 1.0
Household income (%) Household income (%)
 <$10,000-$29,999 5.3  <$10,000-$29,999 0.0
 $30,000-$59,999 12.8  $30,000-$59,999 7.0
 $60,000-$89,999 10.6  $60,000-$89,999 9.3
 More than $90,000 63.8  More than $90,000 76.7
 Declined to answer 7.4  Declined to answer 7.0
Pain characteristics
 Intensity out of 10 5.60 (1.78)
 Interference T-score 55.27 (9.35)
 Duration in years 3.38 (3.25)
Pain location (%)
 One location 54.1
 Multiple locations 43.1
 Headache 67.9
 Musculoskeletal 25.7
 Abdominal 18.3
 Leg 13.8
 Chest 11.0
 Other 25.7
SDs in parentheses.
*Statistically significant group difference (P < 0.05) based on independent-groups t-tests or χ2 test.

Descriptive statistics for the key variables of interest are shown in Table 2. Youth in the chronic pain group had lower scores on the ACS (M = 49.45; SD = 8.61) than youth in the control group (M = 54.15; SD = 9.03), t(152) = 3.17, P < 0.01. Given the impact of chronic pain on attention and cognitive processing posited by theoretical models,17 and as evidenced in previous research with adults,15,16,35 this difference in ACS scores was expected. The 2 groups did not differ with respect to age, t(148) = 1.58, P = 0.118.

Table 2 - Descriptive statistics (mean, SD) of key variables.
Variable Overall sample (N = 155) M (SD) Chronic pain group (N = 102) M (SD) Control group (N = 53) M (SD)
First fixation bias (low pain face) 0.50 (0.10) 0.51 (0.10) 0.47 (0.10)
First fixation bias (moderate pain face) 0.51 (0.09) 0.52 (0.09) 0.50 (0.09)
First fixation bias (high pain face) 0.56 (0.10) 0.56 (0.10) 0.55 (0.09)
Total fixation bias (low pain face) 46.63 (268.59) 50.58 (243.99)* 39.04 (312.94)*
Total fixation bias (moderate pain face) 115.65 (310.95) 92.06 (328.90) 161.06 (270.26)
Total fixation bias (high pain face) 218.10 (335.33) 242.72 (340.84) 170.72 (322.35)
ACS total score 51.06 (9.01) 49.45 (8.61)* 54.15 (9.03)*
Flanker task (congruent trials) 490.16 (107.33) 489.58 (103.15) 491.21 (115.44)
Flanker task (incongruent trials) 525.65 (112.94) 524.26 (107.02) 528.16 (123.88)
*Statistically significant difference between the chronic pain group and the control group (P < 0.05).
ACS, attentional control scale; M, mean.

The flanker response times were analyzed using a 2 (Group: Chronic Pain, Control) by 2 (Trial Type: Congruent, Incongruent) mixed-model ANOVA. As expected, there was a significant main effect of Trial Type, F(1, 146) = 306.49, P < 0.001, partial η2 = 0.68 such that response times for incongruent trials (M = 525.65, SD = 112.94) were slower than response times for congruent trials (M = 490.16, SD = 107.33), indicative of attentional interference due to distracting stimuli. The main effect of Group was not significant, F(1, 146) = 0.02, P = 0.883, partial η2 < 0.01, nor was the interaction between Group and Trial Type, F(1, 146) = 0.31, P = 0.580, partial η2 < 0.01. For both groups, response times for incongruent trials (Mpain = 524.26, SD = 107.02; Mcontrol = 528.16, SD = 123.88) were slower than response times for congruent trials (Mpain = 489.58, SD = 103.15; Mcontrol = 491.21, SD = 115.44), and there was no indication of a group difference in the interference effect (Mpain = 34.68 ms; Mcontrol = 36.95 ms).

3.4. First fixation bias

The data were analysed using a 2 (Group: Chronic Pain, Control) by 3 (Pain Expressiveness: Low, Moderate, High) mixed-model ANOVA. There was a significant main effect of Pain Expressiveness, F(2, 152) = 18.50, P < 0.001, partial η2 = 0.11 (Fig. 3), indicating that initial orienting bias differed as a function on pain expressiveness (Mlow pain = 0.50 SD = 0.10; Mmoderate pain = 0.51, SD = 0.09; Mhigh pain = 0.56, SD = 0.10). The main effect of Group was not significant, F(1, 153) = 3.05, P = 0.083, partial η2 = 0.02, nor was there an interaction between Group and Pain Expressiveness, F(2, 152) = 0.65, P = 0.562. Follow up one-sample t-tests indicated that for youth in the chronic pain group and the control group, there was a first fixation bias for high pain faces (Mpain = 0.56; Mcontrol = 0.55) that was significantly greater than chance (0.50), t(101) = 5.93, P < 0.001, d = 0.59, and t(52) = 4.21, P < 0.001, d = 0.58, respectively. First fixation biases for low pain faces (Mpain = 0.51; Mcontrol = 0.47) did not significantly differ from chance for youth in the chronic pain group, t(101) = 0.71, P = 0.480, d = 0.07, or for youth in the control group, t(52) = 1.91, P = 0.062, d = 0.26. The same was true for first fixation biases for moderate pain faces (Mpain = 0.52; Mcontrol = 0.50), t(101) = 1.79, P = 0.076, d = 0.18, and t(52) = 0.34, P = 0.737, d = 0.05, respectively.

Figure 3.
Figure 3.:
First fixation bias for the chronic pain group and the control group. **Statistically significant first fixation bias for high pain faces, as compared to the low and moderate pain faces (P < 0.001). Error bars depict one standard error.

3.5. Total fixation bias

Paired t-tests were used to determine whether the mean total fixation time for each pain facial expression, for each group, was significantly greater than the mean total fixation time for neutral faces. For youth in the chronic pain group, total fixation time for high pain faces (1220.34 ms), moderate pain faces (1146.36 ms), and low pain faces (1115.36 ms) was significantly longer than total fixation time for the paired neutral faces, t(101) = 7.19, P < 0.001, d = 0.71, t(101) = 2.83, P = 0.006, d = 0.28, and t(101) = 2.09, P = 0.039, d = 0.21, respectively. For youth in the control group, total fixation time for high pain faces (977.62 ms) and moderate pain faces (1054.30 ms) was significantly longer than for paired neutral faces, t(52) = 3.86, P < 0.001, d = 0.53, t(52) = 4.34, P < 0.001, d = 0.60, respectively, whereas the bias for low pain faces (1064.79 ms) was not t(52) = 0.91, P = 0.368, d = 0.12. Thus, for youth in the chronic pain group, there was an attentional bias for all of the pain faces, in contrast to youth in the control group who exhibited an attentional bias only for high pain and moderate pain faces. This result suggests that youth in the chronic pain group were more perceptive to pain stimuli than youth in the control group because they attended to low pain faces more than neutral faces, which was not the case for youth in the control group.

The total fixation bias data were then analysed using a 2 (Group: Chronic Pain, Control) by 3 (Pain Expressiveness: Low, Moderate, High) mixed-model ANOVA to determine whether there were within- and between-group differences in the magnitude of the biases. Figure 4 shows the total fixation bias data for each group for each level of pain expressiveness (low, moderate, and high). There was a significant main effect of Pain Level, F(2, 152) = 23.06, P < 0.001, partial η2 = 0.23. For the entire sample, total fixation bias for pain faces increased with each level of pain expressiveness (ie, Mlow pain = 46.63, SD = 268.59; Mmoderate pain = 115.65, SD = 310.95; Mhigh pain = 218.10, SD = 335.33). There was no main effect of Group, F(1, 153) = 0.01, P = 0.909, partial η2 < 0.01. There was a significant interaction between Group and Pain Expressiveness, F(2, 152) = 3.39, P = 0.036, partial η2 = 0.04. This interaction was followed up using t-tests to examine within-group differences in total fixation bias between pain expressiveness. For the chronic pain group, total fixation bias for high pain faces (M = 242.72, SD = 340.84) was significantly greater than total fixation bias for low pain faces (M = 50.58, SD = 243.99), t(101) = 6.91, P < 0.001, d = 0.68, and moderate pain faces (M = 92.06, SD = 328.90), t(101) = 4.78, P < 0.001, d = 0.47. Total fixation bias for low and moderate pain faces did not differ, t(101) = 1.36, P = 0.177, d = 0.13. For the control group, there was a different pattern of biases. Specifically, like the chronic pain group, total fixation bias for high pain faces (M = 170.72, SD = 322.35) was significantly greater than total fixation bias for low pain faces (M = 39.04, SD = 312.94), t(52) = 3.41, P = 0.001, d = 0.47. Unlike the chronic pain group, however, total fixation bias for high and moderate pain faces (M = 161.06, SD = 270.27) did not differ, t(52) = 0.22, P = 0.826, d = 0.03, and total fixation bias for moderate and low pain faces were significantly different, t(52) = 2.88, P = 0.005, d = 0.39. These results indicate that youth with chronic pain attended to moderate and high pain faces differently, whereas youth without chronic pain did not, and that youth without chronic pain attended to low and moderate pain faces differently, whereas youth with chronic pain did not.

Figure 4.
Figure 4.:
Total fixation bias for the chronic pain group and the control group. *Statistically significant difference, P < 0.01; **statistically significant difference, P < 0.001; error bars depict one standard error.

3.6. Correlations between attentional bias scores and attentional control

Bivariate Pearson correlations between the attentional bias measures (ie, the first fixation biases and the 3 total fixation biases), the ACS total score, and the flanker task conflict score are listed in Table 3 for the chronic pain group and in Table 4 for the control group. For the chronic pain group, there were no statistically significant correlations between the attentional bias measures and the 2 measures of attentional control (the ACS total score and the flanker task conflict score; all ps > 0.05). Interestingly, there was no correlation between the ACS total scores and the flanker task conflict scores (r = −0.03, P = 0.750).

Table 3 - Correlations between attentional bias and attentional control variables (chronic pain group).
Variable 1 2 3 4 5 6 7 8
1. First fixation bias (low pain)
2. First fixation bias (moderate pain) −0.03
3. First fixation bias (high pain) 0.12 0.20*
4. Total fixation bias (low pain) −0.09 0.11 0.07
5. Total fixation bias (moderate pain) 0.05 0.22* 0.05 0.44**
6. Total fixation bias (high pain) −0.18 0.14 0.14 0.61** 0.56**
7. ACS total score −0.16 −0.04 −0.12 −0.00 −0.01 −0.04
8. Flanker conflict score 0.06 −0.09 0.05 −0.03 0.14 −0.01 −0.03
*P < 0.05,**P < 0.01, two-tailed test.
ACS, attention control scale.

Table 4 - Correlations between attentional bias and attentional control variables (control group).
Variable 1 2 3 4 5 6 7 8
1. First fixation bias (low pain face)
2. First fixation bias (moderate pain face) 0.17
3. First fixation bias (high pain face) 0.31* 0.20
4. Total fixation bias (low pain) 0.33* −0.00 0.44**
5. Total fixation bias (moderate pain) 0.10 0.29* 0.25 0.47**
6. Total fixation bias (high pain) 0.04 0.18 0.36** 0.57** 0.40**
7. ACS total score −0.09 0.12 −0.20 −0.30* −0.11 −0.19
8. Flanker conflict score −0.02 0.32* 0.07 −0.14 0.16 −0.02 0.06
*P < 0.05, **P < 0.01, two-tailed test.
ACS, attention control scale.

For the control group, total fixation bias for low pain faces was negatively correlated with total score on the ACS (r = −0.30, P = 0.030). First fixation bias for moderate pain faces was positively correlated with the flanker task conflict score (r = 0.32, P = 0.020). There were no other significant correlations between the attentional bias measures and the attentional control measures (all ps > 0.05). As was the case for the chronic pain group, there was no correlation between the ACS total scores and the flanker task conflict scores (r = 0.06, P = 0.677).

3.7. Attentional control as a moderator of attentional biases

To test for a moderating effect of attentional control on attentional biases, regression analyses were conducted using the PROCESS macro for SPSS.22 The ACS total score and the flanker task conflict score were used as moderators (in separate analyses), with group membership (ie, chronic pain or control) predicting total fixation bias to low, moderate, and high pain faces. The results of the moderation analyses are shown in Table 5. The interaction effect of ACS total score by group membership on total fixation bias to low pain faces was statistically significant, b = −10.25, P = 0.047; however, the overall model was not statistically significant, F(3, 150) = 2.16, P = 0.096, and thus was not interpreted further. For the other analyses, none of the models or interaction effects were statistically significant (all ps > 0.05). Although we had no reason to hypothesize or expect that first fixation biases would be moderated by attentional control, these moderation analyses were also conducted and none were found to be significant (all ps > 0.05).

Table 5 - Moderation analyses with attentional control variables.
b (95% CI) SE t P
Outcome: total fixation bias (low pain faces) F(3, 150) = 2.16, P = 0.096, R 2 = 0.04
 Constant 58.58 (14.86, 102.30) 22.13 2.65 0.009
 ACS total score −3.60 (−8.47, 1.28) 2.47 1.46 0.147
 Group 18.77 (−74.36, 111.91) 47.14 0.40 0.691
 ACS total score × group −10.25 (−20.38, −0.13) 5.12 2.00 0.047
Outcome: total fixation bias (moderate pain faces) F(3, 150) = 0.729, P = 0.536, R 2 = 0.014
 Constant 118.74 (67.36, 170.12) 26.00 4.57 <0.001
 ACS total score −1.32 (−7.05, 4.41) 2.90 0.45 0.650
 Group 80.02 (−29.44, 189.49) 55.40 1.45 0.151
 ACS total score × group −3.01 (−14.91, 8.89) 6.02 0.50 0.618
Outcome: total fixation bias (high pain faces) F(3, 150) = 1.13, P = 0.338, R 2 = 0.02
 Constant 221.93 (166.85, 277.01) 27.88 7.96 <0.001
 ACS total score −3.35 (−9.50, 2.79) 3.11 1.08 0.238
 Group −46.14 (−163.48, 71.19) 59.38 0.78 0.438
 ACS total score x group −5.24 (−17.99, 7.52) 6.46 0.81 0.418
Outcome: total fixation bias (low pain faces) F(3, 144) = 0.02, P = 0.997, R 2 < 0.001
 Constant 42.61 (−2.13, 87.36) 22.64 1.88 0.062
 Flanker conflict −0.01 (−1.91, 1.90) 0.96 0.01 0.994
 Group −6.06 (−99.41, 87.29) 47.23 0.13 0.898
 Flanker conflict × group 0.35 (−3.46, 4.15) 1.93 0.18 0.858
Outcome: total fixation bias (moderate pain faces) F(3, 144) = 1.72, P = 0.166, R 2 = 0.03
 Constant 107.72 (57.80, 157.64) 25.26 4.27 <0.001
 Flanker conflict 1.73 (−0.40, 3.86) 1.08 1.61 0.110
 Group 79.01 (−25.14, 183.17) 52.69 1.50 0.136
 Flanker conflict × group 0.11 (−34.14, 4.36) 2.15 0.05 0.960
Outcome: total fixation bias (high pain faces) F(3, 144) = 0.64, P = 0.591, R 2 = 0.01
 Constant 216.67 (160.75, 272.58) 28.29 7.66 <0.001
 Flanker conflict 0.80 (−1.59, 3.18) 1.21 0.66 0.659
 Group −72.45 (−189.10, 44.19) 59.01 1.23 0.222
 Flanker conflict × group −0.64 (−5.40, 4.11) 2.41 0.27 0.790
95% confidence intervals (CI) in parentheses.
b, regression coefficient (unstandardized).

4. Discussion

This study used eye-gaze tracking to assess attentional bias for pain-related stimuli in a clinical sample of youth with chronic pain and a control sample of youth without chronic pain. In what follows, we review the major findings of our study and their implications for our understanding of attentional biases in youth with chronic pain.

4.1. Initial orienting bias

For both groups, youth exhibited a first fixation bias for high pain faces. This result is consistent with that of Heathcote et al.,24 who found a first fixation bias for pain faces in a sample of healthy youth. Taken together, findings to date indicate that an initial orientating bias to pain faces is present in both youth with chronic pain and their pain-free peers. Consequently, we view this orienting bias to pain stimuli as reflecting a common underlying information processing bias, as opposed to a consequence of chronic pain. This interpretation is consistent with the notion of an evolutionary predisposition for pain stimuli to capture attention.17 However, given the cross-sectional nature of our data, this conclusion is necessarily speculative at this time.

4.2. Total fixation bias

For both groups, total fixation bias increased with each level of pain expressiveness, with the strongest biases exhibited for high pain faces. This finding is consistent with the results of Heathcote et al.24 In this study, youth with chronic pain exhibited an attentional bias for pain faces at all expression levels. As compared to neutral faces, youth in the control group exhibited an attentional bias for moderate and high pain faces, but not for low pain faces. Youth with chronic pain seem to be more vigilant toward all levels of pain expressiveness as compared to youth without chronic pain, who did not seem to differentiate between low pain and neutral facial expressions.

For total fixation biases, there was an interaction between group and pain expressiveness. For the chronic pain group, total fixation bias for low and moderate pain faces did not differ, whereas the bias for high pain faces was significantly greater than the biases for low and moderate pain faces. Conversely, in the control group, total fixation bias for high and moderate pain faces did not differ, but both were greater than the bias for low pain faces. One explanation for this result is that youth with chronic pain may be habituated to moderate levels of pain imagery, given the regular presence of pain in their lives, resulting in low and moderate pain images being attended to similarly. Although there were group differences in the magnitude of the attentional biases to the different levels of pain imagery, it is important to keep in mind that, overall, youth with chronic pain group did not attend to pain faces significantly more than youth in the control group. Instead, youth with chronic pain were distinguished from youth without chronic pain by their attentional bias for low pain faces, and by their heightened attention to high pain faces relative to moderate pain faces. Thus, it is not the presence of an attentional bias to pain stimuli that differentiates youth with chronic pain, but rather the degree to which they attend to different levels of pain expressions. Overall, our results are consistent with theoretical models that posit that attentional bias to pain stimuli is characterized by preferential attending to pain-related information, or hypervigilance.1,2,26,45,54

4.3. The role of attentional control

An unexpected finding was that both self-report and behavioural measures of attentional control were largely unrelated to attentional bias measures. Moreover, attentional control did not moderate attentional biases for any of the pain faces in either group. These results suggest that attentional control, in and of itself, may not play an important role in moderating attentional biases in youth with chronic pain, nor in youth without chronic pain. However, a limitation of our study is that we focused solely on the association between attentional bias and attentional control without factoring in other potential interacting mechanisms.24,25 Previous studies that have found a moderating role of attentional control in youth24,25 have typically examined it as a moderator of the link between cognitive-affective variables and attentional bias. It may be that, on its own, attentional control does not play as influential a role in moderating attentional biases to pain when other threat-related factors are not considered. For this reason, future research should examine attentional control within the context of goal-directed behaviour and other contextual factors.

Interestingly, youth with chronic pain had lower scores on the ACS than youth in the control group, yet performance on the flanker task did not differ. Moreover, across both groups, there were no statistically significant associations between attentional control and performance on the flanker task. This finding aligns with recent research suggesting that, although widely used to measure attentional control, the ACS may not necessarily correlate with behavioural indicators of attentional control, and, may in fact, more accurately assess perceived ability, rather than actual ability.41 This may, in part, explain why attentional control did not moderate attentional biases in our study. Future research examining the validity of the ACS as a measure of attentional control in behavioral tasks, particularly in pediatric samples, will help clarify whether the ACS can be used as a proxy for behavioral tasks.

4.4. Limitations and directions for future research

This study had limitations that highlight important avenues for future research. First, like many studies on pediatric chronic pain, the sample was fairly homogeneous, comprised primarily of middle-class White youth. Although this is representative of samples from tertiary-level pediatric chronic pain clinics,20,39,46 it limits the generalizability of the findings and underscores the importance of conducting similar research with more diverse samples. Second, although the inclusion of a control group allowed us to compare attentional biases of youth with chronic pain and pain-free peers, youth in the control group were not age- or sex-matched to the clinical sample. Furthermore, the face stimuli used in the current investigation depicted neutral faces and 3 levels of pain expressiveness. Although this allowed us to assess for possible differences in attentional bias as a function of pain expressiveness, it did not allow for a comparison with other negative facial expressions (eg, anger, sadness, and fear). Future studies should seek to differentiate attentional biases to pain from those to other negative expressions such as anger, sadness, and fear. It is also important to keep in mind that the faces used in this study were evaluated and validated as indicative of acute pain. Given the unique experience of chronic pain, these stimuli may not necessarily generalize in the same way to a clinical sample.

Although eye-tracking paradigms overcome certain limitations of the dot-probe task, this methodology is not without its own limitations. As with other attention tasks, eye-tracking data are typically collected in laboratory settings.48 Recently, researchers have begun to question the ecological validity of assessing attention in this way.32,47,48 Indeed, the integrated functional-contextual framework48 posits that cognitive biases (including attentional biases) are context-dependent, dynamic, and interrelated. If true, assessing attentional biases in a laboratory setting may not accurately capture the bias because it manifests outside the laboratory and in relation to the individual's own pain experience, motivations, context, and goals. This recognition has fueled a call for new paradigms and technology such as ecological momentary assessment and augmented reality to increase ecological validity and as a way to account for personally salient, contextual factors, such as motivation and goal-driven behaviour.32,48 Another goal for future research is to examine relationships between attentional bias and theoretical antecedents of chronic pain (eg, anxiety sensitivity and pain catastrophizing) to better elucidate how attentional biases may factor into existing and future models of chronic pain.23 Ultimately, to better understand the clinical utility of attentional biases to pain, prospective research is necessary to examine how such mechanisms relate to or influence pain-related outcomes and commonly co-occurring mental health symptoms.

5. Conclusion

Attentional biases are hallmark factors in theoretical models of chronic pain development and maintenance in both adult and pediatric populations.2,12,54 Yet, experimental research to date has yielded mixed findings, with a paucity of research in youth as compared to adult populations. In this study, youth with and without chronic pain showed an initial orienting bias and a sustained attention bias to pain faces. Group differences emerged when comparing sustained attention to low pain faces. Attentional control, assessed through self-report and a behavioural task, did not moderate attentional biases between or within groups. Although research in this area is nascent, these findings underscore the utility of assessing attentional bias using eye tracking, as differences between the groups in their patterns of attending emerged only when attention was measured over the entire 3-second presentation time. With advances in paradigms and conceptualizations of attentional biases as being dynamic and contextually influenced, future research will further elucidate the role and clinical relevance of attentional biases in pediatric chronic pain.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Supplemental audio content

An audio abstract associated with this article can be found at http://links.lww.com/PAIN/B28.

Acknowledgments

This research was supported by funding awarded to Dr Noel by the Vi Riddell Pediatric Pain Initiative (RSO #1036777), the Alberta Children's Hospital Foundation and Research Institute (Shaikh Award; RSO #1042861), and funding from the Canadian Institutes of Health Research Strategy for Patient Oriented Research “Chronic Pain Network” (RSO #1041605).

References

[1]. Asmundson GJ, Katz J. Understanding the co-occurrence of anxiety disorders and chronic pain: state-of-the-art. Depress Anxiety 2009;26:888–901.
[2]. Asmundson GJ, Noel M, Petter M, Parkerson HA. Pediatric fear-avoidance model of chronic pain: Foundation, application and future directions. Pain Res Management 2012;17:397–405.
[3]. Beck JE, Lipani TA, Baber KF, Dufton L, Garber J, Smith CA, Walker LS. Attentional bias to pain and social threat in pediatric patients with functional abdominal pain and pain-free youth before and after performance evaluation. PAIN 2011;152:1061–7.
[4]. Beveridge JK, Neville A, Wilson AC, Noel M. Intergenerational examination of pain and posttraumatic stress disorder symptoms among youth with chronic pain and their parents. PAIN Rep 2018;3:e667.
[5]. Boyer MC, Compas BE, Stanger C, Colletti RB, Konik BS, Morrow SB, Thomsen AH. Attentional biases to pain and social threat in children with recurrent abdominal pain. J Pediatr Psychol 2006;31:209–20.
[6]. Brookes M, Sharpe L, Kozlowska K. Attentional and interpretational biases toward pain-related stimuli in children and adolescents: a systematic review of the evidence. J Pain 2018;19:1091–101.
[7]. Canadian Institutes of Health Research, Natural Sciences and Engineering Research Council of Canada, and Social Sciences and Humanities Research Council, Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans, December 2018. Available at: https://ethics.gc.ca/eng/policy-politique_tcps2-eptc2_2018.html
    [8]. Castarlenas E, Jensen MP, von Baeyer CL, Miró J. Psychometric properties of the numerical rating scale to assess self-reported pain intensity in children and adolescents. Clin J Pain 2017;33:376–83.
    [9]. Christ SE, Kester LE, Bodner KE, Miles JH. Evidence for selective inhibitory impairment in individuals with autism spectrum disorder. Neuropsychology 2011;25:690–701.
    [10]. Cisler JM, Koster EH. Mechanisms of attentional biases towards threat in anxiety disorders: an integrative review. Clin Psychol Rev 2010;30:203–16.
    [11]. Coffelt TA, Bauer BD, Carroll AE. Inpatient characteristics of the child admitted with chronic pain. Pediatrics 2013;132:e422–429.
    [12]. Crombez G, Eccleston C, Van Damme S, Vlaeyen JW, Karoly P. Fear-avoidance model of chronic pain: the next generation. Clin J Pain 2012;28:475–83.
    [13]. Crombez G, Van Ryckeghem DM, Eccleston C, Van Damme S. Attentional bias to pain-related information: a meta-analysis. PAIN 2012:475–83.
    [14]. Derryberry D, Reed MA. Anxiety-related attentional biases and their regulation by attentional control. J Abnorm Psychol 2002;111:225–36.
    [15]. Dick B, Eccleston C, Crombez G. Attentional functioning in fibromyalgia, rheumatoid arthritis, and musculoskeletal pain patients. Arthritis Care Res Official J Am Coll Rheumatol 2002;47:639–44.
    [16]. Dick BD, Rashiq S. Disruption of attention and working memory traces in individuals with chronic pain. Anesth Analgesia 2007;104:1223–9.
    [17]. Eccleston C, Crombez G. Pain demands attention: a cognitive-affective model of the interruptive function of pain. Psychol Bull 1999;125:356–66.
    [18]. Enders CK. Applied missing data analysis. New York, NY: Guilford Press, 2010.
    [19]. Eriksen BA, Eriksen CW. Effects of noise letters upon the identification of a target letter in a nonsearch task. Percept Psychophys 1974;16:143–9.
    [20]. Gold JI, Mahrer NE, Yee J, Palermo TM. Pain, fatigue, and health-related quality of life in children and adolescents with chronic pain. Clin J Pain 2009;25:407–12.
    [21]. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap): a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–81.
    [22]. Hayes AF. Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York: Guilford Press, 2013.
    [23]. Heathcote LC, Jacobs K, Van Ryckeghem DML, Fisher E, Eccleston C, Fox E, Lau JYF. Attention bias modification training for adolescents with chronic pain: a randomized placebo-controlled trial. PAIN 2018;159:239–51.
    [24]. Heathcote LC, Lau JYF, Mueller SC, Eccleston C, Fox E, Bosmans M, Vervoort T. Child attention to pain and pain tolerance are dependent upon anxiety and attention control: an eye-tracking study. Eur J Pain 2017;21:250–63.
    [25]. Heathcote LC, Vervoort T, Eccleston C, Fox E, Jacobs K, Van Ryckeghem DM, Lau JY. The relationship between adolescents' pain catastrophizing and attention bias to pain faces is moderated by attention control. PAIN 2015;156:1334–41.
    [26]. Holley AL, Wilson AC, Noel M, Palermo TM. Post-traumatic stress symptoms in children and adolescents with chronic pain: a topical review of the literature and a proposed framework for future research. Eur J Pain 2016;20:1371–83.
    [27]. Irwin DE, Stucky B, Langer MM, Thissen D, Dewitt EM, Lai JS, Varni JW, Yeatts K, DeWalt DA. An item response analysis of the pediatric PROMIS anxiety and depressive symptoms scales. Qual Life Res 2010;19:595–607.
    [28]. Judah MR, Grant DM, Mills AC, Lechner WV. Factor structure and validation of the attentional control scale. Cogn Emot 2014;28:433–51.
    [29]. Kamper SJ, Henschke N, Hestbaek L, Dunn KM, Williams CM. Musculoskeletal pain in children and adolescents. Braz J Phys Ther 2016;20:275–84.
    [30]. Kashikar-Zuck S, Carle A, Barnett K, Goldschneider KR, Sherry DD, Mara CA, Cunningham N, Farrell J, Tress J, DeWitt EM. Longitudinal evaluation of patient-reported outcomes measurement information systems measures in pediatric chronic pain. PAIN 2016;157:339–47.
    [31]. King S, Chambers CT, Huguet A, MacNevin RC, McGrath PJ, Parker L, MacDonald AJ. The epidemiology of chronic pain in children and adolescents revisited: a systematic review. PAIN 2011;152:2729–38.
    [32]. Lau JYF, Heathcote LC, Beale S, Gray S, Jacobs K, Wilkinson N, Crombez G. Cognitive biases in children and adolescents with chronic pain: a review of findings and a call for developmental research. J Pain 2018;19:589–98.
    [33]. Little R. A test of missing completely at random for multivariate data with missing values. J Am Stat Assoc 1988;83:1198–202.
    [34]. Merskey H, Bogduk N. Classification of chronic pain: descriptions of chronic pain syndromes and definition of pain terms. Task Force Taxonomy Int Assoc Study Pain 1994. pp. 209–214.
    [35]. Moriarty O, McGuire BE, Finn DP. The effect of pain on cognitive function: a review of clinical and preclinical research. Prog Neurobiol 2011;93:385–404.
    [36]. Neville A, Soltani S, Pavlova M, Noel M. Unravelling the relationship between parent and child PTSD and pediatric chronic pain: the mediating role of pain catastrophizing. J Pain 2018;19:196–206.
    [37]. Noel Ma, Groenewald CBbc, Beals-Erickson SEb, Gebert JTd, Palermo TMbefg. Chronic pain in adolescence and internalizing mental health disorders: a nationally representative study. PAIN 2016;157:1333–8.
    [38]. Ólafsson RP, Smári J, Guðmundsdóttir F, Ólafsdóttir G, Harđardóttir HL, Einarsson SM. Self-reported attentional control with the Attentional Control Scale: factor structure and relationship with symptoms of anxiety and depression. J Anxiety Disord 2011;25:777–82.
    [39]. Palermo TM, Fonareva I, Janosy NR. Sleep quality and efficiency in adolescents with chronic pain: relationship with activity limitations and health-related quality of life. Behav Sleep Med 2008;6:234–50.
    [40]. Palermo TM, Valenzuela D, Stork PP. A randomized trial of electronic versus paper pain diaries in children: impact on compliance, accuracy, and acceptability. PAIN 2004;107:213–19.
    [41]. Quigley L, Wright CA, Dobson KS, Sears CR. Measuring attentional control ability or beliefs? Evaluation of the factor structure and convergent validity of the Attentional Control Scale. J Psychopathology Behav Assess 2017;39:742–54.
    [42]. Rueda MR, Fan J, McCandliss BD, Halparin JD, Gruber DB, Lercari LP, Posner MI. Development of attentional networks in childhood. Neuropsychologia 2004;42:1029–40.
    [43]. Savedra MC, Tesler MD, Holzemer WL, Wilkie DJ, Ward JA. Pain location: validity and reliability of body outline markings by hospitalized children and adolescents. Res Nurs Health 1989;12:307–14.
    [44]. Schoth DE, Godwin HJ, Liversedge SP, Liossi C. Eye movements during visual search for emotional faces in individuals with chronic headache. Eur J Pain 2015;19:722–32.
    [45]. Sharp TJ, Harvey AG. Chronic pain and posttraumatic stress disorder: mutual maintenance? Clin Psychol Rev 2001;21:857–77.
    [46]. Simons LE, Claar RL, Logan DL. Chronic pain in adolescence: parental responses, adolescent coping, and their impact on adolescent's pain behaviors. J Pediatr Psychol 2008;33:894–904.
    [47]. Todd J, van Ryckeghem DML, Sharpe L, Crombez G. Attentional bias to pain-related information: a meta-analysis of dot-probe studies. Health Psychol Rev 2018;12:419–36.
    [48]. van Ryckeghem DML, Noel M, Sharpe L, Pincus T, Van Damme S. Cognitive biases in pain: an integrated functional-contextual framework. PAIN 2019;160:1489–93.
    [49]. Verstraeten K, Vasey MW, Claes L, Bijttebier P. The assessment of effortful control in childhood: questionnaires and the test of everyday attention for children compared. Personal Individual Differences 2010;48:59–65.
    [50]. Vervoort T, Caes L, Crombez G, Koster E, Van Damme S, Dewitte M, Goubert L. Parental catastrophizing about children's pain and selective attention to varying levels of facial expression of pain in children: a dot-probe study. PAIN 2011;152:1751–7.
    [51]. Vervoort T, Goubert L, Crombez G. The relationship between high catastrophizing children's facial display of pain and parental judgment of their child's pain. PAIN 2009;142:142–8.
    [52]. Vervoort T, Trost Z, Van Ryckeghem DM. Children's selective attention to pain and avoidance behaviour: the role of child and parental catastrophizing about pain. PAIN 2013;154:1979–88.
    [53]. Vinall J, Pavlova M, Asmundson GJG, Rasic N, Noel M. Mental health comorbidities in pediatric chronic pain: a narrative review of epidemiology, models, neurobiological mechanisms and treatment. Children 2016;3:40.
    [54]. Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. PAIN 2000;85:317–32.
    [55]. Walker LS, Sherman AL, Bruehl S, Garber J, Smith CA. Functional abdominal pain patient subtypes in childhood predict functional gastrointestinal disorders with chronic pain and psychiatric comorbidities in adolescence and adulthood. PAIN 2012;153:1798–806.
    [56]. Wilson AC, Samuelson B, Palermo TM. Obesity in children and adolescents with chronic pain: associations with pain and activity limitations. Clin J Pain 2010;26:705–11.
    Keywords:

    Chronic pain; Pediatric; Attentional bias; Attentional control; Eye-tracking

    © 2020 International Association for the Study of Pain