Dimensions of pain-related parent behavior: Development and psychometric evaluation of a new measure for children and their parents

Pain 137 (2008) 689–699 www.elsevier.com/locate/pain

Dimensions of pain-related parent behavior: Development and psychometric evaluation of a new measure for children and their parents Christiane Hermann a,b,*, Katrin Zohsel a, Johanna Hohmeister a, Herta Flor a a

Department of Cognitive and Clinical Neuroscience, Ruprecht-Karls University Heidelberg, Central Institute of Mental Health, Mannheim, Germany b Department of Clinical Psychology, Justus-Liebig-University Giessen, Otto-Behaghel-Strasse, 10F, D-35394 Giessen, Germany Received 24 August 2007; received in revised form 21 March 2008; accepted 31 March 2008

Abstract The development and maintenance of chronic pain are influenced by its social context, and especially by the responses of family members. For children, very few instruments are available that measure pain-related parental behavior. Using the Multidimensional Pain Inventory for adults (MPI; [Kerns RD, Turk DC, Rudy TE. The west haven-yale multidimensional pain inventory (WHYMPI). Pain 1985;23:345–356.]) as a model, we developed and evaluated a child and parent versions of the Pain-related Parent Behavior Inventory (PPBI). Here, we specifically studied maternal pain-related behavior as perceived by the child and self-reported by the mother. As substantiated by exploratory factor analysis in a mixed sample of 193 children and adolescents (8–16 years) either suffering from recurrent pain of different origin or being healthy controls, both PPBI versions entail the identical subscale solicitousness, distracting behaviors and discouraging/ignoring responses. Child and parent PPBI subscales were internally consistent and were not substantially related to age or gender. Validity analyses yielded a pattern of correlations with measures of depression, trait anxiety, pain activity, and pain-related cognitions that is consistent with the psychometric data for the adult MPI and findings on the social context of chronic pain. Child-perceived maternal behavior was significantly related to overall parenting and to mothers’ actual behavior as observed during a cold pressor test. Finally, the PPBI was sensitive to differences in mothers’ responses depending on the specific nature of the child’s pain. Child and parent reports of parental behaviors were modestly correlated and were differentially related to the validity measures, hence supporting the importance of assessing the social context of pediatric pain independently of both the child’s and the parent’s perspectives. Ó 2008 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. Keywords: Assessment; Children; Social responses; Parent behavior; Pain; Solicitousness; Discouragement; Distraction

1. Introduction According to Fordyce’s operant model [10], social responses provide contingencies of reinforcement and

* Corresponding author. Address: Department of Clinical Psychology, Justus-Liebig-University of Giessen, Otto-Behaghel-Strasse, 10F, D-35394 Giessen, Germany. Tel.: +49 641 99 26080; fax: +49 641 99 26099. E-mail address: [email protected] (C. Hermann).

punishment, thus shaping pain behavior. Experimental and clinical data suggest that solicitousness is associated with higher reported pain, more pain behaviors and greater disability in adults and children [7,22,25,33]. Contrary to a strictly operant perspective, in adults, punishing behaviors of significant others do not necessarily reduce pain and pain behaviors. Rather, such behavior is also related to pain, catastrophizing, disability and low acceptance (e.g., [4,21]). This suggests a broader perspective of the social context that takes into

0304-3959/$34.00 Ó 2008 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2008.03.031

690

C. Hermann et al. / Pain 137 (2008) 689–699

account the communicative aspect of patients’ and significant others’ pain-related behavior [19,23,29]. From a developmental perspective, social influences such as parental behaviors are particularly important in forming children’s responses to and beliefs about pain [6,8]. For children, there is a lack of assessment instruments despite their clinical usefulness as part of a multidimensional assessment of pediatric pain. Recently, the Inventory of Parent/Caregiver Responses to the Children’s Pain Experience (IRPEDNA; [15]) and that of the Adult Responses to Children’s Symptoms (ARCS; [32]) have become available that assess parent-reported parental behavior to children’s pain. Both instruments were designed using the significant others’ response scales of the MPI – significant others’ version (SOMPI; [16]) and interviews with parents of pediatric pain patients as item sources. Factor analyses confirmed a three-scale structure (ARCS: protect, minimize, distract & monitor; IRPEDNA: solicitousness, discouragement, promotion of well-behaviors and coping) which is consistent with the three-dimensions of significant others’ pain-related responses (‘solicitousness’, ‘distraction’, and ‘punishment’) originally identified in adult pain patients. Thus far, no multidimensional instrument has been validated to assess children’s perceptions of parental behaviors. In adults, patients and their significant others’ do not necessarily agree. Typically, correlations are small to medium [24,28]. In children with recurrent abdominal pain, Walker et al. [34] observed moderate agreement between mothers’ self-reported protective behaviors and children’s report (r = .33). Evidence is growing that patients’ and spouses’ perceptions provide differential information about the social context, may be differentially related to patients’ pain-related distress and disability, and are differentially influenced by the patients’ and the informants’ emotional functioning and the relationship between patient and informant (e.g., [5,19,23]). We aimed at developing and validating a parent and child version of a questionnaire for assessing painrelated parent behavior. Similar to the IRPEDNA and ARCS, part B of the MPI [17] and the significant others’ version of the MPI served as a model. In the present study, only maternal pain-related behavior was investigated. To this end, children with and without recurrent pain problems and their mothers were recruited. The factor structure and scale intercorrelations were examined and reliability was determined. Construct validity was evaluated using measures for examining convergent and discriminant validity and using external criteria. This included testing the relationship with the child’s emotional well-being and pain experiences as well as with core dimensions of parenting behavior and with the observation of mothers’ actual behavior in an experimental pain situation. Also, the relationship between the child’s and the mother’s reports was determined.

2. Methods 2.1. Participants One hundred and ninety-three children and adolescents (108 girls, 85 boys) between the ages of 8 and 16 years (M = 11.22, SD = 1.84) and their mothers participated. The sample included 61 pain-free children (30 girls, 31 boys; age: M = 10.92, SD = 1.66, range: 9–15 years), 71 children suffering from migraine or tension headache (38 girls, 33 boys; age: M = 11.07, SD = 1.78, range: 8–15 years), 47 children suffering from gastrointestinal pain (28 girls, 19 boys; age: M = 11.39, SD = 1.87, range 8–15 years) and a heterogeneous pain group of 14 children and adolescents suffering from widespread musculoskeletal or other pains (12 girls, 3 boys; age: M = 12.60, SD = 2.26, range 9–16 years). The pediatric pain patients had been suffering from pain for at least 6 months. On the average, a pain history of 4.38 years (SD = 2.76) was reported. The subjects were either participating in our own ongoing experimental studies on pediatric pain or received treatment in local outpatient clinics or a pediatric rheumatology clinic. Patients for the experimental studies were recruited by announcements in the local media. Outpatients and inpatients were approached at their intake on a consecutive basis, if they did not meet exclusion criteria. Exclusion criteria were (1) younger than 8 years or older than 16 years, (2) pain due to cancer or serious disease, (3) pain duration less than 6 months, and (4) problems with the German language. All eligible inpatients and outpatients who were approached agreed to participate, although they did not always comply with completing all of the additional questionnaires used for the validity analyses (see below). All children and adolescents and one of their parents signed informed consent. Inpatients and outpatients were informed that their treatment was not contingent upon completion of the questionnaires and that completion of the questionnaires was part of a study. Except for the cold pressor study described later, fathers were not a priori excluded from participation. However, since we anticipated that few fathers would participate, and since we wanted to keep the burden on the children at a minimum, the children were only asked to report on the mother’s behavior.

2.2. Development of the PPBI-C/-P The PPBI was developed based on part B of the MPI for adults ([17]; German: [9]) and the respective version for significant others (SO-MPI; [16]). Specifically, items reflecting three dimensions of pain-related parent behavior were included: solicitous responses, discouraging responses, and distracting responses. The MPI items were reformulated and adapted in content for use in children and their parents. Also, some items were dropped or added. The item ‘getting pain medication’ was not retained since there are several factors that may limit this item’s validity as an index of solicitousness. Whether or not medication is given to a child depends on many other factors such as parents’ attitude about pain medication for their child, children’s own attitude, and the type of the pain problem. The item ‘getting angry’ was also removed since based on our clin-

C. Hermann et al. / Pain 137 (2008) 689–699

691

ical experience such responses have a very low rate of occurrence, at least when assessed by self-report. Instead, three items were added that reflect distancing and minimizing parental behaviors (‘not taking pain seriously’, ‘being hesitant about believing the child’, and ‘not having sufficient time to attend to the pain’). The items of the child and parent versions of the PPBI are identical except that the wording reflects either the child’s or the parent’s perspective. Since it was expected that primarily mothers and their children would participate, the child version used in this study refers to the mother. For all items, respondents are asked to indicate how often the described parent behavior occurs when the child is in pain on a 5-point Likert scale (1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = very often).

man-speaking children and adolescents. Here, only the scale ‘social support’ was used for determining construct validity of the PPBI-C.

2.3. Measures for the validity analysis

2.3.5. Pain complaints They were assessed using an adapted version of the Children’s Somatization Inventory (CSI; [11]), which determines frequency and intensity of 36 somatic complaints including pain. Both frequency (‘‘how often do you have. . .”) and subjective intensity (‘‘how bothersome is. . .”) are rated on a 5point Likert scale (frequency: 0 = never. . . 4 = almost always; intensity: 0 = not at all. . . 4 = extreme). The CSI includes 7 pain items (headache, stomach ache, back pain, muscle pain, joint pain, limb pain, and chest pain). Average frequency and intensity of these pain complaints were used for evaluating the validity of the PPBI-C/-P.

In order to allow a comprehensive construct validity analysis, several measures were used. Specifically, for testing discriminant validity, measures of depressed mood and trait anxiety were included in order to demonstrate that painrelated maternal behavior is largely independent of the child’s emotional problems. For testing convergent validity, a measure of pain-related social support and a questionnaire assessing parenting behavior were used. Construct validity was further evaluated using external criteria such as pain complaints, pain-related cognitions and observed maternal behavior during an experimental pain test. 2.3.1. Depressed mood The Depression Test for Children (DTC; [27]) is a wellestablished German self-report screening instrument for depressive symptoms in children and adolescents between 9 and 15 years of age. The DTC contains 55 items to be answered on a dichotomous response scale (yes/no). The items form three subscales, ‘dysphoric mood’, ‘agitated behaviors’, and ‘exhaustion/somatic symptoms’, as verified by factor analysis. The DTC scales were shown to be internally consistent, temporally stable (1–5 week interval) and valid (factorial, convergent, and discriminative validity). 2.3.2. Trait anxiety The Children’s Anxiety Test-II (CAT-II; [31]) consists of 18 dichotomous (yes/no) items that address feelings of anxiousness, phobic fears, and fear-related somatic symptoms. Its reliability and validity have been demonstrated for children between the ages of 9 and 15 years and it is one of the most widely used instruments to measure trait anxiety in Germanspeaking children and adolescents. 2.3.3. Pain experience questionnaire – child version (PEQ-C) The PEQ-C was developed based on part A of the German MPI for adults [9] and measures the psychosocial impact of chronic pain in children. It contains the subscale pain severity, interference, affective distress, and social support. For each item, ratings are made on a 7-point Likert scale. Internal consistency, factorial, construct and external validity were recently demonstrated in a sample of pediatric pain patients and their mothers [13]. The PEQ-C is the first questionnaire validated for assessing the psychosocial impact of chronic pain in Ger-

2.3.4. Parenting behavior In order to evaluate construct validity of the PPBI-C, a subsample of children was asked to complete the Parental ChildRearing Inventory (PCI; ‘Erziehungsstilinventar’; [18]). The PCI consists of six subscales that represent the dimensions ‘praise’, ‘blame’, ‘inconsistency’, ‘support’, ‘restrictive behaviors’ and ‘punishment intensity’ of parental behaviors. The PCI has been comprehensively validated for children and adolescents (10–14 years) and is one of the most established German questionnaires on parenting behavior.

2.3.6. Pain-related cognitions They were determined by administering the Pain-Related Cognitions Questionnaire for children and adolescents (PRCQ-C; [12]). The PRCQ-C consists of three subscales (catastrophizing, positive self-statements, and problem-solving) that assess cognitions children may have when they are in pain. For each item a frequency rating is obtained using a 5-point Likert scale (1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = very often). Its reliability and validity have recently been demonstrated [12]. The PRCQ-C is the first questionnaire available for measuring pain-related cognitions in German speaking children. 2.3.7. Parental behavior during a cold pressor test A subsample of the children and their mothers participated in a cold pressor test (CPT). After an initial baseline (5 min) and anticipatory phase (30 s), children were prompted to immerse their arm into a cold water tank (10 °C) and keep it immersed as long as possible (maximal duration: 5 min). The cold water was not circulated. The immersion phase was followed by a recovery phase (4 min). The mother of the child was present during all four phases. During the immersion phase, the mother was instructed to interact with her child and encourage him/her to keep the arm immersed as long as possible. In addition, she was asked to obtain pain ratings from the child at regular intervals until the child removed the hand from the water. The interaction between mother and child was videotaped and scored off-line using the ChildAdult-Medical-Procedure-Interaction-Scale Revised [2,3]. The CAMPIS-R was originally developed to categorize parental verbal behavior during painful medical procedures, but has also been used to code maternal behavior during an

692

C. Hermann et al. / Pain 137 (2008) 689–699

experimental symptom provocation procedure (e.g., [35]). Maternal behavior was coded using the following categories: (1) neutral behaviors such as statements about the child’s current physical or emotional state or about the procedure itself; (2) coping-promoting behaviors such as suggestions on how to cope with the pain or distracting remarks; (3) distress-promoting behaviors such as expressing own worries, critical comments, or reassurance. The videotapes were scored by two independent raters (master level students) who were trained (8–10 h) and supervised by an experienced clinical psychologist (KZ). For the training, training videos were used and interrater agreement was checked prior to the coding of the data obtained in this study. Interrater agreement was good (intraclass correlation coefficients: .84–.94). 2.4. Procedure Children and their mothers completed the PPBI and the other questionnaires either as part of the study protocol or as part of their pre-treatment assessment. The PPBI of one child and those of two mothers were incomplete and, therefore, had to be excluded from analysis. Children who were treated as outpatients or inpatients and their parents did not always comply with completing all of the provided questionnaires (DTC, CAT-II, and CSI), despite repeated reminders and despite having consented earlier. In addition, some treatment providers were reluctant in asking children and their parents to complete a comprehensive battery of psychological questionnaires since a psychological assessment is not part of their routine assessment and was not considered as important. In these cases, we had to keep the number of administered questionnaires to a minimum. The validity analyses are based on: Pain-free group – N = 61 (CAT-II, DTC, PRCQ-C), N = 57 (CSI); Pain group – N = 94 (CAT-II), N = 93 (DTC), N = 77 (CSI child report), N = 79 (CSI parent report), N = 130 (PRCQ-C), and N = 131 (PEQ-C). A subsample of 24 pain-free children (13 girls, 11 boys; age: M = 11.29, SD = 1.55) and 23 children with recurrent abdominal pain (13 girls, 10 boys; age: M = 11.48, SD = 1.55) participated in the cold pressor test and completed the questionnaire on parenting behavior (PCI). These children had been recruited specifically for this experimental pain test. Only mothers and their children were invited because, in previous studies, we were not able to recruit a sufficient number of fathers willing to participate. The data for the main research questions of this experimental pain test will be reported elsewhere.

for the total sample and separately for the children with recurrent pain and the pain-free children, since one may expect differences in the strength of this relationship due to differences in the rate at which pain-related parent behavior occurs. In an exploratory fashion, for both PPBI-C and PPBI-P, age effects were determined by computing Pearson productmoment correlation coefficients for each of the subscales. In addition, gender differences were evaluated by multivariate analyses of variance (MANOVAs). These analyses were computed for the total sample and both subsamples of pain-free and pain-suffering children. In a first set of validity analyses based on the total sample, construct validity was determined by analyzing the relationship between all PPBI-C/-P subscales and the child’s emotional well-being (depressed mood, anxiety), pain complaints, and pain-related cognitions using Pearson product-moment correlations. These analyses were performed separately for the subgroup of healthy children and the children with recurrent pain since between-group differences with regard to the strength of these correlations could not be ruled out. In the pain group only, the relationship between child-reported pain-related parental behavior and child-perceived pain-related social support (PEQ-C) was evaluated. In order to control for the risk of an inflated Type I error due to the number of correlations between questionnaire measures, a conservative significance level of p < .01 was used for these validity analyses. Additional analyses revealed that, despite the fact that not all validity measures were validated for eight-year-olds, the same pattern of findings emerged regardless of whether or not these children were included in the analysis. Therefore, only the findings based on all participants are reported. In a second step, construct validity was further investigated in a subsample of the participants by evaluating the relationship between the PPBI-C subscales and observed maternal behavior in an experimental pain situation as well as parenting behavior in general (PCI). Since the behavioral observation data were not normally distributed, Spearman rank-correlations were computed. In light of the a priori hypotheses for these analyses, the significance level was set at .05. Finally, concurrent validity was explored by computing between-group differences with regard to the subscale scores between the pain-free group and the subgroups of children suffering from recurrent headaches, gastrointestinal pains and heterogeneous pain were evaluated using one-way analyses of variance (ANOVAs). For these analyses, the significance level was set at .05.

3. Results 2.5. Data analysis

3.1. PPBI-C PPBI-C and PPBI-P were each subjected to a principal component analysis with varimax rotation and the number of factors to be extracted was set to be three. Prior to factor analysis, sample adequacy was checked by performing Bartlett’s sphericity test and computing the Kaiser-MeyerOlkin-criterion (KMO). The distribution of the scale scores was evaluated using the one-sample Kolmogorov–Smirnov test. Scale intercorrelations were analyzed using Pearson product-moment correlation based on the total sample. The relationship between child and parent reports was determined

3.1.1. Exploratory factor analysis First, adequacy of the data for factor analysis was evaluated. A significant Bartlett’s sphericity test (v2(136) = 1103.97, p < .001) and a KMO value of .81 indicated that the variables in the population matrix were correlated. The hypothesized three-factor solution accounted for 52.73% of the variance (see Table 1). All items showed satisfactory factor loadings of at least .5 on one of the factors.

C. Hermann et al. / Pain 137 (2008) 689–699

693

Table 1 Results of the exploratory factor analysis of the PPBI-C in the total sample (N = 192) Items: When I am in pain, my mother. . ..

Factors

rit

Discouraging responses . . .becomes irritated (3) . . .tells me to pull myself together (4) . . .quickly becomes impatient with me (14) . . .doesn’t really listen to me (15) . . .doesn’t really believe me (11) . . .doesn’t have the time to take care of me (12) . . .doesn’t take it seriously (9) . . .takes special care of me (17) . . .is especially nice to me (6) . . .gets me something to eat or drink (2) . . .takes over my chores and duties (8) . . .makes me get some rest (7) . . .asks how she can help me (1) . . .suggests that I may play, read or do something else (5) . . .suggests doing something nice (10) . . .talks with me about something else in order to distract me (13) . . .turns on the TV or the radio (16) Accounted variance (%) Total variance accounted for (%)

Solicitous responses

Distracting responses

.75 .73 .73 .69 .67 .62 .56

.62 .60 .60 .54 .52 .49 .42 .63 .60 .58 .53 .51 .54

.82 .82 .64 .63 .58 .56

(.43) 19.59 52.73

18.79

.77 .74

.51 .64

.66 .62

.59 .51

14.36

Note: Only factor loadings >.40 are shown. The item numbers refer to the order of the items in the questionnaire. rit = item correlation coefficient, i.e., correlation between the item and the respective subscale with this item removed.

Two items (#1, #13) had double-loadings on the solicitousness and distraction subscales which differed by less than .3. These items were retained on the scale with the higher loading. With two exceptions (item 9: rit = .42; item 12: rit = .49), the item-discrimination coefficients (i.e., the correlation between the item and the respective factor with this item removed) of all items were satisfactory or good (.51 6 rit 6 .64). 3.1.2. Reliability As suggested by the respective Cronbach’s Alpha coefficients, all subscales were demonstrated to have satisfactory internal consistency: .80 (discouragement), .81 (solicitousness), and .77 (distraction). 3.1.3. Distribution of scale scores Normality of the scale scores was tested using the one-sample Kolmogorov–Smirnov test. The respective test statistic revealed that for the subscales ‘solicitousness’ (Z = 1.2) and ‘distraction’ (Z = 1.24), the null hypothesis of a normal distribution could not be rejected. The scores of the subscale ‘discouragement’ (Z = 2.46) were positively skewed. 3.2. PPBI-P 3.2.1. Exploratory factor analysis The suitability of the data for factor analysis was suggested by a significant Bartlett’s sphericity test

(v2(136) = 1003.47, p < .001) and a KMO value of .75. Principal component analyses revealed a three-factor solution accounting for 49.14% of the variance (see Table 2). The magnitude of the factor loadings of all items (P.5) was satisfactory. With one exception (#4), the items did not show pronounced double loadings with the loading on the primary factor being at least .3 higher than the loading on the remaining scales. Item #4 was retained on the scale with the higher-factor loading (difference = .15) which was also consistent with the factor solution obtained for the PPBI-C. Item-discrimination coefficients were moderate to good (.416 rit 6 .63), and three items (#1, #7, #8) had lower values. 3.2.2. Reliability All three subscales were shown to be internally consistent as indicated by the respective Cronbach Alpha values: .79 (discouragement), .72 (solicitousness), and .76 (distraction). 3.2.3. Distribution of scale scores Similar to the results for the PPBI-C, the null hypothesis of normal distribution of the scores on the subscales ‘solicitousness’ (Z = 1.0) and ‘distraction’ (Z = 1.16) could not be rejected. The distribution of the scores of the subscale ‘discouragement’ (Z = 1.25) showed a trend to being positively skewed.

694

C. Hermann et al. / Pain 137 (2008) 689–699

Table 2 Results of the exploratory factor analysis of the PPBI-P in the total sample (N = 191) Items: When my daughter/my son is in pain. . .

Factors

rit

Discouraging responses . . .I become impatient (14) . . .I become rather irritated (3) . . .I have little time to take care of him/her (12) . . .I do not always listen carefully (15) . . .tell him/her to pull him/herself together (4) . . .I do not take it seriously at first (9) . . .I find it hard to believe him/her (11) . . .I take special care of him/her (17) . . .I treat him/her especially nicely (6) . . .I try to make him/her get some rest (7) . . .I get him/her something to eat or drink (2) . . .I take over his/her chores and duties (8) . . .I ask how I can help him/her (1) . . .I talk with him/her as a distraction (13) . . .I suggest playing, reading or doing something else (5) . . .I encourage him/her to do something nice (10) . . .I turn on the TV or the radio (16)

Solicitous responses

Distracting responses

.81 .77 .69 .66 .55 .50 .49 .75 .70 .66 .59 .53 .49 .77 .75 .75 .59

Accounted variance (%) Total variance accounted for (%)

18.71 49.14

15.48

.63 .59 .48 .52 .49 .45 .43 .63 .57 .39 .41 .33 .39 .61 .57 .61 .45

14.94

Note: Only factor loadings >.40 are shown. The item numbers refer to the order of the items in the questionnaire. rit = item correlation coefficient, i.e., correlation between the item and the respective subscale with this item removed.

3.3. PPBI-C/-P scale intercorrelations The scales ‘solicitousness’ and ‘distraction’ were moderately, but significantly intercorrelated (see Table 3). No other significant scale intercorrelations were observed. The scale ‘solicitousness’ correlated significantly positively with the scale ‘distraction’ and significantly inversely with the scale ‘discouraging responses’. Moreover, a significant positive relationship between the scale ‘solicitousness’ and the scale ‘discouraging responses’ was noted (see Table 3).

– r(190) = .32 (p < .001). Subgroup analyses revealed that the relationship between parent- and child-reported discouragement and distracting behavior, but not solicitousness was slightly lower in the pain-free group than in the pain group (discouragement: pain group – r(130) = .29, p < .001, pain-free group – r(60) = .19, p = .15; solicitousness: pain group – r(130) = .26, p = .003, pain-free group – r(60) = .26, p = .045; distracting behavior: pain group – r(130) = .36, p < .001, pain-free group – r(60) = .18, p = .16). 3.5. Exploratory analysis of age and gender effects

3.4. Relationship between parent and child report In the total sample, for all three-dimensions of parental behavior significant, but small-sized positive correlations between child and maternal reports were obtained: discouragement – r(190) = .27 (p < .001), solicitousness – r(190) = .25 (p < .001) and distracting behavior Table 3 PPPBI-C and PPBI-P scale intercorrelations in the total sample Solicitous responses Solicitous responses

**

Distracting responses .51**

Distracting responses

.35**

Discouraging responses

-.21**

Discouraging responses -.15* -.003

.24**

In the total sample, distracting parent behavior as reported by the mother (PPBI-P) slightly decreased with increasing age of the child (r(191) = .15, p = .037). None of the PPIB-C or the remaining PPBI-P subscales was found to be significantly related to age in the total sample and either subsample of pain-free children and children with recurrent pain ( .14 < all rs < .008). Consistent across child and parent reports, the respective MANOVAs did not yield a significant gender main effect (PPBI-C: p > .3; PPBI-P: p > .5). 3.6. Construct validity analyses I: Relationship with measures of emotional well-being (discriminant validity), pain-related social support (convergent validity), pain complaints (external criterion), and pain-related cognitions (external criterion)

*

p < .01, p < .05. Note: Intercorrelations for the PPBI-C (N = 192) are shown to the right above the diagonal and for the PPBI-P (N = 191) to the left below the diagonal.

3.6.1. PPBI-C/-P subscale ‘solicitous responses’ According to both child and mother reports, maternal solicitousness was unrelated to the child’s level of

C. Hermann et al. / Pain 137 (2008) 689–699

695

Table 4 Correlations between PPBI-C/PPBI-P subscales, trait anxiety and depressive symptoms in the pain-free children and the children with recurrent pain PPBI-C1

PPBI-P1

Solicitous responses

Distracting responses

Discouraging responses

Solicitous responses

Distracting responses

Discouraging responses

Painfree

Pain group

Painfree

Pain group

Painfree

Painfree

Pain group

Painfree

Pain group

Painfree

Pain group

.05 .13 .03 .05

.10 .15 .10 .01

.07 .01 .09 .07

.05 .05 .08 .16

.06 .03 .05 .04

.00 .13 .09 .04

.17 .11 .06 .19

.05 .15 .14 .09

.25(*) .09 .12 .27*

.06 .14 .06 .02

.18 .23(*) .26* .06

Depressive symptoms (DTC) Dysphoric mood Agitated behavior Exhaustion/somatic complaints Anxiety (CAT-II)

Pain group .31* .14 .26* .34*

p < .01; (*)p < .05; Note: The significance level was set at p < .01 in order to control for multiple comparisons. The validity analyses are based on: Pain-free group– N = 61 (CAT-II, DTC); Pain group – N = 94 (CAT-II), N = 93 (DTC). Abbreviations: DTC = Depression Test for Children [27]; CAT-II = Children’s Anxiety Test II [31].

* 1

anxiety, depressive symptoms and pain complaints (see Table 4). Child-perceived maternal solicitous behavior was significantly positively correlated with positive self-statements and problem-solving (PRCQ-C) in the pain group and, as a trend, in the pain-free group (see Table 5). In the children with recurrent pain, maternal report of solicitous responses and the child’s level of pain catastrophizing were significantly positively associated. In the pain-free group, self-reported solicitousness of the mother showed a trend for being significantly related with problem-solving. In the pain group, childreported maternal solicitousness was significantly positively correlated with pain-related social support (PEQ-C, r = .67, p < .01). 3.6.2. PPBI-C/-P subscale ‘distracting responses’ In both groups, no significant association between child-reported distracting behavior of the mother and the child’s level of anxiety and depressive symptoms was observed. Similarly, in the pain-free group,

mother-reported distracting behavior was unrelated to the child’s emotional well-being. In the pain group, mothers reported significantly more pain-related distractive behavior when the child was highly anxious and more dysphoric (trend) (see Table 4). In both groups, children reporting a greater frequency of maternal distracting responses also indicated significantly higher levels of positive self-statements and pain-related problem-solving (PRCQ-C) (see Table 5). A higher frequency of mother-reported distracting behavior was significantly positively associated with a higher frequency and intensity of pain complaints primarily in the pain group (see Table 5). Mothers of children with recurrent pain problems were significantly more likely to endorse distracting responses when the child engaged in pain-related catastrophizing (see Table 5). Finally, in children with recurrent pain, child-reported distracting responses of the mother were significantly correlated with pain-related social support (PEQ-C; r = .36, p < .01).

Table 5 Correlations between PPBI-C/PPBI-P subscales, pain complaints and pain-related cognitions in the pain-free children and the children with recurrent pain PPBI-C

PPBI-P

Solicitous responses

Distracting responses

Discouraging responses

Solicitous responses

Painfree

Pain group

Painfree

Pain group

Painfree

Pain group

Pain complaints (CSI child/parent version)1 Frequency .08 .11 Intensity .18 .16

.22 .05

.22(*) .19

.04 .13

.20 .17

.05 .04

.06 .01

.27(*) .20

Pain-related cognitions (PRCQ-C) Catastrophizing .18 Positive self-statements .30(*) Problem-solving .21

.15 .51* .47*

.16 .34* .29*

.27(*) .26(*) .05

.20(*) .05 .01

.18 .24 .28(*)

.25* .03 .07

.14 .27(*) .01

Painfree

Pain group

.12 .31* .23*

Distracting responses

Discouraging responses

Painfree

Painfree

Pain group

.41* .38*

.24 .22

.07 .07

.37* .03 .12

.06 .20 .02

.18(*) .01 .07

Pain group

p<.01; (*)p < .05; Note: The significance level was set at p < .01 in order to control for multiple comparisons. PPBI-C was correlated with CSI child report, and PPBI-P was correlated with CSI parent report. The analyses were based on PPBI-C – N = 57 (pain-free group) and N = 77 (pain group), PPBI-P – N = 57 (pain-free group) and N = 79 (pain group). Abbreviations: CSI = Children’s Somatization Inventory [11]; PRCQ-C = Pain-related Cognition Questionnaire for children [12]. * 1

696

C. Hermann et al. / Pain 137 (2008) 689–699

3.6.3. PPBI-C/-P subscale ‘discouraging responses’ In the pain-free group, no significant relationship between maternal discouraging responses as reported by either the child or the mother and the child’s anxiety level and depressive symptoms was observed (see Table 4). By contrast, in the children with recurrent pain, greater anxiety, higher levels of dysphoric mood and exhaustion/somatic complaints were significantly positively associated with more frequent child-perceived maternal discouraging responses (see Table 4). Moreover, mothers reported significantly more pain-related discouraging responses when the child endorsed higher levels of somatic symptoms (DTC). As indicated by child and mother report in both groups, frequency and intensity of pain complaints did not correlate significantly with pain-related maternal discouragement (see Table 5). In both groups of children, child-perceived discouraging maternal behavior showed a trend towards being significantly positively associated with painrelated catastrophizing (see Table 5). There was also a trend for mother’s self-reported discouragement to be positively correlated with children’s catastrophizing. In the pain-free group only, positive self-statements showed a trend to be significantly positively associated with discouraging maternal behavior as indicated by the child. Finally, in the pain group, greater maternal discouragement (PPBI-C) was significantly associated with lower levels of perceived social support (PEQ-C) (r = .32, p < .01). 3.7. Construct validity analyses II: Relationship with parenting behavior (PCI) (convergent validity) and observed maternal behavior during a cold pressor test (external criterion) 3.7.1. Parenting behavior (PCI) Pain-related solicitous behavior of the mother was significantly positively correlated with supportive parenting behavior in general (r(46) = .59, p < .001) and, as a trend, with parental praising (r(46) = .28, p = .061). Distracting pain-related maternal behavior was significantly positively associated with supportive parenting (r(46) = .45, p < .01). Pain-related discouraging maternal behavior was significantly positively

related to mothers’ general tendency to engage in blaming (r(46) = .53, p < .001), inconsistent parenting behavior (r(46) = .59, p < .001) and restrictive behavior (r(46) = .53, p < .001). The remaining correlations did not reach statistical significance. 3.7.2. Maternal behavior during a cold pressor test Pain-related distracting behavior of the mother (PPBI-C) was significantly positively related to maternal coping-promoting behavior during the CPT (r(44) = .31, p = .04). Moreover, maternal discouraging behavior (PPBI-C) was significantly positively correlated with distress-promoting behavior of the mother during the CPT (r(44) = .33, p = .03). Also, a significant inverse relationship between maternal solicitousness (PPBI-C) and neutral behavior of the mother during the CPT was observed (r(44) = .32, p = .03). No other significant correlations between actual maternal behavior in the experimental pain situation and child-reported pain-related maternal behavior were observed. 3.8. Concurrent validity: between-group comparisons of PPBI-C/-P subscale scores For both PPBI-C and PPBI-P, one-way ANOVAs revealed significant group differences with regard to parental distracting responses (PPBI-C: F(3, 188) = 5.48, p = .001; PPBI-P: F(3, 187) = 8.78, p < .001; see Table 6). Post-hoc Tukey tests showed that children suffering from recurrent headaches and their mothers reported significantly less maternal distracting behavior as compared to the control and the other pain groups (p’s < .05). Discouraging and solicitous maternal behaviors as reported by either the child or the mother did not significantly differ between groups. 4. Discussion The present study sought to evaluate parallel versions of the PPBI for children and parents that assess painrelated parental responses. Based on the MPI, the PPBI-C was first designed and then the parent form as its counterpart. As substantiated by exploratory factor analyses, both versions consist of three subscales that

Table 6 PPBI-C and PPBI-P scores (M, SD; scale range: 1–5) for the different subgroups of children Subgroup

Controls (PPBI-C: N = 61; PPBI-P: N = 60) Recurrent headache (PPBI-C: N = 70; PPBI-P: N = 71) Recurrent GI pain (PPBI-C: N = 46; PPBI-P: N = 45) Heterogeneous pains (PPBI-C/-P: N = 15)

PPBI-C

PPBI-P

Solicitous responses

Distracting responses

3.42 3.56 3.82 3.37

3.00 2.53 3.16 3.23

(.70) (.92) (.83) (1.05)

Note: Means with different superscripts differ significantly from one another.

(.93)b (.99)a (.93)b (.89)b

Discouraging responses

Solicitous responses

Distracting responses

1.52 1.52 1.75 1.42

3.59 3.52 3.42 3.33

2.70 2.19 2.83 3.05

(.52) (.58) (.77) (.49)

(.61) (.72) (.75) (.72)

(.72)b (.89)a (.78)b (.94)b

Discouraging responses 1.79 1.80 2.05 1.71

(.52) (.59) (.63) (.65)

C. Hermann et al. / Pain 137 (2008) 689–699

assess solicitous, discouraging and distracting behaviors with the same items loading on each subscale. The scale model is consistent with the MPI and both ARCS and IRPEDNA. At the item level, the PPBI scales ‘solicitousness’ and ‘discouragement’ closely match the respective ARCS (‘protect’, ‘minimize’) and IRPEDNA (‘solicitousness’, ‘discouragement’) scales. For the PPBI-C/-P, factor-loadings and item-discrimination coefficients were good and allowed each item to be assigned unambiguously to one scale. Internal consistency of the subscales is satisfactory and matches well with the values for the MPI (.77–.81; [17]) and the SOMPI (.75–.8, [16]; .78–.91, [28]). Moreover, despite being about half the length of the ARCS, the internal consistency of the PPBI-P subscales is of comparable magnitude (.67–.87, [32]). The PPBI subscales are not independent. Similar to MPI, ARCS and IRPEDNA scales, solicitous and distracting parental behaviors were substantially correlated (PPBI-C: .51; MPI [9,17,20]: .40–.53; PPBI-P: .35; SOMPI [16]: .54; ARCS: .42; IRPEDNA: .40). In accordance with previous findings, the scales ‘solicitousness’ and ‘discouragement’ showed a small inverse correlation reaching statistical significance only for the PPBI-P (PPBI-P: .21; SO-MPI: .04; ARCS: .22; IRPEDNA: .17). As for the MPI, the PPBI-C scales ‘discouragement’ and ‘distraction’ were unrelated. Unlike the respective SO-MPI (.08), ARCS ( .13) and IRPEDNA ( .17) scales, the PPBI-P scales ‘discouragement’ and ‘distraction’ were significantly positively correlated (.24). This finding is most likely accounted for by the specific scale contents. Contrary to the SO-MPI, the PPBI-P scale ‘discouragement’ emphasizes parents’ not taking the child’s pain seriously rather than expressing frustration. On the other hand, in contrast to the MPI and the PPBI-C/-P scale ‘distracting behavior’, the ARCS scale ‘encourage and monitor’ and the IRPEDNA scale ‘promoting well behavior’ not only assess distracting behaviors, but also include items reflecting parents’ attending to the child’s pain and prompting active coping. Possibly, parents more willing to ignore the child’s pain may also tend to distract the child. Age and gender of the child were not substantially related to the PPBI-C/-P subscales. This corresponds with the respective data for the ARCS [32]. Similar to the IRPEDNA ( .13), with increasing age of the child, parents were somewhat less likely to report distracting responses (PPBI-P: .15). The lack of a pronounced age effect is not surprising if the fundamental nature of the assessed social responses is taken into account. For example, when interviewed, adult patients and their spouses also report solicitous and distracting responses as most frequent [23]. For the IRPEDNA, Huguet et al. [15] observed that parents of boys as compared to girls endorsed higher levels of solicitous and distract-

697

ing responses. This IRPEDNA gender effect is most likely accounted for by the specific sample composition (e.g., age, gender proportion, and presence/absence of recurrent pain) and the parental informant. Mothers as compared to fathers endorsed more promoting wellbehaviors and solicitousness, yet the interaction between parental informant and child’s gender was not specifically tested [15]. Clearly, from a developmental perspective, more research is needed to unravel the complex pattern of sex- and gender-role related factors that modulate the response to pain in others [1]. Here, age and gender effects were only investigated in an exploratory manner. Evidence for construct validity was also obtained. For children without recurrent pain, pain-related parental behaviors (PPBI-C/-P) were unrelated to the child’s level of trait anxiety, depressive symptoms, and pain complaints, hence supporting the discriminant validity of the PPBI. Children with recurrent pains who were more anxious, dysphoric, and suffered from more somatic symptoms also reported more discouraging parental responses. Hence, similar to adult chronic pain sufferers [24], emotionally distressed pediatric pain patients may be more sensitive to parents’ ignoring and minimizing the child’s pain. Alternatively, children may express emotional distress if their mothers show such distancing behaviors. Or, the mothers of these children may get increasingly frustrated. The mothers in the pain group endorsed more discouraging behavior when their children indicated more somatic symptoms. Moreover, the higher the frequency and intensity of the child’s pain complaints, the more likely the mother was to report distractive responses. This pattern may reflect these parents’ experience with recurrent pain episodes and, potentially, increased efforts for directing attention away from the pain. Most importantly, the different correlational patterns for PPBI-C and PPBIP underline the importance of measuring the child’s and parent’s perceptions. This conclusion is further supported by the observed modest relationship between the children’s (PPBI-C) and the parents’ answers (PPBI-P) which replicates previous observations in adults [24] and in children [34]. For solicitousness, the relationship between child and parent reports was of comparable size in children with and without recurrent pain. For discouragement and distracting behavior, these correlations were lower in the pain-free group. Most likely, this pattern is accounted for by the base rates of parental pain-related behavior with solicitousness being the primary parental response. Especially if pain frequency is low, other parental behaviors are likely to occur rarely and less consistently. Convergent validity was supported by strong positive correlations between the PPBI-C subscales ‘solicitousness’ and ‘distraction’ and child-reported pain-related

698

C. Hermann et al. / Pain 137 (2008) 689–699

social support (PEQ-C). Also, child-perceived parental discouragement was significantly related to lower levels of pain-related social support. Thus, perceived painrelated social support seems to reflect the overall pattern of caregivers’ situation-specific behaviors [23]. The relationship between parental behavior and painrelated cognitions as an external criterion was also evaluated. Whether or not they were suffering from recurrent pain, children endorsing more positive selfstatements and problem-solving also indicated significantly greater maternal solicitousness and distracting behaviors. Possibly, such supportive maternal behaviors may help children to engage in approach-oriented coping. Children’s proneness to catastrophizing showed a trend to be associated with child-perceived maternal discouragement. This pattern is inconsistent with the communal model of catastrophizing which postulates that catastrophizing may be a means to elicit solicitous responses from others [30]. Rather, it matches recent data in adults showing that punishing responses of others are associated with more catastrophizing [4] and reduced pain acceptance [21] and that dissatisfying spousal responses enhance catastrophizing [14]. On the other hand, primarily in the pain group, children’s catastrophizing was significantly associated with greater levels of mother-reported solicitousness, distraction, and as a trend with discouragement. Possibly, children who express more catastophizing serve as a potent cue for triggering maternal responses. Clearly, the direction in which social responses and patient’s catastrophizing interact and mutually influence each other remains to be elucidated. Rather than assuming a linear relationship, there are good reasons for postulating curvilinear relationships between children’s catastrophizing and parental responses. Construct validity of the PPBI-C was further evaluated in a subsample of children with RAP and controls undergoing a CPT in the presence of their mothers. Mothers were more likely to display coping-promoting behaviors when their children had indicated greater levels of maternal distracting behavior. Furthermore, mothers were significantly more prone to interact with their child in a distress-promoting manner if, according to the child’s report, they tended to punish and discourage the child’s pain behaviors. Finally, higher levels of child-reported solicitousness were associated with a significantly lower frequency of neutral maternal behaviors during the CPT. It seems that children’s reported parent behaviors (PPBI-C) do in fact correspond to parents’ actual responses when the child is in pain. Moreover, pain-related parental responses reflect overall parenting. Specifically, supportive parenting was significantly associated with greater parental solicitousness and distraction. Maternal pain-related discouragement was highly related to blaming, inconsistent, and restrictive parenting. Thus, consistent with the findings in adults (e.g.,

[9,16,20,23,28]), the parent–child interaction and the quality of the parent–child communication seem to mold the pattern of parents’ behaviors when the child is in pain. Finally, concurrent validity was explored. PPBI-C/P did not reveal evidence for pronounced differences in pain-related parent behavior between children with or without recurrent pain, in accordance with the findings for the IRPEDNA [15] and during a pain-inducing exercise task [26]. Yet, children with recurrent headaches, most of them suffering from migraine, and their mothers concurred that mothers engage less in distractive behaviors. Typically, during a migraine attack, patients suffer from photo-/phonophobia and prefer resting undisturbed. Mothers may therefore refrain from distracting activities. It appears that the PPBI is sufficiently sensitive to differentiate such pain-specific social responses. Taken together, our results suggest satisfactory psychometric properties of PPBI-C and PPBI-P. Clearly, additional validation is necessary. One limitation is that retest reliability could not be examined. Second, the validity analyses need to be replicated using additional samples. Also, it will be crucial to investigate fathers’ pain-related responses. Third, predictive validity and sensitivity to change await to be demonstrated. Despite these limitations, both PPBI-C/-P are promising for a multidimensional assessment of the social context of pain in children and adolescents. It can provide important information for clinicians when trying to understand a child’s pain problem and when tailoring treatment. For example, the PPBI could be used for deciding whether or not a parent training is needed. Our data emphasize the need for assessing the social context of pediatric pain both from the child’s as well as the parent’s perspective.

Acknowledgements We thank Sabrina Klimm and Stefanie Greiner for their help in conducting the cold pressor test and Sandra Herrmann for her help in collecting the questionnaire data. We also thank Friedrich Ebinger, MD, Children’s University Hospital Heidelberg, the staff of the outpatient clinic at the Children’s University Hospital Mannheim, the pediatric migraine program at the Department of Child and Adolescent Psychiatry at the University Heidelberg and the staff of the Children’s Rheumatology Clinic Garmisch-Partenkirchen for their help in recruiting the patients. The study was funded by the Deutsche Forschungsgemeinschaft (Clinical Research Unit 107 ‘‘Neuronal plasticity and learning in chronic pain: Mechanisms, prevention and therapy”, Project He 2784/5). The authors do not have any conflicts of interest.

C. Hermann et al. / Pain 137 (2008) 689–699

References [1] Bernardes SF, Keogh E, Lima ML. Bridging the gap between pain and gender research: a selective literature review. Eur J Pain 2008;12:427–40. [2] Blount RL, Cohen LL, Frank NC, Bachanas PJ, Smith AJ, Manimala MR, et al. The Child–adult medical procedure interaction scale-revised: an assessment of validity. J Pediatr Psychol 1997;22:73–88. [3] Blount RL, Sturges JW, Powers SW. Analysis of child and adult behavioral variations by phase of medical procedure. Behav Ther 1990;21:33–48. [4] Boothby JL, Thorn BE, Overduin LY, Ward LC. Catastrophizing and perceived partner responses to pain. Pain 2004;109:500–6. [5] Cano A, Gillis M, Heinz W, Geisser M, Foran H. Marital functioning, chronic pain, and psychological distress. Pain 2004;107:99–106. [6] Chambers CT. The role of family factors in pediatric pain. In: McGrath PJ, Allen Finley G, editors. Pediatric pain: biological and social context. Seattle: IASP Press; 2003. p. 99–130. [7] Chambers CT, Craig KD, Bennett SM. The impact of maternal behavior on children’s pain experiences: an experimental analysis. J Pediatr Psychol 2002;27:293–301. [8] Craig KD, PillaiRiddell RR. Social, influences, culture and ethnicity. In: McGrath PJ, Finley GA, editors. Pediatric pain: biological and social context. Seattle: IASP Press; 2003. p. 159–82. [9] Flor H, Rudy TE, Birbaumer N, Streit B, Schugens MM. Zur Anwendbarkeit des West Haven-Yale Multidimensional Pain Inventory im deutschen Sprachraum. Der Schmerz 1990;4:82–7. [10] Fordyce WE. Behavioral methods in chronic pain and illness. St. Louis, MO: Mosby; 1976. [11] Garber J, Walker LS, Zeman J. Somatization symptoms in a community sample of children and adolescents: further validation of the children’s somatization inventory. Psychol Assess 1991;3:588–95. [12] Hermann C, Hohmeister J, Zohsel K, Ebinger F, Flor H. The assessment of pain coping and pain-related cognitions in children and adolescents: current methods and further development. J Pain 2007;10:802–10. [13] Hermann C, Hohmeister J, Zohsel K, Tuttas M, Flor H. The impact of chronic pain in children and adolescents: development and initial validation of a child and parent version of the pain experience questionnaire. Pain 2008;135:251–61. [14] Holtzman S, DeLongis A. One day at a time: the impact of daily satisfaction with spouse responses on pain, negative affect and catastrophizing among individuals with rheumatoid arthritis. Pain 2007;131:202–13. [15] Huguet A, Miro J, Nieto R. The inventory of parent/caregiver responses to the children’s pain experience (IRPEDNA): development and preliminary validation. Pain 2008;134:128–39. [16] Kerns RD, Rosenberg R. Pain-relevant responses from significant others: development of a significant-other version of the WHYMPI scales. Pain 1995;61:245–9. [17] Kerns RD, Turk DC, Rudy TE. The west haven-yale multidimensional pain Inventory (WHYMPI). Pain 1985;23:345–56.

699

[18] Krohne HW, Kiehl GE, Neuser KW, Pulsack A. Das ‘Erziehungsstil-Inventar’ (ESI): Konstruktion, Psychometrische Kennwerte Gu¨ltigkeitsstudien. Diagnostica 1984;30:299–318. [19] Leonard MT, Cano A, Johansen AB. Chronic pain in a couples context: a review and integration of theoretical models and empirical evidence. J Pain 2006;7:377–90. [20] Lousberg R, Van Breukelen GJ, Groenman NH, Schmidt AJ, Arntz A, Winter FA. Psychometric properties of the multidimensional pain inventory, dutch language version (MPI-DLV). Behav Res Ther 1999;37:167–82. [21] McCracken LM. Social context and acceptance of chronic pain: the role of solicitous and punishing responses. Pain 2005;113:155–9. [22] Merlijn VP, Hunfeld JA, van der Wouden JC, HazebroekKampschreur AA, Passchier J, Koes BW. Factors related to the quality of life in adolescents with chronic pain. Clin J Pain 2006;22:306–15. [23] Newton-John TR, Williams AC. Chronic pain couples: perceived marital interactions and pain behaviours. Pain 2006;123:53–63. [24] Pence L, Cano A, Thorn B, Ward LC. Perceived spouse responses to pain: the level of agreement in couple dyads and the role of catastrophizing, marital satisfaction, and depression. J Behav Med 2006;29:511–22. [25] Peterson CC, Palermo TM. Parental reinforcement of recurrent pain: the moderating impact of child depression and anxiety on functional disability. J Pediatr Psychol 2004;29:331–41. [26] Reid GJ, McGrath PJ, Lang BA. Parent–child interactions among children with juvenile fibromyalgia, arthritis, and healthy controls. Pain 2005;113:201–10. [27] Rossmann P. DTK-Depressions test fu¨r Kinder [DTC-Depression Test for Children]. 2nd edition. Bern: Huber Verlag; 2005. [28] Sharp TJ, Nicholas MK. Assessing the significant others of chronic pain patients: the psychometric properties of significant other questionnaires. Pain 2000;88:135–44. [29] Sullivan MJ, Martel MO, Tripp D, Savard A, Crombez G. The relation between catastrophizing and the communication of pain experience. Pain 2006;122:282–8. [30] Sullivan MJ, Thorn B, Haythornthwaite JA, Keefe F, Martin M, Bradley LA, et al. Theoretical perspectives on the relation between catastrophizing and pain. Clin J Pain 2001;17:52–64. [31] Thurner R, Tewes P. Kinder-Angst-test (KAT II) [children’s anxiety test]. Go¨ttingen: Hogrefe; 2000. [32] Van Slyke DA, Walker LS. Mothers’ responses to children’s pain. Clin J Pain 2006;22:387–91. [33] Walker LS, Claar RL, Garber J. Social consequences of children’s pain: when do they encourage symptom maintenance? J Pediatr Psychol 2002;27:689–98. [34] Walker LS, Levy RL, Whitehead WE. Validation of a measure of protective parent responses to children’s pain. Clin J Pain 2006;22:712–6. [35] Walker LS, Williams SE, Smith CA, Garber J, Van Slyke DA, Lipani TA. Parent attention versus distraction: impact on symptom complaints by children with and without chronic functional abdominal pain. Pain 2006;122:43–52.