Mother-child concordance for pain location in a pediatric chronic pain sample

Abstract

Body maps have long been used to assess pain location in adult and pediatric chronic pain patients. Assessing agreement between parent and child reports of pain location using such maps may help establish a unified picture of children’s pain experience. However, few studies have examined the extent of agreement between mothers and children on the location of the child’s pain. Using kappa coefficients and other determinants of the magnitude of kappa we assessed mother-child concordance in pain location using body maps with 21 standardized areas in 41 children with chronic pain (65.9% female, mean age = 14.60) and their mothers. The highest level of agreement was found for the abdominal region; agreement for the head region was moderate and not superior to the other body areas. Approximately half of the body map areas yielded poor to fair mother-child agreement, while the other half yielded moderate or better agreement. There was more agreement between mothers and sons than between mothers and daughters on the total number of body areas considered painful, but there were no effects of pubertal status, race, and ethnicity on agreement. Our results are consistent with previous studies indicating that parent assessments of children’s pain do not necessarily mimic their child’s report. Future research should test additional psychosocial factors that may contribute to parent-child discordance regarding the location of the child’s pain.

Introduction

Pain perception in children is complex and often difficult to assess. Although parents are often seen as the “experts” regarding their child’s pain, (1) it is now recognized that children themselves can provide accurate reports of their own pain. (2-4) In accord, the majority of recent studies have examined the child’s own account of his/her pain in addition to reports from the child’s parents. A key question posed by researchers and clinicians alike is the extent to which parents and children agree regarding the child’s pain. Extant work in this area has focused primarily on indicators such as pain intensity or pain frequency with few studies examining parent-child agreement on the location of the child’s pain. Despite the usual focus on the chief presenting complaint during clinical encounters, children often experience pain in more than one location. (5) Therefore, understanding parent-child agreement on the location(s) of the child’s pain may enhance the accuracy of pain assessment and treatment in pediatric chronic pain.

Body maps (i.e. pain charts, manikins) have long been used as part of the multidimensional assessment of pain in pediatric samples, and such maps can be completed without assistance by children as young as 8 years of age. (5) Only three prior studies have utilized body maps to assess parent-child agreement regarding the child’s pain. Using a body map with 17 different locations in pediatric spina bifida patients, one such investigation found that children who reported more intense pain also reported more pain locations; there was higher parent-child concordance for pain intensity in locations where pain of severe intensity was reported. (6) However, parent-child agreement regarding specific pain locations was not examined. Another investigation in Duchene (DMD) and Becker (BMD) muscular dystrophy patients found poor parent-child concordance for pain duration and frequency using a body map with 9 distinct areas, but agreement on pain location was not assessed. (7) Graumlich et al, (8) in their study of pediatric patients with sickle cell disease (SCD), reported parent-child concordance rates using a body map with 14 body locations. The percent agreement between parents and children regarding pain location ranged from 58.3% to 95.8% across all body sites, but agreement for individual areas was not reported.

Recently, von Baeyer, et al (5) proposed the use of a body map developed by the SUPER-KIDZ research group (9) containing 21 standardized and demarcated areas. Use of a standardized body map facilitates comparisons across studies, and pain locations may be more accurately assessed using a body map than by interview. (5) The aim of the current study was to examine agreement between mothers and children regarding the location of the child’s pain using the SUPER-KIDZ body map in a sample of pediatric chronic pain patients. Because prior studies have not done so, we tested mother-child agreement for each of the body regions. As noted by Chambers et al(10), previous reports showing high parent-child concordance have mostly relied on correlational analyses which may over-estimate parent-child agreement. Thus, following Chambers, we calculated kappa coefficients to assess mother-child agreement as well as additional statistics, i.e., prevalence index, bias index, and the maximum obtainable kappa, to aid in the interpretation of kappa. Since headaches and abdominal pain are the most common pain complaints in children,(11) we hypothesized that agreement for these areas would be greater than that of the other body regions. Based on the findings of Chambers(10) and others,(6-9) we expected the overall extent of agreement to be moderate. We also conducted exploratory analyses to examine sex and racial/ethnic differences in overall mother-child agreement regarding the number of painful locations.

Methods

Participants

All recruitment and study procedures were approved by the University of California, Los Angeles Institutional Review Board (IRB). Subjects were part of a larger study on sex and pubertal differences in laboratory pain responses. The study sample consisted of children and adolescents with chronic pain aged 8-17 years (n = 46), as well as their mothers (n = 46). All fathers were also invited to participate; however, only a small number of fathers did so (n = 5). Therefore, only data on mothers and children will be presented. The sample was primarily recruited through a multidisciplinary tertiary clinic specializing in pediatric chronic pain. Prior to the patient’s initial consultation with the clinic physicians, the patient and his/her parents met with a trained research associate who gave an initial overview of the study. If the patient and his/her family were interested in participating, the research associate then asked a series of questions to ascertain study eligibility. Inclusion and exclusion criteria are further discussed below.

Seven participating families were originally recruited as well-child control participants for the larger laboratory pain study but were found to meet criteria for chronic pain (15.2% of the sample). These participants were recruited through advertisements, community events, and via referrals from previous participants (who earned an additional $25 for each referred family that completed the study). Study advertisements were posted on online forums (e.g. Craigslist) and at physical locations (e.g. libraries, pediatricians’ offices, etc.). Participants were also recruited by study staff at community events (festivals/fairs, etc.). The study PI (LKZ) reviewed each of these cases and re-assigned them to the chronic pain group. Seven additional participating families (15.2% of the sample) were recruited from other local clinics and by physicians at private practices and university-based clinics. Physicians and clinic staff provided patients with the study flyer to enable interested families to contact the study coordinator regarding participation. Each of these cases was reviewed by the study PI to determine eligibility.

Inclusion criteria followed the commonly accepted definition of chronic/recurrent pain as pain that has persisted for three months or longer. (12) Subjects were excluded from the study if they presented with a developmental delay, autism, or significant anatomic impairment that could preclude understanding of or participation in study procedures. Siblings of the child/adolescent participants were also excluded. Mothers could be biological or adoptive parents of the child; however, the child had to reside at least half of the time in the household with their mother for her to be eligible to participate. Each participating family member was compensated with $50 for their participation.

Of the original 46 mother-child pairs enrolled in the study, 5 pairs (10.9%) were excluded because participants did not complete a body map (i.e. they reported that the child did not experience pain in the last month). Therefore, data from 41 children (65.9% female, 75.6% late puberty, mean age = 14.60, age range = 9.33) and 41 mothers were analyzed in this study. See Table 1 for additional demographic information. Table 1 also displays the presenting pain complaints for girls, boys, and the total sample. As shown in Table 1, over 60% of the patient sample presented with more than one pain complaint.

Table 1.

: Demographic characteristics and presenting pain complaints for girls, boys, and the total sample.

	Boys (N = 14)	Girls (N = 27)	Total Sample (N = 41)
Child mean age in years (SD)	13.5 (2.5)	15.3 (2.3)	14.6 (2.5)
Child pubertal state [Late Puberty – n (%)]	8 (57.1)	23 (85.2)	31 (75.6)
Child race [Caucasian – n (%)]	8 (57.1)	18 (66.7)	26 (63.4)
Child ethnicity [Hispanic – n (%)]	3 (21.4)	6 (22.2)	9 (22.0)
Mother mean age in years (SD)	45.2 (5.0)	45.47 (7.8)	45.38 (6.9)
Mother race [Caucasian – n (%)]	9 (64.3)	19 (70.4)	28 (68.3)
Mother ethnicity [Hispanic – n (%)]	3 (21.4)	3 (11.1)	6 (14.6)
Pain Diagnosis
Headaches	9 (64.3)	15 (55.6)	24 (58.5)
Abdominal Pain	4 (28.6)	16 (59.3)	20 (48.8)
CRPS	1 (7.1)	2 (7.4)	3 (7.3)
Fibromyalgia	2 (14.3)	9 (33.3)	11 (26.8)
Joint Pain	1 (7.1)	2 (7.4)	3 (7.3)
Neck Pain	1 (7.1)	1 (3.7)	2 (4.9)
Shoulder Pain	0 (0.0)	1 (3.7)	1 (2.4)
Back Pain	3 (21.4)	6 (22.2)	9 (22.0)
Leg Pain	1 (7.1)	1 (3.7)	2 (4.9)
Chest Pain	0 (0.0)	1 (3.7)	1 (2.4)
Hand Pain	0 (0.0)	1 (3.7)	1 (2.4)
Foot Pain	1 (7.1)	1 (3.7)	2 (4.9)
Multiple Pain Diagnoses	7 (50.0)	19 (70.4)	25 (61.0)

Note: Frequencies for pain diagnoses sum to more than 100% due to multiple pain diagnoses.

CRPS = complex regional pain syndrome, type 1 or type 2.

Procedure

Mothers provided written informed consent and children provided written assent. Mothers and children each independently completed a body map assessing the location of the child’s pain in separate rooms; each participant was also interviewed separately by a trained research assistant about the child’s pain. The procedure for completing the body maps conformed to that recommended by von Baeyer et al. (5) Participants were first asked by a trained research assistant whether the child had experienced pain in the past month. Those who answered affirmatively were given a body map (described below; see Measures section) and asked to shade all areas where the child had experienced pain in the past month. Subjects were instructed to fill in the entirety of a particular demarcated area, even if the pain site only manifested in a small section of it. Markings were completed using a dark colored pen.

Completed body maps were independently scored by two trained research assistants. Any mark in an area was recorded as positive for pain in that area. Any area that was not marked was recorded as negative for pain in that area. Laterality was ignored (i.e. a mark on the left foot, right foot, or both feet all scored as positive for pain in the feet). For the 1,722 body map areas scored (i.e. 21 body map areas for 82 subjects – 41 mothers and 41 children), there were a total of 28 discrepancies between the two scorers (0.15% error rate). These discrepancies were addressed and resolved by two additional independent researchers through collaborative conferencing on each discrepancy.

Measures

Demographics

A questionnaire designed for the larger laboratory pain study was used to assess the child’s/parent’s ethnicity and race, child’s/parent’s age, and child’s biological sex.

Pain Interview

A series of open-ended questions about the child’s pain were administered using an interview developed for the laboratory pain study. The interview included questions about pain intensity, pain duration, and which pain bothered them the most.

Stage of Puberty

Children completed a self-report measure based on diagrams representing different stages of pubertal development. (13) This instrument consisted of schematic drawings with written descriptions of five stages of secondary sexual characteristics on two separate dimensions based on the Tanner’s Sexual Maturity Scale. (14) Such self-assessment ratings correlate with ratings based on physical examination by physicians. (13, 15) Subjects were given sex-appropriate measures and asked to rate her/himself on each of two dimensions by selecting the drawing closest to his/her stage of development. A single score, ranging from I (prepubertal) to V (adult) level of development, was computed by averaging the two ratings.

Body Map

The body map used in this study is the pain chart recommended by the Childhood Arthritis and Rheumatology Research Alliance (CARRA; CA, USA) for studies of recurrent and chronic pain and first adopted as part of the SUPER-KIDZ pain assessment project. (9) There are 21 distinct body areas identifiable on the map, and both front and back views are represented. As noted above, participants were asked to shade all areas in which the child had experienced pain in the past month.

Statistical Analyses

Kappa coefficients were calculated for each of the 21 body map areas. Kappa statistics are typically used to indicate the amount of agreement between two raters (e.g., clinicians) regarding the presence or absence of a disease state or diagnosis. In the current study, we calculated kappa coefficients to assess the extent of concordance between mothers and children. Kappa (κ) is defined as a measure of “true” agreement. It indicates the proportion of agreement beyond that expected by chance (that is, the achieved agreement beyond chance as a proportion of the possible agreement beyond chance). Kappa values range from −1 to 1, with 1 signifying perfect agreement between raters, 0 signifying agreement no better than that expected by chance, and negative numbers signifying agreement worse than that expected by chance. The latter only rarely occurs in smaller samples as a result of sampling variation.(16, 17)

Kappa values were categorized using the scale Landis and Koch (18) proposed as standards for strength of agreement: <0.000 = poor, 0.001-0.200 = slight, 0.201-0.400 = fair, 0.401-0.600 = moderate, 0.601-0.800 = substantial, and 0.801-1.000 = almost perfect agreement. Although these standards are controversial, they provide a simple heuristic by which the magnitude of kappa may be evaluated. To reduce the possibility of Type 1 error in evaluating the statistical significance of the kappa values, a Bonferroni correction for 21 tests was performed; p < 0.002 was considered significant (i.e., agreement was significantly better than would be expected by chance alone).

As suggested by Sim and Wright (2005), the following statistics were also calculated to assist in interpreting the magnitude of kappa. As noted by these authors, the interpretation of the value of kappa on its own is problematic without taking into account key factors such the prevalence of an attribute (base rates) and inter-rater bias. (19)

Prevalence Index (PI)

This statistic measures the proportions of positive and negative responses from each observer and determines if they are equally probable or if their probabilities vary. (20) If the PI is high (i.e. the prevalence of a positive rating is either very high or very low), chance agreement is also high, and kappa is reduced.(21) Thus, when PI is large, kappa is lower than when PI is low or zero; the effect of PI on kappa is greater for large values of kappa than for small values.(22)

Bias Index (BI)

The extent to which two observers differ in their assessment of the occurrence of a condition is referred to as bias.(22) BI measures the extent to which the raters disagree on the proportion of positive (or negative) cases. As BI increases, the proportion of agreement expected by chance decreases, and the value of kappa increases. (19, 22) Therefore, a large BI has the effect of inflating kappa. (19, 22)

Kappa Max (κ_Max)

This statistic refers to the maximum attainable kappa for each body map area. To calculate κ_Max, the proportions of positive and negative judgments by each rater are taken as fixed, and the distribution of paired ratings is then adjusted to represent the greatest possible agreement. κ_Max serves to gauge the strength of agreement while preserving the proportions of positive ratings given by each rater. In effect, it provides a reference value for kappa that preserves the individual’s overall propensity for a positive endorsement. (19) Greater the discrepancy between kappa and κ_Max indicates poorer agreement than kappa alone would suggest.

Additional exploratory analyses used independent t-tests to test for differences based on the sex of the child, the child’s pubertal status (early vs. late puberty), and the race/ethnicity of the child (Caucasian vs. non-Caucasian and Hispanic vs. non-Hispanic, respectively). The following dependent variables were evaluated: 1) the total number of body areas endorsed by the child; 2) the total number of body areas endorsed by the mother; 3) the total number of mother-child body areas endorsed by both mother and child (the number of mother-child agreements); 4) the total number of mother “positive” and child “negative” occurrences (i.e., the mother indicated that the child experienced pain in a body area(s), but the child did not); 5) the total number of mother “negative” and child “positive” occurrences (i.e., the mother indicated that the child did not experience pain in a body area(s), but the child did). For these exploratory tests, a standard probability level of p < 0.05 (two-tailed) was used to evaluate the results.

Results

Table 2 displays the descriptive data for the 21 body map areas for mothers and children. Table 3 shows the kappa, PI, BI, and κ_Max statistics, indicating the extent of agreement between mothers and children for the 21 body map areas. The mean number of painful areas endorsed by children was 5.51 (SD: 3.802; Range: 1-17) and 4.42 for mothers (SD: 2.585; Range: 1-12). Only 6 children (14.6%) and 2 mothers (4.9%) reported a single body area as being painful.

Table 2.

Descriptive statistics for mother-child pairs

	Mother- Child Disagreement	Mother- Child Both “Positive”	Mother- Child Both “Negative”	Mother “Positive”, Child “Negative”	Child “Positive”, Mother “Negative”
Upper Arm	4 (9.8%)	0 (0.0%)	37 (90.2%)	1 (2.4%)	3 (7.3%)
Forearm	5 (12.2%)	0 (0.0%)	36 (87.8%)	1 (2.4%)	4 (9.8%)
Face	6 (14.6%)	0 (0.0%)	35 (85.4%)	2 (4.9%)	4 (9.8%)
Groin	1 (2.4%)	1 (2.4%)	39 (95.1%)	0 (0.0%)	1 (2.4%)
Wrist	9 (22.0%)	1 (2.4%)	31 (75.6%)	5 (12.2%)	4 (9.8%)
Elbow	4 (9.8%)	2 (4.9%)	35 (85.4%)	2 (4.9%)	2 (4.9%)
Thigh	5 (12.2%)	2 (4.9%)	34 (82.9%)	1 (2.4%)	4 (9.8%)
Calf	9 (22.0%)	2 (4.9%)	30 (73.2%)	4 (9.8%)	5 (12.2%)
Chest	10 (24.4%)	2 (4.9%)	29 (70.7%)	3 (7.3%)	7 (17.1%)
Hand	4 (9.8%)	3 (7.3%)	34 (82.9%)	1 (2.4%)	3 (7.3%)
Hip	6 (14.6%)	4 (9.8%)	31 (75.6%)	1 (2.4%)	5 (12.2%)
Ankle	10 (24.4%)	4 (9.8%)	27 (65.9%)	4 (9.8%)	6 (14.6%)
Upper Back	14 (34.1%)	4 (9.8%)	23 (56.1%)	3 (7.3%)	11 (26.8%)
Foot	9 (22.0%)	6 (14.6%)	26 (63.4%)	2 (4.9%)	7 (17.1%)
Neck	9 (22.0%)	6 (14.6%)	26 (63.4%)	1 (2.4%)	8 (19.5%)
Shoulder	15 (36.6%)	7 (17.1%)	19 (46.3%)	6 (14.6%)	9 (22.0%)
Mid-Back	9 (22.0%)	11 (26.8%)	21 (51.2%)	1 (2.4%)	8 (19.5%)
Knee	11 (26.8%)	11 (26.8%)	19 (46.3%)	3 (7.3%)	8 (19.5%)
Lower Back	8 (19.5%)	12 (29.3%)	21 (51.2%)	2 (4.9%)	6 (14.6%)
Abdomen	6 (14.6%)	16 (39.0%)	19 (46.3%)	5 (12.2%)	1 (2.4%)
Head	11 (26.8%)	20 (48.8%)	10 (24.4%)	7 (17.1%)	4 (9.8%)

Table 3.

κ, κ_Max, prevalence index, and bias index for the 21 body map areas

	κ	κ Max	Prevalence Index (PI)	Bias Index (BI)	κ Significance Level (p)
Forearm	−0.041	0.376	0.878	0.063	0.739
Upper Arm	−0.038	0.788	0.902	0.049	0.776
Wrist	0.056	0.895	0.732	0.024	0.717
Chest	0.153	0.661	0.659	0.098	0.298
Upper Back	0.171	0.526	0.463	0.195	0.215
Calf	0.178	0.909	0.683	0.024	0.252
Shoulder	0.204	0.841	0.293	0.073	0.185
Ankle	0.291	0.858	0.561	0.049	0.060
Face	0.32	0.774	0.756	0.049	0.035
Thigh	0.384	0.631	0.78	0.073	0.008
Head	0.433	0.845	0.244	0.073	0.005
Foot	0.435	0.686	0.488	0.122	0.003
Elbow	0.446	1.00	0.805	0.00	0.004
Knee	0.451	0.75	0.195	0.122	0.003
Neck	0.445*	0.568	0.488	0.171	0.002*
Hip	0.492*	0.661	0.659	0.098	0.001*
Hand	0.547*	0.774	0.756	0.049	<0.001*
Mid-Back	0.547*	0.648	0.244	0.171	<0.001*
Lower Back	0.594*	0.797	0.22	0.098	<0.001*
Groin	0.655*	0.655	0.927	0.024	<0.001*
Abdomen	0.709*	0.806	0.073	0.098	<0.001*

Note:

^*= p < .002

Poor Mother-Child Agreement (κ < 0.000): Upper Arm, Forearm, Wrist

κ values for the upper arm, forearm and wrist were extremely low or negative, indicating poor agreement by the Landis and Koch scale. Moreover, the large discrepancies between κ and κ_Max support the categorization of poor mother-child agreement. Although high PI values have the effect of lowering κ, this effect is greater for large values of κ than for small values. Given the very low values for kappa, the effect of high PI values on agreement for these regions is likely to be minimal. BI values were very low suggesting little impact on κ values. None of the kappa values in this group were statistically significant.

Slight Mother-Child Agreement (0.001 < κ < 0.200): Calf, Chest, Upper Back

κ values for calf, chest and upper back were low, suggesting slight agreement by the Landis and Koch scale. Very large discrepancies between κ and κ_Max support the classification of only slight agreement. Since κ values were low, the likely effects of PI values on κ are small. BI values were low indicating minimal impact on κ. None of the kappa values for these areas were statistically significant.

Fair Mother-Child Agreement (0.201 < κ < 0.400): Shoulder, Ankle, Face, Thigh

Moderately low κ values were found for the shoulder, ankle, face and thigh, suggesting fair agreement by the Landis and Koch scale. Similar to the results for the poor and slight categories above, there were large discrepancies between κ and κ_Max as well as very low BI values across all areas. High PI values for the face and thigh suggest that chance agreement was also high and the kappa values for these regions were somewhat reduced. None of the kappa values in this group were statistically significant.

Moderate Mother-Child Agreement (0.401 < κ < 0.600): Head, Foot, Neck, Elbow, Knee, Hip, Hand, Mid-back, Low-Back

Although the κ values for head, foot, neck, elbow, knee, hip, hand, mid-back and low-back all suggested moderate agreement, examination of the other agreement statistics indicates that the extent of agreement for these regions may be further categorized into two distinct sub-groups.

The subgroup with the poorer agreement statistics included the elbow, head, knee, and foot regions. κ values for this group ranged from 0.43 to 0.45, and all the regions in this subgroup had low BIs (ranging from 0 to 0.12). The difference between κ and κ_Max for the regions in this subgroup ranged from 0.55 to 0.25, indicating a substantial divergence between the obtained agreement and the maximum possible agreement. Although the elbow region had a high PI, this is offset by the fact that it also had the greatest difference between its κ and κ_Max value. None of the kappa values for the areas in this subgroup were statistically significant.

The subgroup with better agreement statistics included the neck, hand, hip, mid- and low-back regions. The κ values for these five areas were similar ranging from 0.45 to 0.59, and the discrepancies between their κ and κ_Max values were lower than that of the above subgroup (ranging from 0.10 to 0.22), indicating better agreement. High PI values for the hip and hand suggested that chance agreement was high and that kappa values for these regions are somewhat reduced. Low BI values across all areas suggest little impact on κ values. All of the κ values for these areas were statistically significant.

Substantial Mother-Child Agreement (0.601 < κ < 0.800): Groin, Abdomen

The groin and abdomen both had high κ values that were statistically significant and reflected substantial agreement according to Landis and Koch’s scale. The discrepancy between the κ and κ_Max values was very small for the abdomen (0.1) and zero for the groin region. The abdomen region had a low PI, indicating low chance agreement resulting in a higher kappa. The groin region had a high PI suggesting that chance agreement was high and that kappa was therefore reduced. Even though there was substantial agreement for the groin region, it should be noted that there were very few cases of groin pain—one case in which both mother and child agreed the pain was present and one case in which the child endorsed pain in this region but the mother denied that the child experienced pain in this region (see Table 2). Low BI values again indicated minimal effects on kappa values.

Overall Mother-Child Agreement: Effects of Gender, Puberty, and Race/Ethnicity

Results of the t-tests indicated that there was a trend for girls to endorse more total body areas than boys (p = 0.054), but there was no difference in the total number of body areas endorsed by mothers of sons versus mothers of daughters. There was significantly more agreement on the total number of body areas between mothers and sons (M = 18.5; SD = 2.0) than between mothers and daughters (M = 16.2; SD = 3.0) (t(39) = 2.62, p < 0.02), but there were no other significant differences based on child sex. There were no significant differences on any of the variables based on pubertal status, race, or ethnicity.

Discussion

We hypothesized that mother-child agreement regarding the location of the child’s pain would be greater for the head and abdominal regions compared to other body regions when assessed using a standardized body map with 21 demarcated areas.(5) Our hypothesis was partially supported in that agreement was greatest for the abdominal region. Although agreement was also high for the groin region, there were only 2 cases endorsing this area as painful. Contrary to our hypothesis, agreement for the head region was not superior to the other body areas. We also hypothesized that the overall magnitude of mother-child agreement would be moderate. This hypothesis was also partly supported in that roughly half of the body map areas showed moderate or better agreement (head, foot, neck, elbow, knee, hip, hand, mid-back, low-back, groin, abdomen), whereas the other half showed poor to fair agreement (forearm, upper arm, wrist, chest, upper back, calf, shoulder, ankle, face, thigh). Our exploratory analyses found that agreement regarding the total number of painful body areas was higher between mothers and sons than between mothers and daughters. However, there were no differences in mother-child agreement based on child race/ethnicity or pubertal status.

The current methodology allowed for a more nuanced examination of mother-child agreement and represents a significant advance over prior work. Since reliance on kappa values alone to assess agreement between observers is controversial,(19) we calculated prevalence index (PI), bias index (BI), and maximum obtainable kappa (κ_Max) to aid the interpretation of kappa. High PI values indicate that the prevalence of a positive rating is either very high or very low, chance agreement is also high and therefore kappa is reduced; the effects of PI on kappa are greater for larger values of kappa.(19, 22) Thus, the high PI values for the groin, hip and hand regions suggest that the obtained kappas for these areas were attenuated. Smaller differences between κ and κ_max indicate better agreement and thus, the large discrepancies between κ and κ_Max values for the head, foot, elbow and knee regions indicated poorer agreement than that suggested by kappa alone. Even though these areas met criteria for moderate agreement, their kappa coefficients were not statistically significant, indicating that agreement was no better than that expected by chance alone. Within the same moderate agreement category, the discrepancy between κ and κ_Max values was much smaller for the neck, hip, hand, mid- and low-back, indicating that these regions evidenced relatively better agreement; in accord, the kappas for these regions were statistically significant. Low BI values across all body regions indicated little bias in the sample (i.e., disagreement on the proportion of positive or negative cases).

Only one prior study by Graumlich et al(8) in pediatric SCD patients has utilized a body map to examine parent-child agreement regarding pain location. (Two other studies using body maps focused on other indices such as pain intensity.(6, 7)) It is difficult to draw comparisons between the Graumlich study and current one for several reasons. Although Graumlich reported parent-child percent agreement for pain location ranged from 58.3% to 95.8%, they did not report parent-child agreement for each individual body area. Moreover, these authors did not detail the method used to assess agreement other than to indicate that parent-child concordance rates were examined by calculating “percent agreement ratios” for all pain sites. As discussed by Chambers et al(1), the use of correlational analysis does not take into account error variance between parents and children and may result in overestimates of these associations. Moreover, the Graumlich sample included patients who presented with a single disease (SCD) whereas our sample consisted of patients with various chronic pain complaints, which likely yielded more heterogeneity in the body locations endorsed as painful and, consequently, a lower likelihood of mother-child agreement for specific body areas. Nevertheless, our findings regarding the overall magnitude of agreement are at least partially consistent with previous research showing moderate mother-child agreement for pain indices such as pain severity and frequency. (1, 6-9)

Epidemiological research indicates that headaches and abdominal pain are the first and second most common pain complaints among children.(11) In line with these findings, the most common presenting pain complaints in the current sample were head pain (58.8%) and abdominal pain (48.8%). Notably, mothers were more likely than children to endorse each of these areas as painful (see Table 2). It is unclear why mother-child agreement was much better for the abdomen than for the head region. During pain interviews conducted after completion of the body maps, we asked mother and children separately which pain bothered the child the most (if more than one location was considered painful). Whereas the number of children (n = 8) and mothers (n = 9) who indicated the abdomen was most bothersome was similar, there was a somewhat greater discrepancy between the number of children (n = 9) and mothers (n = 12) indicating that head pain was the most bothersome. The interviews also ascertained the duration of the most bothersome pain by both child and mother report. The duration of pain was much longer for the abdomen (mean = 6.6 years by child report and 5.3 years by mother report), than for head pain (mean = 2.9 years by child report and 3.0 years by mother report). Thus, it is possible that the longer duration of abdominal pain relative to head pain may explain the better agreement for the former compared to the latter body region.

Our finding of greater agreement between mothers and sons than between mothers and daughters on the total number of painful body areas is new and somewhat unexpected. We previously found stronger associations between mother’s fear of anxiety symptoms and girls’ acute laboratory pain responses compared to boys’. (23) Others have similarly reported that the impact of mothers’ pain-related behavior on children’s laboratory pain responses is more pronounced among girls than boys. (24, 25) Prior work on mother-child concordance for pain using body maps has not examined sex differences in agreement. Our findings of sex-based differences in mother-child agreement cannot be explained by variations in the prevalence of pain as there were no sex differences in the total number of body sites considered painful by either child or parent report. However, girls were substantial older than boys (mean age boys = 13.4 vs. girls = 15.3), and a possible explanation is that mother-child agreement may be generally higher among younger compared to older children.

The current findings have clear implications for the assessment and treatment of chronic pain in children. In clinical practice, the use of interviews to assess pain is a common approach with emphasis typically being placed on the chief presenting complaint.(5) The utility of body maps to assess pain location is underscored by our findings that nearly all of the patients in our sample reported more than one pain location. Only 2 mothers (4.9%) and 6 children (14.6%) endorsed a single area as being painful. Our results support the current recognition that the optimal route to acquiring an accurate portrayal of a child’s pain experience is to obtain both parent and child report. However, given the observed discrepancy between mother and child reports even for common pain locations such as the head, additional work is needed to develop better methods to integrate child and parent information.

Limitations of this study should be mentioned. First, we only included participants who indicated that the child had experienced pain in the past month, and thus 5 of the 46 enrolled families did not complete body maps. Given the fluctuating and cyclical nature of chronic/recurrent pain, asking about usual pain location(s) may have yielded different findings. However, the one month time frame was used to temper the possible effects of memory bias over an extended time period. Second, we were not able to assess father-child agreement due to the small number of enrolled fathers. Further work may examine father-child agreement as well as the extent of agreement between mothers and fathers regarding their child’s pain. Third, our results may not generalize to other pediatric samples. As noted previously, the extent of mother-child agreement may have been diluted by the large number of different pain complaints represented. It is possible that higher levels of agreement may be obtained in patient populations with more focal pain problems or specific disease states.

In sum, in this sample of pediatric chronic pain patients, agreement between mothers and children regarding the location of the child’s pain was greatest for the abdomen, with moderate or better agreement for approximately half of the 21 areas assessed using a standardized body map. Future research may assess whether factors such as mother and/or child psychological characteristics (e.g. anxiety) may contribute to differences in mother-child agreement, as well as further examining the potential impact of pain duration, child development and sex-based effects. Additional work may test whether the current results are applicable to other patient populations as well as healthy populations. Such studies may enhance efforts to understand the underlying reasons for parent-child disagreement regarding the child’s pain.