Studies evaluating the efficacy of treatments for recurrent abdominal pain (RAP) in children have used a wide range of methods, causing difficulty in the comparison of results. An expert panel on functional gastrointestinal disorders recently made recommendations regarding the standardization of study methods for childhood RAP, but many of their recommendations remain untested or lack supportive evidence.
During completion of a pilot study and randomized controlled trial for childhood RAP, baseline data were collected regarding the child and parent reports of abdominal pain frequency and intensity, type of abdominal pain, missed activities due to abdominal pain, psychological factors for the parent and child, parenting styles, and referral source (pediatric gastroenterologist vs general pediatrician).
Children and parent pain reports showed good agreement in children younger than 13 years (weighted κ, 0.77; 95% confidence interval [CI], 0.71-0.84), but only marginal agreement in children 13 years or older (weighted κ, 0.37; 95% CI, 0.30-0.45). We found no significant differences in pain characteristics or psychological factors between children referred by pediatric gastroenterologists in a tertiary care center and those referred by community-based primary care pediatricians. However, children with symptoms consistent with nonspecific functional abdominal pain were reported by their parents to have less frequent pain (P= .003) and fewer missed activities (P= .003) than children with symptoms of irritable bowel syndrome or functional dyspepsia.
Subjects referred by gastroenterologists and general pediatricians were similar, but the subtype of functional gastrointestinal disorder might be an important baseline characteristic of subjects in future RAP studies. We suggest that future interventional studies of childhood RAP measure 2 outcomes with pain reports obtained directly from children. Any child with fewer days of pain and missed activities due to pain after therapy would be considered improved, and those with no missed activities and 4 or fewer days of pain per month at follow-up would be considered healed.
ALTHOUGH PHYSICIANS frequently evaluate recurrent abdominal pain (RAP) in children, no sound evidence exists to guide the therapeutic approach to these patients. Recent systematic reviews of the literature pertaining to therapies for RAP found methodological concerns with many of the published investigations.1- 3 Studies have recruited subjects using multiple strategies from various referral sources and measured many different outcomes at various times before and after interventions. Such a lack of standardization has made it difficult to compare the efficacy of interventions between studies. In addition, although psychological factors have been shown to vary among healthy control subjects and children with RAP and their parents, these factors have rarely been measured or considered as potential confounders.4- 6 Although some interventional studies have excluded children with psychiatric illness, they have not described the methods used to accurately diagnose such disorders.7,8
Because of these methodological variances within the medical literature, a group of international experts in functional gastrointestinal disorders met for a second time in Rome, Italy, and recently published their recommendations for the standardization of research methodology for interventional studies for RAP.9 The Rome II consensus group recommended that pain evaluations should be based on patient reports rather than those of clinicians or researchers; that efficacy evaluation should be based on the percentage of children meeting a predefined clinical outcome rather than the statistical significance of differences between groups or periods of time; and that psychological measures should be performed at baseline for possible use as covariates in the analysis. In addition to these recommendations, others have suggested that the primary outcome should measure the amount of pain and the level of disability,10 because the original definition of RAP by Apley and Naish11 required at least 3 bouts of pain severe enough to affect the child's normal activities within the preceding 3 months.
Fertile research areas were also identified by the Rome II group where there was a lack of sufficient evidence available to allow for consensus. It is not known if disease severity or the presence of comorbidities varies depending on whether subjects are recruited from primary or tertiary care settings.9 Regarding studies in children, it is unclear whether pain reports are best obtained from the children or their parents. Although there was agreement that the primary efficacy evaluation should be based on the percentage of subjects meeting a predetermined clinical outcome, there was no agreement as to what that outcome measurement should be.9 It was also suggested that children with RAP be subdivided into subgroups of abdominal pain, such as functional dyspepsia, irritable bowel syndrome, functional abdominal pain, abdominal migraine, and aerophagia.12 However, fewer data are available to suggest whether children with these different subtypes actually vary in their severity of pain, psychological factors, referral source, or likelihood of achieving predefined clinical outcomes.
We analyzed baseline data after the completion of a pilot study and a randomized controlled trial of the therapeutic effect of guided imagery for childhood RAP in an effort to clarify some of these methodological issues. This report seeks to test the following 3 hypotheses and explore a related research question:
Children's self-reports of abdominal pain will be greater than their parents' report. In addition, the agreement of abdominal pain reports between children and their parents will vary by the child's age.
Children referred to these studies from pediatric gastroenterologists in a tertiary care center will report greater abdominal pain, have more psychological comorbidities, and suffer more from irritable bowel syndrome or functional dyspepsia than those referred from community primary care pediatricians.
Children with symptoms of irritable bowel syndrome or functional dyspepsia will have greater pain and more frequent psychological comorbidities than those with nonspecific functional abdominal pain.
What, if any, definition(s) of clinical responder can be proposed for future research on childhood RAP?
From October 14, 1998, through June 30, 2002, 46 children were recruited using identical methods for a pilot study (n = 15) and then a randomized controlled trial (n = 31) investigating the use of guided imagery for the treatment of RAP. Recruitment letters explaining the study and an enrollment form were sent to all licensed pediatricians and pediatric gastroenterologists in Pima County, Arizona. Enrollment criteria for the studies were children aged 5 to 18 years, with 3 or more episodes of abdominal pain during the previous 3 months that interrupted normal daily activities and who did not have evidence of an underlying disease to explain their symptoms. All children underwent a complete history and physical examination performed by their pediatrician or pediatric gastroenterologist and a minimal laboratory evaluation, including a complete blood cell count, measurement of erythrocyte sedimentation rate, and urinalysis. All other diagnostic tests were performed at the discretion of the treating physician. Children were recruited from pediatric gastroenterologists within the Department of Pediatrics, University of Arizona, Tucson, and general pediatricians throughout the greater Tucson metropolitan area. The studies were approved by the institutional review board of the University of Arizona. Depending on the child's age, all children signed an assent form, in which case 1 parent also signed an informed consent, or co-signed an informed consent form with 1 of their parents.
After enrollment into 1 of the studies, children and parents were required to complete pain diaries for a minimum of 2 weeks before beginning the intervention. Children were instructed to complete their diary each day using the Faces Pain Scale13 to rate the intensity of any abdominal pain on a scale from 0 to 6 at 7 AM, 2 PM, and 6 PM. Parents were instructed to use the same scale to rate the intensity of the worst stomach pain they had observed in their child during the day and to document whether the child was unable to participate in normal activities (ie, school, friends, and others) that day because of stomach pain. Parents were instructed to grade the child's pain through the use of verbal and behavioral cues (eg, crying, grimacing, and sighing) at 7 AM, 2 PM, and 6 PM on each day.
At enrollment, all children completed a bowel-symptom questionnaire, which was adapted for children from the survey instrument validated by Talley et al.14 Parents assisted those children too young to complete the questionnaire independently. Using this survey, children were classified by the hierarchal approach of the Rome II criteria12 into 1 of the following 3 diagnostic categories: irritable bowel syndrome, functional dyspepsia, or functional abdominal pain. Children classified as having irritable bowel syndrome answered yes to 2 or more of the following questions: (1) "Is the pain often made better by having a bowel movement?" (2) "Do you often have more bowel movements when this pain begins?" and (3) "Do you often have looser bowel movements (stools) when this pain begins?" All children not classified as having irritable bowel syndrome and who reported "pain as usually occurring above the navel that is in the upper belly" were classified as having functional dyspepsia. In addition, children who reported pain as occurring in both the upper and lower belly and answered yes to 1 or more of the following questions were also classified as having functional dyspepsia: (1) "Is this pain often made better (relieved) by burping (bringing up air through the mouth)?" (2) "Do you often feel bloated and actually see your belly swell up?" and (3) "How many times have you actually thrown up (vomited) in the last year?" (Children answering "about once a week" or more were considered to have the condition.) All children not classified as having irritable bowel syndrome or functional dyspepsia were classified as having functional abdominal pain.
On enrollment into 1 of the studies, children and parents completed a number of standard and valid psychological questionnaires, including the Multidimensional Anxiety Scale for Children, Childhood Depression Inventory, and the Symptom Checklist-90.15- 17 The Multidimensional Anxiety Scale for Children was completed by the child and assesses anxiety symptoms across clinically significant symptom domains. For the purposes of this study, we used the following subscales (the range of possible scores are given parenthetically): somatic symptoms (0-18), anxious coping (0-15), perfectionism (0-12), and separation/panic (0-27). The Childhood Depression Inventory was completed by the child and assesses the degree of depression in children. It has a possible score of 0 to 54. The Symptom Checklist-90 was completed by the parent as a self-assessment and is a brief, multidimensional inventory designed to screen for a broad range of psychological problems and symptoms of psychopathology. For the purposes of this study, we used the following symptom subscales (the range of possible scores are given parenthetically): somatization (0-48), depression (0-52), and anxiety (0-40).
Because review of previous studies demonstrated differences between children with RAP and healthy controls without abdominal pain, the following child psychological factors were included in this analysis4: anxiety, depression, somatization, separation anxiety, and perfectionism. Parent psychological factors regarding the parent attending the initial enrollment session included anxiety, depression, and somatization. Parenting styles for mothers and fathers were also assessed using the Parent Bonding Instrument.18 The Parent Bonding Instrument is used to measure dimensions of parent-child bonding and has scores for caring (possible score, 0-36) and overprotectiveness (possible score, 0-39).
Analyses were completed using appropriate parametric tests after transformation of nonnormally distributed continuous variables when appropriate. An α of less than .05 was considered statistically significant. The agreement between parent and child reports of abdominal pain was tested with the κ statistic for dichotomous categories and a weighted κ for pain intensity scores with 7 possible scores (0-6). Because families were asked to continue to complete pain diaries until they received the intervention, the number of actual days of completed diaries varied among subjects, ranging from 14 to 60 days. Therefore, the number of days of pain and number of days with missed activities were standardized to 30 days, allowing the presentation of the number of days during the previous month.
A total of 46 children were enrolled in the 2 studies. However, 4 subjects withdrew from the study after initially consenting owing to family illness (n = 2) or unwillingness of the child to continue participation (n = 2). In addition, data were not complete for 3 children who did not complete the baseline psychological questionnaires or the bowel-symptom questionnaire. The characteristics of the final study population are summarized in Table 1.
We first hypothesized that children's self-reports of abdominal pain will be greater than their parents' report of their pain, and that the agreement of abdominal pain reports between children and their parents will vary by the child's age. The reported number of days with pain and the intensity of pain varied between children and their parents (P<.001 and P = .04, respectively), with children reporting many more days of more intense pain than their parents (Table 1). Regardless of whether the parents used verbal complaints of pain or behavioral cues such as crying, grimacing, or sighing to judge their child's pain, the reports were virtually identical. When matching daily pain reports from the child and parent were analyzed dichotomously (pain vs no pain), the agreement was good (κ, 0.75; 95% confidence interval [CI], 0.69-0.81). However, these results varied significantly by the child's age. The agreement was excellent in children younger than 10 years (n = 10) (κ, 0.85; 95% CI, 0.76-0.95) and in children 10 years or older but younger than 13 years (n = 19) (κ, 0.87; 95% CI, 0.77-0.97), but only marginal in children 13 years and older (n = 13) (κ, 0.28; 95% CI, 0.16-0.40). A similar effect was noted when the child and parent pain reports were analyzed using the daily maximum pain intensity reports, which ranged from 0 to 6. Overall, agreement between parent and child reports were good (weighted κ, 0.68; 95% CI, 0.64-0.72). However, in children younger than 13 years, the agreement was excellent (weighted κ, 0.77; 95% CI, 0.71-0.84), whereas it was only marginal in children 13 years or older (weighted κ, 0.37; 95% CI, 0.30-0.45).
The maximum intensity of abdominal pain reported by the children and parents during each day was compared with the likelihood of missing an activity the same day (Figure 1). Although the risk for missing a normal activity was very low for child-reported pain intensities of 0 or 1, it increased with a rating of 2 (relative risk [RR], 11.5; 95% CI, 6.5-20.5) to 4 (RR, 15.9; 95% CI, 9.2-27.7) and again increased with ratings of 5 (RR, 20.6; 95% CI, 11.3-37.6) or 6 (RR, 19.3; 95% CI, 10.1-36.6). A similar but more marked trend was seen with parent ratings of intensity, with an increased risk for missing an activity seen with a rating of 2 (RR, 35.0; 95% CI, 11.0-111.7) to 4 (RR, 48.1; 95% CI, 14.9-155.0) and with ratings of 5 (RR, 87.1; 95% CI, 27.1-280.0) to 6 (RR, 104.2; 95% CI, 33.0-329.7) compared with ratings of 0. Nevertheless, even on days when children reported a pain episode of the maximal intensity of 6, they missed a normal activity only 26.7% of the time; when parents reported their child as having an episode of the maximal intensity, the child missed a normal activity 56.1% of the time.
Risk for a missed activity by the maximum daily intensity of abdominal pain episodes. Asterisk indicates differences were significant (P<.001, χ2 test for trend).
Our second hypothesis was that children referred to these studies from pediatric gastroenterologists in a tertiary care center will report greater abdominal pain, have more psychological comorbidities, and suffer more from irritable bowel syndrome or functional dyspepsia than those referred from community primary care pediatricians. There were no apparent differences between children and parents by their referral source except for the parents' ethnicity (P<.001) (Table 2). The more diverse ethnic breakdown of the patients referred by community primary care practices is more reflective of the entire referral community in Tucson, rather than that seen by the tertiary care center. Children referred by pediatric gastroenterologists had a higher anxiety score of borderline significance than those referred by general pediatricians (P = .08). However, the statistical significance of this finding must be interpreted cautiously because of the large number of comparisons completed in Table 2.
Third, we hypothesized that children with symptoms of irritable bowel syndrome or functional dyspepsia will have greater pain and more frequent psychological comorbidities than those with nonspecific functional abdominal pain. We found few differences among children presenting with symptoms of irritable bowel syndrome, functional dyspepsia, or functional abdominal pain (Table 3). However, children with functional abdominal pain had fewer days of pain by parental report using behavioral and verbal cues (P = .003 for both) and fewer missed activities (P = .003) than children with irritable bowel syndrome and functional dyspepsia. Individual comparisons between specific abdominal pain subtypes were completed using the Bonferroni multiple-comparison test. These findings were confirmed, as children presenting with symptoms consistent with functional abdominal pain had fewer days of abdominal pain by parent report using verbal cues than children with functional dyspepsia (P = .004) or irritable bowel syndrome (P = .03). The differences were similar when parents reported the child's pain using behavioral cues; ie, children presenting with symptoms consistent with functional abdominal pain had fewer days of abdominal pain than children with functional dyspepsia (P = .005) or irritable bowel syndrome (P = .02). Children with functional abdominal pain were also reported to have fewer days with missed activities due to abdominal pain than children with functional dyspepsia (P = .004) or irritable bowel syndrome (P = .02).
What, if any, definition of clinical responder can be proposed for future research on childhood RAP? Since a suitable outcome measure should include information regarding the frequency of pain and the associated disability, further analyses were completed to evaluate the likelihood of such measures at entry into these studies. During the baseline period, 32 children (76%) missed 1 or more activity due to abdominal pain. Overall, 1 enrollee (<5%) reported fewer than 4 days of pain during the previous month. There were no children enrolled into the study with 4 or fewer days of pain who had not missed an activity due to abdominal pain during the preceding month.
The primary finding of this study was that agreement between child and parent reports of childhood abdominal pain was inversely related to the age of the child. In addition, we found little difference between children referred to our studies by tertiary or primary care physicians. In contrast, children referred with symptoms consistent with nonspecific functional abdominal pain were reported by parents to have fewer days of pain and fewer missed activities than children with symptoms of irritable bowel syndrome or functional dyspepsia.
Most investigations of childhood RAP report both child and parent measures of pain episodes. Although some have recommended using reports from both children and parents, it adds to the multiple outcomes already reported in studies of childhood RAP.10 For example, 1 systematic review of the literature found that 39 treatment outcomes were reported in 10 randomized controlled trials of childhood RAP.1 In addition, those who support using reports from both sources have not discussed how discordant reports from children and parents should be analyzed. Although it seems intuitive that the Rome II consensus recommendation that self-reports of pain obtained directly from subjects be extended to adolescents, it is not clear whether this should be done for younger children. Our findings suggest that pain outcomes may be reported directly from subjects as young as 5 years of age, in conjunction with their parents. Parents' and children's reports showed good agreement in younger children, an age at which researchers might be concerned about the accuracy of children's reports. However, the greatest disagreement between parent and child reports occurred during adolescence, when reports directly from the patient would be most consistent with the Rome II recommendations. Nevertheless, inferences about accuracy based on the κ statistic are difficult to draw, especially since errors between the child and parent might not have been independent, and therefore the most important conclusion from our data is that reports over time should be obtained from the same person.19 Our findings are consistent with other studies that also found that parent ratings tend to underestimate their child's level of pain.20,21 However, unlike these previous studies, which had parents and children complete their assessments independently, we encouraged parents of younger children to help their child in the completion of the diaries. Therefore, it is unclear whether the relationship between the child and parent reports of pain and age is explained by parents of younger children being more in touch with their child's daily events or simply by their presence with their child at the time of diary completion. Although we gathered information about specific episodes of pain, we preferred to report the number of days with pain, because it is difficult to compare the relative clinical importance of a daily report of, eg, 2 short episodes of abdominal pain with 1 long episode.
The consensus of the Rome II recommendations was that the primary efficacy evaluation should be based on the percentage of children meeting a predetermined clinical outcome. A recent review of childhood RAP therapies found that although virtually all of the 10 studies reviewed used the number of pain episodes as a primary outcome, only 1 measured missed activities due to pain.1 In addition, this study only measured missed activities due to school absences, which are insufficient for use much of the time.22 Instead, most studies reported the pain severity or intensity. Although our findings showed a clear association between pain intensity and risk for missing a normal activity, only 27% of the most painful episodes reported by the children and 56% of the most painful episodes reported by the parents resulted in a missed activity. Therefore, the reliance on the number and intensity of abdominal pain episodes is clearly not consistent with the original description by Apley and Naish11 of childhood RAP that has been used for more than 40 years. We suggest the use of 2 outcomes that will be practical, clinically important, and consistent with the initial description of RAP11 and will allow comparisons across interventional studies. The first outcome would define a child as improved if he or she had fewer days of abdominal pain and fewer missed activities after the intervention than at baseline. This would seem to be a pragmatic approach to detecting any minimal, clinically significant, therapeutic effect. The second outcome would define a child as healed if they experienced no missed activities and 4 or fewer days of abdominal pain within a given month during or after the intervention. This outcome can easily be measured within the study design supported by the Rome II panel, which recommended a minimal treatment duration of 8 to 12 weeks with follow-up at 6 to 12 months for those therapies found to be successful in the short term.9 The numbers of days with pain and missed activities are important baseline descriptive measures of study populations. Although the strict use of 4 or fewer days of abdominal pain per month is open to debate, the creation of new scores from categories with arbitrary values, which are not previously validated, is unnecessary, confusing, and difficult to interpret clinically.8,23
It was somewhat surprising that the children referred by pediatric gastroenterologists in a tertiary care center did not differ significantly from those referred by community primary care pediatricians with respect to the amount and type of abdominal pain, psychological factors, their parents' psychological factors, and parenting styles. All study subjects were referred by physicians only, and no recruitment messages were directed to the community as has been done in some previous studies.22 It is unknown whether subjects recruited with such strategies and not undergoing evaluation by pediatricians would vary from our 2 study populations.
The Rome II panel also recommended classifying subtypes of RAP.12 At the same time, they recommended not to subclassify subjects during interventional studies because symptoms can fluctuate within individuals over time. Although the clinical significance of the recommended classification scheme has not been validated, we believed it was important to evaluate for potential differences between these subtypes. The primary difference noted was that parents perceived children with functional abdominal pain to have less pain and fewer missed activities than children with irritable bowel syndrome or functional dyspepsia. The children's reports of pain did not differ between these groups; therefore, the significance of this finding is unclear. Although the use of the bowel-symptom questionnaire has not been validated as a tool to classify these subtypes of RAP, our preliminary findings suggest that this might be important for the baseline description of future populations of research subjects.
Although we have attempted to add clarity to some of the conundrums faced by researchers in this field, there are 2 possibly important weaknesses of this study. The moderate sample size limited the power of the subgroup analyses, ie, between RAP subtypes. Although we sought referrals from all community pediatricians, we were unable to ensure communitywide sampling. We do not know how many families were offered enrollment into the study and to what extent enrollment into a trial of relaxation techniques might have led to adverse selection of our sample.
Since the initial description of RAP in children by Apley and Naish,11 our understanding of functional bowel disorders has grown substantially. Recent findings have suggested that although many children with RAP respond well to reassurance, a significant number may continue to suffer into adulthood.24 The paucity of efficacious evidence-based treatments noted in recent systematic reviews of RAP therapies emphasizes the need for continued research in this area.1- 3 However, as noted in those same reviews, the methods must be standardized and allow for comparisons across trials. This is especially critical because the biopsychosocial model has been adopted to guide the therapeutic approach to children with these disorders, and it is therefore likely that effective therapies will need to be explored from various disciplines, such as psychology and nutrition, as well as pharmacy.25,26 This integrated approach to therapy will necessitate a common evidence-based methodology across a number of disciplines if we are to alter the course of these potentially long-term disabling illnesses.
Corresponding author: Thomas M. Ball, MD, MPH, Department of Pediatrics, University of Arizona College of Medicine, 1501 N Campbell Ave, PO Box 245073, Tucson, AZ 85724-5073 (e-mail: firstname.lastname@example.org).
Accepted for publication March 6, 2003.
This study was supported by National Center for Complementary and Alternative Medicine grant 5 P50-AT00008 from the National Institutes of Health, Bethesda, Md.
We would like to acknowledge the participation of the study families and the work of the research associates, Patricia K. Brown and Kathryn S. Chavez, and of Russell P. Roberge for his assistance with data analysis. In addition, we acknowledge Sari Acra, MD, and Patrick Burke, MD, PhD, for their initiation of this project.
Recent reviews of therapies for RAP in children have reported a paucity of methodologically sound studies. In addition, an international panel of experts has recently published their recommendations in an effort to improve the standardization of research in this field. The present study reports findings pertinent to a number of unresolved issues in the study of treatments for childhood RAP. We found that we were able to obtain accurate abdominal pain reports from the children, that measuring the intensity of abdominal pain episodes is not a sufficient substitute for the measurement of missed activities due to pain, and that children referred for enrollment to an investigation by pediatric gastroenterologists in a tertiary care center did not differ from those referred by community primary care pediatricians. However, children with functional abdominal pain were reported by their parents to have fewer episodes of pain and missed activities than children with irritable bowel syndrome or functional dyspepsia.
Ball TM, Weydert JA. Methodological Challenges to Treatment Trials for Recurrent Abdominal Pain in Children. Arch Pediatr Adolesc Med. 2003;157(11):1121-1127. doi:10.1001/archpedi.157.11.1121