Copyright 2006 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2006
Among adults with functional gastrointestinal disorders, psychological distress influences who consults a physician, but little is known about predictors of consultation when the patient is a child.
To determine the relative contributions of psychological symptoms of the mother, psychological symptoms of the child, severity of child abdominal pain, and family stress to consultation.
Health maintenance organization.
Two hundred seventy-five mothers of 334 children who had abdominal pain in the past 2 weeks, as per child self-report.
Main Outcome Measures
Mothers completed questionnaires about themselves (Symptom Checklist 90–Revised) and their children (school absences, medication use, and the Child Behavior Checklist). Children completed the Pain Beliefs Questionnaire to assess perceived pain severity.
Thirty-nine children had been taken to the clinic for abdominal pain symptoms at least once in the past 3 months (consulters), whereas 295 were nonconsulters. Logistic regression analyses revealed that both the child's self-report of perceived pain severity (P<.001) and maternal psychological symptoms (P = .006) predicted consultation. Although children who visited physicians had significantly more psychological symptoms, this was not a significant predictor of consultation after adjusting for maternal psychological symptoms. Family stress did not predict consultation.
The decision to take a child to the clinic for abdominal pain is best predicted by maternal psychological distress and the child's perceived pain severity.
Among adults with unexplained (functional) abdominal pain, it has been reported that consulting behavior is influenced by the severity of abdominal pain,1 the presence of distressing psychological symptoms,2- 4 and exposure to stressful life events.5 A variety of studies have investigated the determinants of physician consultation when the patient is a child. In some of these studies, predictors of physician consultation included child demographic characteristics (younger age or lower educational level), pain intensity, pain frequency, and school absenteeism.6,7 Because the actual decision to consult is made by parents, a related body of research has focused on the psychological traits of the parents instead of, or in addition to, the sociodemographic characteristics, traits, and symptoms of the child. This research has suggested that child health care utilization is related to parental anxiety, somatization, or maternal health care utilization.8- 15 However, limitations exist in all of these studies, including (1) relatively small sample sizes; (2) reliance on volunteers from the community rather than randomly selected participants, which could lead to biased sampling; (3) parent rather than child report of children's symptoms, potentially confounding parental perceptions and psychological distress; and (4) failure to assess the relative contribution of parental, family, and child distress. Furthermore, most of this research has focused on general consulting rather than abdominal pain in children, a common reason for seeking health care for children and the most frequent reason for visits to pediatric gastroenterologists.16 Therefore, the aim of this study was to assess the relative contribution of the following variables to physician consultation for a child's abdominal pain: severity of the child's self-reported abdominal pain, the child's psychological symptoms, the mother's psychological symptoms, and family stress.
This study was part of a larger study17 looking at children of parents who either did or did not have irritable bowel syndrome (IBS). Families with IBS were identified by reviewing the automated database of a large health maintenance organization to identify all women who had received a diagnosis of IBS or abdominal pain at a clinic visit during the preceding 2 years. Non-IBS families were selected from the same database by identifying women who did not have a clinic visit during the preceding 2 years for IBS, abdominal pain, constipation, or diarrhea. Additionally, only families who had 1 or more children between the ages of 8 and 15 years were selected from the database. All potential study subjects were further screened by trained telephone interviewers to determine whether the mother met Rome II criteria for IBS diagnosis.18 Other inclusion criteria ascertained in the telephone interview were that (1) the mother was the child's legal guardian and the child had lived with her for at least half of the time during the preceding 2 years, (2) no interviewed child had a developmental disability requiring full-time special education, and (3) no family member had a diagnosis of ulcerative colitis or Crohn disease.
This process resulted in 1447 families being initially selected from the database. Of these, 789 (54.5%) agreed to be further screened by telephone, and 857 (59.2%) of the initial 1447 families were then determined to be eligible. The participation rate was 97% among those determined to be eligible during the telephone interview. This resulted in a final sample of 334 children and 275 mothers.
The study was reviewed and approved by the human subjects review boards for the Group Health Cooperative, the University of Washington, and the University of North Carolina at Chapel Hill (the data management site).
All assessments were conducted by means of a single interview in the home. Child assessments were administered orally, face to face. These assessments were conducted by graduate-level mental health professionals who received extensive training in standardized questionnaire administration. Mothers independently completed a battery of pencil-and-paper questionnaires (in the home while the child was being interviewed). Measures are described as follows.
This symptom was assessed with an item from the Children's Somatization Inventory,19,20 a measure of children's nonspecific somatic symptoms. Ratings were made on a 0-4 scale: not at all (0), a little (1), some (2), a lot (3) and a whole lot (4) for the single item of “pain in your stomach or abdomen.” As described previously, the sample was limited to those children who provided a rating of 1 or higher; that is, those who experienced at least “a little” pain in the past 2 weeks. Stated conversely, children with no pain were excluded. The Children's Somatization Inventory has excellent reliability and validity.
The Pain Beliefs Questionnaire21 contains 32 statements about stomachaches. Sample items are “My stomachaches mean I have a serious illness,” and “When I have a bad stomachache, it usually lasts a long time.” Children were asked to respond to each statement by rating how true it was for their stomachaches (in general). There were 5 possible answers ranging from “not at all true” to “very true.” The 32 questions are scored for 5 different subscales that describe different dimensions of symptom severity (eg, condition seriousness, condition duration, condition frequency, episode intensity, and episode duration) and 2 subscales that describe how the child copes (eg, problem-focused coping potential and emotion-focused coping potential). The 5 symptom severity dimensions can be summed together to form a total score; this summary index was used as the measure of severity of the child's abdominal pain. The Pain Beliefs Questionnaire was found to have adequate validity and reliability.21
Mothers were asked to complete the anxiety, depression, and somatization subscales of the Symptom Checklist 90–Revised (a total of 35 items).22 These 3 subscales were summed together to create a single index of overall psychological distress, as is recommended in the Brief Symptom Inventory-18, which is the short form of the Symptom Checklist 90–Revised. The Global Symptom Index of the Brief Symptom Inventory-18, computed in this way, was shown to be valid and reliable.22
Mothers completed the Child Behavior Checklist for children aged 4 to 18 years.23,24 This is a valid, reliable, and frequently used scale to assess behavioral problems in children. The Child Behavior Checklist is conventionally scored for 2 broad dimensions of behavioral problems: internalizing and externalizing. The Internalizing scale is the sum of 3 subscales (withdrawn, somatic complaints, and anxious/depressed) that correspond to the dimensions measured in the mothers. For the purposes of this study, the Internalizing scale was used as an index of psychological distress in the child, and this score was converted to a sex-appropriate T score to facilitate pooling of boys and girls.
The Family Inventory of Life Events is a parental report form that assesses potentially stressful negative life events experienced by family members in the previous year. McCubbin et al25 report that internal consistency reliability is 0.81 and 4-week test-retest reliability is 0.80. Agreement between family members on the occurrence of events supports the validity of the Family Inventory of Life Events.
Mothers were asked, “In the past 3 months, how many times did your child go to the doctor for abdominal pain?” Choices were “none,” “1-3 days,” “4-6 days,” “7-9 days,” and “10 or more days.” Children were classified as consulting a physician for abdominal pain if the response was anything other than “none.”
Unpaired t tests and χ2 tests were used to compare consulters (those taken to the medical clinic for gastrointestinal symptoms at least once in the past 3 months) and nonconsulters (those not taken to the medical clinic for gastrointestinal symptoms in the past 3 months) with respect to demographic, clinical, and psychosocial variables. Bivariate correlations were used to examine relationships among the predictor variables. Logistic regression was used to determine which of 4 variables (abdominal pain symptom severity, maternal psychological distress, child psychological distress, and family stress) contributed significantly to consulting behavior. This was a 2-step process. Each variable was first analyzed via univariate regression, with just that predictor entered into the model. Those variables that significantly predicted consultation in univariate regressions were then analyzed via stepwise regression. For all analyses, α = .05.
Demographic and clinical characteristics of the sample are listed in Table 1. Mothers were, on average, 42 years old. Most mothers were white (82%), as were their children (81%). Just under one half of the mothers had received at least a 4-year college degree (45%, with 27% having attended graduate school). Employment status was mixed, with 19% unemployed, 25% employed part-time, and 55% employed full-time. Finally, 80% of mothers were married or cohabiting with a partner.
The mean child age was 12.1 years; sex distribution was fairly even (55% female). Children varied in their reports of abdominal pain during the past 2 weeks: 207 (62%) reported “a little pain”; 80 (24%), “some pain”; 41 (12%), “a lot of pain”; and just 6 (2%), “a whole lot of pain.”
Table 2 compares clinical and psychosocial characteristics of consulters and nonconsulters. Children who were brought to the clinic for stomachaches perceived greater pain severity than children with stomachaches who were not brought to the clinic, as per the Pain Beliefs Questionnaire (P = .003). They were also more likely to miss school and to take medicine for abdominal pain (P<.001 for both). Mothers of consulters reported greater psychological distress than did mothers of nonconsulters (P = .002). Similarly, consulting children were rated higher on the Child Behavior Checklist (our measure of child psychological distress) by their parents as compared with nonconsulting children (P = .002). Consulters and nonconsulters did not differ with respect to family stress (P = .14).
Table 3 displays Pearson correlations among the predictor variables. Significant positive correlations were found among maternal psychological distress, child psychological distress, and family stress (all rated by the mother). In addition, perceived pain severity (as rated by the child) was significantly related to child psychological distress (as rated by the mother). Perceived pain severity was unrelated to both maternal psychological distress and family stress.
Results of the logistic regressions are displayed in Table 4. In univariate regressions, maternal psychological symptoms, child psychological symptoms, and perceived pain severity all predicted consultation. However, in stepwise regression analysis, only maternal psychological symptoms and perceived pain severity were found to be significant predictors (P = .006 and P < .001, respectively).
To further support our findings, we conducted the same regressions using a slightly different sample based on maternal vs child ratings of pain severity in the past 2 weeks (again, excluding children with no pain). This sample included 323 children: 58 consulters and 265 nonconsulters. Results were the same. Only maternal psychological symptoms and perceived pain severity were entered into the final model (β = .28; odds ratio = 1.33; P = .001 for the former and β = .04; odds ratio = 1.04; P = .02 for the latter).
A principal finding of this study is that a child's visit to the clinic for abdominal pain is best predicted by symptoms of psychological distress in the mother and severity of abdominal pain as reported by the child, rather than psychological distress in the child or family stress. The fact that most of the children in our sample were only in “a little” pain adds interest to this finding. It suggests that these parents are seeking health care even when their children's complaints are minor.
The correlations among the predictor variables are also of interest. Mothers' psychological distress did not correlate with child-reported pain severity, yet both were related to utilization. This lends support to the regression findings; both factors independently predict utilization.
As stated earlier, previous studies have found that some parental psychological factors as well as parental health care utilization are related to children's health care utilization. This study has demonstrated that these findings were replicated in the area of abdominal pain, a chronic and common childhood complaint. Furthermore, more recently, Janicke and Finney26 refined the investigations of general children's health care utilization by determining that parental self-efficacy, in interaction with parental stress, contributed to children's utilization. They, along with others, have recommended that interventions targeted to reducing family stress were therefore a reasonable direction to reduce health care utilization. Our study suggests that these efforts might be more efficient when directed at parents, rather than efforts to reduce child distress or overall family stress factors, since neither child distress nor family stress appeared to be significant determinants of health care utilization, at least within this population.
One limitation of our study is that whereas other research has demonstrated that the father may have a significant influence on the illness behavior of the child,8,27 the current study was limited by our ability to obtain data only from mothers. It is possible that had we measured the psychological traits of fathers in the group of parents, our findings may have been more robust.
Another area where a question might be raised concerns the subsample used and whether these findings would generalize to children with more severe pain. As mentioned previously, most children (62%) reported “a little pain,” followed by “some pain” (24%), “a lot of pain” (12%), and “a whole lot of pain” (2%). To address this potential issue, we repeated the regression analyses and included all children with at least some pain (n = 127). The findings did not change; maternal psychological distress and perceived pain severity both significantly predicted consultation.
In conclusion, this study adds support to the mounting evidence that children's health care utilization cannot be seen in isolation from environmental influences. Consulting for health care is based on a myriad of factors, from the child's response to physical symptoms to the way those in his or her environment react to this response. Ultimately, a parent's decision to consult is subjective, and psychological state can have a major effect on this decision. When the parent is distressed, this factor can have paramount influence on this decision. Our study has presented further evidence of which factors are the most important to focus on in assessing children's health care utilization. Future research should assess how addressing parental psychological distress could affect children's health care use patterns.
Correspondence: Rona L. Levy, MSW, PhD, MPH, School of Social Work, University of Washington, 4101 15th Avenue NE, Box 354900, Seattle, WA, 98105 (firstname.lastname@example.org).
Submitted for Publication: December 22, 2005; final revision received March 28, 2006; accepted May 7, 2006.
Author Contributions:Study concept and design: Levy, Walker, and Whitehead. Acquisition of data: Levy, Walker, and Feld. Analysis and interpretation of data: Levy, Langer, Walker, and Whitehead. Drafting of the manuscript: Levy, Langer, and Walker. Critical revision of the manuscript for important intellectual content: Levy, Langer, Walker, Feld, and Whitehead. Statistical analysis: Levy, Langer, and Whitehead. Obtained funding: Levy. Administrative, technical, and material support: Levy, Feld, and Whitehead. Study supervision: Levy.
Funding/Support: This study was supported by grants RO1 HD36069, RO1 DK31369, and R24DK67674 from the National Institutes of Health.
Levy RL, Langer SL, Walker LS, Feld LD, Whitehead WE. Relationship Between the Decision to Take a Child to the Clinic for Abdominal Pain and Maternal Psychological Distress. Arch Pediatr Adolesc Med. 2006;160(9):961-965. doi:10.1001/archpedi.160.9.961