OBJECTIVE: To investigate the demographic and clinical factors of children who present to the pediatric emergency department (ED) with abdominal pain and their outcomes.
METHODS: A review of the electronic medical record of patients 1 to 18 years old, who presented to the Children’s Hospital of Pittsburgh ED with a complaint of abdominal pain over the course of 2 years, was conducted. Demographic and clinical characteristics, as well as visit outcomes, were reviewed. Subjects were grouped by age, race, and gender. Results of evaluation, treatment, and clinical outcomes were compared between groups by using multivariate analysis and recursive partitioning.
RESULTS: There were 9424 patient visits during the study period that met inclusion and exclusion criteria. Female gender comprised 61% of African American children compared with 52% of white children. Insurance was characterized as private for 75% of white and 37% of African American children. A diagnosis of appendicitis was present in 1.9% of African American children and 5.1% of white children. Older children were more likely to be admitted and have an operation associated with their ED visit. Appendicitis was uncommon in younger children. Constipation was commonly diagnosed. Multivariate analysis by diagnosis as well as recursive partitioning analysis did not reflect any racial differences in evaluation, treatment, or outcome.
CONCLUSIONS: Constipation is the most common diagnosis in children presenting with abdominal pain. Our data demonstrate that no racial differences exist in the evaluation, treatment, and disposition of children with abdominal pain.
- CI —
- confidence interval
- CT —
- computed tomography
- ED —
- emergency department
- EMR —
- electronic medical record
- IV —
- OR —
- odds ratio
What’s Known on This Subject:
Abdominal pain is a frequent complaint in pediatric emergency departments, with a broad differential diagnosis. The impact of demographic and clinical characteristics of patients on the evaluation and management of these children is not well known.
What This Study Adds:
The most common cause of abdominal pain is constipation, which rarely requires hospital admission. Demographic factors, in particular race, do not seem to affect evaluation and management.
Abdominal pain is a common problem in the community and a frequent chief complaint in the pediatric emergency department (ED) with an extensive differential diagnosis that spans the entire spectrum of disease severity.1 Consequently, the evaluation and outcomes for these children vary widely. For example, a 12-year-old girl with right lower quadrant abdominal pain could be suffering from a surgical problem like acute appendicitis or ovarian torsion. Or, she could have a nonsurgical, non–life-threatening disease like constipation or mesenteric adenitis. Differences in the prevalence of some diseases based on age, gender, and race further complicate the picture.2 Consequently, pediatric EDs expend a large number of resources in the evaluation and management of children presenting with abdominal pain.
We sought to define the demographic characteristics, along with visit outcomes, of children with abdominal pain presenting to our ED. We speculated that these characteristics would affect the evaluation and management of these children, as well as their ultimate outcome, and sought to describe this relationship.
A retrospective review of the electronic medical record (EMR) of patients who presented to the Children’s Hospital of Pittsburgh ED with a chief complaint of abdominal pain listed in the EMR between January 1, 2008, and June 30, 2010, was conducted. Children’s Hospital of Pittsburgh is an urban, tertiary care center with an annual ED volume of >70 000 patients. Children older than 1 year and younger than 18 years were included.
The following data were gathered: age, gender, race, type of insurance, month of visit, day of week of visit, final diagnosis (both ED and, for admitted patients, discharge), whether the patient had an operation during the visit, treatments administered, radiology tests performed, and final disposition (admit to acute care floor, admit to PICU, discharge, or death). For analysis purposes, subjects were divided into 3 groups based on race: white, African American, and other. In our institution, race is assigned by the registration clerk in consultation with family members. This information was taken directly from the EMR. Race information was entered by registration personnel at the time of ED check in.
Fever was defined as a recorded temperature in the ED of 38.5°C or greater. Rectal, axillary, and oral temperatures are all used at our institution based on the child’s age and development. Children with no temperature recorded were assumed to be afebrile.
Diagnoses were based on recorded ED and hospital discharge diagnoses. For subjects with multiple diagnoses listed, a single diagnosis was assigned to each subject based on the following priority: (1) appendicitis, (2) renal calculus, (3) urinary tract infection, (4) constipation, (5) gastroenteritis, (6) abdominal pain, and (7) other. For study purposes, these diagnoses were mutually exclusive; each subject could have only 1 diagnosis for a particular visit.
The data were analyzed to ascertain the impact of race on the evaluation and management of abdominal pain. To determine this, modeling was performed by using race as the dependent variable. Predictor variables included the following: imaging study performed (computed tomography [CT] scan, ultrasound, and x-ray), therapy received (narcotics, ketorolac, antiemetics, intravenous [IV] fluids, and enemas), disposition, whether the child had surgery performed, and final diagnosis. Recursive partitioning was performed by using these same factors.
Statistics were calculated by using SPSS Statistics version 18 (IBM SPSS Statistics, IBM Corporation, Chicago, IL). Data were analyzed by using χ2, contingency tables, Student’s t test, and 1-way analysis of variance as appropriate. Modeling was performed by using backward-stepping binary logistic regression. Recursive partitioning was performed by using CART Salford Predictive Miner v6.6 (Salford Systems, San Diego, CA). This activity was approved by the University of Pittsburgh Institutional Review Board. Informed consent was waived by the Institutional Review Board and patient data were de-identified by an honest broker before being made available to the investigators.
During the study period, there were 10 121 patient visits to the ED for abdominal pain representing 6.2% of 164 083 overall patient encounters. Of these, 128 were of children <1 year old and 569 were of children >18 years old. These visits were excluded, leaving 9424 subject visits for analysis. Of these visits, 5493 had multiple diagnoses listed. The leading diagnosis was selected by using the rank order we established. The races of children in the “other” category are summarized in Supplemental Table 10.
Demographic data are summarized in Table 1. Notably, there were differences in mean age, gender, and type of insurance across the 3 race categories. African American children in the study tended to be slightly older, were more likely to be female, more likely to have public insurance, and less likely to be febrile.
On univariate analysis, the racial categories had significant differences in how they were evaluated and their outcomes. As summarized in Table 2, African American children were less likely to have an abdominal CT scan and/or ultrasound compared with white children. African American children were also less likely to receive IV fluids, narcotics, and antiemetics. Additionally, African American children were less likely to be admitted to the hospital and/or have an operation connected with their ED encounter.
Overall, appendicitis accounted for 4.3% of patient visits for abdominal pain. The diagnosis was more commonly seen in white children (5.1%) as compared with African American children (1.9%) with abdominal pain (P < .001, odds ratio [OR] 0.35, 95% confidence interval [CI] 0.25–0.50). A diagnosis of constipation (P = .002, OR 1.25, 95% CI 1.10–1.41), gastroenteritis (P < .001, OR 1.60, 95% CI 1.31–1.95), or urinary tract infection (P = .018, OR 1.44, 95% CI 1.10–1.87) was more common in African Americans.
Subanalysis was performed for 4 of the diagnoses: appendicitis, constipation, abdominal pain, and other. The results of univariate analysis are shown in Tables 3, 4, 5, and 6, respectively. For subjects diagnosed with appendicitis, there were no differences in interventions and outcomes between the races. For subjects diagnosed with constipation, African American children were less likely to have a CT scan (OR 0.10, 95% CI 0.01–0.74) or ultrasound (OR 0.53, 95% CI 0.34–0.86) and to receive IV fluids (OR 0.50, 95% CI 0.35–0.70) as compared with white children. However, the differences in rates of CT scan and ultrasound were not significant on multivariate analysis. The differences in IV fluid administration remained (adjusted OR 0.50, 95% CI 0.35–0.70).
In subjects diagnosed with undifferentiated abdominal pain, African American children were less likely to have a CT scan performed as compared with white children. Additionally, they were less likely to receive narcotics, antiemetics, and IV fluids. Moreover, African American children were less likely to be admitted to the hospital and to have surgery performed (Table 6).
Table 7 displays ORs with 95% CIs for the univariate factors as well as for binary logistic regression models that attempt to adjust for potential confounders. In the adjusted model, there was no longer a significant difference in the administration of an antiemetic. Moreover, African American children were actually more likely to receive ketorolac and have an abdominal x-ray performed. However, for all other factors (rates of CT scan, ultrasound, IV fluid administration, narcotic administration, admission, and operation), the modeling did not completely account for the racial differences.
For children diagnosed with undifferentiated abdominal pain, the difference in IV fluid and narcotic administration seen on univariate analysis remained, with African American children only two-thirds as likely as white children to receive either (Table 8). However, because African American children were more likely to receive ketorolac, multivariate analysis revealed no difference between white and African American children receiving either narcotics or ketorolac in this subgroup.
Table 9 is a summary of the multivariate analysis for the 3695 white and African American children whose diagnosis did not fit into one of the analyzed categories. This category included a disparate group of diagnoses. The most common was “vomiting alone,” with 78 children. Among these children, there were 74 unique discharge diagnoses of varying apparent severity. Significant differences were seen among racial groups, with African American children being less likely to be admitted or have surgery, a CT scan, an ultrasound, or any imaging in general. Additionally, African American children were more likely to receive ketorolac but less likely to receive IV fluids.
Further exploration of these factors was performed by using recursive partitioning. Focus was primarily on those factors that demonstrated a difference across races on multivariate analysis. These included admission rates, surgery rates, use of imaging (CT scan and ultrasound), pain medications, and IV fluids. This analysis was performed on the complete dataset. A representative decision tree for administration of narcotics is shown in Fig 1. Notably, race was not a predictor for any of these outcomes of interest in the first several layers of any of the decision trees.
Disease Burden and Diagnoses
Abdominal pain was a frequently encountered chief complaint in our ED during the study period. As reported, patient demographic characteristics influenced the evaluation and treatment of these patients, as well as their disposition. The most common diagnosis seen was constipation, with almost 20% of all patients and >25% of children 5 to 12 years of age receiving this diagnosis.
Constipation has been studied in the community setting. Van den Berg and colleagues3 performed a systematic review of the epidemiology of constipation in children. They reported a wide range in prevalence (0.7%–29.6%). However, they did not comment on what proportion of these children sought care in a pediatric ED.
Liem and colleagues4 reported a prevalence of constipation of 1.1% during a 2-year period, using the Medical Expenditure Panel Survey database. They found a mean of 0.52 ED visits per child with constipation. This was greater than the 0.16 ED visits per member of the general pediatric population; however, they did not comment on the proportion of ED visits that were directly related to constipation.
Our results support that constipation is a frequently confronted problem in the pediatric ED. In our study, 92% of children eventually diagnosed with constipation were imaged. Most commonly (90%) this was with an abdominal x-ray. This implies that there were objective data that led to the diagnosis of constipation as the cause of abdominal pain in most of these children. Of note, some sources recommend against imaging for constipation.5 Fewer than 3% of constipated children were admitted to the hospital, indicating that most of these patients can be managed as outpatients in the primary care setting once the diagnosis is confirmed. Usually, the diagnosis can be made with a thorough history and complete physical examination (including rectal examination), with or without an abdominal radiograph. All of these are readily accessible outside of the ED in most community practice settings.
There are few published studies that have defined the demographic, clinical, and etiological characteristics of children presenting to the ED with abdominal pain. Reynolds and Jaffe6 performed a prospective study looking at the causes of abdominal pain in children presenting to an ED. Their study of 377 ED patients found that 135 received a discharge diagnosis of abdominal pain, 7% had a diagnosis of constipation, and 2% had otitis media. They did not report associated clinical and demographic characteristics of patients. Scholer and colleagues7 retrospectively examined 1141 children aged 2 to 12 years who presented to their doctor for a nonscheduled visit or to the ED. They found that 5.1% of all visits during the study period were because of abdominal pain; 18.6% had a final diagnosis of upper respiratory infection and/or otitis media, 16.6% had group A streptococcal pharyngitis, 16.0% had viral syndrome, and 15.6% had abdominal pain. Only 2% were diagnosed with constipation. Fewer than 1% of children had an illness requiring surgery. Interestingly, neither of the preceding studies reported a single case of intussusception as an etiology. Our results are drastically different compared with these 2 studies, with our data showing much higher rates of both constipation and surgery. These studies are >15 years old and there may have been a shift in utilization or diagnoses causing the disagreement.
On initial univariate analysis, several of the studied interventions and outcomes appear to show significant racial differences, including disposition, imaging, and therapy. However, univariate analysis also demonstrated significant differences in gender, age, temperature, and diagnosis across race categories. Factoring these characteristics into a multivariate analysis made these differences smaller, but many of them remained.
Interestingly, when univariate analysis was performed on individual diagnoses, most of the racial differences were no longer statistically significant. For example, when the 404 children diagnosed with appendicitis were studied as a subset of the entire data set, there were no differences seen in imaging performed, analgesia and other therapy administered, or disposition, with regard to race. Suspected appendicitis has a well-defined evaluation and management protocol at most institutions, including ours. These data support that the evaluation and management of these children is performed without regard to race.
Interestingly, this contrasts with results of other studies of appendicitis in children. Multiple authors have reported that rates of rupture vary across racial lines, with minorities generally having higher rates.2,8–10 These differences have not been explained by differences in insurance or socioeconomic status.11 However, others have shown that racial differences in ruptured appendicitis disappeared after adjusting for access to care.12 Still others have reported no racial disparities in the rate of rupture.13,14 We did not look specifically for rates of rupture in our study, and the data overall are mixed.
Our finding of decreased incidence of appendicitis in African American children, as compared with white children, is consistent with previous population-based studies.10,15,16 For example, Kokoska and colleagues10 reported an OR of 0.39 (95% CI 0.38–0.41) for African American children to develop appendicitis as compared with white children. Their study was population-based and reviewed 2 national databases. This finding is consistent with our reported OR of 0.35, even though we reviewed visits only for abdominal pain.
Although appendicitis is generally a painful surgical condition that frequently requires narcotic therapy, constipation usually does not. When analyzed separately, children with constipation received pain medications infrequently and with equal rates across all races. However, African American children with constipation did not receive IV fluids, CT scans, or ultrasound as frequently as their white counterparts.
Univariate analysis of the “catch-all” diagnosis of abdominal pain revealed several factors pertaining to racial differences. Multivariate analysis explained many of these. IV fluids were not given to African American children as frequently as white children in this group. Interestingly, whereas African American children were less likely to receive narcotics, as compared with white children they were more likely to receive ketorolac. Because of this, there was no significant difference between the 2 groups in the rate of receiving narcotics and/or ketorolac.
The diagnosis category, “other,” demonstrated differences across races for many factors including admission rate, surgery rate, CT scan, ultrasound, ketorolac, and IV fluids. Interestingly, there was no difference in the rate of narcotic administration or narcotics and/or ketorolac. This other category was a broad group of diagnoses, some of which are very severe and others that are minor. This makes it difficult to determine the influence of the individual diagnosis on reported differences.
Recursive partitioning further elucidated these characteristics. When factors that demonstrated a racial difference were examined using this process, race was found not to be an important factor in any of the decision trees. Clinical and demographic factors like fever, diagnosis, age, and gender were much more important and this held true using multiple methods to analyze the data.
Chamberlain and colleagues first reported on racial differences in admission rates for children seen in the ED.17 Reviewing nearly 9000 children across 13 sites, they found that white children in the 2 lowest illness severity quintiles were 1.5 to 2.0 times more likely to be admitted to the hospital as compared with all other races with similar illness severity. However, this study did not look at diagnosis as a factor affecting admission.
There have been other recent reports of racial differences in the administration of analgesia in the pediatric ED for abdominal pain. One recent study examined a national database of pediatric visits to the ED for abdominal pain. The authors looked at racial groups and reported pain scores as well as other confounders and concluded that African American children were less likely to receive analgesia in the ED as compared with white children.18 Again, final diagnosis was not included as a potential confounder. This may explain why our results differ. Another study using a national database focused specifically on CT use and again found decreased use in African American children.19 Although this study included the diagnosis of appendicitis in multivariate analysis, other diagnoses were not included. Moreover, the authors did not report results for the cohort of children who were diagnosed with appendicitis.
When we included specific diagnoses in our study’s analysis, rates of admission and analgesia administration for subjects with appendicitis, constipation, and abdominal pain were found not to be different among different races. Confounding factors, such as access to care, comorbidities, and response to pain, are not directly accounted for in the study data. However, recursive partitioning confirmed that there were no racial differences across all of the diagnoses we examined. Recursive partitioning is less prone to confounding caused by these types of residual confounders. This further emphasizes the importance of diagnosis as a predictive factor in these kinds of studies. It is difficult to draw conclusions about differences among racial groups when important clinical factors are not accounted for.
Our data were obtained retrospectively from the EMR. It is likely that a few children were miscoded, particularly with regard to chief complaint and final diagnosis. We believe this to be an exceedingly small proportion of the total visits. Race was assigned by the registration clerk in conjunction with the patient’s family members. There may be a very small proportion of subjects who were misclassified. We do not believe that this would substantially affect our conclusions. Additionally, we had to select a single leading diagnosis for each visit. Conceivably, using a different rank-ordering system could have yielded slightly different results; however, we believe that any small differences would not change our conclusions. Additionally, other factors not captured in the data set, for example duration of pain, previous surgeries, or past medical history, may have influenced the evaluation and management of individual patients.
Overall, 70.5% of children during the study period had some form of imaging. Other institutions may have higher or lower rates of utilization. These data were drawn from a single center and reflects the local standards and coding habits of those physicians. Our facility had no “abdominal pain pathway” during the study period; evaluation and management was at the discretion of the treating provider. Although we believe our conclusions reflect the general practice nationally, we cannot definitively prove this.
Our study data did not include the route of administration for pain medications. It is possible that there may be differences across racial groups in this regard. The mechanism of action of narcotics is not dependent on the route of administration. Moreover, when given in recommended doses, narcotics are equi-analgesic irrespective of how they are administered.
Our database also did not include the timing of any of the interventions studied. We are unable to demonstrate, for example, when during the evaluation process a particular child received narcotics. It is possible that children in different race groups received narcotics before diagnosis at different rates. Although we do not think any difference would be significant, we do not have the data available to prove this. Such an analysis is outside of the scope of this study.
Abdominal pain is a common chief complaint in the pediatric ED. Constipation was the most frequent diagnosis and it rarely required hospital admission. Multivariate and recursive partitioning analysis demonstrated that, after classifying by diagnosis, no racial differences existed in the evaluation, treatment, and disposition of children presenting with abdominal pain at our institution.
- Accepted March 11, 2013.
- Address correspondence to Kerry Caperell, MD, Division of Pediatric Emergency Medicine, University of Louisville, 571 S Floyd St, Suite 300, Louisville, KY 40059. E-mail:
Dr Caperell conceptualized and designed the study, carried out the analyses, drafted the initial manuscript, and revised the manuscript; Dr Pitetti guided the study design and the initial analyses, and reviewed and revised the manuscript; Dr Cross performed the recursive partitioning portion of the analysis, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
- Russy R
- Scholer SJ,
- Pituch K,
- Orr DP,
- Dittus RS
- Smink DS,
- Fishman SJ,
- Kleinman K,
- Finkelstein JA
- ↵Livingston EH, Fairlie RW. Little effect of insurance status or socioeconomic condition on disparities in minority appendicitis perforation rates. Arch Surg. 2012;147(1):11–17
- Nwomeh BC,
- Chisolm DJ,
- Caniano DA,
- Kelleher KJ
- Addiss DG,
- Shaffer N,
- Fowler BS,
- Tauxe RV
- ↵Johnson T, Weaver M, Borrero S, Davis E, Myaskovsky L, Zuckerbraun N. Racial and ethnic disparities in the management of pediatric abdominal pain. Paper presented at: Pediatric Academic Societies; April 28, 2012, 2012; Boston, MA.
- Copyright © 2013 by the American Academy of Pediatrics