Prevalence, characteristics, and management of childhood functional abdominal pain in general practice

Abstract Objective To (i) describe the proportion of children presenting with abdominal pain diagnosed by the GP as functional abdominal pain (GPFAP); (ii) evaluate the association between patient and disease characteristics and GPFAP; (iii) describe diagnostic management by the GP in children presenting with abdominal pain, and (iv) evaluate whether children with GPFAP fulfill diagnostic criteria for functional abdominal pain (FAP) as described in current literature: chronic abdominal pain (CAP) and the Rome III criteria (PRC-III) for abdominal pain-related functional gastrointestinal disorders (FGID). Design Cross-sectional study. Setting General practices in the Netherlands. Subjects 305 children aged 4–17 years consulting for abdominal pain. Main outcome measures GPFAP, CAP, FGIDs. Results 89.2% of children were diagnosed with GPFAP. Headaches and bloating were positively associated with GPFAP whereas fever and > 3 red flag symptoms were inversely associated. Additional diagnostic tests were performed in 26.8% of children. Less than 50% of all children with GPFAP fulfilled criteria for CAP and FGIDs; in 47.9% of patients the duration of symptoms at presentation was less than three months. Conclusions In almost 90% of children included in this study the GP suspected no organic cause for the abdominal pain. GPs diagnose FAP in children without alarm symptoms and order diagnostic testing in one out of four children presenting with abdominal pain. No difference was found in GPs’ management between children with a diagnosis of GPFAP and other diagnoses. Only about half of the children with a GP diagnosis of FAP fulfilled time-criteria of FAP as defined in the literature.

pain (CAP) [5]. The " Pediatric Rome Criteria III " (PRC-III) classifi ed abdominal pain-related functional gastrointestinal disorders (FGIDs) using a symptombased approach [6]. We assume that children suspected of FAP in general practice have comparable characteristics to children receiving a diagnosis of FAP in other settings.
To fulfi ll these defi nitions organic diseases need to be excluded. The extent of diagnostic testing is left to the decision of the clinician. Wanting to avoid unnecessary medical testing in children presents GPs with the diffi cult task of reassuring themselves and their patients that no signifi cant causes are missed.

Background
Abdominal pain accounts for 5% of childhood consultations in general practice [1]. It has a major impact on the child ' s well-being and the healthcare system [2,3]. Usually, this complaint is not associated with organic disease and is labeled as functional abdominal pain (FAP).
There are different approaches for defi ning childhood FAP. In 1958, Apley described recurrent abdominal pain as Ն 3 bouts of pain, severe enough to affect activities, over a period of at least three months [4]. Von Baeyer added criteria for impact on daily functioning, and called it chronic abdominal To date, no symptoms, signs, or tests have been reported to help discriminate between organic and functional abdominal pain [7]. We therefore assume that in general practice, diagnostic testing in children presenting with abdominal pain will be directed towards ruling in or out organic disease, rather than towards diagnosing FAP.
The present study investigates the proportion of FAP in children presenting with abdominal pain in general practice according to different defi nitions, evaluates the association between patient and disease characteristics and a GP diagnosis of FAP (GPFAP), and describes GPs ' diagnostic management.

Design and setting
We performed a cross-sectional analysis of baseline data of the HONEUR abdominal pain cohort. Fiftythree GPs, together comprising a population of 16.000, children aged 4 -17 years, were recruited in Rotterdam, a multicultural city, and its rural surroundings. GPs recruited consecutive children consulting for abdominal pain during a two-year period. A child was eligible if the consultation was not preceded by a consultation for this complaint in the previous three months. After written informed consent, a research nurse visited the children within one week and collected data. Included children were younger than eligible non-included children (mean 8.5 versus 9.2 years); fewer children diagnosed with " gastroenteritis " and more with " generalized abdominal pain " were included. Therefore, our cohort represents young school-aged children consulting their GP with a new episode of abdominal pain not obviously related to gastroenteritis [8].

Selection of determinants
We evaluated the association between GPFAP and characteristics reported to be associated with FAP in a systematic review [7]. In addition we evaluated red fl ag symptoms reported to be associated with organic disease in (inter)national guidelines [7,9,10]. Demographic data, additional symptoms, comorbidity, and family history were recorded on structured questionnaires. For assessment of somatization, we used the somatic syndrome scale of the Child Behavior Checklist [11 -13]. Pain intensity was determined on an 11-point numerical rating scale (NRS) in children aged 8 -17 years; and on a six-point pain faces scale for children Ͻ 8 years [14]. The GP ' s management was recorded in 16 structured and one open question. Description of the data collection is described in detail elsewhere [8].

Primary outcome GPFAP
GPs coded consultations according to the International Classifi cation of Primary Care (ICPC) [15]. These codes were extracted from the medical records three months after inclusion. Of the children that consulted the GP more than once for abdominal pain in this three-month period, in 19 children the initial code was changed. For this analysis the last given ICPC code was used. We considered the following diagnoses as GPFAP: " abdominal pain, general " (D01), " epigastric pain " (D02), " abdominal pain, localized other " (D06), " constipation " (D12) and " irritable bowel syndrome (IBS) " (D93). A study in the Netherlands showed an agreement for " generalized abdominal pain " between GPs and experts of 85% [16].

Defi nitions of FAP as defi ned in the literature
CAP: The occurrence of abdominal pain at least 1.
once each month in the past three months, severe enough to stay home from school, terminate or avoid play, take medication for the pain, or to be rated as moderate to severe ( Ն 3/10 on the NRS) [5]. FGIDs: IBS, functional dyspepsia, functional 2.
abdominal pain, and functional abdominal pain syndrome [6]. The PRC-III defi ne a time period of two months in which the symptoms had to occur at least once a week; however, we used the timeline as proposed by Von Baeyer (Supplementary Appendix I available online at http://informahealthcare.com/doi/abs/10.3109/ 02813432.2013.844405).

Statistical analyses
Data are presented as means with standard deviations (SD) or percentages of the number of patients Although children with functional abdominal pain (FAP) are mainly managed in primary care, not much is known about FAP in this setting. This study showed that: In almost 90% of children presenting with • abdominal pain included in this study, the general practitioner (GP) suspects no organic cause. GPs diagnose FAP in children without red • fl ag symptoms and order diagnostic testing in one out of four children presenting with abdominal pain. Only 50% of children with a diagnosis of • FAP by their GP fulfi ll the time-criteria for FAP as defi ned in the literature.
responding per item. Factors associated with GPFAP were identifi ed by logistic regression analyses adjusted for age. The association between GP management and GPFAP was evaluated by logistic regression analyses and adjusted for potentially relevant confounders identifi ed in the bivariate analysis. Results are expressed as odds ratios (OR) and 95% confidence intervals (95%CI). Analyses were performed using SPSS version 17.0.

Study sample
In total 305 of 348 invited children (87.6%) gave informed consent and participated in the study. Mean age was 8.3 years and 62.0% were female. Eight consultations were not given an ICPC code and it was not possible to determine the diagnosis based on information in the medical records. In 297 children a diagnosis was available, of which 265 were diagnosed as GPFAP (89.2%) ( Table I).

Characteristics associated with GPFAP
The chance of a diagnosis of GPFAP increased with increasing age. Headaches and bloating were associated with GPFAP. Fever and a UTI in the past year were negatively associated with GPFAP; vomiting, blood on stool, and intra-abdominal comorbidity showed a trend towards an inverse association.
Having Ͼ 3 red fl ag symptoms showed a signifi cant inverse association with GPFAP (see Table I).

Management by GP
Additional diagnostic testing was performed in 26.3% of children; 10.1% of children were referred to specialist care. No differences were observed in diagnostic management by the GP between children with and without GPFAP and between children with or without Ͼ 3 red fl ag symptoms. The association between GP management and GPFAP remained insignifi cant when adjusting for potentially relevant confounders (Table II). Of 265 children with GPFAP the GP ordered blood sampling in 23.0% of children, abdominal X-rays or ultrasonography in 8.3%, and both in 4.5%.

Relation between defi nitions
Of 265 children with GPFAP, 130 (50.6%) fulfi lled FGID criteria: 53.8% fulfi lled criteria for FAP, 38.5% for IBS, and 7.7% for functional dyspepsia (see Table II). All children with GPFAP not fulfi lling FGID criteria lacked the time criterion of three months. Of children with GPFAP, 47.9% fulfi lled criteria for CAP; of children with GPFAP not fulfi lling CAP criteria 92.0% lacked the time criterion.

Discussion
The present study showed that in 90% of children presenting with abdominal pain, the GP suspected FAP. Older age and the coexistence of headaches and bloating were associated with GPFAP. Fever, a UTI in the past year and having Ͼ 3 red fl ag symptoms were inversely associated with GPFAP. The GP ordered additional testing in one out of four children. No differences in GPs ' management were observed between children with and without GPFAP. Only about 50% of children with GPFAP fulfi lled the criteria for CAP or an FGID. This discrepancy was due to the shorter duration of complaints than the 3 months in the respective defi nitions.
In 90% of children the GP suspects FAP; this is consistent with the fi ndings of Apley who found somatic causes in 6 -8% of children with recurrent abdominal pain in population-based studies [17]. Recent studies found organic abnormalities in 45 -88% [18,19]; however, these studies were performed in specialistcare settings and selection of patients and excessive testing may have yielded higher proportions of organic abnormalities. Furthermore, abnormal fi ndings are not necessarily causally related to abdominal pain.
The observation that age was signifi cantly associated with GPFAP is in accordance with the fi ndings of others that the prevalence of chronic pain increases with age [20]. Headache and bloating were positively associated with GPFAP. Headache is, together with limb and abdominal pain, the most frequently reported functional complaint in children [21]. Bloating is a common, not well-defi ned symptom in adults related to IBS [22]. Somatization and a family history of GI complaints have been found by others to be associated with FAP. Although somatization showed a tendency toward a statistically signifi cant association with GPFAP (p ϭ 0.08) we could not fully confi rm these fi ndings, possibly due to a lack of power. Given that children with GPFAP have a shorter duration of symptoms compared with children with FAP in specialist care, our fi nding may indicate that these characteristics are related to the duration of symptoms rather than to symptoms of FAP.
The fi nding that red fl ag symptoms were inversely associated with GPFAP is in accordance with (inter) national guidelines in which red fl ag symptoms are associated with a higher risk of organic disease [7,23]. These results support our hypothesis that the diagnostic management of children with abdominal pain is directed towards the in-or exclusion of organic causes.
According to guidelines, organic disease needs to be excluded before a diagnosis of FAP can be made. However, in a population with a small prior probability of organic disease, the risk of false positive fi ndings will be relatively high. Testing will therefore not be cost-effi cient and introduces unnecessary parental worries. From this perspective additional testing in one out of four children seems high.  1.11 (0.30 -4.19) Notes: * Age, Ͼ 3 red fl ag symptoms, and intra-abdominal comorbidity.
As the point of entry to healthcare, many children will visit the GP with short-term symptoms. Together with the low prior probability of somatic pathology, this enforces GPs to consider FAP in an early stage. Given the high percentage of GPFAP in children lacking chronicity, the GP seems confi dent in giving a symptom-diagnosis after exclusion of organic disease. He might presume the patient will return in the case of persistence of complaints.
Compared with others, we found a higher proportion of children fulfi lling the PRC-III for FAP [24,25], which might indicate subgroup misclassifi cation. A possible explanation is that we were not able to assess abdominal migraine as its criteria were considerably revised in the updated criteria. Furthermore, we did not use the sub-classifi cation of functional constipation as this is not considered an abdominal pain-related FGID. Therefore, children otherwise fulfi lling criteria for abdominal migraine or functional constipation were " misclassifi ed " as FAP.

Study limitations
First, the presence of patient determinants was assessed using standardized questionnaires and we are not sure whether the GP used these determinants in his diagnostic reasoning. However, the red fl ag symptoms were selected from (inter)national guidelines and we may therefore reasonably assume the GPs did include them in their diagnostic reasoning. Second, we were not able to use the " Rome III Diagnostic Questionnaire for Pediatric FGIDs " as this study was ongoing at the time of its publication; nevertheless, we used equivalent questions used for the assessment of FGIDs (Supplementary Appendix I available online at http://informahealthcare. com/doi/abs/10.3109/02813432.2013.844405). This might have led to subgroup misclassifi cation as was pointed out in the discussion section.
Third, we used different criteria for required symptom duration than proposed by the PRC-III. Although the required duration of two months includes more children fulfi lling FGID criteria, the less stringent criterion for symptom frequency is more likely to have led to an overestimation of the prevalence of FGIDs in our cohort.
Fourth, the power of our study was lower than the expected 80% [8]. Given a prevalence of GPFAP of 90% and a distribution of determinants between children with and without GPFAP of 90% versus 10%, we had a power of 72% to detect an OR of 2.5 with an α of 0.05. There is a chance of 28% that our results were found by chance. We feel, however, that a loss of power of 8% is not enough to infl uence our conclusions.

Clinical implications and future research
Additional diagnostic testing is performed in one out of four children with abdominal pain. However, we found heterogeneity in the kind of tests asked for, and there is a lack of evidence for their diagnostic value in primary care. Therefore, studies on the diagnostic value of additional diagnostic tests in children with abdominal pain in primary care are essential. To our knowledge these studies are lacking, which might be explained by the fact that evaluating diagnostic tests in a population with low prior probabilities of disease is a methodological challenge.
In a prior Dutch cohort of children with abdominal pain, it was found that the consultation rate after fi rst presentation of abdominal pain was 21.9% [1], indicating that active follow-up of children with FAP is not common practice. Instead of managing FAP by focusing on the exclusion of organic disease, the GP could more often use active follow-up to monitor the course of the complaints. Our fi nding that only half of the children with GPFAP had chronic complaints, whereas FAP in referred children seems a chronic condition, makes follow-up even more warranted. To date, there is a lack of knowledge on the prognosis of FAP and its determinants emphasizing that studies on this topic are highly recommended.

Conclusion
In our study GPs suspected FAP in almost 90% of children visiting with abdominal pain. Only about 50% of children with a diagnosis of GPFAP fulfi lled the criteria for CAP or an FGID due to a short duration of their complaints at presentation. GPs suspect FAP in children without red fl ag symptoms and order additional tests in one out of four. It remains inconclusive whether a child suspected of FAP in general practice has comparable characteristics to a child with FAP diagnosed in other settings. For better understanding of FAP and its prognosis in primary care further studies in this setting are needed.