Gijsbers, Carolien F.M.*; Benninga, Marc A.†; Schweizer, Joachim J.‡; Kneepkens, C.M. Frank§; Vergouwe, Yvonne||; Büller, Hans A.||
In 1958, Apley and Naish (1) published their landmark article on recurrent abdominal pain (RAP) in children. RAP, defined as “at least 3 bouts of pain, severe enough to affect the daily activities, over a period of not less than 3 months,” was found in 108 (10.8%) of 1000 schoolchildren. As no organic diagnoses were suspected in these 108 children (no ancillary investigations being performed), Apley and Naish concluded RAP to be mainly attributed to psychological factors. Alarm symptoms were advocated to be used as indicators of organic disease. Several versions of these so-called red flags were published, addressing Crohn disease, peptic ulcer, and urinary tract disease (2).
In the 1970s, the Manning criteria for irritable bowel syndrome (IBS) in adults were developed to obtain a positive diagnosis of IBS without the need of diagnostic testing (3). These criteria were followed by the expert-based Rome criteria, first published in the early 1990s, providing criteria for functional gastrointestinal disorders in adults and children (4,5). The Rome criteria were meant to structure the area of functional symptoms by classification, leading to more homogenous patient groups, which could be helpful in patient care and could form the basis of research on pathophysiological mechanisms as well as clinical research. The Rome Working Teams pointed out that the validation of the diagnostic criteria is needed because they were adopted by consensus rather than by data analysis (4–6). Therefore, we set out to validate the Rome III criteria as to their capacity to differentiate between organic and functional causes of abdominal pain and to assess the role of alarm symptoms with respect to their capacity to differentiate between organic disease and functional causes of abdominal pain.
Between May 2002 and May 2004, all consecutive patients (ages 4–16 years) presenting with RAP according to the Apley criteria and referred by general practitioners to a secondary center (C.F.M.G.), were included. The patients were prospectively evaluated and diagnosed in 2 independent ways.
First, they were evaluated for a clinical diagnosis. In this first phase, they were evaluated in a standardized way as described earlier (Fig. 1) (7,8). Standardized history and physical examination were followed by a diagnostic workup, consisting of a blood cell count, chemistry, urine screening, fecal tests, hydrogen breath tests, plain abdominal radiograph, and abdominal ultrasound, resulting in the establishment of preliminary diagnoses, such as Helicobacter pylori (Hp) gastritis, and diagnostic clues, such as carbohydrate malabsorption.
A causal relation of these preliminary diagnoses or diagnostic clues with the abdominal pain was only assumed, when the pain fulfilled the criteria described in the online-only appendix (http://links.lww.com/MPG/A302) (7,8). Therefore, therapeutic interventions were performed in the second phase. If intervention was needed for more than 1 preliminary diagnosis in a patient, stepwise interventions were performed as shown in Figure 1. When therapeutic interventions for these preliminary diagnoses were not successful, patients with persistent abdominal pain had a trial with laxatives. According to our definition, children were considered to have (occult) constipation when they became pain free with laxative therapy and remained so during a follow-up period of at least 6 months, thereby including both children fulfilling the Rome criteria for functional constipation and children not fulfilling these criteria who are diagnosed with “occult constipation,” as described in the online-only appendix (http://links.lww.com/MPG/A302) (8,9). In case of persistent pain after a trial with laxatives, other investigations were performed as indicated.
The resulting clinical diagnoses were independently reviewed by 2 pediatric gastroenterologists (C.M.F.K. and J.J.S.) and subsequently categorized as “organic,” “functional,” “both organic and functional,” “spontaneous recovery,” and “rest group.”
Second, a fourth pediatric gastroenterologist (M.A.B.) independently categorized the symptom clusters of the patients based on the patients charts according to the Rome III criteria as “irritable bowel syndrome” (IBS), “functional dyspepsia” (FD), “functional abdominal pain” (FAP), “functional abdominal pain syndrome” (FAPS), or “no Rome pain syndrome.”
Pain waking the child from sleep, localization in right upper or right lower quadrant, rectal blood loss, fever (by history), weight loss (by history), family history of inflammatory bowel disease or celiac disease or (first-degree family) peptic ulcer, hemoglobin concentration <7.0 mmol/L, and erythrocyte sedimentation rate (ESR) >20 mm were considered alarm symptoms (4). Alarm symptoms were provided by the parents without verification through diaries or otherwise, except for hemoglobin and ESR that were tested as part of the diagnostic workup. As indicated by the Rome criteria, a Rome diagnosis was made if a patient presented with the symptom cluster of a functional gastrointestinal pain syndrome in the absence of alarm symptoms (4).
After these 2 diagnostic categorizations, validation of the Rome diagnoses took place with the clinical diagnoses as the reference standard (Fig. 2). The study was approved by the regional medical ethics committee.
To validate the Rome III criteria (Rome diagnoses, based on symptom clusters of the Rome pain syndromes and absence of alarm symptoms) as to their capacity to recognize functional causes of abdominal pain, descriptive statistics were performed, using SPSS Statistics 17.0 (IBM, Armonk, NY). The sensitivity and specificity as well as the positive predictive value (PPV) and negative predictive value (NPV) and the positive and negative likelihood ratio (LR pos, LR neg) of the Rome diagnoses were estimated, using the clinical diagnoses as the reference standard.
The role of absence of alarm symptoms with respect to their capacity to exclude organic disease, and consequently to recognize functional causes, was similarly analyzed.
A total of 220 patients entered the study protocol, of whom 20 were lost to follow-up. Table 1 shows the final clinical diagnoses after at least 6 months follow-up of the remaining 200 patients (113 girls, 87 boys, mean age 8.8 [range 4.1–16.0] years). Organic causes were found in 34 patients (17%), functional causes in 79 (40%), both organic and functional cause in 18 (9%), and spontaneous recovery in 54 patients (27%), the remaining group of 15 patients (8%) consisted of patients with no certain diagnosis (13) and 2 unresolved cases. For statistical analysis we included the group of spontaneously recovered patients and the rest group in the functional group because most of them may be supposed to have had functional pain. Thus, the “functional group” consisted of 148 patients (74%). Likewise, the children with both organic and functional causes were included in the organic group. Thus, the “organic group” consisted of 52 patients (26%). Table 1 also depicts the symptom clusters of the functional abdominal pain syndromes, present at presentation, in relation with the clinical diagnoses.
One or more alarm symptoms were found in 115 patients (57.5%) (Table 2). Alarm symptoms were found in 28 patients (54%) in the organic group and in 87 patients (59%) in the functional group. The statistics of absence of alarm symptoms to find a functional disorder (clinical diagnosis) are based on the 2 × 2 table depicted in Table 3: sensitivity 0.41 (95% confidence interval 0.33–0.49), specificity 0.54 (0.40–0.67), PPV 0.72 (0.62–0.81), NPV 0.24 (0.17–0.32), LR pos 0.89 (0.63–1.27), and LR neg 1.09 (0.82–1.45).
Table 4 shows the Rome diagnoses, based on symptom clusters at presentation and absence of alarm symptoms. As can be seen from the table, a functional diagnosis according to the Rome criteria did not exclude the possibility of an organic cause of the abdominal pain. At the end, the clinical diagnoses made clear which patients had an organic and which patients had a functional diagnosis.
The performance of the Rome criteria (Rome diagnosis) to detect a functional disorder (clinical diagnosis) is based on the 2 × 2 table depicted in Table 5: sensitivity 0.35 (95% confidence interval 0.27–0.43), specificity 0.60 (0.46–0.73), PPV 0.71 (0.61–0.82), NPV 0.24 (0.17–0.32), LR pos 0.87 (0.59–1.29), LR neg 1.09 (0.85–1.40).
This is the first study to validate the Rome criteria for the functional pain syndromes as to their capacity to differentiate between organic and functional causes of abdominal pain in children and to assess the significance of alarm symptoms in children with abdominal pain.
Our main finding is that the Rome III criteria are not able to differentiate between organic and functional causes of abdominal pain and that alarm symptoms suggesting organic causes of RAP are present in the same percentage in patients with a clinical diagnosis of a functional condition as in patients with an organic condition, implicating insufficient capacity to differentiate between organic disease and functional causes of abdominal pain.
Alarm symptoms are considered a reason for more extensive clinical investigations (2,4,10). In this study we found 57.5% of the patients presenting with 1 or more alarm symptoms. Remarkably, alarm symptoms were found as frequently in patients with functional disorders as in patients with organic disease. Although alarm symptoms are important for the recognition of more severe conditions such as Crohn disease, they are not helpful in the diagnostic process of RAP. In this population of 200 patients, we did not find any patient with Crohn disease because at presentation in this secondary care center, the patients with Crohn disease had other main symptoms than merely RAP, and therefore could not be included in the study.
It is noteworthy that the individual alarm symptoms all showed roughly the same distribution in the various clinical diagnosis groups. Rectal blood loss is not unusual in children with constipation; in most patients blood loss was probably caused by fissures, which, however, were not seen at physical examination. Other alarm symptoms in children with a functional diagnosis are more difficult to understand. In a study on the frequency of Rome III pain diagnoses in children with abdominal pain, Helgeland et al (11) found 7% to have alarm symptoms, also suggesting that a better definition of alarm symptoms is needed.
The low sensitivity of the Rome criteria is inherent to the Rome procedure and the high prevalence of alarm symptoms: before a Rome diagnosis of a functional disorder can be made, patients with alarm symptoms need to be investigated for organic disease. Unfortunately, alarm symptoms appear to be at least as frequent in functional as in organic disorders, and many patients with alarm symptoms end up to have a functional cause of their pain.
The Rome symptoms are developed to enable the diagnosis of a functional disorder in the absence of alarm symptoms with sufficient certainty, without the need of further investigations. The PPV indicates the reliability of this approach. Organic disease would have been missed (1-PPV) in 21 of 73 patients (29%) with a Rome diagnosis. This percentage is higher than that in the study by Helgeland et al (11), who found that 8 of 152 children (5.3%) with organic diagnoses would have been missed had the investigations been limited to medical history, clinical examination, and stool examination for blood.
The history of the Rome criteria starts with the Manning criteria for adult IBS, published in 1978 (3). In 1990, Talley et al (12) evaluated the validity of the Manning criteria for the discrimination of patients with IBS from healthy controls, from patients with other gastrointestinal diseases and from patients with nonulcer dyspepsia. Sensitivity and specificity of the Manning score with respect to differentiation of IBS from organic disease were 58% and 74%, respectively. Two systematic reviews on the validation of symptom based criteria in adults with IBS concluded that insufficient studies are performed to validate the diagnostic criteria for IBS but that until now none of the symptom-based criteria can accurately exclude organic disease (13,14).
It is noteworthy that the process of validation of the Rome criteria is complicated by the requirement of “No evidence of an inflammatory, anatomic, metabolic, or neoplastic process that explains the subject's symptoms” as a prerequisite for a functional diagnosis. Obviously, it is unclear how to establish “no evidence” of organic disease. The Rome Working Team advises diagnosis of IBS when symptoms meet the IBS criteria, in the presence of a normal physical examination and growth curve with the absence of alarm symptoms; for FAP and FD comparable advices are given (4,10). Our study shows that more organic causes can be found when more thorough investigation is performed. As we have shown earlier, the majority of children with an organic cause can be identified with a limited number of tests (8). We are not the only ones drawing this conclusion. Several authors disputed the clinical utility of the Rome criteria because of insufficient differentiation between functional and organic disease and insufficient help with identification of individualized therapeutic strategies that are based on etiologic mechanisms (15,16).
Additionally, this study provides no evidence that differentiation between the functional pain syndromes makes any difference with respect to therapy or outcome. For clinical use, the added value of the Rome classification is limited. This is reflected in the use of the Rome criteria in children with abdominal pain by only 39% of members of North American Society of Pediatric Gastroenterology, Hepatology, and Nutrition (17). With respect to research, the gain of the Rome classification for functional gastrointestinal pain syndromes lies in the more restricted symptom clusters as opposed to the broad Apley criteria. More uniformity of presentation presumably is a better basis for research, especially with respect to pathophysiology; however, diagnostic disagreement with respect to Rome classification is a recognized problem, in view of the considerable inter- and intraobserver variability and the limited diagnostic agreement between physicians, parents, and children, with respect to both Rome II and Rome III criteria (18–21).
The strength of this study lies in the way of establishing the diagnoses, based on preset criteria including therapeutic intervention to establish a causal relation in every patient, as a basis for validation of the Rome criteria for the functional pain syndromes, and the independent scoring of the Rome criteria.
The limitations lie first in the difficulty to prove causal relationships with certainty. Therefore, we strictly adhered to 3 criteria for the assessment of a clinical diagnosis: the patients should be pain free after the intervention; there should be an appropriate, diagnosis-specific interval between therapeutic intervention and the patient becoming pain free; and the therapeutic effect should last at least 6 months. Thus, we aimed to evade the risk of considering spontaneous recovery or temporary relief as caused by the intervention, as well as that of confusing the treatment result with a placebo effect—which has been shown to seldom last as long as 6 months (8,22). A second limitation is that the categorization of Rome diagnoses was performed by 1 specialist only. This approach was chosen, however, because the observer in question was himself a member of the Rome III committee, and because we were less interested in differentiating between the various Rome diagnoses—which is the main cause of interobserver variation—than in their presence or absence. Obviously, this has had no influence with respect to the major effect of alarm symptoms.
In conclusion, we provide evidence that the clinical symptom-based Rome III criteria are not specific enough to rule out organic causes of abdominal pain. Moreover, the presence of alarm symptoms does not differentiate between organic and functional abdominal pain.
1. Apley J, Naish N. Recurrent abdominal pains: a field survey of 1,000 school children. Arch Dis Child
2. Oberlander TF, Rappaport LA. Recurrent abdominal pain during childhood. Pediatr Rev
3. Manning AP, Thompson WG, Heaton KW, et al. Towards positive diagnosis of the irritable bowel. Br Med J
4. Rasquin A, Di Lorenzo C, Forbes D, et al. Childhood functional gastrointestinal disorders: child/adolescent. Gastroenterology
5. Hyman PE, Milla PJ, Benninga MA, et al. Childhood functional gastrointestinal disorders: neonate/toddler. Gastroenterology
6. Thompson WG. The road to Rome. Gut
1999; 45 (suppl 2):II80.
7. Gijsbers CFM, Benninga MA, Büller HA. Clinical and laboratory findings in 220 children with recurrent abdominal pain. Acta Paediatr
8. Gijsbers CFM, Kneepkens CMF, Schweizer JJ, et al. Recurrent abdominal pain in 200 children: somatic causes and diagnostic criteria. Acta Paediatr
9. Gijsbers CFM, Kneepkens CMF, Vergouwe Y, et al. Occult constipation: faecal retention as a cause of recurrent abdominal pain in children. Eur J Pediatr
2013; (in press).
10. Rasquin-Weber A, Hyman PE, Cucchiara S, et al. Childhood functional gastrointestinal disorders. Gut
1999; 45 (suppl 2):II60–II68.
11. Helgeland H, Flagstad G, Grotta J, et al. Diagnosing pediatric functional abdominal pain in children (4-15 years old) according to the Rome III Criteria: results from a Norwegian prospective study. J Pediatr Gastroenterol Nutr
12. Talley NJ, Phillips SF, Melton LJ, et al. Diagnostic value of the Manning criteria in irritable bowel syndrome. Gut
13. Ford AC, Talley NJ, Veldhuyzen van Zanten SJ, et al. Will the history and physical examination help establish that irritable bowel syndrome is causing this patient's lower gastrointestinal tract symptoms? JAMA
14. Jellema P, van der Windt DA, Schellevis FG, et al. Systematic review: accuracy of symptom-based criteria for diagnosis of irritable bowel syndrome in primary care. Aliment Pharmacol Ther
15. Camilleri M. Do the symptom-based, Rome criteria of irritable bowel syndrome lead to better diagnosis and treatment outcomes? The con argument. Clin Gastroenterol Hepatol
16. Quigley EM. The ‘con’ case. The Rome process and functional gastrointestinal disorders: the barbarians are at the gate!. Neurogastroenterol Motil
17. Schurman JV, Hunter HL, Friesen CA. Conceptualization and treatment of chronic abdominal pain in pediatric gastroenterology practice. J Pediatr Gastroenterol Nutr
18. Saps M, Di Lorenzo C. Interobserver and intraobserver reliability of the Rome II criteria in children. Am J Gastroenterol
19. Schurman JV, Friesen CA, Danda CE, et al. Diagnosing functional abdominal pain with the Rome II criteria: parent, child, and clinician agreement. J Pediatr Gastroenterol Nutr
20. Chogle A, Dhroove G, Sztainberg M, et al. How reliable are the Rome III criteria for the assessment of functional gastrointestinal disorders in children? Am J Gastroenterol
21. Czyzewski DI, Lane MM, Weidler EM, et al. The interpretation of Rome III criteria and method of assessment affect the irritable bowel syndrome classification of children. Aliment Pharmacol Ther
22. Spiller RC. Problems and challenges in the design of irritable bowel syndrome clinical trials: experience from published trials. Am J Med
© 2014 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,