“If all you have is a hammer…”

Posted on by

In my training, one of my mentors would be critical of the mentality “scope first, think second.”  He was concerned that too many gastroenterologists/pediatric gastroenterologists used endoscopy as a tool simply “because if all you have is a hammer, the world looks like a nail.”

A recent publication (Clin Gastroenterol Hepatol 2014; 12: 963-69) on first glance, however, provides ammunition to the “scope first” gastroenterologists and undermines the concept that “functional” GI disorders vastly outnumber “organic” GI disorders.  The key finding was that esophagogastroduodenoscopy (EGD) provided an “accurate” diagnosis in 38%.  This included reflux esophagitis in 21%, eosinophilic esophagitis (EoE) in 5%, eosinophilic gastroenteritis (EGE) in 4%, H pylori in 2%, celiac disease 0.6%, and Crohn’s disease 0.4%. This finding is dramatically higher than previous studies.  In fact, in a recent published study (Understanding Idiopathic Nausea | gutsandgrowth) on idiopathic nausea, a control group of patients with chronic abdominal pain had a normal endoscopy in 100%!

In this prospective study of 290 children (ages 4-18 years with a mean age of 11.9 years), the primary indication for upper endoscopy was chronic abdominal pain.  Of this 290, 216 had at least 1 alarm feature and 125 had at least 2 alarm features.  Alarm features were considered to be the following:

  • Nighttime awakening 33.3%
  • Weight loss 15.6%
  • Family history of IBD 8.4%
  • Vomiting 7.8%
  • Dysphagia 6.9%
  • Nocturnal diarrhea 6.7%
  • Gastrointestinal bleed 5.8%
  • Chronic diarrhea 5.8%
  • Unexplained fever 4.4%
  • Arthralgia 4.0%
  • Growth failure 2.4%
  • Perirectal disease 0.7%
  • Delayed puberty 0.2%

There is little debate that abdominal pain in combination with true alarm symptoms (True red flags in recurrent abdominal pain | gutsandgrowth) merits further evaluation.  The aspect of this report that is worthy of close inspection is the diagnostic yield in the 74 patients without alarm symptoms.  The authors note that 25 (33.7%) had a diagnosis established with EGD including 16 with reflux esophagitis, 4 with EGE, 2 with EoE, 1 with erosive esophagitis, 1 with celiac and 1 with H pylori.  The diagnostic criteria for EGE included ≥10 eosinophils per hpf in the stomach and ≥20 eosinophils per hpf in the duodenum.

The authors note that the diagnostic yield was based on gross endoscopic findings in procedure notes or histologic changes in biopsy reports; “final pathology report on biopsies provided the data source for histologic diagnosis.”

In my opinion there are multiple flaws of this prospective study.

1. There is a very high rate of reflux esophagitis in both the alarm group and the non-alarm group patients with chronic abdominal pain.  Of the entire cohort (n=290), the authors identified reflux esophagitis in 21% and this was “primarily histologic esophagitis.”  Furthermore, the authors state that “the presence or response to PPI therapy was not predictive of esophagitis or GERD.”  So, the obvious problems:

  • Presence of histologic reflux esophagitis varies widely based on the interpreting pathologist.  In a prospective study, more than one pathologist interpreting the histology would be useful.
  • Presence of histologic reflux esophagitis does not exclude the likelihood of coexisting functional disorders (related blog post: Why didn’t patient with documented reflux get better with PPI …).  As a practical matter, “slight” or “focal” esophagitis on histology has questionable real-world relevance in pediatric gastroenterology.
  • The authors acknowledge, “current expert consensus indicates that histology has limited value in evaluating pediatric GERD.”  Yet this diagnostic finding is one of the reasons why the authors claim that EGD is so valuable.

2. In the entire cohort, the authors try to validate their findings by indicating that identification of a diagnosis led to medical therapy that was “effective in approximately 67%” of children with short-term followup. (Only 81% had short-term followup outcomes available).  Yet, there is no control group.  How many children with chronic abdominal pain will improve for a short period without an EGD-based diagnosis?  The answer: a lot of them.

3. Limitations include selection bias toward those with more severe symptoms or alarm symptoms.  In addition, this study included only a small number who were considered to have no alarm symptoms.  Finally, the short-term followup makes conclusions about the response to therapy questionable.

This study will be a useful reference for any pediatric gastroenterologist who wants to justify the need for an endoscopy.  The authors note “the majority of children in our study (93%) met criteria for functional gastrointestinal disorders, and a significant proportion (38%) still had significant histologic findings.  Therefore, we conclude the Rome III criteria alone are not sufficient to identify children who require upper endoscopy, and screening for alarm symptoms has limited utility.”

In my opinion, the reliance on histology as well as selection bias weaken the findings of this study.  If a patient with a histologic diagnosis of reflux (or several other entities) and a presentation of chronic abdominal pain does not improve, the pediatric gastroenterologist should remember that only “a poor carpenter blames his tools.”

Bottomline: EGD remains an important tool in evaluating abdominal pain.  However, I think this study substantially overestimates its utility.

Related blog posts: