Diagnostic Performance of Esophagogastroduodenoscopy, Colonoscopy, and Small Bowel Endoscopy in Thai Adults with Chronic Diarrhea

Background: Gastrointestinal endoscopy is recommended to investigate chronic diarrhea in Western countries, but its benets have infrequently been investigated in Southeast Asia. This study aimed to determine the diagnostic utility of esophagogastroduodenoscopy (EGD), colonoscopy, and small bowel endoscopy in Thai chronic diarrhea. Methods: Medical records of consecutive patients who underwent EGD and/or colonoscopy to investigate chronic diarrhea at our center from 2008 to 2012 were reviewed. We also evaluated consecutive patients with negative EGD and colonoscopy who underwent subsequent small bowel endoscopy, including push enteroscopy, balloon-assisted enteroscopy (BAE), and video capsule endoscopy (VCE), from 2008 to 2018. The diagnostic yield of each endoscopic modality was analyzed. Results: A total of 272 patients underwent EGD and/or colonoscopy. Mean hemoglobin and albumin levels were 11.6 g/dL and 3.8 g/dL, respectively. EGD and colonoscopy were performed in 135 and 269 patients, respectively, and the diagnostic yield was 5.9% for EGD and 42.7% for colonoscopy. No patient with normal colonoscopy had positive EGD ndings. Thirty-nine patients with normal EGD and colonoscopy underwent small bowel endoscopy. Mean hemoglobin and albumin levels were 10.9 and 2.7 g/dL, respectively. Push enteroscopy, BAE, and VCE were performed in 22, 20, and 11 patients with a diagnostic yield of 22.7%, 60.0%, and 45.5%, respectively. Conclusion: Colonoscopy was shown to be an essential investigation in chronic diarrhea. In contrast to western, EGD did not add benet to colonoscopy. Enteroscopy played an important role in the diagnosis of chronic diarrhea when colonoscopy was negative.


Introduction
Diarrhea is de ned by abnormal stool weight (> 200 g/day) or frequency (> 3 times/day). Four-week symptom duration is generally considered to be the cutoff for distinguishing acute from chronic diarrhea. [1,2] Chronic diarrhea was estimated to affect approximately 5% of the Western population. [1] Functional disorder (e.g., irritable bowel syndrome [IBS]) and in ammatory diseases (e.g., in ammatory bowel disease [IBD], microscopic colitis, and celiac disease) are the most common causes of chronic diarrhea in western countries. [3,4] In addition to taking a detailed history and performing a complete physical examination, endoscopic evaluation should be considered in patients with inconclusive diagnosis after routine blood and stool tests or who fail to respond to empirical therapy.
According to the American Society for Gastrointestinal Endoscopy guidelines, a diagnostic colonoscopy should be performed to evaluate chronic diarrhea. [5] The diagnostic yield of colonoscopy in patients with chronic diarrhea ranges from 18-31% in Western countries, and the common diagnoses are IBD or microscopic colitis. [3,4,6] Upper gastrointestinal (GI) evaluation for diseases involving the duodenum should also be considered in patients with chronic diarrhea that have negative ndings on lower endoscopy. [5] Celiac disease, giardia infection, Crohn's disease, eosinophilic gastroenteritis, Whipple's disease, and intestinal amyloid are probable diagnoses in these patients. [5] Among patients with normal esophagogastroduodenoscopy (EGD) and colonoscopy, video capsule endoscopy (VCE) was reported to have a diagnostic yield ranging from 43-54%. [7,8] Deep enteroscopy, which can obtain tissue samples, demonstrated potential for diagnosing small bowel disease in patients presenting with chronic diarrhea [9][10][11][12][13] , but data are limited.
Although gastrointestinal endoscopy is recommended to investigate chronic diarrhea in Western countries, its bene ts have infrequently been investigated in Southeast Asia. Moreover and importantly, the etiologies of chronic diarrhea in Southeast Asia differ from those in Western countries. Southeast Asia has a higher prevalence of gastrointestinal infections but a lower prevalence of celiac and IBD.
Accordingly, the aim of this study was to investigate the diagnostic utility of esophagogastroduodenoscopy (EGD), colonoscopy, and small bowel endoscopy in Thai adults with chronic diarrhea.

Study design and population
We retrospectively reviewed the medical records of consecutive patients aged 18 years or older who underwent EGD and/or colonoscopy to investigate chronic diarrhea at Siriraj Hospital, Bangkok, Thailand, from January 2008 to December 2012. We also evaluated consecutive patients with negative EGD and colonoscopy who underwent subsequent small bowel endoscopy, including push enteroscopy, balloonassisted enteroscopy (BAE), and video capsule endoscopy (VCE), from January 2018 to December 2018. We excluded patients with a personal history of underlying intestinal conditions, such as IBD or short bowel syndrome. All methods were carried out in accordance with the Declaration of Helsinki. The protocol for this study was approved by the Siriraj Institutional Review Board (SIRB) on 17 January 2019 (COA no. 045/2019). The requirement to obtain written informed consent from included patients was waived by the Siriraj Institutional Review Board due to the anonymous retrospective nature of this study.

De nitions and collected data
In this study, the de nition of chronic diarrhea was watery stool ≥ 3 times/day or ≥ 1 occurrence of mucous-bloody stool per day for more than four weeks. Patient demographic data, clinical presentations, blood and stool tests, endoscopic ndings, pathological ndings, and radiological ndings were reviewed from electronic medical records. De nite diagnoses were made based on the ndings of speci c investigations that were ordered based on suspicion of a speci c disease and response to therapy. The diagnostic yield of each endoscopic modality was calculated. Furthermore, since small bowel endoscopy is invasive and not widely available, a decision to perform small bowel endoscopy should be carefully made. Therefore, we evaluated predictive factors for small bowel mucosal diseases in patients with negative EGD and colonoscopy who underwent small bowel endoscopy in which diagnostic endoscopy would be of bene t.
Since this was a retrospective observational study, patient characteristics between those who underwent EGD and/or colonoscopy and those with negative EGD and colonoscopy who underwent small bowel endoscopy were different. As such, the EGD and/or colonoscopy patients and the small bowel endoscopy patients were analyzed separately.

Statistical methods
Continuous data are presented as mean and standard deviation if normally distributed, and as median and range or interquartile range (IQR) if not normally distributed. Categorical variables are presented as frequency and percentage. We assessed the diagnostic yield of each endoscopic modality. Comparison of the diagnostic performance of EGD to colonoscopy, and colonoscopy to colonoscopy without ileal intubation was performed using McNemar's test. Analysis to identify independent predictors of small bowel mucosal diseases in patients with negative EGD and colonoscopy who underwent small bowel endoscopy was performed based on logistic regression. A p-value < 0.05 was considered statistically signi cant. All statistical analyses were performed using SAS Statistics software (SAS, Inc., Cary, NC, USA).

Diagnostic performance of EGD and colonoscopy
A total of 272 consecutive patients who underwent EGD and/or colonoscopy to investigate chronic diarrhea were included in this analysis. Baseline characteristics are outlined in Table 1. The mean age was 54 years, and 111 (40.8%) patients were male. The mean hemoglobin and albumin levels were 11.54 g/dL and 3.76 g/dL, respectively. The median duration of symptoms at presentation was 12 weeks. Of 272 patients, 132 underwent both EGD and colonoscopy, 137 underwent only colonoscopy, and 3 underwent only EGD ( Figure 1). A diagnosis was obtained from endoscopy for 116 of 272 (42.6%) patients. The de nite diagnoses are shown in Table 1. Among the 269 patients who underwent colonoscopy, 115 (42.7%) obtained a de nite diagnosis. Isolated ileal involvement without colonic lesions Table 1 Characteristics of consecutive patients who underwent EGD and/or colonoscopy (N=272)  Table 2, nineteen patients had mucosal diseases. All of those patients were diagnosed by small bowel endoscopy, except one patient with intestinal capillariasis who was diagnosed by repeated stool examination after negative push enteroscopy. Twenty patients had non-mucosal diseases.
As shown in Figure 2, the diagnostic yield of push enteroscopy, BAE, and VCE was 22.7% (5/22), 60.0% (12/20), and 45.5% (5/11), respectively. For 17 patients with negative PE, 3 had mucosal diseases missed by PE, including 2 intestinal capillariasis and 1 small bowel Crohn's disease. Of these three, two were diagnosed by subsequent BAE, and the other one was diagnosed by repeated stool examination. For 8 patients with negative BAE, only one had a mucosal disease. This patient was diagnosed by typical VCE nding and response to anti-parasitic agents. Small bowel imaging, either small bowel follow-through (SBFT) or computed tomography (CT) abdomen, was performed before enteroscopy in 6 of 22 push enteroscopy, and in 17 of 20 balloon-assisted enteroscopy. Among the procedures that had abnormal small bowel imaging performed prior to the procedure, the diagnostic yield was 63.2% (12 of 19). The diagnostic yield was 21.1% (4 of 19) and 25.0% (1 of 4) in the procedures without small bowel imaging and in the procedures with normal small bowel imaging, respectively.
Concerning VCE, as shown in Figure 2, 3 of 11 patients underwent VCE rst. Two of those three patients had positive ndings, and those two patients subsequently underwent BAE and obtained a de nite diagnose. Seven of 11 VCE were performed after small bowel enteroscopy, and 5 of 7 were performed after negative results of small bowel enteroscopy. VCE detected an abnormality in one patient that led to a diagnosis of parasitic infection, and VCE con rmed the diagnosis of non-mucosal disease in 4 patients. The remaining 2 of 7 patients underwent VCE to evaluate disease extension after enteroscopy detected abnormal ndings. The last patient underwent only VCE and had negative ndings. Among the six patients with negative VCE ndings, none had the mucosal disease. Abbreviations: SD, standard deviation; IQR, interquartile range; NSAID, non-steroidal anti-in ammatory drug; SLE, system lupus erythematosus Table 3 shows univariate and multivariate analysis to identify factors that independently predicted small bowel mucosal diseases. Younger age and lower albumin levels were found to be signi cant predictors of mucosal disease in univariate analysis. Although multivariate analysis revealed no independent predictive factors, a trend towards independent prediction was found for lower albumin levels with an odds ratio of 0.52 (95% con dence interval: 0.25-1.08; p=0.08). Table 3 Univariate and multivariate analysis to identify factors that independently predict mucosal disease

Discussion
In this study, we found the etiologies of chronic diarrhea in Thai patients to be different from those in Western patients, particularly in small bowel diseases. Among chronic diarrhea with ileocolonic causes, the common causes in our cohort were infectious diseases and in ammatory bowel diseases, whereas IBD and microscopic colitis are the common causes in Western countries. In small bowel diseases, the most common cause of chronic diarrhea in our cohort was parasitic infections. Interestingly, we found no celiac disease, which is the common cause of small bowel disease in Western countries.
Similar to Western countries, our study showed that colonoscopy had high diagnostic performance in patients with chronic diarrhea. The diagnostic yield was 42.7%, which is comparable to the values reported by several previous studies (range: 10.0% to 49.5%). [3,[14][15][16][17] Furthermore, our study showed that the terminal ileum should be accessed because the diagnostic yield would have been decreased from 42.7% to 39.4% if the terminal ileum had not been intubated. Makkar, et al. reported that the diagnostic yield was 15.0% when colonoscopy was performed without ileal intubation, and the yield increased to 16.9% when performed with ileoscopy. [18] American Society for Gastrointestinal Endoscopy guideline recommends EGD for chronic diarrhea workup due to its potential for diagnosing celiac disease. [5,19] However, the prevalence of celiac disease is low in Southeast Asia [20] ; therefore, EGD had a diagnostic yield of only 5.9% in this study. Furthermore, all positive ndings were also detected if ileocolonoscopy was performed, which means that EGD conferred no additional diagnostic bene t when combined with colonoscopy. This nding suggests that EGD should not be routinely performed to investigate chronic diarrhea in our region.
Interestingly, the involved small bowel segments causing chronic diarrhea in our cohort were at more distal segments that are unreachable by EGD. Nineteen patients with negative EGD and colonoscopy who underwent small bowel endoscopy were found to have mucosal diseases. Of those, 5 were diagnosed by push enteroscopy, 12 by balloon-assisted enteroscopy, 1 by typical VCE nding of intestinal capillariasis, and 1 by repeated stool examination. The diagnostic yield of push enteroscopy, BAE, and VCE was 22.7%, 60.0%, and 45.5%, respectively. Push enteroscopy could be considered instead of EGD when diarrhea from small bowel lesions is suspected in our region despite having the lowest diagnostic yield among the three modalities since most general gastroenterologists can perform it, and it is less invasive and less expensive than BAE. Balloon-assisted enteroscopy, which can more deeply access the small bowel compared to push enteroscopy, was reported the have a diagnostic yield of 55.0% to 73.5% in previous studies. [10][11][12][13] Similarly, the diagnostic yield of BAE in our study was 60%. It is essential to note that small bowel imaging studies could contribute to the high diagnostic yield of BAE since 17 of 20 BAE had small bowel imaging performed before BAE to localize the lesions. Furthermore, in 12 patients whose diagnoses were obtained by BAE, almost all (11 of 12) had lesion localization by small bowel imaging studies before the endoscopy. For VCE, the diagnostic yield in this study was 45%, which is comparable to previous studies. [7,8] The major limitation of VCE is its inability to obtain tissue sampling. However, our study showed that VCE could help to guide the abnormal ndings prior to BAE and exclude small bowel mucosal lesions if the results were normal.
Our study showed that small bowel imaging studies, either SBFT or CT abdomen, should be considered as a supplementary investigation to localize the lesion and guide which endoscopic modality should be performed. The diagnostic yield of small bowel enteroscopy was higher if those procedures were performed with guidance from small bowel imaging. Based on the results of this study, we propose a diagnostic diagram for patients with chronic diarrhea in our region, as shown in Figure 3.

Strengths and limitations
The strength of this study is that it is the rst to report the diagnostic performance of each endoscopic modality in chronic diarrhea in an area with a high prevalence of infections and a low prevalence of celiac disease. We also investigated the role of small bowel imaging in the diagnosis of chronic diarrhea.
The most notable limitation is our study's retrospective design, which made it impossible to perform all evaluated modalities in all patients. Another limitation is that our data were collected from a single center.

Conclusion
In Thailand, colonoscopy is the mainstay investigation in patients with chronic diarrhea. EGD was found to have a low diagnostic yield, and it did not show added bene t when combined with colonoscopy because most of the small bowel lesions were in the jejunum and ileum, which are unreachable by EGD.
Enteroscopy, which has more potential for reaching the abnormal small bowel segments than EGD, Availability of data and materials The datasets generated and/or analyzed during the current study are not publicly available due to our center's patient con dentiality policies, but they may be made available by the corresponding author to appropriate parties upon reasonable request. Proposed diagnostic algorithm for patients with chronic diarrhea