A FEASIBILITY STUDY OF CONDUCTING OUTCOME ANALYSIS FOR PEDIATRIC ULCERATIVE COLITIS USING MULTICENTER ELECTRONIC HEALTH RECORDS

Background: The low fermentable oligosaccharide, disaccharide, monosaccharide, and polyol (FODMAP) diet, which restricts FODMAP intake, currently is used as a treatment for pain-related disorders of gut-brain interaction (DGBI). A FODMAP Gentle diet, which is less restrictive, has been proposed as a pediatric-friendly alternative to a complete low FODMAP diet (PMID: 36297053).
Objectives: In a clinical sample of children with DGBI from the US to: 1) Characterize the high FODMAP foods consumed; and 2) Determine the degree to which a FODMAP Gentle diet would restrict commonly consumed high FODMAP foods
Methods: Children ages 7-13 years with DGBI enrolled in an ongoing randomized controlled trial based in the US completed 3-day baseline food records. High FODMAP foods in the participants’ diets were identified. The foods were characterized further by the food groups/categories to which the FODMAPs belonged. Whether the FODMAP Gentle diet would restrict the consumed food also was determined.
Results: Eighty-six participants (53, 61.6% female) with a median (IQR) age of 10.5 (9.1-12) years were studied. A total of 1609 high FODMAP foods were consumed by the participants and each participant consumed a median (IQR) of 6 (4.7-7.9) high FODMAP foods per day. High FODMAP foods containing fructans were most prevalent in the diet, followed by those containing lactose and fructose (Table 1). Wheat-based high FODMAP foods were the most prevalent food category consumed, followed by dairy, fruits/fruit juices, beverages, and condiments (Table 1). 1078 (67%) of the high FODMAP foods consumed were ultra-processed and 274 (17%) contained high fructose corn syrup as a major ingredient. High FODMAP foods were categorized into 79 food categories with milk, breaded animal protein, and sodas being the most frequently consumed (Table 2). 1,302 (81%) of consumed high FODMAP foods would be restricted on a FODMAP Gentle diet.
Conclusion: Children with DGBI in the US frequently consume high FODMAP foods. The most consumed FODMAPs are fructans followed by lactose and fructose related to the intake of wheat-based foods, dairy, fruits/fruit juices, and beverages. The vast majority of high FODMAP foods commonly consumed would be restricted on a FODMAP Gentle diet.

Background: The use of electronic health records (EHRs) has received attention as a method to conduct novel outcome research and generate real-world evidence at lower cost than traditional clinical trials. Aim: Evaluate utility of routinely collected EHR data from a Patient-Centered-Outcome-Research-Network (PCORnet) network to identify clinical, demographic, and lab characteristics that are associated with increased risk of treatment escalation (TE) in children newly diagnosed with UC. Methods: Multicenter deidentified EHR data (01/2010 – 12/2021) were collected from the Greater Plain Collaborative (GPC) of13 healthcare systems in the Midwest. Eligible pediatric population with UC was selected using ICD codes for UC excluding chronic proctitis. Standard treatment regimens of mesalamine or oral/IV corticosteroids, use of biologics and/or immunomodulators (IM), as well as colectomy (COL) were identified using RXNORM and CPT codes. Missing information on initial therapy was exclusionary. TE defined as a composite endpoint of receiving biologics and/or IM and/or COL after initial standard treatment. Demographic characteristics were compared against the prospective pediatric UC cohort from the PROTECT Study (NCT01536535). Relevant clinical (height, weight, BMI, hospitalization) and lab (hemoglobin, serum albumin, erythrocyte sedimentation rate, C-reactive-protein, platelet count, total white blood cell count and differential) characteristics at baseline were extracted based on domain knowledge using ICD and LOINC codes. Odds ratios of univariate and multivariable logistic regression and hazard ratios of univariate and multivariable cox proportional survival models were applied to identify potential discriminative characteristics associating with TE and time to TE, respectively. Missing lab values were imputed using multivariate imputation by chained equation. Results: 509 eligible pediatric UC patients (4-17 yrs) started with standard treatment were identified and compared to the PROTECT cohort (Table 1) being comparable except for lower rate of hospitalization at baseline. 219/509(42%) had TE at median[P25, P75] days of 95[14,338] since initial treatment, 20/509 (4%) underwent COL at median[P25,P75] days of 85[9,255]. The final logistic regression and cox survival model both showed moderate discriminative power to predict TE. The common positive predictors for TE and time-to-TE were high monocyte proportion and platelet counts. Race of Black/African American, BMI z-score, hemoglobin level, lymphocyte proportion, albumin were negatively associated with TE and time-to-TE (Table 2). Conclusion: This study demonstrates that multicenter EHR data can be used to identify a trial-comparable study sample of potentially larger size and to identify clinically meaningful endpoints for conducting outcome analysis and generating real-world evidence.