First study to assess the reliability of commonly used pain scales in children with disorders of gut‐brain interaction

There are no validated measures to assess chronic abdominal pain (AP) in clinical trials of children with disorders of gut–brain interaction (DGBIs). Currently used AP measures are extrapolated from studies on adults or children with acute AP. The primary aim of the study was to assess the commonly used pain scales in children with DGBIs. The secondary aim of the study was to compare specific pain measures with the overall subjective assessment of AP well‐being in children.


| INTRODUC TI ON
Studies from across the globe have demonstrated that abdominal pain (AP) is a common complaint among school-aged children.
One study from the United States followed school children for 16-24 weeks and found that an average of 38% of them reported AP each week, and more than 50% reported pain that persisted for >4 weeks. 1 AP is also the most common complaint in children presenting to the pediatric gastroenterologist. 2,3 Children with AP are more likely to have increased school absenteeism, poorer quality of life, functional disability, and psychological distress. 4 The long-term outcomes of the management of disorders of gut-brain interaction (DGBIs) in children are also suboptimal. 5 This underscores the need for clinical trials on novel treatments.
A limitation of all randomized controlled trials (RCTs) in children with DGBIs is the absence of validated pain measures. As such, we rely on measures extrapolated from the literature of acute pain in children or adult based studies. 6 Five self-report measures of pediatric pain intensity were recommended for use in systematic reviews or by expert panels. 4 Of these, only three have been used in RCTs for the treatment of chronic pediatric AP. These scales include the numeric rating scale (NRS), the visual analog scale (VAS), and the Faces Pain Scale (FPS), or more recently, the updated version, the Faces Pain Scale Revised (FPS-R). 7 Each of these scales has their own set of rules for their use and limitations and it remains to be answered if they can be used interchangeably, and whether the results of studies using different scales can be interpreted similarly. There have been no studies comparing the results of using each of these scales in children with chronic AP or DGBIs. The paucity or lack of validation of the different pain scales hinders comparisons of outcomes among clinical trials, the combination of data for meta-analysis, and the establishment of evidence-based treatment guidelines. Taking into consideration the great need for establishing recommendations for clinical trials in children with DGBIs, in 2016, the Rome Foundation issued provisional recommendations on the use of pain scales in the pediatric population. 6 These recommendations were based on EMA and FDA guidance for trials on adults with IBS and a very small number of clinical trials in children. 4,6 Some of the shortcomings of this approach included the possibility that children's understanding of each of these scales may differ from adults and may even differ across pediatric age groups. These limitations precluded issuing recommendations based on strong pediatric evidence.

| ME THODS
This was a sub-study using already existing data from a multicenter prospective crossover randomized clinical trial of non-deceptive placebo versus no treatment control group for children with functional AP or irritable bowel syndrome. 8

| Study population
Children 8-21 years of age with functional AP or irritable bowel syndrome defined per Rome III criteria were included. All had normal laboratory test results and negative lactose breath test results, or lack of response to a lactose-free diet for 14 days. Patients with organic disease were excluded.

| Study visits and pain measures
Our study assessed the correlation of daily self-reported pain severity scores between three widely used pain scales in children with chronic AP and the scores of the three pain scales with a self-report global question of well-being ( Figure 1). Two of the pain measures (VAS and NRS) were recommended for their use by the Rome IV criteria, while the FPS-R is commonly used in clinical settings and the pediatric pain literature. 9,10 The NRS is an 11-point (0-10) numerical rating scale of pain intensity, with zero representing no pain and 10 representing the highest level of pain. 11 The VAS scale asks the patient to mark on a horizontal 100-mm line how much pain they have, with the 0-mm endpoint representing no pain and 100-mm endpoint being the highest level of pain. 12 The FPS and FPS-R use 6-7 faces to represent a range from no pain (far left face) to very much pain (far right face). 13 The global improvement question was adapted from a large multicenter pediatric RCT on DGBIs: "Did your belly feel OK today?" The global improvement question was a 0-or 1-point response (yes or no). 8 During the study, children provided daily self-reports of pain using the three pain measures (1 week prior to randomization, 3 weeks placebo, and 3 weeks control for a total of 7 weeks).

| Data analysis
Statistical analysis was performed using Google Colab. Sample size was obtained based on existing data from the parent study. The demographic data and scores for the questionnaires were expressed in the form of the mean ± standard deviation (SD). The self-report pain measures were all normalized to a 10-point measure to facilitate comparisons. The correlations among the scales and the global question were measured using Pearson's correlation. In this study, the Pearson's correlations were defined as statistically significant when p-value was <0.001.

| RE SULTS
A total of 80 patients were assessed for eligibility to the prospective clinical trial. Forty-nine were excluded: 41 did not meet inclusion criteria, five refused to participate, and three were non-compliant.
Of the 31 remaining, one was excluded after being randomized in the parent study due to a car accident. All the remaining 30 participants completed the study. Twenty-five patients were recruited from Boston Children's Hospital, and five were recruited from Nationwide Children's Hospital. Mean age (SD) was 14.1 (3.4) years. The minimum age was 9 years old, and the maximum age was 21 years old. Twentyfour of the participants were female (80%), and six (20%) were male.
There were no clinically meaningful differences between patients with functional AP and those with irritable bowel syndrome, except for the expected differences in bowel movement frequency and laxative use. In total, children completed 4975 of 5880 (84.6%) possible reports.

| Correlations between reported pain scores: VAS, NRS, FPS-R
Breakdown of data points for the three pain measures and the global question as reported by patients and after normalization to a 10-point measure of the three pain scales are provided in Table 1. Correlations between all the pain measures were significant (p < 0.001) ( Table 2). The VAS and NRS had the strongest correla-   Table 2). F I G U R E 1 Daily self-report pain severity scales (VAS, NRS, FPS-R) and a global question of well-being.

| Subanalysis: stratification by age and gender
The data was also analyzed by age groups (9-14 and 15-21 years).
Data was not analyzed by gender, given the small sample size of males (20%). The strong correlations were consistent when stratified by age. This was determined by comparing the difference of Pearson correlations between each group, which was found to be minimal (Table 3).

| DISCUSS ION
To our knowledge, this is the first study to assess some of the most commonly used AP scales in children with DGBIs. Overall, our study found strong correlations between the 11-point NRS, the 0-100 mm VAS, and the FPS-R, which suggests that these three scales can be Several studies have investigated minimal critical important differences for measurement of pain; however, these studies were conducted in children with acute pain or in adults, rather than in children with chronic AP. 6 In adults, the EDA and FDA recommended an improvement in AP of at least 30% on an 11-point NRS as the primary endpoint for clinical trials of IBS. Despite the dearth of data substantiating these endpoints in children, the Rome Foundation subcommittee for pharmacological clinical trials in children with IBS adopted the adult based recommendations. 6 The subcommittee recommended that the endpoints for change should be a decrease in AP intensity of at least 30% from baseline and that this value should be at least equal to the reliable change index (RCI) for that sample. In studies where the RCI cannot be obtained, the pediatric subcommittee recommended a cut-off of 25 mm change in VAS or an 11-point NRS.
Understanding whether the scales can be used in children with chronic AP is relevant as children with recently developed pain may appraise their pain differently than children with months or years of pain, as was the case of the sample of our study that had a mean duration of pain of 3.3 years. 8 For this purpose, we assessed the face and content validity of the pain scales by correlating the pain intensity of each scale with an easy to understand question of whether their "belly felt okay or not", based on a previously used global endpoint. 8 We found a negative correlation between the in-  TA B L E 1 Self-report of pain intensity as reported in the pain scales and global question.
suggests that the global question measures more domains than AP alone and that the inclusion of global questions could add additional value to RCTs in children with DGBIs.
Our study had several limitations. First, the sample size of our study, which includes 30 children, does not ensure external validity. Another limitation was the lack of all data points, as children did not complete all pain assessments every day of the study. However, the attrition rate of this study was only 15.4%, which is within the expected range considering that this 7-week study requested children to complete 588 pain assessments. Moreover, we were able to obtain and analyze more than 4000 data points, a considerable size for a pediatric study.

| CON CLUS ION
In summary, our study fills an important gap in the pain literature and

ACK N OWLED G M ENTS
The authors thank Kevin Kotzen, who assisted in the statistical interpretation of this study.

AUTH O R CO NTR I B UTI O N S
Lee Ginton and Miguel Saps: Conception and design, statistical analysis, interpretation of data, drafting and revision of the manuscript.
Samuel Nurko and Carlo Di Lorenzo: acquisition of data.