Pharmacologic Treatment in Functional Abdominal Pain Disorders in Children: A Systematic Review

PubMed, Medline, Embase, PsycINFO, and Cochrane Library (including Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effect, and Cochrane Central Register of Controlled Trials) were searched from inception to February 2021. To identify unpublished or ongoing studies, the ClinicalTrials.gov register, the World Health Organization International Clinical Trials Registry Platform Search Portal, and the Current Controlled Trials metaRegister of Controlled Trials active registers were searched. To identify relevant articles and reviews missed by the search strategies, the reference lists from reviewed articles were searched by hand. The full search strategies are available on request. The protocol was registered at the International Prospective Register of Systematic Reviews (identifier CRD42020159847).

Two researchers (R.R. and C.M.A.B.) independently reviewed the titles and abstracts yielded by the search using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia). Duplicated records were removed. In case of interresearcher disagreements, a third investigator (M.G.) was consulted. Inclusion and exclusion criteria are presented in Table 1. Outcome measures were identified according to the core outcome set (COS) for FAPDs.¹⁶  There were no language restrictions. All potentially relevant studies were retrieved in full.

Two review authors (R.R. and C.M.A.B.) independently performed data extraction from all included studies, using a predesigned data extraction form containing items on study details (author, publication year, country), participants (subjects, age, sex, disease and definition), inclusion and exclusion criteria of the study, intervention characteristics (type and dose of pharmacologic treatment), control characteristics (no intervention, placebo, or other pharmacologic interventions, including dose and details), total number of patients originally assigned to each intervention group (N patients/controls), outcome measures, instruments and results (type of outcome measures used, time of assessment, and length of follow-up), and adverse events.

The risk of bias of all included studies was independently assessed by the same authors by using the Cochrane risk-of-bias tool.¹⁷  Bias was assessed as a judgment (high, low, or unclear) for individual elements from 5 domains (selection, performance, attrition, reporting, and other). For each outcome, the overall quality of evidence was assessed by using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach (Supplemental Tables 6–15).^18,19  Any interresearcher disagreements in the assessments were resolved by a third investigator (M.G.).

Dichotomous outcomes were reported as odds ratios or relative risks (RRs) along with 95% confidence intervals (CIs). Continuous outcomes were reported as mean differences (MDs) with 95% CIs. In case of crossover trials, only data from the first phase of the study (ie, before the crossover occurred) were extracted.

A total of 1989 potentially relevant articles and abstracts were identified (Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2009 flow diagram, Fig 1). After removal of duplicates (n = 196) and screening of titles and abstracts, 55 full-text articles were assessed for eligibility. In total, 25 articles did not meet inclusion criteria and were excluded for various reasons. We identified 11 systematic reviews^15,20–29  and 6 ongoing trials. Authors of the ongoing trials were contacted. We identified 3 additional relevant articles by searching by hand reference lists from these systematic reviews.^30–32  In total, 17 articles were included for analysis.^30–46  Details of excluded studies are shown in Supplemental Tables 16–18.

Studies reporting 1197 children aged 3 to 18 years of age were included for analysis. Sample sizes ranged from 14 to 132 patients, and follow-up ranged from 2 weeks to 1 year. In 5 trials, researchers investigated antispasmodics compared with placebo^{30,32,33,35,40} ; in 3 trials, researchers evaluated treatment with antidepressants^34,37,42 ; in 2 trials, researchers studied antibiotics^43,44 ; and in 2 trials, researchers compared antihistaminic treatment versus placebo.^36,39  In the remaining studies, researchers evaluated treatment with 5-HT4-receptor agonist (tegaserod),⁴⁶  H₂-receptor antagonist (famotidine),³⁸  antiemetic (domperidone),⁴⁵  melatonine,³¹  and buspirone.⁴¹  No studies were included with laxatives, antidiarrheal agents, analgesics, antimigraine agents, and serotonergic agents as treatment.

In the majority of randomized controlled trials (RCTs), researchers did not present results with absolute numbers; these studies could not be included in the meta-analysis. Because of heterogeneity and limited reporting, no further meta-analysis was possible. The characteristics of the included studies are presented in Table 2.

Overall, 8 of the 17 studies (47%) were at high risk of bias in at least 1 domain. High risk-of-bias ratings were given for lack of blinding of participants and outcome assessors, incomplete outcome data, and selective reporting. Other bias was present in 1 study because the dose was changed if there was no response after 4 weeks of treatment, and after 2 months the intervention was changed.³⁹  Furthermore, risk of bias was evaluated to be unclear as a result of inadequate reporting of methods in 11 of 17 studies (65%). Only 4 studies had low risk of bias across all domains (Figs 2 and 3). Detailed information about the risk of bias for the included studies is presented in Supplemental Tables 19–27.

The overall certainty of evidence based on the GRADE system was very low to low, with reasons for downgrading of certainty presented in Supplemental Tables 6–15.

All reported primary outcomes will be described per pharmacologic intervention. Secondary outcomes of all studies are described in Table 3. All (serious) adverse events are shown in a separate table (Table 4).

Three double-blind randomized placebo-controlled trials including 223 participants met the prespecified inclusion criteria.^34,37,42  Bahar et al⁴²  and Saps et al³⁷  investigated the comparison of amitriptyline versus placebo, and Roohafza et al³⁴  randomly assigned patients to receive either citalopram or placebo.

All studies reported treatment success as the primary outcome. In their study, Bahar et al⁴²  used overall improvement in quality-of-life scores, finding an improvement in 7of 18 in the amitriptyline group versus 0 of 17 in the placebo group; however, the baseline scores in the amitriptyline group were significantly lower than the placebo group. Saps et al³⁷  reported overall response to treatment as reported by patients, with 27 of 46 in the amitriptyline group and 23 of 44 in the placebo group reporting they felt better at the end of the study. Finally, Roohafza et al³⁴  reported an outcome of ≥2 point reduction in pain scores and found 31 of 59 in the citalopram group versus 23 of 56 in the placebo group responded.

Bahar et al⁴²  reported pain frequency scores using a visual analog scale (VAS), in which no significant differences were apparent. Pain intensity scores were reported in the studies of Bahar et al⁴²  and Saps et al,³⁷  with both studies using a VAS. No statistically significant differences were found between the two groups.

Bahar et al⁴²  reported no withdrawals due to adverse events. Saps et al³⁷  reported 3 withdrawals due to adverse events (fatigue [n = 1], rash and headaches [n = 1] [amitriptyline], and dizziness [n = 1] [placebo]), and none of the adverse events were considered to be serious. Roohafza et al³⁴  reported 5 withdrawals due to adverse events (drowsiness [n = 1], dizziness [n = 3], and nausea [n = 1]).

Five RCTs (n = 495, age 4–18 years), of which 2 trials used peppermint oil and 3 used drotaverine, mebeverine, or trimebutine, met the prespecified inclusion criteria and were included in this systematic review.^{30,32,33,35,40} 

In 3 studies, researchers predefined the primary outcome “treatment success.” Karabulut et al³²  reported overall clinical recovery, with 37 of 39 in the trimebutine group and 8 of 39 in the group with no treatment (P < .0001). Kline et al³³  concluded that 71% (peppermint oil group) versus 43% (placebo group), respectively, reported improvements in the change of symptom scale (P < .001). Pourmoghaddas et al³⁵  showed that treatment response was reported in 32 of 59 (41%) in the mebeverine group compared with 23 of 56 (30%) in the placebo group (P = .117). Asgarshirazi et al⁴⁰  and Narang et al³⁰  did not predefine treatment success.

Pain severity was reported by Kline et al.³³  The mean severity of pain symptoms in the peppermint oil group was significantly lower than that in the placebo group (Wilcoxon Signed-Ranks: t [60] = 1.99; P < .03). Narang et al³⁰  reported a significant reduction of pain episodes in the drotaverine group compared with the placebo group (mean [SD]: 10.3 [14] vs 21.6 [32.4]; P = .01). The study by Asgarshirazi et al⁴⁰  revealed that improvement in pain severity in the peppermint oil group (3.11 ± 1.36) was significantly better than in the lactol (3.93 ± 1.06; P = .373) and placebo (4.24 ± 1.33; P = .001) groups. Pain duration and frequency decreased significantly more in the peppermint oil group (respectively, 26.17 ± 11.61 and 2.00 ± 0.98) than in the lactol (respectively, 37.06 ± 25.51 [P = .012] and 2.34 ± 0.87 [P = .0001]) and placebo groups (respectively, 51.60 ± 23.74 [P = .0001] and 3.40 ± 1.41 [P = .0001]).³⁷ 

In 4 studies (n = 377), researchers reported withdrawals due to adverse events.^30,33,35,40  Narang et al³⁰  reported 1 discontinuation due to urticaria in the drotaverine group, the study investigating mebeverine reported 3 withdrawals due to adverse events (drowsiness and nervousness [n = 2], nausea [n = 1]),³⁵  and in the 2 peppermint oil studies,^33,40  no patients discontinued the interventions because of adverse events.

Two double-blind randomized placebo-controlled trials including 115 participants met the prespecified inclusion criteria.^43,44  Collins et al⁴³  investigated rifaximin versus placebo and Heyland et al⁴⁴  randomly assigned patients to receive either trimethoprim-sulfamethoxazole (TMP-SMX) or placebo.

Treatment success was not defined in both studies.^43,44 

Pain intensity scores were reported in both studies by using a VAS.^43,44  Collins et al⁴³  reported no significant differences apparent (absolute scores were lacking). Heyland et al⁴⁴  reported no significant differences between the 2 groups with a decrease from 6.9 to 4.1 in the TMP-SMX group and 7.4 to 3.0 in the placebo group.

In both studies (n = 115), the authors predefined withdrawals due to adverse events as outcome measure. Collins et al⁴³  reported 1 withdrawal from the study because of abdominal pain after taking 1 day of rifaximin. Heyland et al⁴⁴  found no withdrawals due to adverse events.

Two RCTs (n = 52, age 4–13 years), one using cyproheptadine and one using pizotifen, met the prespecified inclusion criteria and were included in this systematic review.^36,39 

Sadeghian et al³⁶  reported treatment success as a child’s assessment on response to treatment defined as “no pain/become better.” A total 13 of 15 children in the cyproheptadine group versus 5 of 14 children in the placebo group reported no pain/become better at the study’s end (P = .005).

After 2 weeks of treatment, children in the cyproheptadine group reported significantly more reduction in pain frequency (86.7 vs 35.7%; P = .002) and pain intensity (86.7 vs 28.6%; P = .001) compared with the placebo group.³⁶  In the study by Symon and Russell,³⁹  children in the pizotifen group reported a significant improvement of the drug on the index of severity and index of misery (severity: MD = −16.21 [95% CI −26.51 to −5.90], P = .005; misery: MD = −56.07 [95% CI −94.07 to −18.07], P = .007) and fewer days of abdominal pain compared with the placebo group (MD = 8.21 [95% CI 2.93 to 13.48], P = .005).

In both studies, researchers reported no withdrawals due to adverse events.^36,39 

One double-blind randomized placebo-controlled trial including 100 participants met the prespecified inclusion criteria. Karunanayake et al⁴⁵  investigated the comparison of domperidone versus placebo.

Although the number of children who reported success was higher in the domperidone group (22 of 50) than in the placebo group (14 of 50), differences were not statistically significant.

The difference in percentage of reduction of pain intensity scores was statistically significant between the 2 groups in favor of the intervention (54% vs 30%; P = .008).

No withdrawals due to adverse events were reported.⁴² 

One RCT including 48 participants met the prespecified inclusion criteria. Khoshoo et al⁴⁶  investigated the comparison of group A (polyethylene glycol 3350 oral solution) versus group B (treatment with polyethylene glycol 3350 and tegaserod [5-HT4 agonist]).

Khoshoo et al⁴⁶  reported a successful outcome as ≥3 point reduction in pain scores and found this success in 14 of 21 in the tegaserod and laxative group (group B) versus 5 of 27 in the polyethylene glycol 3350 group (group A) (P < .05).

Khoshoo et al⁴⁶  reported pain intensity scores using a VAS. The difference in reduction of pain intensity scores was statistically significant between the 2 groups in favor of the tegaserod group (P < .05).

Withdrawals due to adverse events were not adequately reported.⁴⁶ 

In a randomized crossover trial, See et al³⁸  compared famotidine versus placebo. In total, 25 children (aged 5–18 years) with a diagnosis of FD were included.

Treatment success was defined as “become better,” by using a self-reporting global improvement in symptoms scale. In total, 66.7% children receiving famotidine significantly improved compared with 15.4% receiving a placebo (odds ratio 11.0 [95% CI 1.6 to 75.5]; P = .015).

Abdominal pain was assessed by using a combined pain score (pain frequency, pain intensity, and peptic index). No significant difference in abdominal pain scores was found between both groups.³⁸ 

No withdrawals due to adverse events were reported.³⁸ 

One double-blind randomized placebo-controlled trial including 117 participants met the prespecified inclusion criteria. Badihian et al⁴¹  investigated buspirone (azapirone: a group of neuromodulators targeting 5-HT serotonin receptors) versus placebo.

Analysis revealed no significant differences in treatment success rates between the buspirone and placebo groups (47.5% vs 48.3%; P = .929).

Improved pain scores were reported in both groups after 4 weeks of treatment.³⁸  No significant differences in improvement in pain scores were found between the 2 groups at 4 weeks (buspirone 1.50 (±1.24) vs placebo 1.60 (±1.25); P = .708).

Two participants in the placebo group and 3 patients in the buspirone group withdrew from the study.³⁸  No details were given.

One double-blind randomized placebo-controlled crossover trial comparing melatonin with placebo, including 14 participants, met the prespecified inclusion criteria.³¹  Only data before the crossover occurred were extracted.

Because information regarding 1 subject on initial treatment was lacking, only data of 11 participants before the crossover occurred could be extracted. A positive clinical response (a response grade of ≥3) was achieved in 1 of 6 (17%) subjects on initial treatment with melatonin versus 2 of 5 (40%) subjects initially treated with placebo (P = not significant).

Withdrawals due to adverse events were not reported.⁴⁶ 

We systematically reviewed 17 articles to determine the efficacy and safety of pharmacologic interventions in children with FAPDs. This systematic review clearly reveals the scarcity of high-quality, placebo-controlled trials. When treatment success is used as the primary end point, peppermint oil, cyproheptadine, and tegaserod might be potential effective and safe treatments, but well-designed intervention studies are needed before this conclusion can be made. In addition, there was no evidence that any other drug treatment has a significant role in the treatment of FAPDs. Therefore, the current evidence is insufficient to recommend any specific pharmacologic compound to treat FAPDs in children.

The findings of the current study are in line with our previous review.¹⁵  Although the previous systematic review included only 6 studies (compared with 17 studies in this update), evidence to support the use of any pharmacologic compound in daily practice of FAPDs is still lacking. An explanation is that the number of studies in which researchers assess the different compounds is often limited to 1 or 2 intervention trials and that the majority of studies only included a small patient group. Interestingly, efficacy studies in adults with FAPD revealed different results.²⁸  In a systematic review and network meta-analysis, researchers concluded that antispasmodic agents, peppermint oil, and tricyclic antidepressants (TCAs) were significantly more effective than placebo in adults with IBS.⁴⁷  In studies in adults with IBS on TCAs, researchers found a positive treatment effect of this group of drugs, whereas in studies of pediatric IBS, researchers found no evidence to recommend the use of antidepressants.^28,48 

In recent years, a large number of new pharmacologic treatments have been developed and reveal promising results for future treatments of pediatric FAPDs.⁴⁹  In the population of adults with FAPDs, new mixed µ-opioid receptor agonist and δ-opioid receptor antagonist such as eluxadoline or plecanatide, a guanylyl cyclase agonist, have revealed their efficacy and safety for IBS with diarrhea and irritable bowel syndrome with constipation (IBS-C), respectively.^50–52  Moreover, lubiprostone has revealed positive results in adults with IBS-C.⁵³  However, evidence on efficacy and safety of these treatments in the pediatric population is lacking. Recently, oral immunoglobulin revealed promising results in children with IBS-C. However, this agent is beyond the scope of this review and therefore is not included.^54,55  Trials on eluxadoline (NCT03339128), linaclotide (NCT02559817), and fecal microbiota transplant (NCT03074227) are ongoing in children with FAPDs.

A placebo might be a potential effective treatment of children with FAPDs as well.⁵⁶  This is supported by a meta-analysis, in which researchers found a pooled placebo response of 41% in this population.⁵⁷  The significant role of placebo is remarkable yet not surprising because in adult patients with IBS, similar results were found.^58–61  The placebo response consists of the “true placebo effect,” which comprises conditioning and expectations, and other factors such as regression to the mean, methodologic bias, and the natural course of the disease.^62,63  The true placebo effect especially has the potential to enhance the physician-patient relationship resulting in higher rates of treatment response.^64,65  These findings have important implications for clinical practice to improve treatment outcomes.

In 13 of the 17 studies, researchers reported on side effects; in 4 of those 13 studies, no side effects occurred during the intervention. In 9 studies, researchers reported side effects during the treatment period. However, there was no significant difference between the intervention and control groups. This is in line with another review in which researchers concluded that the risk of side effects in pediatric pharmacologic FAPD treatment is low.⁶⁶  These results are supported by a meta-analysis in the adult FAPD population, in which researchers reported that none of the drugs were more likely than placebo to lead to withdrawal from the study because of side effects.⁴⁷  Only when comparing the total number of side effects in TCA treatment compared with placebo, a significantly higher number of adverse events were reported in the intervention group.⁴⁷ 

The major strength of this review is its strict systematic methodology in line with the high-quality standards of Cochrane (previously known as the Cochrane Collaboration). First, the search strategy was developed in consultation with an information specialist from Cochrane. Second, the screening process included 2 independent reviewers. In addition, authors of included studies were contacted for additional data or explanation about their study design. Furthermore, strength of evidence was assessed by using the Cochrane risk-of-bias tool and GRADE to increase the transparency of the grading process and appropriate certainty of data to support readers in interpreting the results.

The limitations of this study are primarily related to the lack of high-quality evidence that is currently available but have no direct relationship with our review process. First, there were few trials with a low risk of bias, and there was evidence of heterogeneity between RCTs in our analyses. This was across multiple dimensions, including treatment and comparators, length of therapy, and most importantly, the choice of outcome measures. Furthermore, no subgroup analyses were possible because of the small sample size. Many of the studies are of such a small size that it is surprising that ethical approval was granted, and this mirrors a wider problem in the field.⁶⁷  Our previous systematic review examining this topic revealed similar findings in terms of the efficacy of pharmacologic treatment.¹⁵  Despite the addition of 11 trials since the publication of our previous review, the overall estimates of the efficacy and safety of these treatments remain almost identical, and this is because of the design and methodologic choices of researchers in the field. Another limitation is that most RCTs included in this review were published before the recommendations for the design of intervention trials on children with FAPDs.⁶⁸  In 2016, the Rome Foundation made recommendations for designing trials in pediatric FAPDs.⁶⁸  This committee recommends considering abdominal pain as the primary outcome, assessed with daily diaries, and this would address the issue described above. Furthermore, a large sample size, a study duration of at least 4 weeks, and a 6-month follow-up period is recommended, similarly addressing the weaknesses uncovered in this updated review. To set up well-designed (double-blind, placebo-controlled) RCTs in children is difficult because of a high placebo effect (ie, difficult to measure true difference between intervention and placebo group) and complex trial designs, in which participants often refuse to participate because of the possibility of receiving placebo.^56,57  Furthermore, funding is often scarce.⁴⁹  Despite these challenges, if studies cannot be done in this fashion, they add nothing to the overall “certainty” of evidence and are unlikely to impact guidelines and practice. Recently, the pediatric FAPD COS was created.¹⁶  This COS aimed to decrease study heterogeneity and increase the comparability of study results by measuring a standardized minimum set of outcome measures. If all future trials will use these outcome measures, as well as associated measurement instruments, it is likely that this may improve the quality of research and information and, finally, GRADE certainty and clinical decision-making.⁶⁹ 

In clinical practice, the first step in management may consist of education, reassurance, and simple dietary advice.¹⁴  Lately, there is increasing evidence for the effectiveness of nonpharmacologic treatment, such as hypnotherapy and cognitive behavior therapy.^27,49,70  These treatments are not hampered by severe adverse events and may be especially effective in children with lasting symptoms. Therefore, based on the majority of current evidence and expert opinion, nonpharmacologic therapy could be the first intervention attempt in pediatric FAPDs. However, because the pathogenesis of these disorders remains unclear in children, the optimal treatment strategy is not known. To date, it is preferable to discuss both pharmacologic therapies and nonpharmacologic options during shared decision consultation. This can be used to make a tailor-made approach for each patient.

On the basis of the current evidence, it is not possible to recommend any specific pharmacologic agent for the treatment of FAPDs in children because of the low quality of included studies. However, agents such as antispasmodics or antidepressants can be discussed in daily practice because of their favorable treatment outcomes and considering the lack of important side effects. Because a large proportion of children continues to experience FAPD-related symptoms in adulthood, high-quality studies on pharmacologic treatments in pediatric FAPDs are necessary to provide adequate treatment. In future intervention trials, researchers must include homogenous outcome measures and instruments, as recommend by the pediatric FAPD COS, a large sample size, and long-term follow-up.

We acknowledge the great work and contributions from Yuhong Yuan. She developed the search strategies (information specialist, Cochrane Gut Group) and performed the literature search. We are grateful to the authors who kindly responded to our request for clarification of the protocol and additional data on the trials in which they were involved.

AM

abdominal migraine

CI

confidence interval

COS

core outcome set

FAPD

functional abdominal pain disorder

FAP-NOS

functional abdominal pain–not otherwise specified

FD

functional dyspepsia

GRADE

Grading of Recommendations, Assessment, Development and Evaluation

IBS

irritable bowel syndrome

IBS-C

irritable bowel syndrome with constipation

MD

mean difference

RCT

randomized controlled trial

RR

relative risk

TCA

tricyclic antidepressant

TMP-SMX

trimethoprim-sulfamethoxazole

VAS

visual analog scale