Probiotika zur Behandlung von funktionellen Bauchschmerzen bei Kindern
Abstract
Background
Functional abdominal pain is pain occurring in the abdomen that cannot be fully explained by another medical condition and is common in children. It has been hypothesised that the use of micro‐organisms, such as probiotics and synbiotics (a mixture of probiotics and prebiotics), might change the composition of bacterial colonies in the bowel and reduce inflammation, as well as promote normal gut physiology and reduce functional symptoms.
Objectives
To assess the efficacy and safety of probiotics in the treatment of functional abdominal pain disorders in children.
Search methods
We searched MEDLINE, Embase, the Cochrane Central Register of Controlled Trials (CENTRAL) and two clinical trials registers from inception to October 2021.
Selection criteria
Randomised controlled trials (RCTs) that compare probiotic preparations (including synbiotics) to placebo, no treatment or any other interventional preparation in patients aged between 4 and 18 years of age with a diagnosis of functional abdominal pain disorder according to the Rome II, Rome III or Rome IV criteria.
Data collection and analysis
The primary outcomes were treatment success as defined by the primary studies, complete resolution of pain, improvement in the severity of pain and improvement in the frequency of pain. Secondary outcomes included serious adverse events, withdrawal due to adverse events, adverse events, school performance or change in school performance or attendance, social and psychological functioning or change in social and psychological functioning, and quality of life or change in quality life measured using any validated scoring tool. For dichotomous outcomes, we calculated the risk ratio (RR) and corresponding 95% confidence interval (95% CI). For continuous outcomes, we calculated the mean difference (MD) and corresponding 95% CI.
Main results
We included 18 RCTs assessing the effectiveness of probiotics and synbiotics in reducing the severity and frequency of pain, involving a total of 1309 patients.
Probiotics may achieve more treatment success when compared with placebo at the end of the treatment, with 50% success in the probiotic group versus 33% success in the placebo group (RR 1.57, 95% CI 1.05 to 2.36; 554 participants; 6 studies; I2 = 70%; low‐certainty evidence).
It is not clear whether probiotics are more effective than placebo for complete resolution of pain, with 42% success in the probiotic group versus 27% success in the placebo group (RR 1.55, 95% CI 0.94 to 2.56; 460 participants; 6 studies; I2 = 70%; very low‐certainty evidence). We judged the evidence to be of very low certainty due to high inconsistency and risk of bias.
We were unable to draw meaningful conclusions from our meta‐analyses of the pain severity and pain frequency outcomes due to very high unexplained heterogeneity leading to very low‐certainty evidence.
None of the included studies reported serious adverse events. Meta‐analysis showed no difference in withdrawals due to adverse events between probiotics (1/275) and placebo (1/269) (RR 1.00, 95% CI 0.07 to 15.12). The results were identical for the total patients with any reported adverse event outcome. However, these results are of very low certainty due to imprecision from the very low numbers of events and risk of bias.
Synbiotics may result in more treatment success at study end when compared with placebo, with 47% success in the probiotic group versus 35% success in the placebo group (RR 1.34, 95% CI 1.03 to 1.74; 310 participants; 4 studies; I2 = 0%; low certainty). One study used Bifidobacterium coagulans/fructo‐oligosaccharide, one used Bifidobacterium lactis/inulin, one used Lactobacillus rhamnosus GG/inulin and in one study this was not stated).
Synbiotics may result in little difference in complete resolution of pain at study end when compared with placebo, with 52% success in the probiotic group versus 32% success in the placebo group (RR 1.65, 95% CI 0.97 to 2.81; 131 participants; 2 studies; I2 = 18%; low‐certainty evidence).
We were unable to draw meaningful conclusions from our meta‐analyses of pain severity or frequency of pain due to very high unexplained heterogeneity leading to very low‐certainty evidence.
None of the included studies reported serious adverse events. Meta‐analysis showed little to no difference in withdrawals due to adverse events between synbiotics (8/155) and placebo (1/147) (RR 4.58, 95% CI 0.80 to 26.19), or in any reported adverse events (3/96 versus 1/93, RR 2.88, 95% CI 0.32 to 25.92). These results are of very low certainty due to imprecision from the very low numbers of events and risk of bias.
There were insufficient data to analyse by subgroups of specific functional abdominal pain syndrome (irritable bowel syndrome, functional dyspepsia, abdominal migraine, functional abdominal pain ‐ not otherwise specified) or by specific strain of probiotic.
There was insufficient evidence on school performance or change in school performance/attendance, social and psychological functioning, or quality of life to draw conclusions about the effects of probiotics or synbiotics on these outcomes.
Authors' conclusions
The results from this review demonstrate that probiotics and synbiotics may be more efficacious than placebo in achieving treatment success, but the evidence is of low certainty. The evidence demonstrates little to no difference between probiotics or synbiotics and placebo in complete resolution of pain. We were unable to draw meaningful conclusions about the impact of probiotics or synbiotics on the frequency and severity of pain as the evidence was all of very low certainty due to significant unexplained heterogeneity or imprecision.
There were no reported cases of serious adverse events when using probiotics or synbiotics amongst the included studies, although a review of RCTs may not be the best context to assess long‐term safety. The available evidence on adverse effects was of very low certainty and no conclusions could be made in this review. Safety will always be a priority in paediatric populations when considering any treatment. Reporting of all adverse events, adverse events needing withdrawal, serious adverse events and, particularly, long‐term safety outcomes are vital to meaningfully move forward the evidence base in this field.
Further targeted and appropriately designed RCTs are needed to address the gaps in the evidence base. In particular, appropriate powering of studies to confirm the safety of specific strains not yet investigated and studies to investigate long‐term follow‐up of patients are both warranted.
PICO
Zusammenfassung in einfacher Sprache
Probiotika für die Behandlung von funktionellen Bauchschmerzen bei Kindern
Probiotika bei Bauchschmerzen von Kindern
Kernaussagen
Probiotika haben möglicherweise bei Kindern mit funktionellen Bauchschmerzen eine bessere Wirkung als Placebo (Scheinbehandlung).
Synbiotika haben möglicherweise bei Kindern mit funktionellen Bauchschmerzen eine bessere Wirkung als Placebo (Scheinbehandlung).
Was sind funktionelle Bauchschmerzen?
Funktionelle Bauchschmerzen kommen bei Kindern häufig vor. Der Begriff funktionelle Bauchschmerzen wird verwendet, wenn keine Ursache für die Symptome gefunden werden kann. Zu diesen Symptomen gehören häufige, seit mindestens sechs Monaten anhaltende Bauchschmerzen, die im täglichen Leben Probleme verursachen.
Was sind Probiotika?
Probiotika sind lebende Bakterien und Hefen, denen verschiedene gesundheitliche Vorteile nachgesagt werden. Sie werden oft als "gute Bakterien" bezeichnet. Es wird angenommen, dass Probiotika das natürliche Gleichgewicht der Bakterien im Darm unterstützen und möglicherweise die Symptome bestimmter Krankheiten verbessern können. Ihnen können auch so genannten Präbiotika (Lebensmittel, die das Wachstum dieser Bakterien und Hefen fördern) zugesetzt werden. Werden Präbiotika und Probiotika in einem Präparat kombiniert, spricht man von Synbiotika.
Was wollten wir herausfinden?
Wir wollten herausfinden, ob Probiotika und Synbiotika zur Behandlung von funktionellen Bauchschmerzen bei Kindern eingesetzt werden können und ob sie sicher sind.
Wie gingen wir vor?
Wir suchten nach Studien, in denen Probiotika oder Synbiotika mit Placebo, keiner Behandlung oder einer anderen Intervention bei Kindern im Alter von 4 bis 18 Jahren mit der Diagnose einer funktionellen Bauchschmerzerkrankung verglichen wurden. Wir fassten die Ergebnisse der Studien mit statistischen Methoden zusammen, verglichen sie und bewerteten die Vertrauenswürdigkeit in die Evidenz anhand von Faktoren wie der Studienmethodik und der Größe der Studien.
Was fanden wir heraus?
Wir fanden 18 Studien mit insgesamt 1309 Kindern, die Probiotika oder Synbiotika mit Placebo verglichen.
Wir stellten fest, dass Probiotika bei Kindern mit funktionellen Bauchschmerzen möglicherweise eine bessere Schmerzlinderung und Linderung anderer Magen‐Darm‐Probleme bewirken können als eine Placebobehandlung. Bei Kindern, die Probiotika einnahmen, wurde die Behandlung häufiger als erfolgreich bewertet als bei Kindern, die ein Placebo einnahmen. Auch bei Synbiotika wurde ein Unterschied zu Placebo festgestellt, allerdings auf der Grundlage einer geringeren Anzahl von Studien. Es lagen nicht genügend Informationen vor, um Veränderungen in der Schmerzhäufigkeit beim Vergleich von Synbiotika und Placebo zu berücksichtigen.
Wir können keine Schlüsse über die Sicherheit ziehen, da die gefundenen Hinweise auf unerwünschte oder schädliche Wirkungen von sehr niedriger Vertrauenswürdigkeit sind.
Was schränkt die Evidenz ein?
Die Evidenz für Synbiotika in diesem Review ist dadurch begrenzt, dass die Ergebnisse aus weniger Studien stammen. Was die Sicherheit betrifft, so gab es nicht genügend Fälle von unerwünschten oder schädlichen Wirkungen, um die Sicherheit von Probiotika und Synbiotika ausreichend einschätzen zu können.
Wie aktuell ist die vorliegende Evidenz?
Es wurden Studien bis Oktober 2021 berücksichtigt.
Authors' conclusions
Summary of findings
| Probiotic compared to placebo for management of functional abdominal pain disorders in children | ||||||
| Patient or population: children (4 to 18 years) with functional abdominal pain disorders | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with probiotics | |||||
| Treatment success (at study end, as reported by study authors) | Study population | RR 1.57 (1.05 to 2.36) | 554 (6 studies) | ⊕⊕⊝⊝ | — | |
| 339 per 1000 | 532 per 1000 | |||||
| Complete resolution of pain (at study end, as reported by study authors) | Study population | RR 1.55 (0.94 to 2.56) | 460 (6 studies) | ⊕⊝⊝⊝ Very low b | — | |
| 272 per 1000 | 422 per 1000 (256 to 696) | |||||
| Severity of pain (at study end, using the Faces Pain Scale) | Severity of pain using the Faces Pain Scale when comparing probiotics versus placebo: SMD ‐0.28 (95% CI ‐0.67 to 0.12) | 665 participants | ⊕⊝⊝⊝ | — | ||
| Frequency of pain (at study end, episodes per week) | Frequency of pain episodes (per week) when comparing probiotics versus placebo: MD ‐0.43 (95% CI ‐0.92 to 0.07) | 605 participants | ⊕⊝⊝⊝ | — | ||
| Withdrawals due to adverse events | Study population | RR 1.00 (0.07 to 15.12) | 544 | ⊕⊝⊝⊝ Very low e | — | |
| 4 per 1000 | 4 per 1000 | |||||
| Serious adverse events | There were no SAEs reported within these studies in either group. | 685 | ⊕⊝⊝⊝ Very low e | — | ||
| Adverse events (any) | Study population | RR 1.00 (0.07 to 15.12) | 489 | ⊕⊝⊝⊝ Very low e | — | |
| 4 per 1000 | 4 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aDowngraded one level due to inconsistency (I² = 59% for both outcomes) and one level for risk of bias. bDowngraded three levels due to very high inconsistency (I² = 70%) and risk of bias (allocation concealment, attrition and reporting bias). cDowngraded three levels due to very high inconsistency (I² = 70%) and risk of bias (reporting bias). dDowngraded one level due to risk of bias. eDowngraded two levels due to imprecision because of very low numbers of adverse event cases and one level due to risk of bias. | ||||||
| Synbiotic compared to placebo for management of functional abdominal pain disorders in children | ||||||
| Patient or population: children (4 to 18 years) with functional abdominal pain disorders | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with synbiotic | |||||
| Treatment success (at study end, as reported by study authors) | Study population | RR 1.34 (1.03 to 1.74) | 310 | ⊕⊕⊝⊝ | — | |
| 350 per 1000 | 469 per 1000 | |||||
| Complete resolution of pain (at study end, as reported by study authors) | Study population | RR 1.65 (0.97 to 2.81) | 131 (2 studies) | ⊕⊕⊝⊝ | — | |
| 319 per 1000 | 405 per 1000 | |||||
| Severity of pain (at study end, using the Faces Pain Scale) | Severity of pain measured using the Faces Pain Scale for synbiotics versus placebo: MD ‐0.21 (95% CI ‐0.78 to 0.37) | 319 (4 studies) | ⊕⊝⊝⊝ | — | ||
| Frequency of pain (at study end, episodes per week) | The mean in the placebo group was 3.4 | MD 1.26 lower | — | 80 (1 study) | ⊕⊝⊝⊝ | — |
| Withdrawals due to adverse events | Study population | RR 4.58 | 302 | ⊕⊝⊝⊝ | — | |
| 7 per 1000 | 31 per 1000 | |||||
| Serious adverse events | There were no SAEs reported within these studies in either group | 302 | ⊕⊝⊝⊝ | — | ||
| Adverse events (any) | Study population | RR 2.88 | 189 | ⊕⊝⊝⊝ | — | |
| 11 per 1000 | 30 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aDowngraded one level for imprecision due to low participant numbers. bDowngraded one level due to risk of bias. cDowngraded two levels due to very serious unexplained heterogeneity, and one level due to risk of bias. dDowngraded two levels for severe risk of bias, due to unclear/high risk of bias for the single study that provided data for this outcome. eDowngraded two levels due to very serious imprecision from very low event numbers, and one level due to risk of bias. | ||||||
Background
Description of the condition
The term 'recurrent abdominal pain' was introduced by Apley 1958 and describes clinically apparent, non‐organic, chronic or recurrent abdominal pain in children, with three or more episodes within three months that are severe enough to interfere with daily activities. Drossman 2006 replaced recurrent abdominal pain with the term "abdominal functional gastrointestinal disorders" (AP‐FGIDs) in the Rome III system, and described AP‐FGIDs as "chronic or recurrent abdominal pain without evidence of an organic cause". In 2016, the Rome III criteria were replaced with the more recent Rome IV criteria (Drossman 2016; Drossman 2017), updating the nomenclature to "functional abdominal pain disorders" (FAPDs) (Hyams 2016).
The Rome IV criteria divide FAPDs into the following subcategories (Hyams 2016):
-
Functional dyspepsia
-
Irritable bowel syndrome (IBS)
-
Abdominal migraine (AM)
-
Functional abdominal pain ‐ not otherwise specified (FAP‐NOS)
The diagnosis of functional dyspepsia must include one or more of the following for at least four days per month (Hyams 2016):
-
Bothersome postprandial fullness
-
Bothersome early satiation
-
Bothersome epigastric pain not associated with defecation
-
Bothersome epigastric burning not associated with defecation
-
After appropriate evaluation, the symptoms cannot be fully explained by another medical condition
These criteria should be fulfilled for the last two months before diagnosis.
The diagnosis of irritable bowel syndrome must include all of the following (Hyams 2016):
-
Abdominal pain at least four days per month associated with one or more of the following:
-
related to defecation;
-
a change in frequency of stool; and
-
a change in form (appearance) of stool.
-
-
In children with constipation, the pain does not resolve with resolution of the constipation (children in whom the pain resolves have functional constipation, not irritable bowel syndrome).
-
After appropriate evaluation, the symptoms cannot be fully explained by another medical condition.
These criteria should be fulfilled for the last three months with symptom onset at least six months before diagnosis of irritable bowel syndrome.
The diagnosis of abdominal migraine must include all of the following (Hyams 2016):
-
Paroxysmal episodes of intense, acute periumbilical, midline or diffuse abdominal pain lasting one hour or more (should be the most severe and distressing symptom).
-
Episodes are separated by periods of usual health lasting weeks to months.
-
The pain is incapacitating and interferes with normal activities.
-
Stereotypical pattern and symptoms in the individual patient.
-
The pain is associated with two or more of the following:
-
anorexia;
-
nausea;
-
vomiting;
-
headache;
-
photophobia;
-
pallor.
-
-
After appropriate evaluation, the symptoms cannot be fully explained by another medical condition.
These criteria should be fulfilled two or more times in the past 12 months.
The diagnosis of functional abdominal pain ‐ not otherwise specified (FAP‐NOS) must be fulfilled at least four times per month and include all of the following (Hyams 2016):
-
Episodic or continuous abdominal pain that does not occur solely during physiologic events (e.g. eating, menses).
-
Insufficient criteria for irritable bowel syndrome, functional dyspepsia or abdominal migraine.
-
After appropriate evaluation, the abdominal pain cannot be fully explained by another medical condition.
These criteria should be fulfilled at least two months before diagnosis.
Functional abdominal pain disorders (FAPDs) are common in children and adolescents with a worldwide pooled prevalence of 13.5% (Korterink 2015). Paediatric FAPDs have a major impact on daily life, resulting in a significantly lower quality of life and higher rates of school absenteeism (Assa 2015; Varni 2015). Moreover, patients are at higher risk for developing anxiety or depressive disorders compared to healthy school‐aged children (Newton 2019). The pathophysiological mechanisms underlying FAPDs are poorly understood and are thought to be multifactorial. Psychosocial, genetic and physiological factors, such as inflammation, poor gastric emptying, increased rectal sensitivity and altered gut microbiota, have been suggested to contribute to the development of functional abdominal pain by influencing the visceral sensitivity, gastrointestinal motility and gut‐brain axis (Korterink 2015). Paediatric FAPDs are now labelled as 'disorders of gut‐brain interaction' given that their bio‐psychosocial basis encompasses complex interactions within the gut‐brain axis (Drossman 2016). More recently, the latter is entitled as the 'microbiota‐gut‐brain axis' to reflect an increase in our understanding of the magnitude, complexity, role and interactions of the microbial populations hosted within the lumen of the gastrointestinal tract.
The management of paediatric FAPDs consists of non‐pharmacological and pharmacological interventions. The first step of treatment includes 'standard medical treatment', which contains explanation, reassurance, and simple dietary and behavioural advice (Schurman 2010). Non‐pharmacological interventions consist of dietary interventions and psychosocial interventions such as cognitive behavioural therapy (CBT) and hypnotherapy.
Description of the intervention
Probiotics are micro‐organisms which, when ingested, are thought to have beneficial effects on a person’s health. Research is ongoing into the use of probiotics for the treatment of various gastrointestinal illnesses including inflammatory pathological disorders, functional disorders and chronic non‐pathological disorders. In infants, it has been proposed that supplying probiotic bacteria can redress the balance of intestinal bacteria and provide a healthier intestinal microbiota landscape with resulting impact on transit through the gut (Savino 2013). In the context of constipation, these mechanisms have been proposed to enhance colonic peristalsis and shorten whole gut transit time (Waller 2011).
How the intervention might work
The use of micro‐organisms might change the composition of bacterial colonies in the bowel and reduce inflammation, as well as promoting normal gut physiology and thereby reducing functional symptoms. Some probiotics may influence colonic motility by softening the stool, changing secretion and absorption of water and electrolytes, modifying smooth muscle cell contractions, increasing the production of lactate and short‐chain fatty acids, and lowering intraluminal pH (Waller 2011). Additionally, as essentially a food supplement, probiotics are generally perceived as having a good safety profile, particularly when compared with other treatments.
Why it is important to do this review
As interest in probiotics for the treatment of gastrointestinal disorders is relatively new, until recently there has been a general paucity of research on the use of these agents. In adults, the evidence has been synthesised previously (Moayyedi 2010). This systematic review found that probiotics appear to be efficacious in irritable bowel syndrome, but the magnitude of benefit and the most effective species and strain remained uncertain.
A previous Cochrane Review in children found only three studies examining probiotics (Huertas‐Ceballos 2009). This review was updated in 2017 with a total of 13 probiotic studies identified, although it focused not only on probiotics but also other dietary interventions (Newlove‐Delgado 2017). This review found moderate‐ to low‐quality evidence to suggest that probiotics may be effective in improving pain in children, with issues around risk of bias, imprecision and inconsistency impacting the certainty of evidence. Additionally, the new Rome IV criteria have simplified and clarified the nomenclature and diagnostic categories in such conditions (Drossman 2017). A number of new trials have also been published.
A new review is indicated, to align with the new classifications within children and update the evidence.
Objectives
To assess the efficacy and safety of probiotics in the treatment of functional abdominal pain disorders in children.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs).
Types of participants
Participants were children between 4 and 18 years of age with a diagnosis of functional abdominal pain disorder. This is in line with the Rome IV criteria, which do not cover infants or toddlers: there is a separate set of diagnostic criteria that address this age group (Hyams 2016). Participants could include children with irritable bowel syndrome, abdominal migraine or functional abdominal pain as defined by the Rome IV criteria (Hyams 2016). We also included participants who met earlier Rome criteria. Studies including children with Hirschsprung’s disease, previous bowel surgery or complex congenital disorders were not included.
Types of interventions
We considered for inclusion studies that assessed probiotic preparations in any form (powder, liquid, capsule) through any route (either oral or rectal) as a single species or as a cocktail of multiple species or treatments (for example, symbiotic) compared to placebo, no treatment or any other interventional preparation. We also considered studies with probiotics as adjunct therapy for inclusion. Studies involving prebiotics alone were not included, as they fall into the more broad scope of dietary interventions, which is covered by another review (Newlove‐Delgado 2017).
Types of outcome measures
Primary outcomes
-
Global improvement or treatment success as defined by the primary studies.
-
Complete resolution of pain.
-
Severity of pain or change in the severity of pain.
-
Frequency of pain or change in the frequency of pain.
Secondary outcomes
-
Serious adverse events.
-
Withdrawal due to adverse events.
-
Adverse events.
-
School performance, or change in school performance or attendance.
-
Social and psychological functioning, or change in social and psychological functioning.
-
Quality of life, or change in quality life, measured using any validated measurement tool.
Search methods for identification of studies
Electronic searches
We identified relevant trials by searching the following electronic sources, from the inception of each database to 1 October 2021 (Appendix 1):
-
Cochrane Central Register of Controlled Trials (CENTRAL 2021, Issue 9) (from inception to 1 October 2021) (via Ovid);
-
MEDLINE (from 1946 to 1 October 2021) (via Ovid);
-
Embase (from 1974 to 1 October 2021) (via Ovid).
We did not restrict the searches by date or language. Studies published in a non‐English language were professionally translated in full.
Searching other resources
Reference checking
We searched the references of all included studies and relevant systematic reviews to identify studies missed by the search strategies.
Personal contacts
We contacted leaders in the field to try and identify other relevant studies. We also contacted manufacturers of probiotic agents to try and identify other studies.
Trials registries
We searched ClinicalTrials.gov (www.clinicaltrials.gov) (Appendix 1) and the World Health Organization International Clinical Trials Registry Platform (ICTRP; https://trialsearch.who.int/) (Appendix 1) to identify ongoing studies, by combining terms related to probiotics and functional abdominal pain in children.
Grey literature
We searched Google, Google Scholar and the OpenGrey Repository using the main search terms. We handsearched conference proceedings from Digestive Disease Week, United European Gastroenterology Week and the European Society for Paediatric Gastroenterology, Hepatology and Nutrition annual scientific meeting (from 2019 to 2021) to identify other potentially relevant studies that may not have been published in full. Concerns have been raised regarding the accuracy of data reported in abstract publications (Pitkin 1999). Therefore, where references to relevant unpublished or ongoing studies were identified, we made attempts to collect sufficient extra information to allow inclusion in this systematic review. Studies from the grey literature were included if sufficient data were reported to judge eligibility for inclusion. If data were incomplete, we contacted the study authors in order to verify the eligibility of the study, and we only included the study if suitable data to assess quality and outcomes were supplied.
Data collection and analysis
Selection of studies
Two authors independently screened titles, abstracts and full reports for eligibility against the inclusion criteria.
Specifically, they:
-
collated the search results using reference management software and removed any duplicate records;
-
examined titles and abstracts to remove results that were not relevant;
-
retrieved the full texts of potentially relevant reports;
-
linked together multiple reports that were found for the same study;
-
examined full‐text reports for studies that met the inclusion criteria;
-
corresponded with primary study investigators to clarify study eligibility when needed; and
-
at all stages, the authors recorded the reasons for inclusion and exclusion of studies, resolving any disagreements through reaching consensus. When consensus could not be reached, we consulted with a third author (AA).
Data extraction and management
We developed data extraction forms a priori to extract information on the relevant features and results of the included studies. Two authors independently extracted and recorded data on a predefined checklist. Extracted data included the following items:
-
characteristics of patients (age, gender, disease distribution, disease duration, activity index);
-
inclusion and exclusion criteria of studies;
-
total number of patients originally assigned to each intervention group;
-
intervention: type and amount of probiotics;
-
control: no intervention, placebo or other interventions;
-
concurrent medications; and
-
outcomes: time of assessment, length of follow‐up, type of symptom score used and adverse events.
Assessment of risk of bias in included studies
Two review authors independently assessed the risk of bias in the included studies using the Cochrane risk of bias tool (Higgins 2011). We assessed the following items: sequence generation; allocation concealment; blinding of participants, parents and health professionals; blinding of outcome assessment; incomplete outcome data; selective outcome reporting; and other potential threats to validity. We judged each domain as being at 'low', 'high' or 'unclear' risk of bias. We compared the judgements, and discussed and resolved any inconsistencies in the assessments. A third review author resolved any disagreements.
Sequence generation for randomisation
We only considered RCTs for inclusion in the review. We assessed randomisation as being at low risk of bias where the procedure for random sequence generation was explicitly described. Examples include computer‐generated random numbers, a random numbers table or coin‐tossing. Where no description was given, we contacted the authors for further information.
Allocation concealment
We assessed concealment of treatment allocation as being at low risk of bias if the procedure was explicitly described and adequate efforts were made to ensure that intervention allocations could not have been foreseen in advance of, or during, enrolment. Examples include centralised randomisation, numbered or coded containers, or sealed envelopes. Procedures considered to have a high risk of bias include alternation or reference to case record numbers or dates of birth. Where no description was given of the method of allocation concealment, we contacted the study authors and, where we did not receive a response, we assigned a judgement of unclear risk of bias.
Blinding of participants, parents and health professionals
In this context, the intervention is administered by parents as well as directly by children so, in effect, we considered them both the targets of the blinding procedures. We primarily assessed the risk of bias associated with the blinding of participants based on the likelihood that such blinding is sufficient to ensure they had no knowledge of which intervention they received. We noted the blinding of health professionals, if reported.
Blinding of outcome assessment
For each included study, we described the methods used, if any, to blind the outcome assessors from knowledge of which intervention a participant received. We judged studies to be at low risk of bias if outcome assessors were blinded, or where we considered that the lack of blinding could not have affected the results. If blinding was not done or was not possible because of the nature of the intervention, we judged the study to be at high risk of bias because it was possible that the lack of blinding influenced the results. If no description was given, we contacted the study authors for more information and if we did not receive a response we assigned a judgement of unclear risk of bias.
Incomplete outcome data
Incomplete outcome data essentially included attrition, exclusions and missing data.
We assigned a judgement of low risk of bias in the following instances:
-
If participants included in the analysis were exactly those who were randomised into the trial; missing outcome data were balanced in terms of numbers across intervention groups, with similar reasons for missing data across groups; or if there were no missing outcome data.
-
If, for dichotomous outcome data, the proportion of missing outcomes compared with the observed event risk was not sufficient to have a clinically relevant impact on the intervention effect estimate.
-
If, for continuous outcome data, the plausible effect size (mean difference) among the missing outcomes was not sufficient to have a clinically relevant impact on observed effect size.
-
If missing data have been imputed using appropriate methods.
We assigned a judgement of high risk of bias in the following instances:
-
When reasons for missing outcome data were likely to be related to the true outcome, with either an imbalance in numbers or reasons for missing data across intervention groups.
-
When, for dichotomous outcome data, the proportion of missing outcomes compared with the observed event risk was sufficient to induce clinically relevant bias in the intervention effect estimate.
-
When, for continuous outcome data, the plausible effect size (mean difference) among missing outcomes was sufficient to induce clinically relevant bias in the observed effect size.
-
When an 'as‐treated' analysis was carried out in cases where there is a substantial departure of the intervention received from that assigned at randomisation.
-
When there was a potentially inappropriate application of simple imputation.
We will assign a judgement of unclear risk of bias in the following instances:
-
When there was insufficient reporting of attrition or exclusions, or both, to permit a judgement of low or high risk of bias.
-
When the study reported incomplete outcome data.
-
When the trial did not clearly report the numbers randomised to intervention and control groups.
Selective outcome reporting
We assessed the reporting of outcomes as being at low risk of bias if all outcomes pre‐specified in the study protocol were reported in the study manuscript or secondary publications. If no protocol existed or if trial registration was retrospective, we assigned a rating of unclear risk of bias if the authors report on the outcomes described in the methods section of the study manuscript. We evaluated all study publications (primary and secondary) to ensure that there was no evidence of selective outcome reporting. If no description was given, we contacted the authors for more information and, if we did not receive a response, we assigned a judgement of unclear risk of bias. If there was evidence of selective reporting (deviation from protocol, key planned outcomes not reported), we assigned a judgement of high risk of bias.
Other potential threats to validity
We considered other potential sources of bias including early trial termination (e.g. if a study was stopped early due to a data‐dependent process) and baseline imbalance between treatment groups. We assessed the study as being at low risk of bias if it appeared to be free from such threats to validity. When the risk of bias was unclear from the published information, we attempted to contact the study authors for clarification. If this was not forthcoming, we assessed these studies as being at unclear risk of bias.
Measures of treatment effect
Dichotomous outcomes
For dichotomous outcomes, we calculated the risk ratio (RR) and corresponding 95% confidence interval (CI).
Continuous outcomes
For continuous outcomes, we calculated the mean difference (MD) and corresponding 95% CI.
Unit of analysis issues
Where cross‐over trials were included, we extracted data from the first phase of the study, if they were reported (i.e. before the cross‐over occurred). We conducted separate analyses for comparisons between probiotics versus placebo, and probiotics versus active comparator (e.g. lactulose). To deal with repeated observations on participants, we determined appropriate fixed intervals for follow‐up for each outcome. To deal with events that may re‐occur (e.g. adverse events), we reported on the proportion of participants who experienced at least one event. If we encountered multiple treatment groups (e.g. different probiotic dose groups or different probiotic species), we divided the placebo group across the treatment groups or we combined probiotic groups to create a single pair‐wise comparison as appropriate.
Dealing with missing data
Where data were missing, we contacted the corresponding authors of included studies to supply any unreported data. For all outcomes in all studies, we carried out analyses as far as possible on an intention‐to‐treat (ITT) basis; that is, we attempted to include all participants randomised to each group in the analyses, and we analysed all participants in the group to which they were allocated regardless of whether or not they received the allocated intervention. For missing continuous data, we estimated standard deviations from other available data, such as standard errors, or we imputed them using the methods suggested in Higgins 2021. We conducted analyses for continuous outcomes based on participants completing the trial, in line with available case analysis; this assumes that data were missing at random. If there was a discrepancy between the number randomised and the number analysed in each treatment group, we calculated and reported the percentage lost to follow‐up in each group. When it was not possible to obtain missing data, we recorded this on the data collection form, reported it in the risk of bias table, and discussed the extent to which the missing data could alter the results and hence the conclusions of the review. We conducted sensitivity analyses to explore the impact of including studies with high levels of missing data on the overall estimate of treatment effect.
Assessment of heterogeneity
We assessed heterogeneity among trial results by visual inspection of forest plots and by calculating the Chi2 test (a P value of 0.10 is regarded as statistically significant heterogeneity). We also used the I2 statistic to quantify the effect of heterogeneity (Higgins 2021). We conducted sensitivity analyses as appropriate to investigate heterogeneity. For example, if a pooled analysis showed statistically significant heterogeneity and a visual inspection of the forest plot identified studies that may have contributed to this heterogeneity, the analysis was repeated excluding these studies to see if this could be explained.
Assessment of reporting biases
If an appropriate number of studies were pooled for meta‐analysis (≥ 10 studies), we planned to investigate the possibility of publication bias through the construction of funnel plots (trial effects versus trial size).
Data synthesis
We combined data from individual trials for meta‐analysis when the interventions, patient groups and outcomes were deemed to be sufficiently similar (determined by consensus). We calculated the pooled RR and corresponding 95% CI for dichotomous outcomes. We calculated the pooled MD and corresponding 95% CI for continuous outcomes that were measured using the same units or when combining change‐from‐baseline and post‐intervention value scores (Higgins 2021). We calculated the pooled standardised mean difference (SMD) and 95% CI when different scales were used to measure the same underlying construct. We carried out meta‐analysis using a random‐effects model. We used Review Manager software for data analysis (RevMan 2020). We analysed data according to the ITT principle. We assumed patients with final missing outcomes to be treatment failures. We grouped analyses by length of follow‐up. We did not pool data for meta‐analysis if we detected a high degree of statistical heterogeneity (I2 > 75%) that was unexplained. In case of a high degree of statistical heterogeneity we investigated whether this could be explained based on clinical grounds or risk of bias, in which case we performed sensitivity analyses. If we could not find any such reasons for the high statistical heterogeneity we presented the results narratively, in detail.
Subgroup analysis and investigation of heterogeneity
We carried out subgroup analyses to further study the effects of a number of variables on the outcomes including:
-
specific probiotic preparation or species;
-
probiotic dose;
-
length of therapy, follow‐up;
-
whether the probiotic was sole therapy or adjunct therapy; and
-
type of functional pain disorder (i.e. irritable bowel syndrome, abdominal migraine or functional abdominal pain, in line with the Rome IV criteria (Hyams 2016)).
Sensitivity analysis
We conducted sensitivity analyses based on the following:
-
random‐effects versus fixed‐effect models (this is based on the approach in the Cochrane Handbook for Systematic Reviews of Interventions Section 13.3.5.6 on sensitivity analysis; Page 2021);
-
studies published in full versus abstract;
-
removing studies judged to be at high risk of bias.
For future updates, if we identify studies of adequate duration we will also explore a sensitivity analysis of dropouts and exclusions by conducting worst‐case versus best‐case scenario analyses, as pre‐specified in our protocol.
Summary of findings and assessment of the certainty of the evidence
We assessed the overall certainty of evidence supporting the primary outcomes (i.e. global improvement or treatment success, complete resolution of pain, severity of pain and frequency of pain) and selected secondary outcomes (serious adverse events, withdrawal due to adverse events, adverse events) using the GRADE approach (GRADEpro GDT; Schünemann 2013), and presented these findings in summary of findings tables for each comparison.
The GRADE approach appraises the certainty of a body of evidence based on the extent to which one can be confident that an estimate of effect, or association, reflects the item being assessed. RCTs start as high‐certainty evidence, but may be downgraded due to overall risk of bias (methodological quality), indirectness of evidence, inconsistency of effect, imprecision (sparse data) and publication bias. Reasons for downgrading are presented in footnotes in the summary of findings tables, with justification. Two review authors independently assessed the overall certainty of the evidence for each outcome after considering each of these factors and graded them as follows:
-
high certainty: further research is very unlikely to change confidence in the estimate of effect;
-
moderate certainty: further research is likely to have an important impact on confidence in the estimate of effect, and may change the estimate;
-
low certainty: further research is very likely to have an important impact on confidence in the estimate of effect, and is likely to change the estimate; or
-
very low certainty: any estimate of effect is very uncertain.
Results
Description of studies
Key characteristics of the included studies can be found in Table 1 and Table 2.
| Study ID | Interventional agent | Dosage (amount and frequency) | Control | Dosage (amount and frequency) | Trial registered(prospective/retrospective/none) | Trial registry outcomes published? | Conflicts of interest |
|---|---|---|---|---|---|---|---|
| Synbiotic group: Bifidobacterium coagulans + fructo‐oligosaccharide | 150 million spores of Bifidobacterium coagulans + fructo‐oligosaccharide twice daily | Peppermint group: peppermint oil (Colpermin)
Placebo group: folic acid | Peppermint group: 187 mg 3 times daily
Placebo group: 1 mg once daily | Prospective | Yes | None declared | |
| Synbiotic group: Bifidobacterium lactis B94 + inulin | 5 × 109 CFU Bifidobacterium lactis 900 mg inulin twice daily | Probiotic group: Bifidobacterium lactis
Prebiotic group: inulin
| Probiotic group: 5 × 109 CFU twice daily
Prebiotic group: 900 mg twice daily
| None | NA | None declared, no financial support received | |
| Synbiotic group: Lactobacillus GG + inulin | 1 x 1010 bacteria/capsule twice daily | Prebiotic group: inulin | Dose unstated (1 capsule twice daily) | None | NA | None declared | |
| Probiotic group: Lactobacillus reuteri | 1 x 108 CFU (5 drops per day) | Placebo group: unidentified placebo | Unstated | Retrospective | Yes | None declared, financial support from Zanjan University of Medical Sciences | |
| Probiotic group: Lactobacillus rhamnosus GG | 3 x 109 CFU twice daily | Placebo group: inert powder | Unstated dose twice daily | Retrospective | Yes | None declared, no financial support received | |
| Probiotic group: Lactobacillus rhamnosus GG | 3 x 109 CFU twice daily | Placebo group: powder | Unstated dose twice daily | None | NA | None declared, financial support from the Medical University of Warsaw | |
| Giannetti 2017 (cross‐over) | Probiotic group: Bifidobacterium longum BB536/ Bifidobacterium infantis M‐63/Bifidobacterium breve M‐16V | 1 sachet daily (3 billion/1 billion/1 billion per bacterium) | Placebo group: unidentified placebo | Unstated (1 sachet daily) | Retrospective | Yes | None declared |
| Guandalini 2010 (cross‐over) | Probiotic group: a patented probiotic preparation, which contains live, freeze‐dried lactic acid bacteria, at a total concentration of 450 billion lactic acid bacteria per sachet, comprising 8 different strains: Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium infantis, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus casei, Lactobacillus bulgaris and Streptococcus thermophilus | 1 sachet once daily if 4 to 11 years old or twice daily if 12 to 18 years old | Placebo group: unidentified placebo | 1 sachet once daily if 4 to 11 years old or twice daily if 12 to 18 years old) | None | NA | None declared, funding from locally available grants; no industry support other than providing probiotic and placebo products |
| Probiotic group: Lactobacillus reuteri DSM 17938 (tablet also containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid) | 1 x 108 CFU once daily (1 x 450 mg chewable tablet) | Placebo group: tablet containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid | Once daily (1 x 450 mg chewable tablet) | Prospective | Yes | None declared, no industry support other than providing probiotic and placebo products | |
| Probiotic group: Lactobacillus reuteri DSM 17938 (tablet also containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid) | 1 x 108 CFU once daily (1 x 450 mg chewable tablet) | Placebo group: tablet containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid | Once daily (one x 450 mg chewable tablet) | Prospective | Yes | Three contributing authors (Iva Hojsak, Sanja Kolacek, Zrinjka Misak) received either payment/honoraria for lectures or consultation, travel grants or lecture fees from several industry sources. All other authors declare no conflict of interest. | |
| Synbiotic group: Lactobacillus GG + inulin | 1 x 1010 CFU capsule twice daily | Prebiotic group: inulin | Unstated dose (1 capsule twice daily) | Prospective | Yes | None declared, funding received from Mashhad University of Medical Sciences, Iran | |
| Probiotic group: Lactobacillus reuteri DSM 17938 | 2 x 108 CFU (in the form of 2 chewable tablets once daily) | Placebo group: unidentified placebo | Unstated (2 chewable tablets once daily) | Prospective | Yes | Three contributing authors received research grants from BioGaia, 2 authors have been speakers for Biogaia and the remaining author had no conflicts to declare | |
| Synbiotic group | Not stated | Placebo group: unidentified | Not stated | None | NA | Abstract only, none declared | |
| Probiotic group: Lactobacillus reuteri | 1 x 108 CFU twice daily in the form of chewable tables | Placebo group: unidentified | Unstated dose (twice daily in the form of chewable tablets) | None | NA | None declared, funding from research centre | |
| Probiotic group: Lactobacillus reuteri DSM 17938 (product also containing sunflower oil, medium‐chain triglyceride oil from coconut oil) | 1 x 108 CFU twice daily in the form of a 10 mL bottle | Placebo group: product containing sunflower oil, medium‐chain triglyceride oil from coconut oil | 10 mL bottle twice daily | None | NA | None declared | |
| Probiotics group: Lactobacillus GG | Unstated dose | Placebo group: unidentified placebo | Unstated dose | None | NA | Abstract only, none declared | |
| Synbiotic group: Bacillus coagulans + fructo‐oligosaccharide | 150 million spores + fructo‐oligosaccharides 100 mg twice daily in the form of tablets | Placebo group: unidentified placebo | 1 tablet twice daily | Prospective | Yes | None declared, funding from Isfahan University of Medical Sciences | |
| Probiotic group: Lactobacillus reuteri DSM 17938 | 1 x 108 CFU once daily in the form of chewable tablet | Placebo group: unidentified placebo | Once daily in the form of chewable tablet | Prospective | Yes | One author (Zvi Weizman) has been a speaker for Biogaia AB which supplied the probiotic. No other conflicts of interest declared, and statement that Biogaia had no role in 'conception, design, and conduct of the study'. |
CFU: colony‐forming unit
NA: not applicable
| Study ID | Methods of diagnosis | FAPD diagnosis | Separate data per sub‐diagnosis reported (yes/no) | Age range | Number of participants | Length of intervention | Time points of outcome measurements |
|---|---|---|---|---|---|---|---|
| Rome III | FAP/IBS/FD | No | 4 to 13 | 54 | 1 month | End of intervention | |
| Rome III | IBS | Not relevant as they only included one | 4 to 16 | 76 | 4 weeks | End of intervention | |
| Rome II | IBS | Not relevant as they only included one | 5 to 17 | 50 | 6 weeks | End of intervention | |
| Rome III | FAP | Not relevant as they only included one | 4 to 16 | 80 | 4 weeks | End of intervention; 4 weeks after end of intervention | |
| Rome II | FAP/IBS | Yes | 5 to 14 | 136 | 8 weeks | End of intervention; 8 weeks after end of intervention | |
| Rome II | FAP/IBS/FD | Yes | 6 to 16 | 104 | 4 weeks | End of intervention | |
| Giannetti 2017 (cross‐over) | Rome III | IBS/FD | Yes | 8 to 17 | 48 | 2 week run‐in period 6 weeks pre‐cross‐over phase 2 weeks washout 6 weeks post‐cross‐over phase | At the end of each period/phase and data combined at the end per intervention |
| Guandalini 2010 (cross‐over) | Rome II | IBS | Not relevant as they only included one | 4 to 18 | 59 | 2 week run‐in period 6 weeks pre‐cross‐over phase 2 weeks washout 6 weeks post‐cross‐over phase | Every 2 weeks and data combined at the end per intervention |
| Rome III | FAP/IBS | No | 4 to 18 | 55 | 12 weeks | 1 month into intervention; end of intervention; 1 month after end of intervention | |
| Rome III | FAP/IBS | No | 4 to 18 | 46 | 12 weeks | 1 month into intervention; end of intervention; 1 month after end of intervention | |
| Rome III | IBS | Not relevant as they only included one | 4 to 18 | 52 | 1 month | Weekly until end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 5 to 16 | 48 | 4 weeks | At 2 weeks and end of intervention | |
| Rome III | FAP/FD | No | Not stated | 80 | 8 weeks | End of intervention | |
| Rome III | FAP/IBS/FD/abdominal migraine | Yes | 6 to 16 | 125 | 4 weeks | At 2 weeks and end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 16 | 60 | 4 weeks | End of intervention; 4 weeks after end of intervention | |
| Unstated | FAP | Not relevant as they only included one | Unstated | 61 | 6 weeks | End of intervention; 4 weeks after end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 18 | 115 | 4 weeks | End of intervention; 8 weeks after end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 15 | 101 | 4 weeks | End of intervention; 4 weeks after end of intervention |
FAP: functional abdominal pain
FAPD: functional abdominal pain disorder
FD: functional dyspepsia
IBS: irritable bowel syndrome
Results of the search
A literature search conducted on 1 October 2021 identified 1712 records. After duplicates were removed a total of 757 records remained for review of titles and abstracts. Two authors independently reviewed these titles and abstracts, and discarded 687 records. We selected 70 potentially relevant reports on the use of probiotics for the management of functional abdominal pain disorders in children for full‐text review (see Figure 1). We excluded 29 studies (30 records), with reasons (see Excluded studies). Five studies (six records) are awaiting classification (see below and Characteristics of studies awaiting classification). We identified two ongoing studies (see Characteristics of ongoing studies).
Included studies
We selected a total of 18 studies (32 records) involving 1309 patients for inclusion (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Romano 2014; Sabbi 2012; Saneian 2015; Weizman 2016).
Age of participants
Participants in all included studies were between the ages of 4 and 18 years. Six of the studies had a more restrictive age range than this (Asgarshirazi 2015; Baştürk 2016; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017). Bauserman 2005 specified an age range of 5 to 21 years old for participants, but no included participants were above the age of 17 based on the tables provided. Otuzbir 2016 and Sabbi 2012 did not provide age information.
Diagnosis
Four of the studies based the diagnosis of functional abdominal pain on the Rome II criteria (Bauserman 2005; Francavilla 2010; Gawrońska 2007; Guandalini 2010), whilst all others based the diagnosis on the Rome III criteria, except for Sabbi 2012, which did not provide this information.
Eight studies included more than one diagnosis within the definition of functional abdominal pain disorders (Asgarshirazi 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Jadrešin 2017; Jadrešin 2020; Otuzbir 2016; Rahmani 2020). Four of them provided separate data per diagnosis assessed (Francavilla 2010; Gawrońska 2007; Giannetti 2017; Rahmani 2020).
Functional abdominal pain was studied in 13 studies (Asgarshirazi 2015; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Romano 2014; Sabbi 2012; Saneian 2015; Weizman 2016). Irritable bowel syndrome was studied in 11 studies (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Rahmani 2020). Functional dyspepsia was studies in five studies (Asgarshirazi 2015; Gawrońska 2007; Giannetti 2017; Otuzbir 2016; Rahmani 2020). Abdominal migraine was studied in one study (Rahmani 2020).
Length of the interventions and time points of outcome measurements
Five studies measured outcomes solely at the end of the length of their given interventions: Asgarshirazi 2015, Baştürk 2016 and Gawrońska 2007 at four weeks, Bauserman 2005 at six weeks and Otuzbir 2016 at eight weeks.
In Eftekhari 2015, Romano 2014 and Weizman 2016, interventions lasted four weeks and the outcomes were measured at the end of the intervention and four weeks after the end.
In Kianifar 2015, the intervention lasted four weeks and outcomes were measured at the end of every week until the end of the intervention.
In Maragkoudaki 2017 and Rahmani 2020, the interventions lasted four weeks; outcomes were measured at the end of the second week and at the end of the intervention.
The intervention in Saneian 2015 lasted four weeks and outcomes were measured at the end of the intervention and eight weeks after the end.
In Sabbi 2012, the intervention lasted six weeks and outcomes were measured at the end of the intervention and four weeks after the end.
The intervention in Francavilla 2010 lasted eight weeks and the outcomes were measured at the end of the intervention and eight weeks after the end.
In Jadrešin 2017 and Jadrešin 2020, interventions lasted 12 weeks and outcomes were measured at four weeks into the interventions, at the end of the interventions (12 weeks) and four weeks after the end of the intervention.
Giannetti 2017 had a cross‐over design that included a two‐week run‐in period, six intervention weeks for the pre‐cross‐over phase, followed by a two‐week washout period, and six intervention weeks for the post‐cross‐over phase. Outcomes were measured at the end of each period/phase of the study.
Guandalini 2010 had the same design and length for their intervention and measured outcome data every two weeks until the end of the intervention.
Intervention arms
All studies had two intervention arms except for two, which had three intervention arms (Asgarshirazi 2015; Baştürk 2016). Baştürk 2016 had a synbiotic, a probiotic and a prebiotic (placebo) group; Asgarshirazi 2015 had a synbiotic, a peppermint and a placebo group. We did not use the data for the peppermint group in our analysis as this is beyond the scope of this review.
Intervention and placebo agents
All studies compared probiotics or synbiotics to a placebo (including prebiotics).
Thirteen studies compared probiotics to placebo (Baştürk 2016; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Maragkoudaki 2017; Rahmani 2020; Romano 2014; Sabbi 2012; Weizman 2016). Seven studies used Lactobacillus reuteri (Eftekhari 2015; Jadrešin 2017; Jadrešin 2020; Rahmani 2020; Romano 2014; Maragkoudaki 2017; Weizman 2016). Three studies used Lactobacillus rhamnosus GG (Francavilla 2010; Gawrońska 2007; Sabbi 2012). Baştürk 2016 used Bifidobacterium lactis B94. Giannetti 2017 used a combination of three stains of bifidobacteria. Guandalini 2010 used a combination of eight strains of bifidobacteria, lactobacilli and Streptococcus.
Six studies compared synbiotics to placebo (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Kianifar 2015; Otuzbir 2016; Saneian 2015). Asgarshirazi 2015 and Saneian 2015 used Bifidobacterium coagulans combined with fructo‐oligosaccharide. Bauserman 2005 and Kianifar 2015 used Lactobacillus rhamnosus GG combined with inulin. Baştürk 2016 used Bifidobacterium lactis B94 combined with inulin. Otuzbir 2016 did not provide any information.
Eleven studies used unidentified placebos (Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Sabbi 2012; Saneian 2015; Weizman 2016). Two of them described the placebo as an inert powder (Francavilla 2010; Gawrońska 2007). Jadrešin 2017 and Jadrešin 2020 used a tablet containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid as placebo. Romano 2014 used a product containing sunflower oil and medium‐chain triglyceride oil from coconut oil as placebo. Asgarshirazi 2015 used folic acid as placebo. Baştürk 2016, Bauserman 2005 and Kianifar 2015 used the prebiotic inulin as placebo. For the purposes of our analysis we decided to group inulin together with the other placebos, despite its theoretically potential active role as a prebiotic, because its role in the improvement of functional abdominal pain disorder symptoms is unknown.
All agents were taken orally. Information on dosages can be found in Table 1.
Reporting of primary outcomes
Global improvement or treatment success as defined by the primary studies
Our primary dichotomous outcome of patient global improvement or treatment success as defined by the primary studies was reported in 11 studies (Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020 ; Saneian 2015). In the other seven studies, the outcome was either unclear or not reported (Asgarshirazi 2015; Giannetti 2017; Guandalini 2010; Kianifar 2015; Romano 2014; Sabbi 2012; Weizman 2016).
Complete resolution of pain
Our primary dichotomous outcome of complete resolution of pain was reported in five studies (Baştürk 2016; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Otuzbir 2016).
Severity of pain or change in the severity of pain
Our primary continuous outcome of severity of pain/change in the severity of pain was reported in 13 studies (Asgarshirazi 2015; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Rahmani 2020; Romano 2014; Saneian 2015; Weizman 2016).
Asgarshirazi 2015 measured pain on 0 to 10 numerical rating scale. Bauserman 2005 used a four‐point Likert scale (0 to 3). Eftekhari 2015 and Saneian 2015 measured pain according to the Wong‐Baker six‐point scale (0 to 5). Francavilla 2010 used a combination of 0 to 10 visual analogue scale (VAS) and the 1‐ to 6‐point Faces Pain Scale (FPS). Gawrońska 2007 and Romano 2014 used the Faces seven‐point (0 to 6) pain scale. Jadrešin 2017 and Jadrešin 2020 used the 0 (no hurt) to 10 (hurts worst) Wong‐Baker FACES Pain Rating Scale. Kianifar 2015 measured pain on a (0 to 4) five‐point Likert scale. Maragkoudaki 2017 used an unspecified Wong‐Baker FACES Pain Rating Scale. Weizman 2016 used the face scoring system by Hicks (each of the six faces in the scoring system ranked 0, 2, 4, 6, 8 or 10, where 0 = no pain (relaxed face) and 10 = very severe pain (miserable face)).
Frequency of pain or change in the frequency of pain
Our primary continuous outcome of frequency of pain/change in the frequency of pain was reported in nine studies (Asgarshirazi 2015; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Maragkoudaki 2017; Rahmani 2020; Romano 2014; Weizman 2016).
Asgarshirazi 2015, Eftekhari 2015, Francavilla 2010, Gawrońska 2007, Maragkoudaki 2017 and Weizman 2016 measured this as pain episodes per week. Jadrešin 2017 and Jadrešin 2020 measured this as days without pain. Rahmani 2020 measured it as frequency of repetitive pain per day. Romano 2014 measured pain as episodes per day.
Reporting of secondary outcomes
Serious adverse events
Our secondary outcome of serious adverse events was reported in 12 studies (Asgarshirazi 2015; Bauserman 2005; Eftekhari 2015; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Romano 2014; Saneian 2015; Weizman 2016).
Withdrawals due to adverse events
Withdrawals due to adverse events were reported in 14 studies (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Eftekhari 2015; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Romano 2014; Saneian 2015; Weizman 2016).
Adverse events
Adverse events were reported in 12 studies (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Eftekhari 2015; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Romano 2014; Weizman 2016).
School performance
Our secondary outcome of school performance was reported in three studies (Gawrońska 2007; Maragkoudaki 2017; Weizman 2016).
Social and psychological functioning
Social and psychological functioning was reported in one study (Kianifar 2015).
Quality of life
Quality of life was reported in one study (Guandalini 2010).
Notes on data availability
We noted during data extraction that there were a number of studies with concerning data that could not be interpreted:
In Rahmani 2020, the outcome data were inversed between the text and the tables for several outcomes, with no consistent pattern. It was therefore not possible to confirm which were the appropriate figures and we received no response from the contact author or the editor of the journal after numerous attempts at contact (this is a pre‐publication manuscript that has not been copy‐edited and so we also attempted to contact the journal with no response received). In the end, as some of the data made no mathematical sense if they were taken from the table (negative standard deviation (SD)), we elected to use the data that were reported in the text.
Eftekhari 2015 provided many conflicting results for their complete resolution outcome, with "no hurt" referring to pain intensity and "no pain" referring to no pain episodes. We did not receive any clarification from the authors after contacting them and we decided to use the figures of no pain episodes per week for our dichotomous outcome as this appeared to be the most homogeneous item.
Jadrešin 2017 and Jadrešin 2020 reported a pooled analysis for the outcomes of days without pain, pain intensity and complete resolution of pain in both their studies at the end of their 2020 paper, which seemingly used different results than those reported in the 'Results' section of both papers. We did not receive any response from the authors after contacting them. We could not use the misreported data for the outcomes of pain frequency and pain severity in our analysis because they did not provide SDs or other variance data to calculate SDs.
Maragkoudaki 2017 also had some concerns with regard to the severity of pain outcome, as the baseline mean for one group was 17, which is greater than the pain scale they reported. They may have used a different scale but as this is not specified it was unclear how this could be accommodated within the rest of the data set. We did not receive any response from the authors after contacting them.
Baştürk 2016 and Kianifar 2015 reported randomised patients discontinuing the study and not being included in the results without mentioning the group to which these patients had been randomised. The authors of Baştürk 2016 responded to our email and informed us about the correct randomisation numbers. The authors of Kianifar 2015 did not respond.
Otuzbir 2016 and Sabbi 2012 were available as abstracts only with extremely limited information provided. The authors did not respond to our emails asking for more information.
Giannetti 2017 and Guandalini 2010 were cross‐over studies and did not provide separate data per intervention and control groups for pre‐ and post‐cross‐over; instead results were analysed in one unique analysis combining pre‐ and post‐cross‐over treatments. The authors did not respond to our emails asking for more information.
Information on the primary and secondary outcome data is illustrated in Table 3 and Table 4.
| Study ID | 1a. Global improvement or treatment success | 1b. Complete resolution of pain | 1c. Severity of pain | 1d. Frequency of pain |
|---|---|---|---|---|
| NR | NR | Intervention group: 3.93 ± 1.06 Control group: 4.24 ± 1.33 | Intervention group: 2.14 ± 0.87 Control group: 3.40 ± 1.41 | |
| NR | Intervention group 1: 9/26 Intervention group 2: 7/25 Control group: 3/25 | NR | NR | |
| Intervention group: 11/32 Control group: 10/32 | NR | Change in pain intervention group: ‐1.3 (± 0.3) Change in pain control group: ‐1.7 (± 0.6) | NR | |
| No pain episodes per week, end of first month Intervention group: 20; control group: 26
No pain episodes per week, end of second month Intervention group: 19; control group: 21 | NR | End of first month intervention group: mean (SD) 2.50 (1.45); control group: mean (SD) 2.08 (1.56)
End of second month intervention group: mean (SD) 2.53 (1.43); control group: mean (SD) 2.25 (1.46) | At first month intervention group: mean (SD) 0.68 (0.76); control group: mean (SD) 0.40 (0.59)
At second month intervention group: mean (SD) 0.70 (0.75); control group: mean (SD) 0.53 (0.59) | |
| Decrease of at least 50% in the number of episodes and intensity of pain at 12 weeks Intervention group: 48/69; control group: 37/67
Decrease of at least 50% in the number of episodes and intensity of pain at end of follow‐up Intervention group: 53/69; control group: 43/67 | NR | At 12 weeks intervention group: mean (SD) 2.3 (1.3); control group: 3.4 (2.1)
At end of follow‐up intervention group: mean (SD) 0.9 (0.5); control group: 1.5 (1.0) | Number of episodes per week at 12 weeks intervention group: 1.1 (0.8); control group: 2.2 (1.2)
At end of follow‐up intervention group: 0.9 (0.5); control group: 1.5 (1.0) | |
| NR | Intervention group: 13/52 (FD 1/10, IBS 6/18, FAP 6/24)
Control group: 5/52 (FD 2/10, IBS 6/18, FAP 2/23) | At 4 weeks Intervention group: mean 2.5 (SD ± 1.9) (FD 2.9 ± 1.5, IBS 2.2 ± 2.1, FAP 2.6 ± 2.0) Control group: mean 2.9 (SD ± 1.5) (FD 1.9 ± 1.3, IBS 3.2 ± 1.5, FAP 3.0± 1.5) | At 4 weeks Intervention group: mean 2.2 (SD ± 1.7) (FD 2.7 ± 1.3, IBS 1.8 ± 1.7, FAP 2.3 ± 1.8) Control group: mean 2.6 (SD ± 1.4) (FD 2.0 ± 1.6, IBS 3.1 ± 1.1, FAP 2.4 ± 1.4) | |
| Giannetti 2017 (cross‐over) | NR | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data |
| Guandalini 2010 (cross‐over) | NR | NR | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data |
| NR | Intervention group: 16/26 Control group: 16/29 | End of first month intervention group/control group: 0.75/0.96 End of second month intervention group/control group: 0.17/0.64 End of third month intervention group/control group: 0.32/0.71 End of fourth month intervention group/control group: 0.21/0.6
Difference in the severity of pain between first and fourth month, Wong‐Baker FACES/day intervention group: median 0.42 (range 0.31 to 2.9); control group: median 0.23 (range 1.2 to 2.2) | Number of days without pain intervention group at 4 months: 89.5 (range 5 to 108); control group at 4 months: 51 (range 0 to 107) | |
| NR | Intervention group: 10/24 Control group: 9/22 | End of first month intervention group/control group: 1.35 (IQR 0.64 to 1.98)/1.1 (IQR 0.76 to 2.04) End of second month intervention group/control group: 1.0 (IQR 0.09 to 2.12)/0.8 (IQR 0.37 to 1.68) End of third month intervention group/control group: 0.83 (IQR 0.025 to 2.26)/0.78 (IQR 0.43 to 2.0) End of fourth month intervention group/control group: 0.035 (IQR 0 to 1.0)/0.81 (IQR 0.2 to 1.48)
Change in severity of pain from 1st to 4th month intervention group: 0.55 (IQR 0.28 to 0.55); control group: median 0.36 (IQR ‐0.14 to 0.36) | Number of days without pain intervention group at 4 months: 90 (IQR 54 to 99); control group at 4 months: 59.5 (IQR 21.5 to 89.25) | |
| NR | NR | 1 week: intervention group/control group 1.5 (1.0)/1.8 (0.6) 2 weeks: intervention group/control group 1.2 (1.1)/1.9 (0.8) 3 weeks: intervention group/control group 1.0 (0.9)/1.8 (0.6) 4 weeks: intervention group/control group 0.8 (0.9)/1.5 (0.8) | NR | |
| Reduction in pain score of greater than 50% at 4 weeks intervention group: 19/27 (70.4%); control group: 16/27 (58.3%)
Reduction in pain score of greater than 50% at 8 weeks intervention group: 17/25 (65.4%); control group: 13/23 (56.5%) | NR | Intervention group/control group: mean (SD) 2 weeks: 10.4 (18.8)/12.2 (17.3) 4 weeks: 4.3 (8.5)/4.0 (5.6) 8 weeks: 7.2 (17.7)/2.5 (3.4) | Intervention group/control group: mean (SD) 2 weeks: 5.6 (8.1)/8.2 (10.7) 4 weeks: 2.9 (4.5)/3.1 (4.1) 8 weeks: 4.8 (9.9)/2.8 (3.3) | |
| NR | Intervention group: 25/39 Control group: 18/41 | NR | NR | |
| Intervention group = 32/65 (FAP 13/28, FD 11/16, IBS 6/15, AM 2/6)
Control group = 8/60 (FAP 8/29, FD 0/13, IBS 0/6, AM, 0/3)
| NR (in Rahmani 2020, treatment success was defined as pain intensity = 0) | Text: severity at 4 weeks in intervention group = 1.3 ± 1.1 (Table 1 reports: 1.1 ± 1.3)
Text: severity at 4 weeks in control group = 1 ± 2 (Table 1 reports: 2 ± 1)
FAP (intervention group/control group): 1.2 ± 1.3; 2 ± 1 FD (intervention group/control group): 0.8 ± 1.5; 2.0 ± 6 IBS (intervention group/control group): 1.4 ± 1.4; 2.8 ± 0.8 AM (intervention group/control group): 1.3 ± 1.5; 2.3 ± 0.5 | Text: frequency of repetitive pain at 4 weeks intervention group 3.6 ± 2.2 (Table 1 reports: intervention group 2.2 ± 3.6)
Text: frequency of repetitive pain at 4 weeks control group 4.6 ± 4.9 (Table 1 reports: control group 4.9 ± 4.6)
FAP (intervention group/control group): 2.1 ± 2.7; 4.1 ± 4.4 FD (intervention group/control group): 1.6 ± 3.0; 6.0 ± 5.0 IBS (intervention group/control group): 3.7 ± 5.5; 6.3 ± 0.8 AM (intervention group/control group): 1.1 ± 0.9; 1.3 ± 0.5 | |
| NR | NR | Mean (SD) as we interpreted it from the figures: Week 4 intervention group/control group: 1.25 (0.9)/2 (0.8) Week 8 intervention group/control group: 1 (0.7)/2 (0.8) | Mean (SD) as we interpreted it from the figures: Week 4 intervention group/control group: 1.4 (1.1)/2.2 (0.5) per day Week 8 intervention group/control group: 2.1 (0.6)/2 (0.5) per day | |
| NR | NR | NR | NR | |
| Response at week 4 intervention group: 27/45; control group: 17/43
Response at week 12 intervention group: 29/45; control group: 23/43 | NR | Change in pain scale from start of intervention to week 4 intervention group: mean ‐1.7 (SD ± 1.5); control group: mean ‐1.6 (SD ± 1.5)
Change in pain scale from start of intervention to week 12 intervention group: mean ‐2.1 (SD ± 1.4); control group: mean ‐1.8 (SD ± 1.4) | NR | |
| NR | NR | Improvement in intensity of abdominal pain at 4 weeks intervention group: mean 4.3 (SD ± 2.7); control group: mean 7.2 (SD ± 3.1)
Improvement in intensity of abdominal pain at end 8 weeks intervention group: mean 4.8 (SD ± 3.3); control group: mean 6.4 (SD ± 4.1) | Number of episodes of pain per week at 4 weeks intervention group: mean 1.9 (SD ± 0.8); control group: mean 3.6 (SD ± 1.7)
Number of episodes of pain per week at 8 weeks intervention group: mean 3.4 (SD ± 2.6); control group: mean 4.4 (SD ± 2.9)
|
Numbers presented as per the original study reports.
AM: abdominal migraine
FAP: functional abdominal pain
FD: functional dyspepsia
IBS: irritable bowel syndrome
IQR: interquartile range
NR: not reported
SD: standard deviation
| Study ID | 2a. Serious adverse events | 2b. Withdrawal due to adverse events | 2c. Adverse events | 2d. School performance | 2e. Social and psychological functioning | 2f. Quality of life |
|---|---|---|---|---|---|---|
| 0 | 0 | 0 | NR | NR | NR | |
| NR | Intervention group 1: 3 | Intervention group 1: 3 | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | 0 | 0 | School absenteeism at end of intervention | NR | NR | |
| Giannetti 2017 (cross‐over) | 0 | 0 | 0 | NR | NR | NR |
| Guandalini 2010 (cross‐over) | 0 | 0 | 0 | NR | NR | Questionnaire of disruption to family life (change in score) Intervention group: mean ‐0.9 (SD ± 0.2); control group: mean ‐0.51 (SD ± 0.3) |
| NR | 0 | NR | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| 0 | 0 | 0 | NR | Functional changes on a 3 point Likert scale at end of intervention Intervention group: mean 2.4 (SD ± 0.5); control group: mean 1.9 (SD ± 0.4) | NR | |
| 0 | 0 | 0 | Average number of school absences per week at end of follow‐up Intervention group: mean 0.0 (SD ± 0.0); control group: mean 0.11 (SD ± 0.52) | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | Intervention group: 5; control group:0 | NR as numbers of people with adverse events
Total number of adverse events intervention group: 45; control group: 43 | NR | NR | NR | |
| 0 | 0 | 0 | Days of school absenteeism over 4 weeks Intervention group: mean 2.7 (SD ± 0.9); control group: mean 1.9 (SD ± 1.1) | NR | NR |
Numbers presented as per the original study reports.
NR: not reported
SD: standard deviation
Excluded studies
We excluded a total of 29 studies (see Characteristics of excluded studies). We excluded10 studies because they were conducted in adult patients (Cha 2012; Choi 2015; Enck 2009; Han 2016; Le Neve 2016; Mezzasalma 2016; Sen 2002; Spiller 2016; Yoon 2014; Yoon 2015). We excluded four studies because they were letters to journals or letters to authors (Anonymous 2010; Chassany 2008; Faber 2003; Pélerin 2016). We excluded one record because it was a comment on an included study (Abu‐Salih 2011). We excluded nine studies because they were review articles and not randomised controlled trials (Anuradha 2005; Berger 2007; Cash 2011; Charrois 2006; Enck 2007; Ford 2012; Kajander 2008; Rose 2011; Schmulson 2011). We excluded one study because it studied antibiotics rather than probiotics (Drossman 2011). We excluded one study because it studied the effect of guar gum rather than probiotics (Comito 2011). We excluded two studies because they looked at the effect of probiotics on functional constipation rather than functional abdominal pain disorders (Baştürk 2017; Wegner 2018). We excluded one study as it was not randomised (NCT04922476).
Studies awaiting classification
A total of five studies are categorised as awaiting classification (Characteristics of studies awaiting classification). We were unable to confirm whether Chao 2011 met our inclusion criteria from the information presented, and we were unable to contact the authors to clarify. Gholizadeh 2021 was found in a pre‐publication update search and will be included in future reviews. Initially, Sudha 2018 was included, but we noted concerns with the outlying data as these were highly positive, as well as significant conflicts from the team. We sought advice from the Cochrane research integrity unit and the Cochrane Gut team and, based on this, we attempted to contact the authors on numerous occasions, as well as the editors of the journal for clarification. No response has been received (two named authors were directly employed by the manufacturer of their interventional agent, and the study was funded by the same manufacturer). This is in line with the Cochrane policy for managing potentially problematic studies. Given the concerns and lack of response from the authoring team or journal, we have moved this study to awaiting classification. NCT00793494 is a trial registration with no corresponding published results, which states that recruitment was terminated due to inability to recruit. NCT02613078 is a trial registration with no corresponding published results and insufficient details for us to be confident that it meets our inclusion criteria.
Ongoing studies
We also identified two ongoing studies (see Characteristics of ongoing studies).
Risk of bias in included studies
The risk of bias analysis for the included studies is summarised in Figure 2.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
In addition, we reviewed each included study for potential conflicts of interest as well as the source of their funding, summarising our findings in Table 1. Fifteen of the18 studies declared no conflict of interest, or did not make a statement on conflicts of interest.
Jadrešin 2020 declared that three named authors had received payment or honoraria in the past for lectures or consultation from industry sources.
Maragkoudaki 2017 declared that three named authors had received research grants from the manufacturer of their interventional agent, and a further two authors had been speakers for the same manufacturer.
Weizman 2016 declared that one author had previously been a speaker for the manufacturer of their interventional agent.
Allocation
Random sequence generation
In 15 of the 18 included studies the method of random allocation of participants to intervention groups (selection bias) was described and judged to be adequate (Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Rahmani 2020; Romano 2014; Saneian 2015; Weizman 2016). We rated these studies as having a low risk of selection bias for random sequence generation.
Two studies stated that they randomly allocated participants, but did not describe how they achieved this, and did not respond to requests for clarification. Therefore, we rated them as having an unclear risk of selection bias (Asgarshirazi 2015; Sabbi 2012).
One study randomised participants according to time of admission into the trial, with no further details on randomisation, and no response to requests for further information. Therefore, we rated this study as having an unclear risk of selection bias (Otuzbir 2016).
Allocation concealment
Allocation concealment was adequately described in 13 of the 18 included studies and we rated them as having a low risk of bias (Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Maragkoudaki 2017; Romano 2014; Saneian 2015; Weizman 2016). The code revealing participant allocation was only revealed by the vendor on completion of the statistical analysis, and those involved in enrolment were unaware of the allocation sequence.
In the remaining five included studies allocation sequence concealment was inadequately described and we thus rated them as having an unclear risk of bias (Asgarshirazi 2015; Kianifar 2015; Otuzbir 2016; Rahmani 2020; Sabbi 2012). In each case we sought further information from the authors but did not receive a reply.
Blinding
Methods for blinding of participants and personnel were described and judged to be low risk of bias for 17 of the 18 included studies (Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Romano 2014; Sabbi 2012; Saneian 2015; Weizman 2016).
Asgarshirazi 2015 described their study as placebo‐controlled and single‐blinded, and stated that the nurse involved in administering the questionnaire was blinded. The authors did not respond to requests for clarification and we therefore rated this study as being at high risk of performance bias.
The method for blinding of outcomes was well described and judged to be at low risk of detection bias in 16 of the 18 studies (Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Romano 2014; Saneian 2015; Weizman 2016).
Two studies did not adequately describe the methods for preventing detection bias, and did not reply to requests for clarification. We subsequently rated them as having an unclear risk of bias (Asgarshirazi 2015; Sabbi 2012).
Incomplete outcome data
We judged 15 of the 18 included studies to be at low risk of attrition bias given the balanced nature of withdrawals for similar reasons and adequately described study flow (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Giannetti 2017; Guandalini 2010; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Romano 2014; Saneian 2015; Weizman 2016).
Two studies gave inadequate information on patient flow through the study and did not respond to requests for clarification. We therefore rated them as having an unclear risk of bias for incomplete outcome data (Otuzbir 2016; Sabbi 2012)
One study did not clarify how many participants were excluded post‐randomisation based on exclusion criteria versus how many were excluded because of poor compliance with treatment. The authors did not respond to our requests for clarification and we therefore rated the study as having an unclear risk of bias for incomplete outcome data (Rahmani 2020).
Selective reporting
We rated five of the 18 included studies as low risk for reporting bias due to the complete reporting of outcomes as per a prospectively registration of the trial, and a full description of adverse events (Asgarshirazi 2015; Francavilla 2010; Kianifar 2015; Maragkoudaki 2017; Weizman 2016).
One study did not provide information on the randomisation and adverse effects for five patients who withdrew from the study post‐randomisation, and did not respond to our requests for clarification. This study also did not have a trial registration. We therefore rated it as having an unclear risk of bias for selective reporting (Baştürk 2016)
One study included conflicting reports for the primary outcomes and had a trial protocol that was registered retrospectively (IRCT2014083018971N1). The authors did not respond to a request for clarification regarding the primary outcome data and we therefore rated the study as having an unclear risk of bias (Eftekhari 2015).
Two studies from the same team included outcomes in their prospective trial registration (NCT01587846) that were not reported in the final study (Jadrešin 2017; Jadrešin 2020). In addition, Jadrešin 2020 included a pooled analysis that does not tally with the data from the individual reports. The authors did not respond to requests for clarification and we therefore rated the studies as having an unclear risk of bias for selective reporting.
One study reported all outcomes appropriately but the methodology was inadequately described, and there was no full text or protocol available to clarify this. In addition, the authors did not respond to our requests for clarification and we therefore rated this study as having an unclear risk of bias for selective reporting (Otuzbir 2016).
One study did not provide data for their outcomes and simply stated that the differences between intervention and control groups were 'significant'. We were unable to contact the authors to request further information, so we rated this study as having an unclear risk of bias for selective reporting (Sabbi 2012).
Other potential sources of bias
There were no concerns about other potential sources of bias for 15 of the 18 included studies and we rated these as low risk of bias (Asgarshirazi 2015; Baştürk 2016; Bauserman 2005; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Kianifar 2015; Maragkoudaki 2017; Otuzbir 2016; Rahmani 2020; Romano 2014; Saneian 2015; Weizman 2016).
Two studies presented results in an unclear fashion by combining pre‐ and post‐cross‐over data and presenting their results in a per condition manner rather than per intervention and control group (Giannetti 2017; Guandalini 2010). They also did not provide randomisation numbers for each therapy. The authors did not respond to our requests for clarification and we therefore rated these studies as having an unclear risk of bias.
There was insufficient information to judge one of the studies and the author did not respond to requests for further information, so we judged it to be at unclear risk of other bias (Sabbi 2012).
Effects of interventions
See: Summary of findings 1 Probiotic compared to placebo for management of functional abdominal pain disorders in children; Summary of findings 2 Synbiotic compared to placebo for management of functional abdominal pain disorders in children
Information on the primary and secondary outcome data we used can be found in Table 3 and Table 4. We planned to conduct several subgroup analyses in our protocol, including probiotic dose and length of therapy, as well as sensitivity analyses (e.g. random‐effects versus fixed‐effect models), but these were not pursued in this review due to a lack of data.
Probiotics versus placebo
Primary outcomes
Treatment success
Six studies with 554 participants provided data for this outcome (Baştürk 2016; Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Maragkoudaki 2017; Rahmani 2020). Meta‐analysis of six studies with 554 participants showed that patients with functional abdominal pain disorders may respond more to probiotics (167/330) than placebo (118/325) (risk ratio (RR) 1.57, 95% confidence interval (CI) 1.05 to 2.36, I2 = 70%) (Analysis 1.1; Figure 3). The evidence is of low certainty due to inconsistency and risk of bias (summary of findings Table 1). After repeating this analysis with a fixed‐effect model, the significant result remained unchanged (RR 1.49, 95% CI 1.23 to 1.80, I2 = 70%).
Subgroup analysis for specific strains revealed low‐certainty evidence for Lactobacillus reuteri, Lactobacillus rhamnosus GG and Bifidobacterium lactis, indicating that there may be no difference to placebo (Analysis 1.1). None of these analyses were statistically significant, but the results were homogenous on visual inspection and the confidence intervals tight. The certainty of the evidence was low for all of these analyses due to serious or very serious imprecision and significant inconsistency. The results of sensitivity analyses using a fixed‐effect model for the probiotics strains Lactobacillus reuteri and Lactobacillus rhamnosus GG were different from the random‐effects analysis, now showing both of these strains as superior to placebo; however, these findings were still of low certainty as described above (Analysis 1.2).
The remaining probiotics studies did not report this outcome.
Complete resolution of pain
Complete resolution of pain was reported in six studies (Baştürk 2016; Eftekhari 2015; Gawrońska 2007; Jadrešin 2017; Jadrešin 2020; Rahmani 2020). Meta‐analysis of the results of these studies did not show a clear difference between probiotics (97/232) and placebo (62/228) (RR 1.55, 95% CI 0.94 to 2.56, I2 = 70%) (Analysis 1.3). The evidence is of very low certainty due to very high inconsistency and risk of bias (summary of findings Table 1).
A sensitivity analysis, removing Rahmani 2020 due to risk of bias for this outcome, again did not show a clear difference between probiotics (65/167) and placebo (54/168) (RR 1.18, 95% CI 0.84 to 1.67, I2 = 30%) (Analysis 1.4).
Severity of pain on completion
This outcome was reported in seven studies with 655 participants (Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Maragkoudaki 2017; Rahmani 2020; Romano 2014; Weizman 2016). We conducted a meta‐analysis; however, we were unable to draw meaningful conclusions due to very high unexplained heterogeneity (Analysis 1.5). The evidence is of very low certainty. In line with our pre‐planned methodology for managing heterogeneity, a narrative synthesis is presented in summary of findings Table 1.
Jadrešin 2017 and Jadrešin 2020 did not provide standard deviations (SDs) for their severity of pain results, and we did not receive a response from the study authors when we requested these. Jadrešin 2017 reported a mean score of 0.21 for the probiotics group and 0.6 for the placebo group at study end. Jadrešin 2020 reported a median score of 0.035 (interquartile range (IQR) 0 to 1) for the probiotics group and 0.81 (IQR 0.2 to 1.48) for the placebo group at study end. We decided not to use the latter for our analysis due to uncertainty as to whether data were skewed in this study, which would make statistical transformation to a mean and SD inappropriate. As stated above, the authors did not respond to our requests for confirmation.
Frequency of pain on completion
Frequency of pain was measured in episodes per week in six studies with 605 patients (Eftekhari 2015; Francavilla 2010; Gawrońska 2007; Maragkoudaki 2017; Rahmani 2020; Weizman 2016). We conducted a meta‐analysis; however, we were unable to draw meaningful conclusions due to very high unexplained heterogeneity (Analysis 1.6). The evidence was of very low certainty. In line with our pre‐planned methodology for managing heterogeneity, a narrative synthesis is presented in summary of findings Table 1.
We conducted a sensitivity analysis, removing Eftekhari 2015 from the analysis due to risk of bias. The results showed that probiotics reduce pain frequency per week when compared to placebo (mean difference (MD) ‐0.58, 95% CI ‐0.81 to ‐0.35, I² = 0%) (Analysis 1.7).
Romano 2014 measured pain frequency in episodes per day. The mean (SD) on completion for the probiotics group was 1 (0.7) and for the placebo group was 2 (0.8).
Jadrešin 2017 reported the median number of days without pain in the probiotics group at four months as 89.5 (range 5 to 108) and the number of days without pain in the placebo group at four months as 51 (range 0 to 107). Jadrešin 2020 reported the median number of days without pain in the probiotics group at four months as 90 (IQR 54 to 99) and the median number of days without pain in the placebo group at four months as 59.5 (IQR 21.5 to 89.25). We decided not to use the latter for our analysis due to the uncertainty as to whether data were skewed in this study, which would make statistical transformation to a mean and SD inappropriate. As stated above, the authors did not answer our requests for confirmation.
Secondary outcomes
Serious adverse events
There were no recorded serious adverse events in any of the included studies within either the probiotics or placebo groups.
Withdrawal due to adverse events
Meta‐analysis of eight studies with 544 participants showed no difference in withdrawals due to adverse events between probiotics (1/275) and placebo (1/269) (RR 1.00, 95% CI 0.07 to 15.12) (Analysis 1.8). The evidence is of very low certainty because of imprecision due to the very low numbers of events and risk of bias (summary of findings Table 1).
Baştürk 2016 reported five post‐randomisation withdrawals from the study, but it was not stated whether these withdrawals came from the intervention or placebo group. However, the author responded to a request to clarification regarding these withdrawals, so these data have now been included in this analysis.
It was not possible to include figures for Giannetti 2017 in this analysis as although the primary study stated that some participants were lost post‐randomisation, the study was a cross‐over trial and did not break down withdrawals by group at randomisation.
It was also not possible to include figures for Guandalini 2010 as the primary study did not specify which groups the withdrawals came from.
Rahmani 2020 did not present any data on either adverse events or withdrawals from the study, and did not respond to our attempts at contact either via the corresponding author or via the journal in which the paper was published. As such, no data from this study are included in this analysis.
Adverse events
We analysed the number of participants experiencing any adverse events if this was explicitly stated in the primary studies or supplied on request from authors. If the total number of events was reported, but it was not clear how many participants experienced these events, we did not include these data for this outcome.
A meta‐analysis of seven studies with 489 participants showed no difference in adverse events between probiotics (1/249) and placebo (1/240) (RR 1.00, 95% CI 0.07 to 15.12) (Analysis 1.9). The evidence is of very low certainty due to imprecision from the very low numbers of events and risk of bias (summary of findings Table 1). These results are identical to the analysis above as all studies reported the same participant numbers for occurrence of adverse events and withdrawals due to adverse events. As reported above, Baştürk 2016 reported five post‐randomisation withdrawals from the study due to adverse events, but it was not stated whether these events occurred in the intervention or placebo group. Following a response to our request for clarification regarding these adverse events, we have now included the data in this analysis.
It was not possible to include figures for Giannetti 2017 in this analysis as although the primary study stated that some participants were lost post‐randomisation, the study was a cross‐over trial and did not break down withdrawals by group at randomisation.
It was also not possible to include figures for Guandalini 2010 as the primary study did not specify which groups the withdrawals came from.
Rahmani 2020 did not present any data on adverse events, and did not respond to our attempts at contact either via the corresponding author or via the journal in which the paper was published. As such, no data are included in this analysis.
School performance or change in school performance or attendance
Due to the different outcomes reported and measures used by the primary studies, it was not possible to perform a meta‐analysis for this outcome.
Gawrońska 2007 reported on school absenteeism at four weeks after the start of intervention, and found one participant absent at four weeks in the placebo group and no children absent in the placebo group.
Maragkoudaki 2017 reported on the average number of school absences per week at both the end of four weeks of intervention and at the end of follow‐up. At the end of four weeks of intervention the average number of days per week absent from school in the probiotics group was 0.07 ± 0.29 and the average number of days absent from school per week in the placebo group was 0.03 ± 0.15 (MD 0.04, 95% CI ‐0.10 to 0.17). At the end of follow‐up the average number of days per week absent from school in the probiotics group was 0.0 ± 0.0 and the average number of days absent from school per week in the placebo group was 0.11 ± 0.52 (MD 0.11, 95% CI ‐0.32 to 0.10). At neither time point were the results statistically significant.
Weizman 2016 reported on the number of days of school missed over the four‐week period of intervention. The average number of days missed in the probiotics group was 1.9 ± 1.1 and the average number of days missed in the placebo group was 2.7 ± 0.9 (P = 0.08).
Social and psychological functioning or change in social and psychological functioning
None of the included studies reported on this outcome.
Quality of life or change in quality life
None of the included studies reported on this outcome. Of note, Guandalini 2010 reported on a measure of quality of life, but did not use a validated measurement tool and so was not included in our review.
Synbiotic versus placebo
Primary outcomes
Treatment success
Meta‐analysis of four studies with 310 participants showed that patients with functional abdominal pain disorders may respond better to synbiotics (74/156) than placebo (54/154) (RR 1.34, 95% CI 1.03 to 1.74, I2 = 0%) (Analysis 2.1) (Baştürk 2016; Bauserman 2005; Otuzbir 2016; Saneian 2015). The evidence is of low certainty due to imprecision and risk of bias (summary of findings Table 2).
The results were consistent when we ran the analysis with a fixed‐effect model (RR 1.36, 95% CI 1.04 to 1.77, I2 = 0%) (Analysis 2.2). We conducted a sensitivity analysis, removing Otuzbir 2016 due to risk of bias, which showed no clear difference between synbiotics (49/117) and placebo (36/113) (RR 1.27, 95% CI 0.88 to 1.82, I2 = 6%) (Analysis 2.3). The evidence remains of low certainty due to serious imprecision.
The remaining synbiotics studies did not report this outcome.
Complete resolution of pain
Complete resolution of pain was reported in two studies with 131 participants (Baştürk 2016; Otuzbir 2016). The results showed no clear difference between synbiotics (34/65) and placebo (21/66) (RR 1.65, 95% CI 0.97 to 2.81) (Analysis 2.4). The evidence is of low certainty due to imprecision and risk of bias (summary of findings Table 2).
We conducted a sensitivity analysis, removing Otuzbir 2016 due to risk of bias. The results showed no clear difference between synbiotics (9/26) and placebo (3/25) (RR 2.88, 95% CI 0.88 to 9.44) (Analysis 2.5).
Severity of pain on completion
Severity of pain was reported in four studies with 319 participants (Asgarshirazi 2015; Bauserman 2005; Kianifar 2015; Saneian 2015). We conducted a meta‐analysis; however, we were unable to draw meaningful conclusions due to very high unexplained heterogeneity (Analysis 2.6). The evidence was of very low certainty. In line with our pre‐planned methodology for managing heterogeneity, a narrative synthesis is presented in summary of findings Table 2). No further conclusions could be drawn after inspection for risk of bias, and visual and clinical heterogeneity.
The remaining studies did not report this outcome or provided unclear results.
Frequency of pain on completion
Only one study with 80 participants reported results for frequency of pain on completion (Asgarshirazi 2015). This study measured frequency in episodes per week (MD ‐1.26, 95% CI ‐1.77 to ‐0.75, I2 = 0%) (Analysis 2.7). The certainty of the evidence is very low due to very high imprecision and risk of bias (summary of findings Table 2).
The remaining synbiotics studies did not report this outcome or presented results in a manner unsuitable for meta‐analysis.
Secondary outcomes
Serious adverse events
There were no recorded serious adverse events in any of the included studies within either the synbiotic or placebo groups.
Withdrawal due to adverse events
Meta‐analysis of four studies with 302 participants showed no difference in withdrawals due to adverse events between synbiotics (8/155) and placebo (1/147) (RR 4.58, 95% CI 0.80 to 26.19) (Analysis 2.8). The evidence is of very low certainty because of risk of bias and imprecision due to the very low numbers of events (summary of findings Table 2).
Baştürk 2016 reported five post‐randomisation withdrawals from the study, but it was not stated whether these withdrawals came from the intervention or placebo group. The author responded to a request to clarification regarding these withdrawals, so we have now included the data in this analysis.
Similarly, Kianifar 2015 reported five post‐randomisation withdrawals from the study, but again it was not stated whether these were from the intervention or placebo groups.
Otuzbir 2016 made no comment on adverse events or post‐randomisation withdrawals within the abstract we were able to review, and did not respond to our requests for further information.
Adverse events
Meta‐analysis of three studies with 189 participants showed no difference in adverse events between synbiotics (3/96) and placebo (1/93) (RR 2.88, 95% CI 0.32 to 25.92) (Analysis 2.9). The evidence is of very low certainty because of risk of bias and imprecision due to the very low numbers of events (summary of findings Table 2).
Baştürk 2016 reported five post‐randomisation withdrawals from the study, but it was not stated whether these withdrawals due to adverse events came from the intervention or placebo group. The author responded to a request to clarification regarding these withdrawals, so we have now included the data in this analysis.
Similarly, Kianifar 2015 reported five post‐randomisation withdrawals from the study due to adverse events (stated as "lack of follow up"), but again it was not stated whether these were from the intervention or placebo groups.
Otuzbir 2016 made no comment on adverse events within the abstract we were able to review, and did not respond to our requests for further information.
Saneian 2015 included a table showing adverse events broken down by symptom. Due to the possibility that some patients fell in to more than one of these categories, and that the data were not broken down by the number of patients suffering adverse events, we could not use these data for meta‐analysis.
School performance or change in school performance or attendance
None of the included studies reported on this outcome.
Social and psychological functioning or change in social and psychological functioning
None of the included studies reported on this outcome. Of note, Kianifar 2015 included disruption of social activities as an outcome in their methods, but did not report on this outcome in their results.
Quality of life or change in quality life
None of the included studies reported on this outcome.
Discussion
Summary of main results
This review includes 18 parallel‐group randomised controlled trials (RCTs). Twelve assessed the effectiveness of probiotics, five assessed the effectiveness of synbiotics, and one assessed the effectiveness of both probiotics and synbiotics in treating functional abdominal pain in childhood.
Probiotics
The results demonstrate that probiotics may achieve greater treatment success at study end than placebo in children with functional abdominal pain (low‐certainty evidence). Subgroup analysis for specific strains revealed low‐certainty evidence for Lactobacillus reuteri, Lactobacillus rhamnosus GG and Bifidobacterium lactis that there may be no difference to placebo. On sensitivity analysis using a fixed‐effect model, Lactobacillus reuteri showed a small increase in treatment success (number needed to treat for an additional beneficial outcome (NNTB) = 7) and Lactobacillus rhamnosus GG a large increase in treatment success (NNTB = 3) when compared to placebo, but this evidence was also of low certainty.
It is not clear whether probiotics are more effective than placebo for complete resolution of pain when compared with placebo (very low‐certainty evidence).
We were unable to draw meaningful conclusions from our meta‐analyses of the pain severity and pain frequency outcomes due to very high unexplained heterogeneity leading to very low‐certainty evidence.
There were insufficient data for subgroup analysis of treatment success or severity of pain on completion of treatment by specific diagnosis of either functional abdominal pain or irritable bowel syndrome.
No studies recorded serious adverse events. Very few withdrawals due to adverse events or adverse events (patient totals) were reported. No conclusions can be made regarding the comparison of probiotics and placebo for any of the adverse event outcomes due to the very low certainty of the evidence.
Synbiotics
Synbiotics may result in more treatment success at study end when compared with placebo for children with functional abdominal pain (low‐certainty evidence).
There may be no difference between synbiotics and placebo for complete resolution of pain.
We were unable to draw meaningful conclusions from our meta‐analyses of pain severity or frequency of pain due to very high unexplained heterogeneity leading to very low‐certainty evidence.
No studies recorded serious adverse events. Very few withdrawals due to adverse events or adverse events (patient totals) were reported. No conclusions can be made in the comparison of synbiotics and placebo for any of the adverse event outcomes due to the very low certainty of the evidence.
There were insufficient data on school performance or change in school performance or attendance, social and psychological functioning, or quality of life.
Overall completeness and applicability of evidence
The evidence is complete in a number of ways. Certainly the use of Rome diagnostic criteria in all studies (as required inclusion criteria) has ensured clinical homogeneity and applicability of the findings (Drossman 2006; Hyams 2016). Additionally, two probiotics in particular have been used in multiple studies, considering safety and efficacy. A range of ages of patients are included in the primary studies and the numbers of participants in the meta‐analyses are appropriate to support the findings. Finally, the range of primary outcomes expected were reported, with reasonable heterogeneity.
However, there are some areas of incomplete evidence. When considering the separate entities of irritable bowel syndrome or functional abdominal pain (Hyams 2016), there are insufficient studies to run these as separate analyses. Many studies considered these sub‐diagnostic categories as one and this is reflected in the analysis. However, this must be considered when applying the findings of this review in clinical practice. Additionally, whilst two particular preparations are the most commonly found, subgroup analysis is still impacted by imprecision due to low event and patient numbers, which reduces the certainty of the evidence. The number of studies is simply smaller when subgrouped by specific strain. The evidence for probiotics allowed meta‐analysis for two specific strains, Lactobacillus reuteri and Lactobacillus rhamnosus GG, but due to the low numbers the certainty of the results was impacted. In this area, the evidence is more complete than in previous reviews (Martin 2017), but further work may be indicated to enhance certainty.
Additionally, the majority of studies had short follow‐up (all less than 20 weeks). Given the chronic nature of these conditions and the length of symptoms needed to qualify for a Rome criteria diagnosis, this evidence does not consider the impact of cessation of therapy or long‐term continuation. This must also be considered when interpreting the evidence.
Finally, the reporting of harms is another area of concern. It is not uncommon to experience difficulties in reporting related to heterogeneity of thresholds of defining serious or severe adverse events, and as such withdrawals due to adverse events is often the most available measure. This is not necessarily the most important outcome for clinicians or patients and represents a gap in the completeness of the evidence that must be considered. This is further compounded by the findings in this review, which found in most cases that the number of withdrawals was identical or very close to the reported number of adverse events. This suggests that very few events occurred that were not at a level of severity that warranted withdrawal. This is of course possible, but it does raise a question about thresholds of reporting in these studies, which may be of interest to patients who may want to know of mild side effects, even if researchers deem them of minimal interest.
The other concern with randomised trial data is this is not necessarily the best method to comprehensively identify all safety issues, particularly rare issues. It has previously been noted that prophylactic use of probiotics in certain conditions has been associated with bowel ischaemia and increased mortality (Besselink 2008), as well as reports of sepsis secondary to Lactobacillus use (Boyle 2006), and other gastrointestinal side effects (Dore 2019). Such rare events are unlikely to be identified in the context of a randomised study, but are nonetheless of interest to prescribers and patients.
Quality of the evidence
We thoroughly assessed the included studies for quality and risk of bias. The evidence is overall at low risk of bias, as shown in Figure 2.
Publication bias could not be examined as there were insufficient studies in each analysis to create funnel plots.
One issue that was apparent was statistical heterogeneity in some of the analyses. Whilst in some analyses this could be explained via sensitivity analyses, for others, despite significant exploration, no reason could be found and therefore we could not pool the data in meta‐analysis. This means that the evidence in the analyses that could be run and that is presented in the summary of findings tables is predominantly of low or very low certainty, but a large proportion of planned analyses were not completed due to heterogeneity and this must be taken into account by readers.
Probiotics versus placebo
The certainty of the evidence for the treatment success outcome in the probiotics versus placebo comparison was compromised due to heterogeneity (I2 = 59%) and risk of bias (predominantly selective reporting) and for this reason we rated the evidence as low certainty. We conducted a sensitivity analysis with a fixed‐effect model, which showed statistical significance for the strains Lactobacillus rhamnosus GG and Lactobacillus reuteri in contrast with the original analysis, indicating further issues with heterogeneity. Given the clinical and methodological context of these studies, it is arguable that they are from subgroups in which a fixed‐effect model would be appropriate, but we did not believe that there was enough patient‐specific detail available to make such a judgement to present this single analysis.
The certainty of the evidence for the complete resolution of pain outcome in the probiotics versus placebo comparison was also severely compromised due to high inconsistency (I2 = 70%) which, when explored in a sensitivity analysis for clinical heterogeneity, was greatly reduced (I2 = 30%).
The severity of pain outcome in the same comparison was also severely impacted by heterogeneity (I2 > 75%). Sensitivity analyses based on clinical heterogeneity, risk of bias, random‐effects versus fixed‐effect models and abstract versus full‐text studies had the same issue, so we were unable to present data for this outcome, as per our pre‐planned methodology for managing very high heterogeneity.
The frequency of pain outcome presented the same issue in this comparison. A sensitivity analysis based on risk of bias showed no inconsistency and we were able to present the results.
We rated the certainty of the evidence for withdrawals due to adverse events and total adverse events for this comparison as very low because of imprecision due to the very low numbers of adverse event cases and risk of bias associated with selective reporting.
Synbiotics versus placebo
The certainty of the evidence for the treatment success outcome in the synbiotics versus placebo comparison was of low certainty due to imprecision and risk of bias from a study for which we only had abstract data (Otuzbir 2016). We rated the results of the risk of bias sensitivity analysis as low certainty because of imprecision due to the low numbers of participants.
We rated the evidence for complete resolution of pain in the synbiotics versus placebo comparison as low certainty because of issues with imprecision due to low participant numbers and risk of bias from a study for which we only had abstract data (Otuzbir 2016).
The severity of pain outcome in the same comparison was also severely impacted by heterogeneity (I2 > 75%), which we could not explain via sensitivity analyses and therefore we could not present the results, in line with the reasoning outlined above.
We rated the evidence for frequency of pain as of very low certainty because of imprecision due to low participant numbers and severe risk of bias from a single open‐label study that did not properly describe randomisation and allocation (Asgarshirazi 2015).
We rated the certainty of the evidence for withdrawals due to adverse events and total adverse events for this comparison as very low because of imprecision due to very low numbers of adverse event cases and risk of bias mainly from a single open‐label study that did not properly describe randomisation and allocation (Asgarshirazi 2015).
The primary evidence for all other secondary outcomes was poorly reported and no conclusions could be reached about them.
Finally, the reporting of adverse events was sparse and so this was also reflected in the GRADE analysis.
Potential biases in the review process
The definitions of the Rome process have changed in small ways over time. The bulk of the included studies used Rome III, with some using Rome II. None used the latest Rome IV criteria, so this must be considered when interpreting the findings of this review.
Some studies reported outcomes as proportions; in order to include the data in the analyses, the numbers of events were calculated by the review authors. We were able to minimise errors by having two independent authors to extract the data. Additionally, some studies reported mean and standard error of the mean (SEM) and thus the standard deviation (SD) had to be calculated. Finally, some studies reported median and range, and again the mean and SD had to be calculated.
We contacted study authors for additional information and clarification; however, some authors failed to reply. We will aim to include any data which become available in future updates.
We identified fewer than the recommended number of studies required to carry out some subgroup analysis, particularly by specific pain disorder. This is a significant issue in the primary literature and future studies should take this into account.
Two studies were only available as abstracts, but we explored their impact in sensitivity analysis when relevant.
Agreements and disagreements with other studies or reviews
A previously published Cochrane Review considered pharmacological treatments for recurrent abdominal pain of childhood (Martin 2017). Whilst this review did not use identical patient groups or disease types, and did not include probiotics as a pharmacological agent, it is worth noting that this review found no evidence to support any of the classes of agents studied (tricyclic antidepressants, antibiotics, 5‐HT4 receptor agonists, antispasmodics, antihistamines, H2 receptor antagonists, serotonin antagonists, selective serotonin re‐uptake inhibitors, a dopamine receptor antagonist and a hormone).
The North American Society for Pediatric Gastroenterology, Hepatology and Nutrition (NASPGHAN) commissioned a review in 2013 (van Tilburg 2013), and at the time there were only two randomised trials included. van Tilburg 2013 concluded that there was promising early evidence regarding probiotics and further research was needed. They in particular requested consideration of specific strains of probiotic and this has been achieved in the current review for both Lactobacillus reuteri and Lactobacillus rhamnosus GG.
There are currently no clear international guidelines for the use of these agents in children.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1: Probiotic versus placebo, Outcome 1: Treatment success
Comparison 1: Probiotic versus placebo, Outcome 2: Treatment success (sensitivity analysis: fixed‐effect model)
Comparison 1: Probiotic versus placebo, Outcome 3: Complete resolution of pain
Comparison 1: Probiotic versus placebo, Outcome 4: Complete resolution of pain (sensitivity analysis: risk of bias)
Comparison 1: Probiotic versus placebo, Outcome 5: Severity of pain
Comparison 1: Probiotic versus placebo, Outcome 6: Frequency of pain (episodes per week)
Comparison 1: Probiotic versus placebo, Outcome 7: Frequency of pain (episodes per week) (sensitivity analysis: risk of bias)
Comparison 1: Probiotic versus placebo, Outcome 8: Withdrawals due to adverse events
Comparison 1: Probiotic versus placebo, Outcome 9: Adverse events
Comparison 2: Synbiotics versus placebo, Outcome 1: Treatment success
Comparison 2: Synbiotics versus placebo, Outcome 2: Treatment success (sensitivity analysis: fixed‐effect model)
Comparison 2: Synbiotics versus placebo, Outcome 3: Treatment success (sensitivity analysis: risk of bias)
Comparison 2: Synbiotics versus placebo, Outcome 4: Complete resolution of pain
Comparison 2: Synbiotics versus placebo, Outcome 5: Complete resolution of pain (sensitivity analysis: risk of bias)
Comparison 2: Synbiotics versus placebo, Outcome 6: Severity of pain
Comparison 2: Synbiotics versus placebo, Outcome 7: Frequency of pain (episodes per week)
Comparison 2: Synbiotics versus placebo, Outcome 8: Withdrawals due to adverse events
Comparison 2: Synbiotics versus placebo, Outcome 9: Adverse events
| Probiotic compared to placebo for management of functional abdominal pain disorders in children | ||||||
| Patient or population: children (4 to 18 years) with functional abdominal pain disorders | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with probiotics | |||||
| Treatment success (at study end, as reported by study authors) | Study population | RR 1.57 (1.05 to 2.36) | 554 (6 studies) | ⊕⊕⊝⊝ | — | |
| 339 per 1000 | 532 per 1000 | |||||
| Complete resolution of pain (at study end, as reported by study authors) | Study population | RR 1.55 (0.94 to 2.56) | 460 (6 studies) | ⊕⊝⊝⊝ Very low b | — | |
| 272 per 1000 | 422 per 1000 (256 to 696) | |||||
| Severity of pain (at study end, using the Faces Pain Scale) | Severity of pain using the Faces Pain Scale when comparing probiotics versus placebo: SMD ‐0.28 (95% CI ‐0.67 to 0.12) | 665 participants | ⊕⊝⊝⊝ | — | ||
| Frequency of pain (at study end, episodes per week) | Frequency of pain episodes (per week) when comparing probiotics versus placebo: MD ‐0.43 (95% CI ‐0.92 to 0.07) | 605 participants | ⊕⊝⊝⊝ | — | ||
| Withdrawals due to adverse events | Study population | RR 1.00 (0.07 to 15.12) | 544 | ⊕⊝⊝⊝ Very low e | — | |
| 4 per 1000 | 4 per 1000 | |||||
| Serious adverse events | There were no SAEs reported within these studies in either group. | 685 | ⊕⊝⊝⊝ Very low e | — | ||
| Adverse events (any) | Study population | RR 1.00 (0.07 to 15.12) | 489 | ⊕⊝⊝⊝ Very low e | — | |
| 4 per 1000 | 4 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aDowngraded one level due to inconsistency (I² = 59% for both outcomes) and one level for risk of bias. bDowngraded three levels due to very high inconsistency (I² = 70%) and risk of bias (allocation concealment, attrition and reporting bias). cDowngraded three levels due to very high inconsistency (I² = 70%) and risk of bias (reporting bias). dDowngraded one level due to risk of bias. eDowngraded two levels due to imprecision because of very low numbers of adverse event cases and one level due to risk of bias. | ||||||
| Synbiotic compared to placebo for management of functional abdominal pain disorders in children | ||||||
| Patient or population: children (4 to 18 years) with functional abdominal pain disorders | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with synbiotic | |||||
| Treatment success (at study end, as reported by study authors) | Study population | RR 1.34 (1.03 to 1.74) | 310 | ⊕⊕⊝⊝ | — | |
| 350 per 1000 | 469 per 1000 | |||||
| Complete resolution of pain (at study end, as reported by study authors) | Study population | RR 1.65 (0.97 to 2.81) | 131 (2 studies) | ⊕⊕⊝⊝ | — | |
| 319 per 1000 | 405 per 1000 | |||||
| Severity of pain (at study end, using the Faces Pain Scale) | Severity of pain measured using the Faces Pain Scale for synbiotics versus placebo: MD ‐0.21 (95% CI ‐0.78 to 0.37) | 319 (4 studies) | ⊕⊝⊝⊝ | — | ||
| Frequency of pain (at study end, episodes per week) | The mean in the placebo group was 3.4 | MD 1.26 lower | — | 80 (1 study) | ⊕⊝⊝⊝ | — |
| Withdrawals due to adverse events | Study population | RR 4.58 | 302 | ⊕⊝⊝⊝ | — | |
| 7 per 1000 | 31 per 1000 | |||||
| Serious adverse events | There were no SAEs reported within these studies in either group | 302 | ⊕⊝⊝⊝ | — | ||
| Adverse events (any) | Study population | RR 2.88 | 189 | ⊕⊝⊝⊝ | — | |
| 11 per 1000 | 30 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aDowngraded one level for imprecision due to low participant numbers. bDowngraded one level due to risk of bias. cDowngraded two levels due to very serious unexplained heterogeneity, and one level due to risk of bias. dDowngraded two levels for severe risk of bias, due to unclear/high risk of bias for the single study that provided data for this outcome. eDowngraded two levels due to very serious imprecision from very low event numbers, and one level due to risk of bias. | ||||||
| Study ID | Interventional agent | Dosage (amount and frequency) | Control | Dosage (amount and frequency) | Trial registered(prospective/retrospective/none) | Trial registry outcomes published? | Conflicts of interest |
|---|---|---|---|---|---|---|---|
| Synbiotic group: Bifidobacterium coagulans + fructo‐oligosaccharide | 150 million spores of Bifidobacterium coagulans + fructo‐oligosaccharide twice daily | Peppermint group: peppermint oil (Colpermin)
Placebo group: folic acid | Peppermint group: 187 mg 3 times daily
Placebo group: 1 mg once daily | Prospective | Yes | None declared | |
| Synbiotic group: Bifidobacterium lactis B94 + inulin | 5 × 109 CFU Bifidobacterium lactis 900 mg inulin twice daily | Probiotic group: Bifidobacterium lactis
Prebiotic group: inulin
| Probiotic group: 5 × 109 CFU twice daily
Prebiotic group: 900 mg twice daily
| None | NA | None declared, no financial support received | |
| Synbiotic group: Lactobacillus GG + inulin | 1 x 1010 bacteria/capsule twice daily | Prebiotic group: inulin | Dose unstated (1 capsule twice daily) | None | NA | None declared | |
| Probiotic group: Lactobacillus reuteri | 1 x 108 CFU (5 drops per day) | Placebo group: unidentified placebo | Unstated | Retrospective | Yes | None declared, financial support from Zanjan University of Medical Sciences | |
| Probiotic group: Lactobacillus rhamnosus GG | 3 x 109 CFU twice daily | Placebo group: inert powder | Unstated dose twice daily | Retrospective | Yes | None declared, no financial support received | |
| Probiotic group: Lactobacillus rhamnosus GG | 3 x 109 CFU twice daily | Placebo group: powder | Unstated dose twice daily | None | NA | None declared, financial support from the Medical University of Warsaw | |
| Giannetti 2017 (cross‐over) | Probiotic group: Bifidobacterium longum BB536/ Bifidobacterium infantis M‐63/Bifidobacterium breve M‐16V | 1 sachet daily (3 billion/1 billion/1 billion per bacterium) | Placebo group: unidentified placebo | Unstated (1 sachet daily) | Retrospective | Yes | None declared |
| Guandalini 2010 (cross‐over) | Probiotic group: a patented probiotic preparation, which contains live, freeze‐dried lactic acid bacteria, at a total concentration of 450 billion lactic acid bacteria per sachet, comprising 8 different strains: Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium infantis, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus casei, Lactobacillus bulgaris and Streptococcus thermophilus | 1 sachet once daily if 4 to 11 years old or twice daily if 12 to 18 years old | Placebo group: unidentified placebo | 1 sachet once daily if 4 to 11 years old or twice daily if 12 to 18 years old) | None | NA | None declared, funding from locally available grants; no industry support other than providing probiotic and placebo products |
| Probiotic group: Lactobacillus reuteri DSM 17938 (tablet also containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid) | 1 x 108 CFU once daily (1 x 450 mg chewable tablet) | Placebo group: tablet containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid | Once daily (1 x 450 mg chewable tablet) | Prospective | Yes | None declared, no industry support other than providing probiotic and placebo products | |
| Probiotic group: Lactobacillus reuteri DSM 17938 (tablet also containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid) | 1 x 108 CFU once daily (1 x 450 mg chewable tablet) | Placebo group: tablet containing isomalt, xylitol, sucrose distearate, hydrogenated palm oil, lemon‐lime flavouring and citric acid | Once daily (one x 450 mg chewable tablet) | Prospective | Yes | Three contributing authors (Iva Hojsak, Sanja Kolacek, Zrinjka Misak) received either payment/honoraria for lectures or consultation, travel grants or lecture fees from several industry sources. All other authors declare no conflict of interest. | |
| Synbiotic group: Lactobacillus GG + inulin | 1 x 1010 CFU capsule twice daily | Prebiotic group: inulin | Unstated dose (1 capsule twice daily) | Prospective | Yes | None declared, funding received from Mashhad University of Medical Sciences, Iran | |
| Probiotic group: Lactobacillus reuteri DSM 17938 | 2 x 108 CFU (in the form of 2 chewable tablets once daily) | Placebo group: unidentified placebo | Unstated (2 chewable tablets once daily) | Prospective | Yes | Three contributing authors received research grants from BioGaia, 2 authors have been speakers for Biogaia and the remaining author had no conflicts to declare | |
| Synbiotic group | Not stated | Placebo group: unidentified | Not stated | None | NA | Abstract only, none declared | |
| Probiotic group: Lactobacillus reuteri | 1 x 108 CFU twice daily in the form of chewable tables | Placebo group: unidentified | Unstated dose (twice daily in the form of chewable tablets) | None | NA | None declared, funding from research centre | |
| Probiotic group: Lactobacillus reuteri DSM 17938 (product also containing sunflower oil, medium‐chain triglyceride oil from coconut oil) | 1 x 108 CFU twice daily in the form of a 10 mL bottle | Placebo group: product containing sunflower oil, medium‐chain triglyceride oil from coconut oil | 10 mL bottle twice daily | None | NA | None declared | |
| Probiotics group: Lactobacillus GG | Unstated dose | Placebo group: unidentified placebo | Unstated dose | None | NA | Abstract only, none declared | |
| Synbiotic group: Bacillus coagulans + fructo‐oligosaccharide | 150 million spores + fructo‐oligosaccharides 100 mg twice daily in the form of tablets | Placebo group: unidentified placebo | 1 tablet twice daily | Prospective | Yes | None declared, funding from Isfahan University of Medical Sciences | |
| Probiotic group: Lactobacillus reuteri DSM 17938 | 1 x 108 CFU once daily in the form of chewable tablet | Placebo group: unidentified placebo | Once daily in the form of chewable tablet | Prospective | Yes | One author (Zvi Weizman) has been a speaker for Biogaia AB which supplied the probiotic. No other conflicts of interest declared, and statement that Biogaia had no role in 'conception, design, and conduct of the study'. | |
| CFU: colony‐forming unit | |||||||
| Study ID | Methods of diagnosis | FAPD diagnosis | Separate data per sub‐diagnosis reported (yes/no) | Age range | Number of participants | Length of intervention | Time points of outcome measurements |
|---|---|---|---|---|---|---|---|
| Rome III | FAP/IBS/FD | No | 4 to 13 | 54 | 1 month | End of intervention | |
| Rome III | IBS | Not relevant as they only included one | 4 to 16 | 76 | 4 weeks | End of intervention | |
| Rome II | IBS | Not relevant as they only included one | 5 to 17 | 50 | 6 weeks | End of intervention | |
| Rome III | FAP | Not relevant as they only included one | 4 to 16 | 80 | 4 weeks | End of intervention; 4 weeks after end of intervention | |
| Rome II | FAP/IBS | Yes | 5 to 14 | 136 | 8 weeks | End of intervention; 8 weeks after end of intervention | |
| Rome II | FAP/IBS/FD | Yes | 6 to 16 | 104 | 4 weeks | End of intervention | |
| Giannetti 2017 (cross‐over) | Rome III | IBS/FD | Yes | 8 to 17 | 48 | 2 week run‐in period 6 weeks pre‐cross‐over phase 2 weeks washout 6 weeks post‐cross‐over phase | At the end of each period/phase and data combined at the end per intervention |
| Guandalini 2010 (cross‐over) | Rome II | IBS | Not relevant as they only included one | 4 to 18 | 59 | 2 week run‐in period 6 weeks pre‐cross‐over phase 2 weeks washout 6 weeks post‐cross‐over phase | Every 2 weeks and data combined at the end per intervention |
| Rome III | FAP/IBS | No | 4 to 18 | 55 | 12 weeks | 1 month into intervention; end of intervention; 1 month after end of intervention | |
| Rome III | FAP/IBS | No | 4 to 18 | 46 | 12 weeks | 1 month into intervention; end of intervention; 1 month after end of intervention | |
| Rome III | IBS | Not relevant as they only included one | 4 to 18 | 52 | 1 month | Weekly until end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 5 to 16 | 48 | 4 weeks | At 2 weeks and end of intervention | |
| Rome III | FAP/FD | No | Not stated | 80 | 8 weeks | End of intervention | |
| Rome III | FAP/IBS/FD/abdominal migraine | Yes | 6 to 16 | 125 | 4 weeks | At 2 weeks and end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 16 | 60 | 4 weeks | End of intervention; 4 weeks after end of intervention | |
| Unstated | FAP | Not relevant as they only included one | Unstated | 61 | 6 weeks | End of intervention; 4 weeks after end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 18 | 115 | 4 weeks | End of intervention; 8 weeks after end of intervention | |
| Rome III | FAP | Not relevant as they only included one | 6 to 15 | 101 | 4 weeks | End of intervention; 4 weeks after end of intervention | |
| FAP: functional abdominal pain | |||||||
| Study ID | 1a. Global improvement or treatment success | 1b. Complete resolution of pain | 1c. Severity of pain | 1d. Frequency of pain |
|---|---|---|---|---|
| NR | NR | Intervention group: 3.93 ± 1.06 Control group: 4.24 ± 1.33 | Intervention group: 2.14 ± 0.87 Control group: 3.40 ± 1.41 | |
| NR | Intervention group 1: 9/26 Intervention group 2: 7/25 Control group: 3/25 | NR | NR | |
| Intervention group: 11/32 Control group: 10/32 | NR | Change in pain intervention group: ‐1.3 (± 0.3) Change in pain control group: ‐1.7 (± 0.6) | NR | |
| No pain episodes per week, end of first month Intervention group: 20; control group: 26
No pain episodes per week, end of second month Intervention group: 19; control group: 21 | NR | End of first month intervention group: mean (SD) 2.50 (1.45); control group: mean (SD) 2.08 (1.56)
End of second month intervention group: mean (SD) 2.53 (1.43); control group: mean (SD) 2.25 (1.46) | At first month intervention group: mean (SD) 0.68 (0.76); control group: mean (SD) 0.40 (0.59)
At second month intervention group: mean (SD) 0.70 (0.75); control group: mean (SD) 0.53 (0.59) | |
| Decrease of at least 50% in the number of episodes and intensity of pain at 12 weeks Intervention group: 48/69; control group: 37/67
Decrease of at least 50% in the number of episodes and intensity of pain at end of follow‐up Intervention group: 53/69; control group: 43/67 | NR | At 12 weeks intervention group: mean (SD) 2.3 (1.3); control group: 3.4 (2.1)
At end of follow‐up intervention group: mean (SD) 0.9 (0.5); control group: 1.5 (1.0) | Number of episodes per week at 12 weeks intervention group: 1.1 (0.8); control group: 2.2 (1.2)
At end of follow‐up intervention group: 0.9 (0.5); control group: 1.5 (1.0) | |
| NR | Intervention group: 13/52 (FD 1/10, IBS 6/18, FAP 6/24)
Control group: 5/52 (FD 2/10, IBS 6/18, FAP 2/23) | At 4 weeks Intervention group: mean 2.5 (SD ± 1.9) (FD 2.9 ± 1.5, IBS 2.2 ± 2.1, FAP 2.6 ± 2.0) Control group: mean 2.9 (SD ± 1.5) (FD 1.9 ± 1.3, IBS 3.2 ± 1.5, FAP 3.0± 1.5) | At 4 weeks Intervention group: mean 2.2 (SD ± 1.7) (FD 2.7 ± 1.3, IBS 1.8 ± 1.7, FAP 2.3 ± 1.8) Control group: mean 2.6 (SD ± 1.4) (FD 2.0 ± 1.6, IBS 3.1 ± 1.1, FAP 2.4 ± 1.4) | |
| Giannetti 2017 (cross‐over) | NR | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data |
| Guandalini 2010 (cross‐over) | NR | NR | Not clear as the authors have combined pre‐ and post‐ cross‐over data | Not clear as the authors have combined pre‐ and post‐ cross‐over data |
| NR | Intervention group: 16/26 Control group: 16/29 | End of first month intervention group/control group: 0.75/0.96 End of second month intervention group/control group: 0.17/0.64 End of third month intervention group/control group: 0.32/0.71 End of fourth month intervention group/control group: 0.21/0.6
Difference in the severity of pain between first and fourth month, Wong‐Baker FACES/day intervention group: median 0.42 (range 0.31 to 2.9); control group: median 0.23 (range 1.2 to 2.2) | Number of days without pain intervention group at 4 months: 89.5 (range 5 to 108); control group at 4 months: 51 (range 0 to 107) | |
| NR | Intervention group: 10/24 Control group: 9/22 | End of first month intervention group/control group: 1.35 (IQR 0.64 to 1.98)/1.1 (IQR 0.76 to 2.04) End of second month intervention group/control group: 1.0 (IQR 0.09 to 2.12)/0.8 (IQR 0.37 to 1.68) End of third month intervention group/control group: 0.83 (IQR 0.025 to 2.26)/0.78 (IQR 0.43 to 2.0) End of fourth month intervention group/control group: 0.035 (IQR 0 to 1.0)/0.81 (IQR 0.2 to 1.48)
Change in severity of pain from 1st to 4th month intervention group: 0.55 (IQR 0.28 to 0.55); control group: median 0.36 (IQR ‐0.14 to 0.36) | Number of days without pain intervention group at 4 months: 90 (IQR 54 to 99); control group at 4 months: 59.5 (IQR 21.5 to 89.25) | |
| NR | NR | 1 week: intervention group/control group 1.5 (1.0)/1.8 (0.6) 2 weeks: intervention group/control group 1.2 (1.1)/1.9 (0.8) 3 weeks: intervention group/control group 1.0 (0.9)/1.8 (0.6) 4 weeks: intervention group/control group 0.8 (0.9)/1.5 (0.8) | NR | |
| Reduction in pain score of greater than 50% at 4 weeks intervention group: 19/27 (70.4%); control group: 16/27 (58.3%)
Reduction in pain score of greater than 50% at 8 weeks intervention group: 17/25 (65.4%); control group: 13/23 (56.5%) | NR | Intervention group/control group: mean (SD) 2 weeks: 10.4 (18.8)/12.2 (17.3) 4 weeks: 4.3 (8.5)/4.0 (5.6) 8 weeks: 7.2 (17.7)/2.5 (3.4) | Intervention group/control group: mean (SD) 2 weeks: 5.6 (8.1)/8.2 (10.7) 4 weeks: 2.9 (4.5)/3.1 (4.1) 8 weeks: 4.8 (9.9)/2.8 (3.3) | |
| NR | Intervention group: 25/39 Control group: 18/41 | NR | NR | |
| Intervention group = 32/65 (FAP 13/28, FD 11/16, IBS 6/15, AM 2/6)
Control group = 8/60 (FAP 8/29, FD 0/13, IBS 0/6, AM, 0/3)
| NR (in Rahmani 2020, treatment success was defined as pain intensity = 0) | Text: severity at 4 weeks in intervention group = 1.3 ± 1.1 (Table 1 reports: 1.1 ± 1.3)
Text: severity at 4 weeks in control group = 1 ± 2 (Table 1 reports: 2 ± 1)
FAP (intervention group/control group): 1.2 ± 1.3; 2 ± 1 FD (intervention group/control group): 0.8 ± 1.5; 2.0 ± 6 IBS (intervention group/control group): 1.4 ± 1.4; 2.8 ± 0.8 AM (intervention group/control group): 1.3 ± 1.5; 2.3 ± 0.5 | Text: frequency of repetitive pain at 4 weeks intervention group 3.6 ± 2.2 (Table 1 reports: intervention group 2.2 ± 3.6)
Text: frequency of repetitive pain at 4 weeks control group 4.6 ± 4.9 (Table 1 reports: control group 4.9 ± 4.6)
FAP (intervention group/control group): 2.1 ± 2.7; 4.1 ± 4.4 FD (intervention group/control group): 1.6 ± 3.0; 6.0 ± 5.0 IBS (intervention group/control group): 3.7 ± 5.5; 6.3 ± 0.8 AM (intervention group/control group): 1.1 ± 0.9; 1.3 ± 0.5 | |
| NR | NR | Mean (SD) as we interpreted it from the figures: Week 4 intervention group/control group: 1.25 (0.9)/2 (0.8) Week 8 intervention group/control group: 1 (0.7)/2 (0.8) | Mean (SD) as we interpreted it from the figures: Week 4 intervention group/control group: 1.4 (1.1)/2.2 (0.5) per day Week 8 intervention group/control group: 2.1 (0.6)/2 (0.5) per day | |
| NR | NR | NR | NR | |
| Response at week 4 intervention group: 27/45; control group: 17/43
Response at week 12 intervention group: 29/45; control group: 23/43 | NR | Change in pain scale from start of intervention to week 4 intervention group: mean ‐1.7 (SD ± 1.5); control group: mean ‐1.6 (SD ± 1.5)
Change in pain scale from start of intervention to week 12 intervention group: mean ‐2.1 (SD ± 1.4); control group: mean ‐1.8 (SD ± 1.4) | NR | |
| NR | NR | Improvement in intensity of abdominal pain at 4 weeks intervention group: mean 4.3 (SD ± 2.7); control group: mean 7.2 (SD ± 3.1)
Improvement in intensity of abdominal pain at end 8 weeks intervention group: mean 4.8 (SD ± 3.3); control group: mean 6.4 (SD ± 4.1) | Number of episodes of pain per week at 4 weeks intervention group: mean 1.9 (SD ± 0.8); control group: mean 3.6 (SD ± 1.7)
Number of episodes of pain per week at 8 weeks intervention group: mean 3.4 (SD ± 2.6); control group: mean 4.4 (SD ± 2.9)
| |
| Numbers presented as per the original study reports. AM: abdominal migraine | ||||
| Study ID | 2a. Serious adverse events | 2b. Withdrawal due to adverse events | 2c. Adverse events | 2d. School performance | 2e. Social and psychological functioning | 2f. Quality of life |
|---|---|---|---|---|---|---|
| 0 | 0 | 0 | NR | NR | NR | |
| NR | Intervention group 1: 3 | Intervention group 1: 3 | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | 0 | 0 | School absenteeism at end of intervention | NR | NR | |
| Giannetti 2017 (cross‐over) | 0 | 0 | 0 | NR | NR | NR |
| Guandalini 2010 (cross‐over) | 0 | 0 | 0 | NR | NR | Questionnaire of disruption to family life (change in score) Intervention group: mean ‐0.9 (SD ± 0.2); control group: mean ‐0.51 (SD ± 0.3) |
| NR | 0 | NR | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| 0 | 0 | 0 | NR | Functional changes on a 3 point Likert scale at end of intervention Intervention group: mean 2.4 (SD ± 0.5); control group: mean 1.9 (SD ± 0.4) | NR | |
| 0 | 0 | 0 | Average number of school absences per week at end of follow‐up Intervention group: mean 0.0 (SD ± 0.0); control group: mean 0.11 (SD ± 0.52) | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | 0 | 0 | NR | NR | NR | |
| NR | NR | NR | NR | NR | NR | |
| 0 | Intervention group: 5; control group:0 | NR as numbers of people with adverse events
Total number of adverse events intervention group: 45; control group: 43 | NR | NR | NR | |
| 0 | 0 | 0 | Days of school absenteeism over 4 weeks Intervention group: mean 2.7 (SD ± 0.9); control group: mean 1.9 (SD ± 1.1) | NR | NR | |
| Numbers presented as per the original study reports. NR: not reported | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Treatment success Show forest plot | 6 | 554 | Risk Ratio (M‐H, Random, 95% CI) | 1.57 [1.05, 2.36] |
| 1.1.1 Lactobacillus reuteri | 3 | 259 | Risk Ratio (M‐H, Random, 95% CI) | 1.57 [0.73, 3.37] |
| 1.1.2 Lactobacillus rhamnosus GG | 2 | 245 | Risk Ratio (M‐H, Random, 95% CI) | 1.57 [0.73, 3.34] |
| 1.1.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Random, 95% CI) | 2.33 [0.68, 8.01] |
| 1.2 Treatment success (sensitivity analysis: fixed‐effect model) Show forest plot | 6 | 554 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.49 [1.23, 1.80] |
| 1.2.1 Lactobacillus reuteri | 3 | 259 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.58 [1.17, 2.12] |
| 1.2.2 Lactobacillus rhamnosus GG | 2 | 245 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.36 [1.07, 1.72] |
| 1.2.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.33 [0.68, 8.01] |
| 1.3 Complete resolution of pain Show forest plot | 6 | 460 | Risk Ratio (M‐H, Random, 95% CI) | 1.55 [0.94, 2.56] |
| 1.3.1 Lactobacillus reuteri | 4 | 306 | Risk Ratio (M‐H, Random, 95% CI) | 1.35 [0.76, 2.41] |
| 1.3.2 Lactobacillus rhamnosus GG | 1 | 104 | Risk Ratio (M‐H, Random, 95% CI) | 2.60 [1.00, 6.77] |
| 1.3.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Random, 95% CI) | 2.33 [0.68, 8.01] |
| 1.4 Complete resolution of pain (sensitivity analysis: risk of bias) Show forest plot | 5 | 335 | Risk Ratio (M‐H, Random, 95% CI) | 1.18 [0.84, 1.67] |
| 1.4.1 Lactobacillus reuteri | 3 | 181 | Risk Ratio (M‐H, Random, 95% CI) | 1.01 [0.76, 1.34] |
| 1.4.2 Lactobacillus rhamnosus GG | 1 | 104 | Risk Ratio (M‐H, Random, 95% CI) | 2.60 [1.00, 6.77] |
| 1.4.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Random, 95% CI) | 2.33 [0.68, 8.01] |
| 1.5 Severity of pain Show forest plot | 7 | 665 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.28 [‐0.67, 0.12] |
| 1.5.1 Faces scales | 6 | 524 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.19 [‐0.61, 0.23] |
| 1.5.2 Combination VAS‐Faces scale | 1 | 141 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.76 [‐1.10, ‐0.41] |
| 1.6 Frequency of pain (episodes per week) Show forest plot | 6 | 605 | Mean Difference (IV, Random, 95% CI) | ‐0.43 [‐0.92, 0.07] |
| 1.6.1 Lactobacillus reuteri | 4 | 360 | Mean Difference (IV, Random, 95% CI) | ‐0.43 [‐1.42, 0.56] |
| 1.6.2 Lactobacillus rhamnosus GG | 2 | 245 | Mean Difference (IV, Random, 95% CI) | ‐0.57 [‐0.81, ‐0.33] |
| 1.7 Frequency of pain (episodes per week) (sensitivity analysis: risk of bias) Show forest plot | 4 | 400 | Mean Difference (IV, Random, 95% CI) | ‐0.58 [‐0.81, ‐0.35] |
| 1.7.1 Lactobacillus reuteri | 2 | 155 | Mean Difference (IV, Random, 95% CI) | ‐0.12 [‐2.80, 2.55] |
| 1.7.2 Lactobacillus rhamnosus GG | 2 | 245 | Mean Difference (IV, Random, 95% CI) | ‐0.57 [‐0.81, ‐0.33] |
| 1.8 Withdrawals due to adverse events Show forest plot | 8 | 544 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.07, 15.12] |
| 1.8.1 Lactobacillus reuteri | 6 | 390 | Risk Ratio (M‐H, Random, 95% CI) | Not estimable |
| 1.8.2 Lactobacillus rhamnosus GG | 1 | 104 | Risk Ratio (M‐H, Random, 95% CI) | Not estimable |
| 1.8.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.07, 15.12] |
| 1.9 Adverse events Show forest plot | 7 | 489 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.07, 15.12] |
| 1.9.1 Lactobacillus reuteri | 5 | 335 | Risk Ratio (M‐H, Random, 95% CI) | Not estimable |
| 1.9.2 Lactobacillus rhamnosus GG | 1 | 104 | Risk Ratio (M‐H, Random, 95% CI) | Not estimable |
| 1.9.3 Bifidobacterium lactis | 1 | 50 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.07, 15.12] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Treatment success Show forest plot | 4 | 310 | Risk Ratio (M‐H, Random, 95% CI) | 1.34 [1.03, 1.74] |
| 2.2 Treatment success (sensitivity analysis: fixed‐effect model) Show forest plot | 4 | 310 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.36 [1.04, 1.77] |
| 2.3 Treatment success (sensitivity analysis: risk of bias) Show forest plot | 3 | 230 | Risk Ratio (M‐H, Random, 95% CI) | 1.27 [0.88, 1.82] |
| 2.4 Complete resolution of pain Show forest plot | 2 | 131 | Risk Ratio (M‐H, Random, 95% CI) | 1.65 [0.97, 2.81] |
| 2.5 Complete resolution of pain (sensitivity analysis: risk of bias) Show forest plot | 1 | 51 | Risk Ratio (M‐H, Random, 95% CI) | 2.88 [0.88, 9.44] |
| 2.6 Severity of pain Show forest plot | 4 | 319 | Mean Difference (IV, Random, 95% CI) | ‐0.21 [‐0.78, 0.37] |
| 2.6.1 Likert scales | 2 | 124 | Mean Difference (IV, Random, 95% CI) | ‐0.13 [‐1.21, 0.94] |
| 2.6.2 Faces scales | 1 | 115 | Mean Difference (IV, Random, 95% CI) | ‐0.30 [‐0.81, 0.21] |
| 2.6.3 Visual analogue scales | 1 | 80 | Mean Difference (IV, Random, 95% CI) | ‐0.31 [‐0.84, 0.22] |
| 2.7 Frequency of pain (episodes per week) Show forest plot | 1 | 80 | Mean Difference (IV, Random, 95% CI) | ‐1.26 [‐1.77, ‐0.75] |
| 2.8 Withdrawals due to adverse events Show forest plot | 4 | 302 | Risk Ratio (M‐H, Random, 95% CI) | 4.58 [0.80, 26.19] |
| 2.9 Adverse events Show forest plot | 3 | 189 | Risk Ratio (M‐H, Random, 95% CI) | 2.88 [0.32, 25.92] |
