Infant Colic—What works

A Systematic Review of Interventions for Breast-fed Infants

Abstract

Objectives: 

To determine the strength of evidence for commonly used interventions for colic in breast-fed and mixed-fed infants younger than 6 months.

Methods: 

Searches of PubMed, CINAHL, Embase, AMED, and Web of Science databases were conducted from July 2014 to July 2015. Included studies were randomised controlled trials involving mothers and their colicky infants younger than 6 months; assessed colic against the Wessel or modified Wessel criteria; and included phytotherapies, prescription medicines, and maternal dietary interventions. Studies with <16 participants were excluded. Meta-analyses were conducted where data were sufficient to enable pooling. Quality was assessed against the Cochrane Risk Bias Assessment Tool.

Results: 

A total of 17 articles met the inclusion criteria for this review. The 6 studies included for subgroup meta-analysis on probiotic treatment, notably Lactobacillus reuteri, demonstrated that probiotics appear an effective treatment, with an overall mean difference in crying time at day 21 of −55.8 min/day (95% CI −64.4 to −47.3, P = 0.001). The 3 studies included for subgroup meta-analysis on preparations containing fennel suggest it to be effective, with an overall mean difference of −72.1 min/day (95% CI −126.4 to −17.7, P < 0.001).

Conclusions: 

Probiotics, in particular L reuteri, and preparations containing fennel oil appear effective for reducing colic, although there are limitations to these findings. The evidence for maternal dietary manipulation, lactase, sucrose, glucose, and simethicone is weak. Further well-designed clinical trials are required to strengthen the evidence for all of these interventions.

What Is Known

• A universally accepted definition of infant colic would enhance the effectiveness of research in this area and enable more studies to be compared for a more meaningful outcome of any further systematic reviews.
• Colic is considered multifactorial in nature.
• Maternal dietary manipulation, probiotics, and some nonprescription remedies may be of benefit for breast-fed colicky infants.

What Is New

• Meta-analysis of 6 randomised controlled trials of Lactobacillus reuteri demonstrated an overall positive effect for colicky breast-fed infants.
• Meta-analysis of preparations containing fennel demonstrated positive effect for colicky breast-fed infants.
• The significant heterogeneity demonstrated here may be addressed through the adoption of standardised case definitions and data collection methodologies.

The term “infant colic” is often used by health care professionals to describe a collection of symptoms, including persistent crying and fussiness, with the affected infant unable to settle or self sooth and sometimes unable to feed properly. It is recognised as a functional gastrointestinal disorder of infancy by the Rome III classification ⁽¹⁾. Persistent infant colic can contribute to parental fatigue and distress and may result in strained parental relationships, and poor parental engagement with their infant ⁽²⁾.

A varied amount of crying daily is considered normal in typically developing infants, with the duration increasing from birth and peaking at approximately 6 weeks ^(3–5). Infant colic has been the subject of many investigations, and since 1954 it has been defined as “crying for >3 hours per day, for >3 days per week, for a period of 3 weeks or longer in otherwise healthy infants”; typically known as Wessel “rule of threes” ^(6,7). The Rome III Criteria of Childhood Functional Gastrointestinal Disorders: Neonate/Toddler defined infant colic as crying for >3 hours per day, for >3 days per week, for at least 1 week, with no failure to thrive ⁽¹⁾.

Establishing an accurate prevalence for infant colic has been problematic due to the various definitions of colic used in research. Canivet et al ⁽⁸⁾ cite a prevalence of 9% to 60% internationally, depending on the definitions and research methodologies used. Interestingly, their research revealed that when 4 distinct definitions were used the prevalence varied from 3.3% to 17.1%, with the prevalence in the Wessel group to be 9.3% ⁽⁸⁾. Other research suggests prevalence varies between 4% and 28% when defined by modified Wessel criteria ^(9,10). More recently, Vandenplas et al ⁽¹¹⁾ in their worldwide survey of paediatric health professionals together with their study of the literature estimated the overall prevalence for colic to be in the order of 20%; however, they concede that the actual prevalence remains uncertain. The importance of using a standardised definition of infant colic for research purposes has been emphasised by Reijneveld et al ⁽⁹⁾, because it is more likely to result in a rigorous study that can be compared with other research through systematic review. These findings strengthen the case for standardising the definition of colic, perhaps with the Rome III criteria, which would enable a more accurate prevalence to be established.

It has been widely recognised that the causes of infant colic are multifactorial, with maternal, paternal, infant, and environmental factors being implicated. Maternal considerations include stress and postnatal depression ⁽¹²⁾, whereas infant factors may include the individual infant's temperament, developmental milestones achieved, the infant's sensory processing capacity, and underlying organic causes ^(7,13). Environmental causes, such as exposure to tobacco smoke, have also been implicated in infant colic ⁽¹⁴⁾. The role of the infant gastrointestinal microbiota has recently come into focus, with several randomised controlled trials investigating the efficacy of Lactobacillus reuteri as a treatment option ^(15–20). Organic causes are thought to include food protein allergy ^(21,22) and/or food intolerances ⁽²³⁾.

Parents of unsettled infants often seek medical assistance to determine the cause of their infant's distress, resulting in high utilisation of health services ^(24,25). There is an excess of information for parents on the Internet; a recent Google search using the search term “infant colic” yielded >1.5 million results; however, much of this advice is conflicting and possibly contributes to parental confusion. The question remains as to how much of the information is evidence based. There have been a number of reviews of treatments for infant colic since 2000 ^(26–37), some narrative and some systematic (Table 1). The review herein differs from previous reviews in that we focussed our review around narrow inclusion criteria, which increases its strength. Our review considers studies that include exclusively breast-fed and mixed-fed infants, whereas 11 of the 12 previous reviews ^{(26–33,35–37)} considered studies of exclusively formula fed in addition to exclusively breast-fed and mixed-fed infants. Our review considers only studies with experimental design, whereas, 7 of the 12 prior reviews ^{(26,27,31–35)} included studies of experimental and observational design, which may reduce their rigour. In addition, our review considers studies which define infant colic as Wessel or modified Wessel criteria, whereas 10 of the 12 previous reviews ^(26–35) considered studies using Wessel, modified Wessel criteria, as well as unspecified or another definitions of colic. Using a narrow definition of infant colic strengthens our review. Our review also includes meta-analyses, whereas, only 2 of the previous reviews ^(28,36) were able to pool data for meta-analysis for probiotics (L reuteri). Our review provides a rigorous examination of commonly used interventions that are easily accessed by parents of colicky infants and aims to provide up to date evidence on the efficacy of common treatments for infant colic for health professionals to convey to the parents of breast-fed and mixed-fed colicky infants.

Table 1: A selection of reviews of interventions for infant colic published since 2000

Table 1: (Continued) A selection of reviews of interventions for infant colic published since 2000

Table 1: (Continued) A selection of reviews of interventions for infant colic published since 2000

METHODS

Inclusion/Exclusion Criteria

This systematic review was compliant with the Preferred Reporting Items of Systematic Reviews and Meta-analysis framework ⁽³⁸⁾. Studies for inclusion were required to be of experimental design, either randomised-controlled or randomised crossover trials, and published after 1 January, 1980. The participants were mothers of colicky fully breast-fed or partially breast-fed infants younger than six months. Studies included in this review defined colic as either Wessel criteria ⁽⁶⁾, or a modified Wessel criteria, as per the Rome III criteria. Included studies also demonstrated the following intervention types: maternal diet; pharmacotherapies, both prescription and over the counter; phytotherapies; and probiotics. Studies were excluded if the sample size was <16 participants, because n = 16 is the minimum number of participants required to show a difference of 1 standard deviation. Outcome measures for included studies were changes in crying duration and response rates as measured by a reduction in symptoms. Table 2 details the Population, Intervention, Comparison and Outcome (PICO) statement for this systematic review, with inclusion and exclusion criteria.

Table 2: PICO inclusion/exclusion criteria

Literature Search

An extensive search of the following 5 databases was conducted from July 2014 to 31 July, 2015: PubMed, CINAHL, Embase, AMED, and Web of Science. Search terms included MeSH headings and search strings, and were informed by prior reading of the literature for treating infant colic. They included the following: Feeding OR Breastfeeding AND Colic OR Irritability OR fussiness AND crying; Infant AND Colic; Maternal diet AND Colic AND Breastfed infants; Infant AND Colic AND Diet; Infant AND Colic AND Reflux; Infant AND Colic AND Medication; Infant AND Colic AND PPIs; Infant AND Colic AND Simethicone; Infant AND Colic AND Probiotics; Infant AND Colic AND Herbal teas; Infant AND Colic AND Sucrose. These search strings were chosen as prior examination of the literature on treating colic suggested that they represent the most common treatment options.

Study Selection Process

Studies were scrutinised based on the inclusion and exclusion criteria by T.H. and M.M. independently. Consensus was reached before excluding any studies at each key step in the selection process. Consensus regarding the final list of studies for inclusion was reached by T.H., M.M., and R.H.

Data Extraction Process and Assessment of Bias

The data extraction process was completed by T.H. and independently verified by M.M. and R.H. Where possible, authors were contacted for any data that were missing. Where studies had reported crying times at baseline and during or after the intervention, these were extracted. Crying times were often reported as mean with standard deviation in minutes per day. Where crying was reported in hours per day, these were converted to minutes per day. Where data were reported in original publications as medians and interquartile ranges, these were converted to means and standard deviations using the formulae in Hozo et al 2005 ⁽³⁹⁾, because these formulae made no assumptions about the original data. In addition, where numbers, or percentages of infants responding to treatment were reported, these data were extracted. The quality of the included studies were assessed using the Cochrane Risk of Bias Assessment Tool, published in the Cochrane Handbook for Systematic Reviews of Interventions⁽⁴⁰⁾.

Data Analysis Process

Separate subgroup meta-analyses were conducted for studies investigating probiotics and studies examining the effectiveness of fennel oil, where these studies reported sufficient outcome data to enable pooling for meta-analysis. Studies of other interventions did not contain sufficient data for subgroup meta-analysis. Mean differences (MDs) with standard deviations for continuous outcomes were calculated from stated baseline and outcome measures of crying duration. For 1 study ⁽⁴¹⁾, the data on herbal tea containing fennel were extracted, compared with the control pre- and postintervention, and these data were then available to be considered in the meta-analysis for tea containing fennel. All of the data were analysed with STATA/SE 13.1 (StataCorp LP, Collge Station, TX) statistical software, using random effects model. Heterogeneity was assessed using the I² statistic. Publication bias was assessed using log odds ratio and funnel plot.

RESULTS

Study Selection

Initial searches resulted in the extraction of 1485 titles. After 380 duplicates were excluded, 1105 articles were reviewed by title. Of these, 935 were excluded by title. The remaining 170 abstracts were reviewed. Of these abstracts, 114 were excluded, leaving 56 full-text articles to be retrieved for review. Of the 56 full-text articles, 39 were excluded on the basis of not meeting the inclusion criteria. This process is detailed in Fig. 1. Studies using dicylcomine HCL and cimetropium bromide were excluded at this stage because it became evident that these products were no longer available in a range of countries, with dicyclomine HCL being contraindicated for infants younger than 6 months in many countries. Studies were also excluded at this stage in which more information was sought from authors, and none was available, or the authors were unable to be contacted.

FIGURE 1: Study selection process.

Study Characteristics and Assessment of Bias Risk

Seventeen studies were included in this systematic review, all examining the effects of a treatment on infant colic in fully or partially breast-fed babies. The interventions included: probiotics/synbiotics (N = 7), maternal dietary interventions (N = 1), preparations containing fennel (N = 3), or other nonprescription remedies (N = 6). Of the 17 studies selected, 12 were randomised controlled trials ^{(15–18,20,41–47)} and the remaining 5 were randomised crossover trials ^(48–52). The randomised crossover trials examined the effects of over the counter remedies including sucrose, glucose, lactase, peppermint, and simethicone. The ingredients of each intervention and placebo are detailed at Table 4. The quality of the 17 included studies were assessed using the Cochrane Risk of Bias Assessment Tool ⁽⁴⁰⁾ (Table 3).

Table 3: Study characteristics

Table 3: (Continued) Study characteristics

Table 3: (Continued) Study characteristics

Table 4: Table of ingredients in the active or placebo preparations used in the studies outlined in Table 3

Table 4: (Continued) Table of ingredients in the active or placebo preparations used in the studies outlined in Table 3

Probiotics/Synbiotics

A subgroup meta-analysis of MDs in crying times for the probiotic L reuteri was performed on data at 21 days of treatment (Fig. 2), because this was the common data collection point for these studies. Overall, L reuteri reduced crying time in the infants studied (pooled MD −55.9 min/day, 95% CI −64.4 to −47.3, P < 0.001) (Figure 2).

FIGURE 2: Subgroup meta-analysis of the effect of probiotics (L reuteri) at 21 days.

The study by Kianifar et al ⁽⁴⁴⁾ examined the effects of a synbiotic mixture that contained 6 species of microbiota (Lactobacillus casei, Lactobacillus acidophilus, Bifidobacterium. infantis, Lactobacillus bulgaricus, Lactobacillus rhamnossus, Streptococcus thermophilus, Bifidobacterium breve) plus fructo-oligosaccharide, and despite the availability of crying data pre and post intervention, it was not included in the subgroup meta-analysis because the preparation did not contain L reuteri and was considered not directly comparable to the studies using L reuteri.

Preparations Containing Fennel

A subgroup meta-analysis was conducted on the 3 studies testing the efficacy of fennel on reducing crying time in infants (Fig. 3). Overall, preparations containing fennel demonstrated effectiveness with a pooled MD of −72.1 min/day (95% CI −126.4 to −17.7, P < 0.01). The Alexandrovich et al ⁽⁴²⁾ study included 125 infants of 2 to 12 weeks, with response rates and cumulative crying being the reported outcomes. The authors reported response to treatment rates for 65% of infants, whereas 23.7% also responded to the placebo, demonstrating a significant placebo effect in this study.

FIGURE 3: Subgroup meta-analysis of the effect of preparations containing fennel at 7 days. CI = confidence interval.

Arikan et al ⁽⁴¹⁾ studied 4 intervention groups and a control group with no intervention. Infants in treatment group 3 (n = 35) were given tea containing fennel. Baseline mean crying time for the fennel tea group was 306.6 min/day (±85.8), and at the end of the intervention it had reduced to 192.0 min/day (±73.8). When compared with the other 2 studies in this subgroup meta-analysis, the MD was −109.2 min/day (95% CI −219.1 to 0.7, P = ns). Because of the large confidence interval and it crossing the line of no effect, the study was, however, considered of limited value.

Savino et al 2005 ⁽⁴⁵⁾ examined the effects of a commercially available phytotherapeutic agent containing fennel (ColiMil), with simethicone used as the comparator (n = 88). The study outcomes were reported as crying durations and response rates. Mean crying duration at baseline in the treatment group was 201.2 min/day (±18.3), reducing to 76.9 min/day (±23.5) at day 7, equating to a mean reduction of crying of −124.3 min/day (±11.8), P < 0.001. This study reported a response rate in the treatment group of 85.4% and 48.9% for the placebo, which is a high placebo response. In the subgroup meta-analysis, the ColiMil containing fennel appeared to be an effective treatment for infant colic, with an MD in crying time of −95.5 min/day, (95% CI −100.9 to −90.2, P < 0.001). ColiMil, however, contains 4 herbal ingredients and it is unclear how much of the preparation's effectiveness was due to the fennel, other ingredients, or the synergistic effect of the combination of the 4 herbal ingredients.

Interestingly, the study by Weizman et al 1993 ⁽⁴⁷⁾ also found that a herbal tea containing fennel was effective compared with a placebo. It was, however, unable to be included in the subgroup analysis because no crying data at baseline or posttreatment were available.

Low Allergen Maternal Diet

Because only 1 study on maternal diet was included in this review ⁽⁴³⁾, it was, therefore, not possible to conduct a subgroup meta-analysis on the effects of maternal dietary elimination on crying time. The authors, however, reported a greater reduction in cry/fuss time in the maternal low allergen diet group, with an adjusted geometric mean ratio of 0.79 (95% CI 0.63–0.97) over a 48-hour period.

Simethicone

Three studies examined simethicone ^(17,49,51). The Alves et al ⁽⁴⁹⁾ and Savino et al ⁽¹⁷⁾ studies used simethicone as the control intervention, whereas the Metcalf et al ⁽⁵¹⁾ study aimed to determine the effect of simethicone on infant colic. The Metcalf et al ⁽⁵¹⁾ study found no significant difference when simethicone was compared with a placebo. Interestingly, the Alves et al study ⁽⁴⁹⁾ showed no differences in the infant's response to the active treatment (peppermint) when compared with simethicone as the control, demonstrating that the treatment and placebo had similar effect. Savino et al ⁽¹⁷⁾ also used simethicone as the control (vs L reuteri), which was found to be of limited use for relieving infant colic. Overall, it appears that simethicone is not an effective treatment for infant colic.

Glucose and Sucrose

Two studies, Arikan et al ⁽⁴¹⁾ and Markestad ⁽⁵⁰⁾, testing the effectiveness of sucrose, and 1 study, Akcam et al ⁽⁴⁸⁾, testing the effectiveness of glucose, were included. The Arikan et al study ⁽⁴¹⁾, a randomised controlled trial, examined the effect of 4 treatments and a control (no intervention), with n = 35 in each group. One treatment arm used a sucrose solution (12% in distilled water), and crying duration in the sucrose group reduced significantly from 342.6 min/day (102.6) to 236.4 min/day (90.6), with an MD of 100.8 min/day (16.8), P < 0.001.

The Markestad study ⁽⁵⁰⁾ was a small crossover trial study (n = 19), again testing the effectiveness of a 12% sucrose solution. The author reported only response rates, and in the sucrose group this was 63%, which appears to show its effectiveness for reducing infant colic.

Akcam et al ⁽⁴⁸⁾ tested the effectiveness of glucose for reducing infant colic with a 30% solution in sterile water in a small crossover trial (n = 25). Outcomes were reported as response rates, with 64% responding to the glucose solution and 48% responding to the placebo.

Lactase

Kanabar et al ⁽⁵²⁾ reported a crossover trial (n = 53), with an intervention period of 10 days duration and included a 5-day washout period. The authors reported outcomes as response rates for both intention to treat (ITT) analysis (26%, 95% CI 12.9–44.4) and for per protocol (PP) analysis for compliant participants (38%, 95% CI 18.8–59.4). The reported reduction in crying duration between the lactase and placebo groups for ITT was 22.4%, P = ns, and for PP 40.4%, P < 0.01.

Heterogeneity

There was considerable heterogeneity between the studies examined here. In terms of subgroups meta-analyses, heterogeneity was considerable for both probiotics (L reuteri) and tea containing fennel, with I² = 77.1%, P = 0.001, and I² = 99.5%, P < 0.01, respectively, albeit much more pronounced for fennel tea; thus random effects models were chosen for these analyses. This was expected because each of the studies within each subgroup analysis considered here differed in many ways, and the limitations will be further discussed in this review.

Publication Bias

Publication bias was assessed with a funnel plot (Fig. 4) and was demonstrated in this review, as evidenced by a high level of heterogeneity between studies ⁽⁵³⁾; differences in methodological quality between studies contributing to the funnel plot's asymmetry and publication bias, that is, some studies ^(41,48,51) examined have been assessed as high risk of bias; 40% of the studies ^{(16,17,19,20,50,51)} are presenting as outliers in the funnel plot; and not all studies in this review reporting their findings in a way that was useful for future meta-analyses. Unpublished studies and grey literature were not considered in this review and may partially explain the reason for the obvious publication bias demonstrated here.

FIGURE 4: Funnel plot with pseudo 95% confidence limits. OR = odds ratio.

DISCUSSION

This systematic review has achieved its aim of determining the strength of evidence for treatments for infant colic in breast-fed and predominately breast-fed infants younger than 6 months. A range of systematic and literature reviews on treatments for infant colic have been published since 2000 ^{(26,28–37,54)}. This review differs greatly from these other previously published reviews with respect to its targeted approach. This systematic review was conducted with a narrow set of inclusion criteria, which examined randomised trials of maternal elimination diets, probiotics/synbiotics, over the counter and prescription remedies, and some herbal preparations. Studies were only included if they defined colic as per Wessel or modified Wessel criteria; studies were conducted with 16 or more participants and involved breast-fed infants, mixed-fed infants, or formula-fed infants in which these infants were given the same intervention as the breast- or mixed-fed infants. Some previous reviews did not report their inclusion criteria, or study selection methods ⁽³²⁾. Although many reviews limited their investigation to studies defining colic as Wessel or modified Wessel criteria ^{(28,29,31,35,36)}, some included studies with varying definitions of colic ^(26,30,33).

With regard to probiotics/synbiotics, our review, with its inclusion of 3 additional studies since previously published meta-analyses of probiotics for treating infant colic ^(28,36), strengthens the evidence for probiotics/synbiotics in alleviating infant colic, in particular, L reuteri. Several earlier reviews ^{(28,31,32,35,36,54)} support this. For example, Sung et al ⁽²⁸⁾ included 3 of the studies that feature in our present review ^(17,18,20). Their meta-analysis concluded that L reuteri was effective for reducing colic. In a more recent review, Sung ⁽³⁷⁾ concluded that L reuteri appears effective for breast-fed infants only, and that only a subset of colicky infants (no formula feedings and no diagnosis of gastro-oesophageal reflux) may benefit from supplementation with L reuteri, and that it cannot be recommended for formula-fed infants at this time. The authors, however, caution that the studies they examined were potentially affected by bias. Our meta-analysis also supports that L reuteri is effective for the treatment of infant colic, for exclusively breast-fed infants, and, the studies meeting our strict inclusion/exclusion criteria showed low risk of bias. We also, however, found that L reuteri was effective for mixed-fed and formula-fed infants.

Further well-designed studies, with standardised definitions of colic are needed to strengthen the evidence base for colic interventions. Presently, there is significant heterogeneity between studies, and several limitations exist within the intervention studies examined herein.

Limitations of Studies Using Probiotics/Synbiotics

The results of the Sung et al study ⁽¹⁹⁾ are contrary to the finding of the others included in this review regarding the usefulness of L reuteri in treating infant colic; this has generated some international controversy. In their original study, Sung et al ⁽¹⁹⁾ suggest that their controversial findings may be due to their study having a larger sample size, compared with previous work in this area; atopy and allergy status of the participants; and/or the possibility of regional differences in endemic gut flora. With respect to the latter, regional differences have been found in gut microbiota, correlated with latitude. The Sung et al study ⁽¹⁹⁾ was conducted in Australia, whereas the other studies are Italian ^(17,18), Canadian ⁽¹⁵⁾, and Polish ⁽²⁰⁾, and apparent regional differences in gut microbiota may serve to explain Sung et al's findings ⁽¹⁹⁾. Interestingly, when we performed a sensitivity analysis, removing the effect of the Sung et al study ⁽¹⁹⁾, this made minimal difference to the overall effectiveness for treating infant colic (MD in crying time −56.4 min/day, 95% CI −64.8 to −47.9, P < 0.001). All of the studies of probiotics/synbiotics examined were assessed as low risk for bias.

Limitations of Studies Using Preparations Containing Fennel

Alexandrovich et al ⁽⁴²⁾ offer a possible mechanism of action for the efficacy of fennel oil. The fennel plant contains many biologically active compounds of unknown effect, and these may have a spasmolytic effect on the smooth muscle of the intestine. Although the authors report a 65% response rate in the fennel group, they also report a higher than expected response rate (23.7%) in the placebo group, noting a significant placebo effect in this study. A similar placebo effect was reported in the Weizman et al study ⁽⁴⁷⁾, in which 26% responded to the placebo, compared with 57% in the fennel group. Interestingly, the response rate for the placebo group of infants in the Savino et al study ⁽⁴⁵⁾ was even higher at 48.9%, compared with 85.4% in the ColiMil group, again, representing a significant placebo effect. The treatment period in all of these studies using preparations containing fennel was 7 days, and perhaps if the treatment period was longer a greater effect in the treatment group may have been observed.

Consideration of the role of the placebo effect is important, because 5 of the studies examined in our review demonstrated little or no more effect than the placebo they were compared with ^{(19,48,49,51,55)}. The placebo effect has been widely investigated, and it has been recognised as neurobiological ^(56,57), meaning that the expectation of a treatment or placebo being effective creates a biological effect. It is feasible that parents with irritable infants will find any treatment or placebo effective because their participation in any investigation may create an expectation. Interestingly, the validity of the placebo effect has been demonstrated in a recently published study by Partty et al ⁽⁵⁸⁾, in which 2 data collection methods were used to collect crying data, notably, parental interviews, and validated baby diaries. They found that when parents were asked at interview whether their infants had responded to treatment, 87% reported yes, whereas analysis of the baby diaries showed 0% responded to treatment; this demonstrates a very high-perceived difference by the parents.

There were several methodological issues with the studies investigating fennel. Randomisation in the study of Alexandrovich et al ⁽⁴²⁾ was conducted on an individual basis, and allocation methods were not clearly reported. This contributed to the bias risk score of medium. The Arikan et al study ⁽⁴¹⁾ was not blinded, and there were several additional aspects of the study that contributed to the overall rating as high risk for bias, for example, allocation was not clearly stated, sample size calculations were not included, and compliance checks were not stated. In the study by Savino et al ⁽⁴⁵⁾, sample size calculations were not specified, compliance checking was not stated, and the balance of ITT and PP analyses were not clear; these issues contributed to the overall rating of medium risk of bias. Weizman et al ⁽⁴⁷⁾ methodology was poorly reported and many aspects contributed to an overall rating of medium risk of bias. For example, blinding was stated as double-blind, but very little information on allocation and concealment was reported to substantiate this. Furthermore, a subjective parental colic improvement score was used, which may be problematic considering the placebo effect, and sample size calculations were not reported. Considering all of the studies examining the effects of preparations containing fennel demonstrated methodological issues, and the aforementioned high placebo response demonstrated that any conclusion regarding the effectiveness of fennel must be drawn with caution.

Limitations of Studies Using Low Allergen Maternal Diet

Two studies on maternal dietary interventions were identified to be meeting the inclusion criteria for this review ^(43,55); however, 1 of these studies also met the exclusion criteria and was removed from the review. Hill et al ⁽⁵⁵⁾ included both breast-fed and formula-fed infants, but applied different interventions for each group. The formula-fed infants were randomised to receive either hydrolysed formula or a placebo of standard cow's milk formula as their dietary intervention, whereas the breast-fed infants underwent maternal dietary intervention. This study was unable to be included in our review because the data were reported for both groups combined, with no information given separately for breast-fed infants. As such, these authors reported findings for 2 completely different interventions, in 2 different populations, as a combined response rate. Therefore, it was not possible to determine whether maternal dietary intervention was an effective treatment option for reducing infant colic in breast-fed infants with the study of Hill et al ⁽⁵⁵⁾.

With respect to a study from Hill et al ⁽⁴³⁾, maternal dietary intervention was found to be effective; however, poor compliance with the dietary intervention in the control group was a major limitation. Only 59% of mothers in the control group were compliant with their intervention, the remaining 41% were partially compliant. This may have skewed the control crying data; however, we cannot be certain of how this has affected the results. In addition, the intervention duration of 1 week may not have been sufficient to eliminate maternal allergens from breast milk, because other studies have shown that some allergens may be present in human breast milk for 9 days and beyond post elimination ^(59–61). It is clear that more well-designed randomised controlled trials of maternal dietary interventions that are of sufficient duration need to be conducted for exclusively breast-fed infants to determine whether this simple noninvasive approach to treating infant colic is effective.

Limitations of Studies Using Lactase

One study investigating lactase, by Kanabar et al ⁽⁵²⁾, was included in this review. Although these authors did not report baseline crying data, they reported a difference in crying times between the treatment and placebo groups. In addition, they reported response rates overall; however, they did not indicate rates for treatment and placebo groups separately, and as such, this study was unable to be included in the overall meta-analysis of any treatment versus placebo. It was included in the narrative section of this review, because it met the inclusion criteria specified by the PICO statement. Although this study was assessed as low risk for bias based on its methodology, the poor reporting of their findings has contributed to insufficient evidence for lactase as an effective treatment for infant colic.

The study of Miller et al ⁽⁶²⁾ also investigated the effectiveness of lactase, however, was excluded because it did not meet the criteria specified by the PICO statement. This study was a small crossover trial (n = 15), with a 7-day intervention period and no washout period. No response rates or differences in crying durations between baseline and at the end of treatment were reported. The results of hydrogen breath testing were, however, reported, and it was found that concentrations of hydrogen in the breath did not differ significantly between the lactase and the control groups, being indicative of a lack of response to treatment. The authors report that no significant difference in crying duration was demonstrated between the lactase group when compared with the placebo group.

Limitations of the Crossover Design Studies

A major limitation with crossover trials is the length of time allocated for any washout period, where inadequate or nonexistent washout periods may cause carry over effects from the initial treatment. Any carryover effects can introduce bias into the study ^(63,64). Of the 5 crossover trials ^(48–52), the Kanabar et al study ⁽⁵²⁾ reported a 5-day washout period, the Alves et al study ⁽⁴⁹⁾ reported a 3-day washout period, and the Metcalf et al study ⁽⁵¹⁾ reported a 1-day washout period, which may not have been adequate for the colic to reappear before the second treatment round. Two studies that did not specify a washout period ^(48,50) involved saccharides, 1 of these studies ⁽⁵⁰⁾ also repeated the crossover intervention and the author conceded that the nature of this repeated crossover design may have contributed to the findings.

Future Directions

Further well-designed randomised controlled trials or randomised crossover trials with appropriate washout periods are needed to strengthen the evidence for colic interventions. Any future research methodology should seek to address the high levels of heterogeneity within the present body of research for infant colic. This could be achieved by ensuring a standardised definition of colic, which could be based on the internationally agreed Rome III criteria. Data collection methods should also be standardised, this could include the use of validated tools that capture infant crying in real time, either paper-based or in electronic format to ensure consistency of the data, and ease of data collection for mothers participating in research. Also, a consistent approach to measuring the rates of responding infants would ensure less heterogeneity, this could be achieved by researchers reaching consensus on defining what is meant by a “responder” to treatment by standardising the percentage of reduced crying time to either 25% and more or 50% and more, for example. In addition, standardised reporting of crying data in mean minutes per day, with standard deviations, would be useful to assist with comparisons across similar interventions.

With regard to probiotics/synbiotics, more well-designed trials with this L reuteri, particularly in the Southern hemisphere, would be useful in addressing the present controversy surrounding the findings of Sung et al ⁽¹⁹⁾, thereby strengthening the emerging body of research.

With regard to future research investigating maternal dietary interventions for colic, heterogeneity can be reduced by ensuring that the infants of participating mothers are exclusive or fully breast-fed, and that the mothers are subject to only 1 intervention, for example, a hypoallergenic diet or another type of diet, and that these diets are well managed by the inclusion of a clinical dietitian in the research team.

CONCLUSIONS

The evidence presented in this review suggests that probiotics containing L reuteri appear to be an effective treatment option for colic in breast-fed infants younger than 6 months. All of the studies selected for this review testing the effectiveness of probiotics were assessed as low risk for bias against the Cochrane Risk Bias Assessment Tool ⁽⁴⁰⁾ and were considered high quality investigations. Although meta-analysis presented evidence for the effectiveness of preparations containing fennel, this evidence must be viewed with caution because the included studies were of variable quality, had methodological issues, and were assessed as medium or high risk of bias. There was limited evidence for the use of glucose, sucrose, lactase, and simethicone as effective treatments for infant colic in breast-fed infants. The studies presented herein were not without methodological issues and were of variable quality. The high level of heterogeneity between the studies examined is a limiting factor to this analysis. There is limited evidence for the efficacy of maternal dietary manipulation through randomised controlled trials, and more well-designed studies are required in this area.