Food allergy , particularly cow’s milk protein allergy, is widely regarded as the most common organic deficit causing prolonged infant crying [49, 50]. Although the incidence is unknown, one estimate is that 2 % of infants overall are food allergic [49]. It is unclear how many such infants cry a lot. Cow’s milk protein allergy probably accounts for less than 5 % of cases of colic [51], and should be considered if the infant has other symptoms such as bloody or mucousy diarrhoea, failure to thrive, poor feeding, significant vomiting, eczema and family history of atopy [52, 53]. In such cases, a limited trial of hypoallergenic formula, or maternal cow’s milk protein elimination diet in exclusively breastfed infants is recommended.

However, the use of hypoallergenic formula or excluding dairy from the mother’s diet cannot be recommended for all infants with colic. The reasons are threefold. First, maternal hypoallergenic diets are difficult to follow and maintain. Second, only a minority of infants with colic respond to cow’s milk protein elimination [49, 54]. Third, the evidence for their use in infant colic is not conclusive. Although at least six randomised trials have suggested extensively hydrolysed formulae—that is, cow’s milk-based formula with the proteins broken down by enzymes—to be effective in managing infant colic [3, 55–59], and only two trials have indicated them to be ineffective [60, 61], all trials have methodological limitations. Many were not blinded [55, 59] or inadequately blinded [56, 60], used inappropriate comparators (e.g. dietary modification versus medication [55], a hydrolysed formula versus another hydrolysed formula [57]), were biased by crossover effects [55], or included infants who may not have had true colic [58]. The majority of studies did not clearly describe the method of generating random allocation sequence or the process of randomisation. The effectiveness of partially hydrolysed formulae is more controversial, with one unblinded trial showing effectiveness [62] and the other not [63]. Two studies suggested a completely hydrolysed formula (that is, amino acid-based formula) to be effective, but both had small sample sizes and neither were randomised trials [64, 65]. Studies examining the elimination of cow’s milk protein from the breastfeeding mother’s diet have also yielded contradictory results, with four indicating effectiveness [56, 60, 66, 67] and two not [55, 68]. Two systematic reviews in 2012 concluded that there was evidence to suggest hydrolysed formulae to be beneficial to infants with colic; however, most of the studies had methodological inconsistencies and biases. They concluded that the role of maternal hypoallergenic diets in breastfed infants was less clear [69, 70].

Soy formula has been shown in one randomised trial to be effective [71], with two other poor quality trials supporting this [55, 72]. However, the methodological flaws of the latter two trials mean the results cannot be conclusive. One trial used dicyclomine hydrochloride rather than a true placebo as the comparator [55]. The other was a small trial of 19 infants, and the study was possibly biased by crossover effects [72]. One other trial found soy formula to be just as effective as partially hydrolysed formula, without comparing them to a cow’s milk-based formula [63]. Without a proper placebo for comparison, the results cannot be conclusive. In addition, soy formula is no longer recommended for infants less than 6 months old due to inconclusive evidence that it may cause harm to young infants (through excessive phytoestrogen compounds) [73].

Excessive intragastrointestinal air, or colonic gas, has often been proposed as a cause of infant colic. However, little evidence has come forward to support this theory. In 1969, Harley et al. demonstrated radiographically normal gastric outlines during colic episodes [74]. Measures to prevent aerophagia are ineffective in preventing colic [75], and the use of gas-reducing agents such as simethicone are also conclusively ineffective [76, 77]. Moreover, intragastrointestinal gas could be a result of swallowed air from crying—that is, an effect, and not a cause [78].

Early authors proposed that colonic hyperperistalsis , spasms or increased rectal pressure could underlie colic [79]. This is based on the evidence that dicyclomine hydrochloride, an agent that may relax colonic smooth muscle and reduce spasms, is effective in treating colic [80–84]. This drug, however, has significant, potentially life-threatening side effects, and therefore is no longer recommended for use in infants. A trial of a herbal tea containing an antispasmodic was also effective [85], but the mixture was associated with possible carcinogenic effects and therefore not recommended [86]. These findings do, however, provide evidence that gut spasms may contribute to colic in some infants. The gut hormones motilin and ghrelin, both stimulators of gastric motility, have been found to be higher in infants with colic compared to controls [87, 88]. Whether these findings have practical applications is not yet clear.

Many authors have suggested carbohydrate malabsorption, lactose overload or lactase insufficiency as causes for infant colic. The excess carbohydrates, such as lactose, in the gut may cause bloating and discomfort. Excess lactose could be a result of lactase insufficiency or lactose overload from excessive consumption of, for example, lactose-rich human foremilk. However, there is insufficient evidence for this as a cause of colic [51]. Studies examining breath hydrogen levels (a by-product of lactose malabsorption) [89–92] and the use of lactase supplementation or lactose elimination have yielded conflicting results [49, 93–96]. The evidence for lactose explanations of colic is weak. Other dietary changes, such as altering the concentrations of fibre [97] and carbohydrates in formulae [98], are ineffective in infants with colic.

The role of gut microbiota in infant colic has recently been under intense scrutiny. The first study that examined this in 1994 found no significant differences in gut microbiota between infants with and without colic, except for Clostridium difficile (C. difficile) which more frequently colonised infants with colic during the time of peak crying when compared to controls [111]. C. difficile is a bacterium known to cause antibiotic-associated diarrhoea.

Over the last 10 years, an Italian group has documented in four separate cross-sectional studies (n = 56–87) that breastfed infants with colic have different gut microbiota compared to breastfed infants without colic. However, these differences have not been consistent. In their 2004 study, breastfed infants with colic were less frequently colonised by Lactobacillus species than those without colic [112]. Yet, in another study the following year, breastfed infants with and without colic had similar overall colonisation rates but different patterns of Lactobacillus species—Lactobacillus brevis and Lactobacillus lactis lactis colonised only those with colic, while Lactobacillus acidophilus colonised only those without colic [113]. The infants with colic were also more likely to have a family history of atopy [113]. This could be a significant confounder, considering that a study in 2010 suggested that infants with cow’s milk protein allergy have higher gut concentrations of anaerobic and Lactobacillus species, and lower gut concentrations of Bifidobacteria , when compared with non-allergic controls [114]. In 2009, the Italian group suggested breastfed infants with colic had more gas-forming coliforms and E. coli concentrations than those without colic [115]. The group’s 2011 study replicated these findings, and suggested that two out of 27 Lactobacillus strains tested had an antimicrobial effect against six gas-forming coliform species isolated from breastfed infants with colic [116]. This finding is interestingly congruent with the group’s 2004 study showing less Lactobacillus species in infants with colic compared to those without [112], further supporting the notion that infants with colic may have more gas-forming coliform species that contribute to gaseous distension and subsequent distress.

Other studies have also suggested a role of gut microbiota in infant colic. In a 2008 Finnish study of 18 breastfed infants with and without colic, those with colic had higher prevalence of indole-producing coliforms, such as Escherichia and Klebsiella species [117]. In a 2009 study of 36 infants with and without colic, those with colic had lower microbial diversity and greater levels of Klebsiella species, while the control group had Enterobacter and Pantoea species that were not detected in the group with colic [118]. A 2012 Finnish study of 89 healthy infants without colic who were at risk of developing allergies suggested Bifidobacterium and Lactobacillus species to be protective against crying and fussing in the first 3 months of life [119]. A 2013 Iranian study of 70 breastfed infants with and without colic found Lactobacillus acidophilus to be present in 20 % of infants without colic, but absent in those with colic [120].

Two recent studies have implicated Helicobacter pylori (H. pylori) in infant colic. A 2012 Saudi Arabian study of 85 infants with and without colic found a higher proportion of infants with colic had positive H. pylori stool antigen compared to those without colic [121], and similar results were found in a 2013 Egyptian study of 100 infants with and without colic [122]. This finding is interesting considering H. pylori is an organism known to cause chronic gastric inflammation [123, 124], and is a well-known cause of gastritis in adults. However, H. pylori colonisation is often asymptomatic. It may be associated with short-term abdominal pain lasting less than 3 months, but not chronic recurrent abdominal pain in children [125]. There is conflicting evidence for its association with epigastric pain in children [125]. In addition, the prevalence of H. pylori infection in children, although not extensively described, is believed to be very low and, therefore, its role in infants is uncertain. The prevalence of H. pylori in children and adolescents in Europe and North America is less than 10 % [126]. In a study of children in Chile, the prevalence of H. pylori was as low as 1 % at 3 months of age, rapidly increasing after 15 months of age to 20 % at 24 months of age—an age by which colic has well and truly resolved [127].

The most exciting recent development in the search of an effective dietary strategy to manage infant colic is the use of probiotics. Probiotics are ‘live micro-organisms which, when administered in adequate amounts, confer a health benefit on the host’ [128, 129]. Probiotics may affect infant crying by altering gut microbiota, leading to changes in gut-mediated pain perception, gut motility, mucosal layers and gut permeability, reduction in gut inflammation and inhibition of gut bacteria. One particular strain of probiotic, Lactobacillus reuteri, has been shown to be effective in reducing infant crying in four randomised trials of exclusively breastfed infants with colic, at a dose of 1 × 10⁸ cfu per day [130–133]. Sample sizes ranged from 50 to 83 in the Italian, Polish and Canadian studies. However, a larger randomised trial of the probiotic in both breastfed and formula-fed infants with colic in Australia (n = 167) concluded it to be ineffective [134]. The most likely reasons for the Australian trial’s controversial results lie in the pragmatic nature of the trial, the use of a more objective and precise measure of primary outcome, the larger sample size involved, and possibly the older age of infants recruited. It is possible that the probiotic was ineffective in infants from Australia because of undetermined differences in infant gut microbiota compared with European or Canadian infants. In addition, one of the Italian studies was not blinded [132], and both Italian studies included exclusively breastfed infants whose mothers were all on dairy-elimination diets [131, 132].

Because of this controversy, a collaboration between the authors of the five aforementioned studies is currently undertaking an individual participant data meta-analysis (IPDMA) to pool data from each individual trial into one dataset for analysis [135]. This creates sufficient power to perform subgroup analyses [136, 137]. Upcoming results from this IPDMA may inform which subgroups of infants with colic can benefit from Lactobacillus reuteri.