Re: Treating infant colic with the probiotic Lactobacillus reuteri: double blind, placebo controlled randomised trial
Probiotics for colic: A hammer/nail issue or a signal-noise problem?
To the editor:
We read with interest the comprehensive study of Sung et al. describing the results of L. reuteri treatment for infants with colic but disagree with Dr. Bennet’s summation of the evidence. A vast lay literature without scientific evidence has emerged over the years to help parents cope with infants with colic. Now that a few placebo-controlled trials have been completed, the evidence is not clear. How can doctors consider probiotic research “a hammer in search of a nail?” Should we “throw in the bib” and devote precious resources to “reassurance, family social support, and the tincture of time?”
Our research group in Houston has investigated (under 3 protocols) more than 130 infants with colic over the past 5 years. We, too, have been plagued by difficult enrollment of infants, many who have been pre-exposed to acid blockers and probiotics in the formula. These problems that may have contributing factor to negative results. Sung et al. have reported that they have used regression models to estimate treatment effects by adjusting for potential confounding factors at baseline. Though it appears that they have considered potential confounding, it is not clear if they have tested for effect modifications in their regression models. Instead they have reported subgroup analyses. The most efficient way is to test for effect modifications in a regression model rather than subgroup analyses by stratification. It is certainly possible the treatment is not uniformly effective for all eligible children.
Although Sung et al. have reported that in their unadjusted analyses, they compared all continuous outcomes by t-test and non-parametric methods, it is not clear why they did not use a nonparametric regression model (e.g., Quantile regression) for their adjusted analyses through multivariable analyses. Because crying time usually has a skewed distribution, it would have been more appropriate to use regression models that are likely to meet the underlying assumptions. Furthermore, for comparisons of their outcomes during the follow up period, is not clear as why they have not taken advantage of longitudinal analyses. For example, Generalized Estimating Equations (GEE) methods are considered more appropriate for comparisons of outcomes during the follow up period.
Despite these concerns, it is not clear if the main results would change if their data were analyzed using more sophisticated techniques. What does seem clear is that pediatricians will continue to be interested in proper studies of this potential treatment for colic, because without them they will likely continue to use the “shotgun” approach. Extreme stress is very evident in the families, and it is palpable. The typical infant with colic cries and fusses for 4-5 hours/day, and the parents get little sleep. Colic has been associated with thoughts of infanticide in ~25% of parents (1). As evidence for the psychosocial impact of infant crying, recordings of infants with colic were used (perhaps inappropriately) for interrogation of suspected terrorists in Guantanamo Bay. (https://www.google.com/search?q=colic+and+guantanamo&oq=colic+and+guanta...).
One biomarker of colic deserves comment. We previously found elevated fecal calprotectin in all infants at this age, with values even higher in those with colic than in age-matched infants (2), but, as the authors mention, Olafsdotir et al. did not make the same observation in 2002. We speculated that gut inflammation may contribute to symptoms in at least some of these infants. This contention is not rejected and is partially supported by the Sung et al. study and unpublished data from our group showing ~50% lower levels of fecal calprotectin after resolution of colic. We are currently studying calprotectin and other biomarkers in a F.D.A.-regulated, N.I.H./NCCAM-sponsored U.S. trial.
With respect to microbiota differences in colic, there are a number of published observations from different groups, mostly concluding that there are changes in the microbiota of infants with colic. We could not find a description of how the researchers measured microbial diversity or E. coli abundance. Ref #34 does not describe microbiota or diversity. Standardization across studies of the approaches used to analyze of the microbiota would be quite helpful in making sense of variations in outcomes.
We agree with the need for evidence-based guidelines in pediatric practice world-wide. There are now well-supported guidelines for treatment of otitis and constipation (3, 4). In contrast, comprehensive, high-quality studies are just emerging in the probiotic field, of which the current study is a good example. There is accumulating evidence (shown in the forest plot of Sung et al. included in their study) that probiotics may indeed have a role in colic. Could “the forest be obscured by the trees” in this report? Fully 33% of all infants had been treated with a formula containing probiotics before the trial. This reflects a huge problem with premature marketing of formulas and probiotics as definitive treatments for infants with colic. We agree with Sung et al. that “It will be valuable to pool data from all existing and ongoing rigorous trials…to investigate with greater certainty which subgroups … could benefit from probiotics.”
Finally, there is evidence to support the contention infants with colic may have long-term effects on gastroenterological functions or behaviors in their later childhood (5).
Reference List
(1) Levitzky S, Cooper R. Infant colic syndrome--maternal fantasies of aggression and infanticide. Clin Pediatr (Phila) 2000 Jul;39(7):395-400.
(2) Rhoads JM, Fatheree NY, Norori J, Liu Y, Lucke JF, Tyson JE, et al. Altered fecal microflora and increased fecal calprotectin in infants with colic. J Pediatr 2009 Dec;155(6):823-8.
(3) Forrest CB, Fiks AG, Bailey LC, Localio R, Grundmeier RW, Richards T, et al. Improving adherence to otitis media guidelines with clinical decision support and physician feedback. Pediatrics 2013 Apr;131(4):e1071-e1081.
(4) Venekamp RP, Sanders S, Glasziou PP, Del Mar CB, Rovers MM. Antibiotics for acute otitis media in children. Cochrane Database Syst Rev 2013;1:CD000219.
(5) Savino F, Castagno E, Bretto R, Brondello C, Palumeri E, Oggero R. A prospective 10-year study on children who had severe infantile colic. Acta Paediatr Suppl 2005 Oct;94(449):129-32.
Competing interests:
The laboratory of Dr. Rhoads received financial support for animal research with L. reuteri from BioGaia, Inc. (Clinical colic funding was from the U.S. National Institutes of Health).
12 April 2014
J. Marc Rhoads
Pediatric Gastroenterologist
Mohammad H Rahbar, Michael Ferris, Yuying Liu, and Nicole Y. Fatheree
Rapid Response:
Re: Treating infant colic with the probiotic Lactobacillus reuteri: double blind, placebo controlled randomised trial
Probiotics for colic: A hammer/nail issue or a signal-noise problem?
To the editor:
We read with interest the comprehensive study of Sung et al. describing the results of L. reuteri treatment for infants with colic but disagree with Dr. Bennet’s summation of the evidence. A vast lay literature without scientific evidence has emerged over the years to help parents cope with infants with colic. Now that a few placebo-controlled trials have been completed, the evidence is not clear. How can doctors consider probiotic research “a hammer in search of a nail?” Should we “throw in the bib” and devote precious resources to “reassurance, family social support, and the tincture of time?”
Our research group in Houston has investigated (under 3 protocols) more than 130 infants with colic over the past 5 years. We, too, have been plagued by difficult enrollment of infants, many who have been pre-exposed to acid blockers and probiotics in the formula. These problems that may have contributing factor to negative results. Sung et al. have reported that they have used regression models to estimate treatment effects by adjusting for potential confounding factors at baseline. Though it appears that they have considered potential confounding, it is not clear if they have tested for effect modifications in their regression models. Instead they have reported subgroup analyses. The most efficient way is to test for effect modifications in a regression model rather than subgroup analyses by stratification. It is certainly possible the treatment is not uniformly effective for all eligible children.
Although Sung et al. have reported that in their unadjusted analyses, they compared all continuous outcomes by t-test and non-parametric methods, it is not clear why they did not use a nonparametric regression model (e.g., Quantile regression) for their adjusted analyses through multivariable analyses. Because crying time usually has a skewed distribution, it would have been more appropriate to use regression models that are likely to meet the underlying assumptions. Furthermore, for comparisons of their outcomes during the follow up period, is not clear as why they have not taken advantage of longitudinal analyses. For example, Generalized Estimating Equations (GEE) methods are considered more appropriate for comparisons of outcomes during the follow up period.
Despite these concerns, it is not clear if the main results would change if their data were analyzed using more sophisticated techniques. What does seem clear is that pediatricians will continue to be interested in proper studies of this potential treatment for colic, because without them they will likely continue to use the “shotgun” approach. Extreme stress is very evident in the families, and it is palpable. The typical infant with colic cries and fusses for 4-5 hours/day, and the parents get little sleep. Colic has been associated with thoughts of infanticide in ~25% of parents (1). As evidence for the psychosocial impact of infant crying, recordings of infants with colic were used (perhaps inappropriately) for interrogation of suspected terrorists in Guantanamo Bay. (https://www.google.com/search?q=colic+and+guantanamo&oq=colic+and+guanta...).
One biomarker of colic deserves comment. We previously found elevated fecal calprotectin in all infants at this age, with values even higher in those with colic than in age-matched infants (2), but, as the authors mention, Olafsdotir et al. did not make the same observation in 2002. We speculated that gut inflammation may contribute to symptoms in at least some of these infants. This contention is not rejected and is partially supported by the Sung et al. study and unpublished data from our group showing ~50% lower levels of fecal calprotectin after resolution of colic. We are currently studying calprotectin and other biomarkers in a F.D.A.-regulated, N.I.H./NCCAM-sponsored U.S. trial.
With respect to microbiota differences in colic, there are a number of published observations from different groups, mostly concluding that there are changes in the microbiota of infants with colic. We could not find a description of how the researchers measured microbial diversity or E. coli abundance. Ref #34 does not describe microbiota or diversity. Standardization across studies of the approaches used to analyze of the microbiota would be quite helpful in making sense of variations in outcomes.
We agree with the need for evidence-based guidelines in pediatric practice world-wide. There are now well-supported guidelines for treatment of otitis and constipation (3, 4). In contrast, comprehensive, high-quality studies are just emerging in the probiotic field, of which the current study is a good example. There is accumulating evidence (shown in the forest plot of Sung et al. included in their study) that probiotics may indeed have a role in colic. Could “the forest be obscured by the trees” in this report? Fully 33% of all infants had been treated with a formula containing probiotics before the trial. This reflects a huge problem with premature marketing of formulas and probiotics as definitive treatments for infants with colic. We agree with Sung et al. that “It will be valuable to pool data from all existing and ongoing rigorous trials…to investigate with greater certainty which subgroups … could benefit from probiotics.”
Finally, there is evidence to support the contention infants with colic may have long-term effects on gastroenterological functions or behaviors in their later childhood (5).
Reference List
(1) Levitzky S, Cooper R. Infant colic syndrome--maternal fantasies of aggression and infanticide. Clin Pediatr (Phila) 2000 Jul;39(7):395-400.
(2) Rhoads JM, Fatheree NY, Norori J, Liu Y, Lucke JF, Tyson JE, et al. Altered fecal microflora and increased fecal calprotectin in infants with colic. J Pediatr 2009 Dec;155(6):823-8.
(3) Forrest CB, Fiks AG, Bailey LC, Localio R, Grundmeier RW, Richards T, et al. Improving adherence to otitis media guidelines with clinical decision support and physician feedback. Pediatrics 2013 Apr;131(4):e1071-e1081.
(4) Venekamp RP, Sanders S, Glasziou PP, Del Mar CB, Rovers MM. Antibiotics for acute otitis media in children. Cochrane Database Syst Rev 2013;1:CD000219.
(5) Savino F, Castagno E, Bretto R, Brondello C, Palumeri E, Oggero R. A prospective 10-year study on children who had severe infantile colic. Acta Paediatr Suppl 2005 Oct;94(449):129-32.
Competing interests: The laboratory of Dr. Rhoads received financial support for animal research with L. reuteri from BioGaia, Inc. (Clinical colic funding was from the U.S. National Institutes of Health).