Updated Meta-analysis of Probiotics for Preventing Necrotizing Enterocolitis in Preterm Neonates

We respectfully disagree with the contention of Damien and colleagues [1] regarding the validity of the analysis of Deshpende and colleagues [2]. We continue to say it is clearly not valid, while Damien and Colleagues [1] claim otherwise. Although much of this argument belongs and has been waged in the recent biostatistics literature, we are writing this rebuttal to clarify this specific application for the Deshpende study [2]. Their analysis cannot be viewed as valid for precisely the reasons stated in Neu and Shuster [3]. The challenge to Shuster [4], regarding problems with empirically weighted methods, were unsuccessfully rebutted in [5,6,7]. We did not reference these papers, as every issue raised in [5,6,7] was listed and dealt with in Shuster and Colleagues [8]. Further, [8] conclusively brings up serious additional issues with empirically weighted random effects methods. It is a mathematical fact that empirically weighted random effects methods tend to understate their sampling error, leading to systemic undercoverage of confidence intervals. Note also in [7], Laird and colleagues acknowledge the bias in estimation. Laird is a pioneer in empirically weighted methods, as her joint article [9] with DerSimonian is amongst the most commonly cited methods for random effects meta-analysis. The standard statistical models for random effects meta-analysis are based on unweighted target population means (See the Methods section of [9]). Why would one expect in a random effects context that a weighted estimate would be unbiased or have trustable error properties for unweighted means? Finally, low event rate empirically weighted methods require fabrication of point estimates for relative risk and their standard errors, generally using a continuity correction whenever exactly one treatment arm has no adverse events. When both arms have no events, these methods must discard the study. Our unweighted methods [10] do not require implementation of either adjustment. Even when both arms of a contributing trial have a small but positive number of events, the asymptotic approximations of the true target within study standard error by the estimated standard error are very tenuous. This makes the weights wildly varying random variables, which are universally analyzed by weighted methods as non-random. Contrary to [1], as far as NEC is concerned, all of the trials qualify as having low event rates.

Conflict of Interest: None

Sincerely,

Josef Neu, M.D. and Jonathan J Shuster, Ph.D.

References

[1] Damien J. Jolley DJ, Green S, Buttery J. Unfounded biostatistical criticism of probiotics/NEC meta-analysis. Pediatrics e-letter July 22.

[2] Deshpande G, Rao S, Patole S and Bulsara M. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics 2010;125:921-930. [3] Neu J and Shuster JJ. Non-administration of routine probiotics unethical – really? Pediatrics (May 6 2010), http://pediatrics.aappublications.org/cgi/eletters/125/5/921#50097 (accessed July 22, 2010). [4] Shuster JJ. Empirical vs natural weighting in random effects meta- analysis Statist. Med. 2010; 29:1259—1267. [5] Thompson S and Higgins J. Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1270—1271. [6] Waksman JA. Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1268—1269. [7] Laird N, Fitzmaurice G and Ding X, Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1266—1267. [8] Shuster JJ, Hatton RC, Hendeles L. and Winterstein AG. Reply to Discussion of “Empirical vs. natural weighting in random effects meta- analysis” Statist. Med. 2010; 29: 1272-1281. [9] DerSimonian R and Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7(3):177-88.

[10 Shuster JJ, Jones LS, Salmon DA. Fixed vs random effects meta- analysis in rare event studies: the rosiglitazone link with myocardial infarction and cardiac death. Statist. Med. 2007; 26:4375–4385.

Conflict of Interest:

None declared

We wish to comment on the methodological discussions in the e- correspondence following the interesting paper on probiotics and necrotizing enterocolitis (NEC) by Deshpande and others [1]. Some contain accusations of error which we believe need to be refuted.

Specifically, the letter from Neu and Schuster [2] calls into question the validity of the statistical methods used in the Deshpande et al [1] paper. They claim that the sequential meta-analysis used “is technically incorrect”, in part because “[The authors] used an empirically weighted analysis which leads to bias and underestimation of standard errors”. In support of this view, the authors cited their own research [3], published in Statistics in Medicine in May 2010. Neu and Schuster [2] chose not to cite the invited commentaries [4-6] in the same issue from eminent biostatisticians, all of whom strongly reject Neu and Schuster’s [3] opinion of empirically weighted analysis methods for random -effects meta-analysis.

Neu and Schuster [2] raise two other biostatistical concerns in their letter. One was that Deshpande et al [1] failed to account for rare event methods in their analysis, and the other that the sequential meta-analysis method used in the original study [1] does not apply in that analytic situation.

It is difficult to agree with the former of these criticisms, given that the event rates reported in the original study were often over 5% for NEC, even in the probiotic groups (average: 2.5%) and higher still for sepsis (average: 15%). The latter criticism arises from Neu & Schuster’s [2] objection to the use of empirical weights, an objection refuted by three independent groups of eminent biostatisticians.

We conclude that the biostatistical criticisms of Neu and Schuster [2] of the original contribution from Deshpande and colleagues [1] are not soundly based. More generally, we believe the methodological incorporation of a variety of different probiotic preparations into the meta-analysis strengthens, rather than weakens the analysis.

The conclusions and clinical implications drawn from the meta- analysis have been robustly and appropriately debated in the original meta -analysis [1], accompanying editorials [7,8], e-correspondence and elsewhere [9].

Yours sincerely

Damien Jolley, School of Public Health and Preventive Medicine, Monash University

Sally Green, School of Public Health and Preventive Medicine, Monash University

Jim Buttery, Infectious Diseases Unit, Monash Children’s Hospital, Murdoch Children’s Research Institute, Dept of Paediatrics, Monash University, Victoria, Australia

References

[1] Deshpande G, Rao S, Patole S and Bulsara M. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics 2010;125:921-930.

[2] Nue J and Schuster JJ. Non-administration of routine probiotics unethical – really? Pediatrics (May 6 2010), http://pediatrics.aappublications.org/cgi/eletters/125/5/921#50097 (accessed July 22, 2010).

[3] Schuster JJ. Empirical vs natural weighting in random effects meta-analysis Statist. Med. 2010; 29:1259—1267.

[4] Thompson S and Higgins J. Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1270—1271.

[5] Waksman JA. Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1268—1269.

[6] Laird N, Fitzmaurice G and Ding X, Comments on ‘Empirical vs natural weighting in random effects meta-analysis’ Statist. Med. 2010; 29:1266—1267.

[7] Soll RF. Probiotics: are we ready for routine use? Pediatrics 2010;125(5):1071-2.

[8] Tarnow-Mordi WO, Wilkinson D, Trivedi A, Brok J. Probiotics reduce all-cause mortality and necrotizing enterocolitis: it is time to change practice. Pediatrics 2010;125(5):1068-70.

[9] Agostoni C, Buonocore G, Carnielli VP, De Curtis M, Darmaun D, Decsi T, et al. Enteral nutrient supply for preterm infants: commentary from the European Society of Paediatric Gastroenterology, Hepatology and Nutrition Committee on Nutrition. J Pediatr Gastroenterol Nutr 2010;50(1):85-91.

Conflict of Interest:

None declared

Probiotics for preterm neonates – time to acknowledge the elephant in the room and call the parents

To the Editor We appreciate the opportunity to respond to the recent letters to the editor in response to our updated meta analysis of probiotics for preventing necrotising enterocolitis (NEC) in preterm neonates.1 We will address the common and important issues. (1) Pooling data from different trials: Given the lack of statistical heterogeneity, the consistent benefits despite significant variations in probiotic strains and protocols across a range of participating units from different parts of the world, reflect strength rather than weakness considering the broader perspective of meta-analysis.2-3 (2) Reproducibility of the results under different set ups is desirable. However significant differences in the set ups of the 11 trials (including 4 adequately powered for NEC) 1 from 9 nations beg a question- how much reproducibility is enough? at what potential cost? (3) Trial sequential analysis (TSA): Researchers have already pointed out the limitations and strengths of TSA.4-6 It must be emphasised that the debate around the use of Empirically Weighted analysis is still unresolved. There are equal arguments for7-9, as well as against10 it. Despite our best efforts we are not convinced that the statistical arguments about TSA are relevant to the real clinical issues raised by our results and their interpretation. We stand by our robust results confirmed with fixed and random effects models, and TSA. An independent response from the developers of TSA and the Cochrane collaboration is needed considering the potential impact of the statistical debate on many neonatal systematic reviews. Appreciating the clinical significance of our results in the context of death and serious disease like NEC is also important. If necessary, realism in evidence-based medicine should relegate statistical analysis to its proper subsidiary place. Even the most robust RCTs only minimize but not eliminate uncertainty. (4) Role of breast milk, baseline event rates, standard of care etc: Despite preferential use of human breast milk (with pro and prebiotics) in most of the units, the incidence of definite NEC has not declined significantly. Units with high standard of care and low baseline event rates in the population at risk are not expected to benefit “significantly” by probiotic supplementation. However even a 10% reduction in death or serious disease is significant. Current/planned placebo controlled trials are not powered enough to detect such effect sizes and pooling their results will only continue the current debate.1 Placebo controlled trials are not necessary to address each and every issue related to probiotics in preterm neonates. Improving access to “safe and effective” probiotics is a priority.1,11-13 Though it has not happened in a proper sequence, probiotics as an intervention have completed a full circle, from basic science14, 15 and cohort studies 16 to conclusive meta analysis1, routine use17, and long term follow up18. It is time to acknowledge the elephant in the room and provide all evidence (for and against) to the parents in a clear, honest, and transparent manner, and let them decide. The belief that we know what is best for our patients needs to be discarded. 1,3Deshpande G, 2,3Rao S, 1,3Patole S, 4Bulsara M 1KEM Hospital for Women, 2Princes Margaret Hospital for Children, 3University of Western Australia, 4University of Notre Dame, Perth, Western Australia

References: 1.Deshpande G, Rao S, Patole S, Bulsara M. Updated Meta-analysis of Probiotics for Preventing Necrotizing Enterocolitis in Preterm Neonates. Pediatrics 2010; 125 (5): 9210-30 2.Gotzsche PC. Why we need a broad perspective on meta-analysis: it may be crucially important for patients. BMJ. 2000; 321(7261): 585–586 3.Stockman JA 3rd. Newborn: probiotics for prevention of necrotising enterocolitis in preterm neonates with very low birthweight— a systematic review of randomised controlled trials. In: Deshpande G, Rao, S, Patole S, eds. Yearbook of Pediatrics. Philadelphia. PA: Elsevier Mosby; 2009:441– 443 4.Nuesch E, Juni P. Commentary: which metaanalyses are conclusive? Int J Epidemiol. 2009;38(1):298 –303 5.Brok J, Thorlund K, Wetterslev J, Gluud C. Apparently conclusive meta- analyses may be inconclusive: trial sequential analysis adjustment of random error risk due to repetitive testing of accumulating data in apparently conclusive neonatal metaanalyses. Int J Epidemiol. 2009;38(1): 287–298 6.Wetterslev J, Thorlund K, Brok J, Gluud C. Trial sequential analysis may establish when firm evidence is reached in cumulative meta-analysis. J Clin Epidemiol. 2008; 61(1):64 –75 7. Laird N, Fitzmaurice G, Ding X. Comments on 'Empirical vs natural weighting in random effects meta-analysis'. Stat Med. 2010;29:1266-7. 8.Thompson SG, Higgins JPT. Comments on ‘Empirical vs natural weighting in random effects meta-analysis.’ Stat Med. 2010; 29: 1270-71. 9.Waksman JA. Comments on Empirical vs natural weighting in random effects meta-analysis Stat Med 2010;29:1268–1269. 10. Shuster JJ, Hatton RC, Hendeles L, Winterstein, AG. Reply to discussion of ‘Empirical vs natural weighting in random effects meta- analysis’. Stat Med. 2010; 29: 1272- 81 11. Lin HC, Hsu CH, Chen HL, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight preterm infants: a multicenter, randomized, controlled trial. Pediatrics. 2008;122(4):693–700 12.Lin HC, Su BH, Chen AC, et al. Oral probiotics reduce the incidence and severity of necrotizing enterocolitis in very low birth weight infants. Pediatrics. 2005;115(1):1–4 13.Bin-Nun A, Bromiker R, Wilschanski M, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight neonates. J Pediatr. 2005;147(2):192–196 14.Siggers RH, Siggers J, Boye M, et al. Early administration of probiotics alters bacterial colonization and limits diet-induced gut dysfunction and severity of necrotizing enterocolitis in preterm pigs. J Nutr. 2008;138(8):1437-44 15. Caplan MS, Miller-Catchpole R, et al. Bifidobacterial supplementation reduces the incidence of necrotizing enterocolitis in a neonatal rat model. Gastroenterology. 1999;117 (3):577-83. 16.Hoyos AB, Reduced incidence of necrotising enterocolitis associated with enteral administration of Lactobacillus acidophilus and Bifidobacterium infantis to neonates in an intensive care unit. Int J Infect Dis. 1999; 3:197-202. 17.Satoh Y, Shinohara K, Umezaki H, et al. Bifidobacteria prevents necrotising enterocolitis and infection. Int J Probiot Prebiot. 2007;2:149–154 18.Chou IC, Kuo HT, Chang JS, et al. Lack of effects of oral probiotics on growth and neurodevelopmental outcomes in preterm very low birth weight infants. J Pediatr. 2010;156 (3):393-6.

Conflict of Interest:

None declared

Neonatologists face an ethical dilemma in balancing three pressing needs: to undertake further research; to disclose unambiguously to parents and institutional review boards (IRBs) current evidence that probiotics halve mortality; (1) and to make ‘open label' probiotics available.(2)

Although Neu et al found trivariate analysis of NEC, sepsis and death inconclusive,(3) for many death is the key safety issue. By Cochrane methods, (4,5) probiotics reduce the relative risk (RR) for death by 58% (p <0.00001).(1) Lack of transparency (6,7) and choice risk loss of trust. Consultation with patient liaison groups (8,9) and guidance from IRBs may be helpful.

Criticisms about pooling interventions of the same class (10) affect Cochrane Reviews of antibiotics, beta blockers, steroids, tocolytics, contraceptives, surfactants, immunoglobulins, ventilation, hypothermia and many others. Lack of heterogeneity justifies pooling probiotics. No neonatal systematic review has a greater effect size for mortality and lower p value. (1) Reversal of this by further evidence would be unprecedented. (11) If pooling is confined to bifidobacterium products, the RR for death is 0.34 (0.20 - 0.55, p<0.0001, n = 1386).

Criticisms of weighting (3,12) also affect Cochrane Reviews. Indeed, Shuster proposes re-calculating all meta-analyses underlying current policies. (13) Others think that this undertaking, which weights trials of 5,000 and 50 equally, would be misleading. (14)

Probiotics are accessible through special access schemes in many jurisdictions and recommended (5,15) or endorsed (16) by various authorities for selected infants. We encourage clinicians to import and, if parents wish, prescribe previously evaluated probiotics using published regimens. (17,18,19,20) We can share procedures for this and for monitoring extraneous organisms and stool colonisation.

Participation in placebo RCTs is encouraged if parents consent after full, balanced disclosure of current evidence, (6,7) particularly if treatment cannot otherwise be accessed. To minimise delays in detecting a conclusive result, researchers could ask independent Data Monitoring Committees to combine interim data with published RCTs of similar products.

The evidence that probiotics reduce mortality (1) calls for major changes in practice. As Edmund Hey wrote, “Do we, knowing what we now know, have the right to deny parents the option ...?” (2)

[A detailed, point by point response to all other issues raised will follow shortly]

References:

1. Deshpande G, Rao S, Patole S, Bulsara M. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics. 2010;125:921–930.

2. Hey E. Probiotics (comment): Has the time come to start using probiotics more widely? Neonatal Formulary 5. Available at: http://www.blackwellpublishing.com/medicine/bmj/nnf5/pdfs/comment/prob_com_jul09.pdf Accessed 15 June, 2010.

3. Neu J, Shuster J. Non-Administration of Routine Probiotics Unethical—Really? Pediatrics 6 May 2010.

4. Deshpande G, Rao S, Patole S. Probiotics for prevention of necrotising enterocolitis in preterm neonates with very low birthweight: a systematic review of randomised controlled trials. Lancet. 2007; 369: 1614 –1620.

5. AlFaleh KM, Bassler D. Probiotics for prevention of necrotizing enterocolitis in preterm infants. Cochrane Database Syst Rev.2008; (1):CD005496.

6. Tarnow-Mordi WO, Wilkinson D, Trivedi A, Brok J. Hypothetical parent information leaflet. Probiotics reduce all-cause mortality and necrotizing enterocolitis: it is time to change practice. Pediatrics 2010; 125: 1068-70. Supplementary information. http://www.pediatrics.org/cgi/content/full/peds.2009-2151/DC1)

7. Wilkinson D. Therapeutic hypothermia and the ‘equal air-time’ solution for controversial randomised trials. J Paeds Child Health 2010 (forthcoming)

8. Partridge N, Scadding J: The James Lind Alliance: patients and clinicians should jointly identify their priorities for clinical trials. Lancet 2004, 364:1923-1924.

9. The James Lind Alliance. http://www.lindalliance.org/ accessed 15 June 2010.

10. Soll R. Probiotics – are we ready for routine use? Pediatrics 2010; 125: 1071-2.

11. Ioannidis J, Lau J. Evolution of treatment effects over time: empirical insight from recursive cumulative metaanalyses. Proc Natl Acad Sci U S A. 2001; 98: 831-6.

12. Shuster JJ. Empirical vs natural weighting in random effects meta -analysis. Statist.Med. 2010; 29 1259-65.

13. Shuster JJ, Hatton RC, Hendeles L, Winterstein, AG. Reply to discussion of ‘Empirical vs natural weighting in random effects meta- analysis’. Statist. Med. 2010; 29: 1272-81.

14. Thompson SG, Higgins JPT. Comments on ‘Empirical vs natural weighting in random effects meta-analysis.’ Statist. Med. 2010; 29: 1270- 71.

15. Danish National Guideline. Rigshospitalet, Copenhagen. http://www.rh- vejledninger.dk/C1256FF60028FE3A/0/FADF649DDFCAFF2FC12576AB0030868D?OpenDocument&Highlight=2, accessed 15 June 2010

16. Tarnow-Mordi W, Wilkinson D, Trivedi A, Sinn J, Dutta S, Parikh T, Lin HC. Re: ESPGHAN commentary and education that probiotics substantially reduce all-cause mortality and necrotizing enterocolitis in preterm infants. J Pediatr Gastroenterol Nutr. 2010; 50: 694; authors’ response: van Goudoever, JB, Koletzko B; Rigo J et al. On behalf of the ESPGHAN Committee on Nutrition pp 694-695.

17. Bin-Nun A, Bromiker R, Wilschanski M, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight neonates. J Pediatr 2005;147:192–196.

18. Lin HC, Su BH, Chen AC, et al. Oral probiotics reduce the incidence and severity of necrotizing enterocolitis in very low birth weight infants. Pediatrics. 2005;115:1– 4.

19. Lin HC, Hsu CH, Chen HL, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight preterm infants: a multicenter, randomized, controlled trial. Pediatrics. 2008;122:693–700.

20. Samanta M, Sarkar M, Ghosh P, et al. Prophylactic probiotics for prevention of necrotizing enterocolitis in very low birth weight newborns. J Trop Pediatr. 2009; 55:128 –131.

Conflict of Interest:

Invited speaker at Hot Topics meeting, Washington DC, December 2010 on "Probiotics reduce mortality"

We read with interest the recent controversial debate on probiotics for preterm infants sparked by the meta-analysis by Deshpande et al [1] and associated commentaries by distinguished authors [2, 3]. Although we are supportive of the therapeutic potential of probiotics for exclusively formula fed VLBW neonates, none of the meta-analyses published to date have extracted data in relation to extremely preterm ELBW infants or those exclusively fed breast milk. As such we would be cautious about encouraging widespread use.

We offer the following evidence base regarding associated confounders:

1) Feed Regimen

Infants exclusively fed maternal or donor breast milk have a significantly lower incidence of NEC in comparison with those formula fed [4,5]. However published trials to date and the recent meta-analysis fail to give sufficient information to allow stratification of this data to the administration of supplementary probiotics to infants receiving EBM. It is well understood that EBM is itself a natural synbiotic, composed of oligosaccharide prebiotics plus probiotics in the form of Lactobacilli and Bifidobacteria, in addition to its other immunomodulatory properties.[6,7,8] This is possibly the most important confounder of these metanalyses, and the fulfilment of specific criteria similar to Peaker and Neville’s Postulates [9] are essential to determine the relevance of this finding to the neonatal population [10].

2) Environmental Cross Contamination

Data from the University of London pilot work has shown that introduction of a probiotic preparation during a randomised controlled trial resulted in 35% of placebo infants being colonised by the probiotic strain. The source of this cross-contamination is not clear, yet the striking rate of placebo infants colonised suggests that the flora of the neonatal unit itself may have been altered by the presence and clinical use of these live cultures.[11] In their observational study of 2007, Mohan et al similarly identified a cohort of infants colonised with a strain of Bifidobacterium only found in commercial probiotic foods.[12] Huge interindividual variations are known to exist in adult life, yet Turnbaugh et al in 2009 identified disparities even in the microbiota of identical twins, strongly suggesting that an external environmental factor or factors alter gut flora very early on in life – and these are maintained into adulthood.[13]

3) Who/What/When/How

Guarner et al noted in 2003 that ‘the effect of a bacterium is strain specific and cannot be extrapolated even to other strains of the same species’ [14] and yet the beneficial effect of probiotics in preventing necrotising enterocolitis and mortality do not appear to be particularly strain specific based on current data. With this in mind, it is essential to see if this benefit is maintained when compared directly with fresh EBM. We support the position that meta-analysis of different bacterial species is an inappropriate technique other than for demonstrating overall benefit [15]. This technique becomes of little use when trying to decide what strain, dose, frequency of administration and length of course when administering probiotics.

4) Long-term concerns

Finally, the role of the intestinal microbiota in human health and disease is of increasing interest with diseases as diverse as inflammatory bowel disease, atopy, obesity and multiple sclerosis all being associated with perturbations in the gut microbiota. Although the identification of rare and idiosyncratic side-effects and complications is not the remit of clinical trials, practicing neonatologists should be aware that the introduction of live bacterial species into a constantly replenishing culture environment (the infant gastrointestinal tract) may result in lifelong carriage of the organism in question. This in addition to the pioneer role that such bacteria may have in influencing host gene expression and hence ongoing bacterial colonisation [16] make the possibility of long-term sequelae a significant concern.

For these reasons, we feel that more evidence is required before probiotics are accepted as an established therapy in our sickest infants.

References (1) Deshpande G, Rao S, Patole S, et al. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics 2010; 125: 921–930. (2) Tarnow-Mordi WO, Wilkinson D, Trivedi A, et al. Probiotics reduce all- cause mortality and necrotizing enterocolitis: It is time to change practice. Pediatrics 2010; DOI: 10.1542/peds.2009-2151. (3) Soll RF. Probiotics: Are We Ready for Routine Use? Pediatrics 2010; DOI: 10.1542/peds.2010-0643. (4) Quigley MA, Henderson G, Anthony MY, et al. Formula milk versus donor breast milk for feeding preterm or low birth weight infants. Cochrane Database Syst Rev. 2007; (4): CD002971. (5) Boyd CA, Quigley M, Brocklehurst P. Donor breast milk versus infant formula for preterm infants: systematic review and metaanalysis. Arch Dis Child Fetal Neonatal Ed 2007; 92: 169-175 (6) Martín R, Jiménez E, Heilig H, et al. Isolation of bifidobacteria from breast milk and assessment of the bifidobacterial population by PCR- denaturing gradient gel electrophoresis and quantitative real-time PCR. Appl Environ Microbiol. 2009; 75(4): 965-9. (7) Martín R, Langa S, Reviriego C, Jimínez E, Marín ML, Xaus J, Fernández L, Rodríguez JM. Human milk is a source of lactic acid bacteria for the infant gut. J Pediatr. 2003;143:754-8. (8) Gueimonde M, Laitinen K, Salminen S, et al. Breast milk: a source of bifidobacteria for infant gut development and maturation? Neonatology. 2007; 92(1): 64-6. (9) Peaker M, Neville MC. Hormones in milk: chemical signals to the offspring? J Endocrinol. 1991;131:1-3. (10) Beattie LM, Weaver LT. Mothers, Babies and Friendly Bacteria. ADC Fetal Neonatal Ed, in press 2010. (11) Costeloe K. PiPS: Trial of probiotic administered early to prevent infection and necrotising enterocolitis protocol- version 3.1. Accessed online at: http://www.hta.ac.uk/protocols/200505010004.pdf on 19th May 2010. (12) Mohan R, Koebnick C, Schildt J, Schmidt S, Mueller M, Possner M, Radke M, Blaut M. Effects of Bifidobacterium lactis Bb12 supplementation on intestinal microbiota of preterm infants: a double-blind, placebo- controlled, randomized study. J Clin Microbiol. 2006 Nov;44(11):4025-31. (13) Turnbaugh PJ, Hamady M, Yatsunenko T et al. A core gut microbiome in obese and lean twins. Nature 2009;22;457(7228):480-4. (14) Guarner F, Malagelada JR. Gut Flora in Health and Disease. Lancet 2003; 361(9356): 512-9. (15) Barclay AR, Stenson B, Simpson JH et al. Probiotics for necrotising enterocolitis: a systematic review. JPGN 2007 Nov;45(5):569-76.

Conflict of Interest:

Dr Beattie has spoken about her research at conferences with travel grants from Danone via The Learning Curve.

An updated meta-analysis of the effects of probiotics for the prevention of necrotizing enterocolitis (NEC) in preterm infants by Deshpande et al. published in this issue of Pediatrics concluded that there are significant benefits of probiotic supplements in reducing death and disease in preterm neonates (1).

One of my major concerns with regard to their meta-analysis is whether it is appropriate to pool data on different probiotic microorganisms. The effects observed in each clinical trial can only be attributed to the individual probiotic strain(s) tested. It could not be inferred to whole species or any other strain of the same species. Meta- analysis by pooling data from different genera, species, strains does not increase statistic power for individual probiotic strain tested.

The data published so far appear to support the potential role in prevention of NEC in preterm infants. However, this is not without controversy. A recent retrospective report failed to show any effects of probiotics (Lactobacillus GG) in prevention of NEC in preterm infants (2).

While further verifying ¡°healthy benefits¡± of any probiotic products for clinical use in the situations such as NEC is needed, one should demonstrate whether the tested strain can survive in significant amount at the action site (large intestine) after going through very acidic (stomach) and alkaline (small intestine) environment and whether the tested strain can colonize at the action site. The safety issue of the probiotics is another concern. The probiotics are generally recognized as safe (GRAS) without significant side effects. A recent randomised, double- blind, placebo controlled clinical trial demonstrated that in patients with advanced acute pancreatitis, probiotics are associated with increased mortality and do not actually reduce the risk of additional infection (3). This may indicate the fact that they are GRAS because they do not survive in significant amount after going through the acidic (stomach) and alkaline(small intestine) environment.

Therefore, before introducing any probiotic products into clinical use, the above questions need to be addressed adequately.

References:

1. Deshpande G, Rao S, Patole S, Bulsara M. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics 2010; 125(5):921-30.

2. Luoto R, Matomaki J, Isolauri E, Lehtonen L. Incidence of necrotizing enterocolitis in very-low-birth-weight infants related to the use of Lactobacillus GG. Acta Paediatr 2010 Mar 8. [Epub ahead of print]

3. Besselink MG, van Santvoort HC, Buskens E, Boermeester MA, van Goor H, Timmerman HM, et al. Probiotic prophylaxis in predicted severe acute pancreatitis: a randomised, double-blind, placebo-controlled trial. Lancet 2008;371(9613):651-9.

Conflict of Interest:

None declared

Deshpande et al [1] presented a meta-analysis of probiotics for preventing necrotizing enterocolitis (NEC) in preterm neonates, the selection criteria of trial in the analysis were randomized, controlled trials (RCTs) of any enteral probiotic supplementation that started within first 10 days and continued for more than 7 days in preterm very low birth weight (VLBW) neonates and reported on stage 2 NEC or higher (Modified Bell Staging). A total of 11 trials were eligible for inclusion in the meta-analysis.[2-12] We reviewed the original articles of these 11 studies with interesting and summarized the data of NEC in the Table that listed the 11 studies as the same order in the table 1 of the meta-analysis. However, only 4 among the 11 studies reported significantly higher risk for NEC in the control group.[5,6,10,11] One reported no NEC developed in their infants,[2] and one did not report the data of NEC.[8] Among the 9 studies reported the incidence of NEC stage 2 or higher, only 4 reported the age at onset of NEC.[3,5,6,10] The ranges of the mean (SD) age at onset of NEC in these 9 studies were from 19.5 (13.4) to 30.8 (14.9) and from 15.5( 9.7) to 22.7 (16.6) days of life for the probiotics group and control group, respectively (Table). Although there was no significant difference between the 2 groups, the age at onset of NEC in the probiotics group seems to have a trend of longer than the control group. Our concern is that do probiotics prevent only the early onset and not the late onset NEC in the preterm neonates?

The cause of NEC is unclear, and its etiology is multifactorial; probiotics may be one of the most effective therapies for the prevention of NEC. The age at onset of NEC is an important issue that should be considered in the strategy of prevention. These trials in the meta- analysis recommended combining NEC plus death in their analysis because death precludes the subsequent development of NEC and, as such, prevents NEC. However, one trial excluded 6 infants from the study who had NEC before 7 days after birth, the age at enrollment were 7.7 (2.0) and 7.9 (2.9) days in probiotics group and control group, respectively.[6] Another trial excluded 29 infants from the study who expired within 2 weeks of life (the causes of death were not declared), however, the age at enrollment were 3.4 (3.7) and 3.8 (3.6) days, and the age at onset of NEC were 20 (7.5) (range, 15-31) and 15.5 (9.7) (range, 10-36) days in probiotics group and control group, respectively.[3]

We performed a retrospective analysis of prospectively collected data of a 10-years cohort on the neurological outcome of ELBW infants at 2-year corrected age in the NICU of China Medical University Hospital. From 1998 to 2007, 458 extremely low birth weight infants (birth weight < 1000 g) were included, the survival rate was 62.9% (288/458). The mean gestational age was 26.3 weeks and the mean birth weight was 818 g (range, 530 -1000 g) among theses 288 infants. The incidence of NEC (Bell's stage 2 or higher) was 8.33% (24/288) (data presented in the May 2010 Annual Conference of Premature Baby Foundation of Taiwan). The mean (SD) age at onset of NEC was 12.2 (10.7) (range, 3-42) days, including 18 infants at before 14 days and 6 infants at after 16 days. Only one study in the meta- analysis reported the detailed days at onset of 12 infants with NEC, the age at onset was before 12 days in 6 infants and after 15 days in 6 infants, probiotics group had 4 infants with NEC and all were diagnosed after 15 days.[3] According to the age at onset of NEC, we defined that the onset before 14 days as early onset NEC and that after 15 days as late onset NEC.

We expect that the ongoing trials may report on the age at onset of NEC and will help answer the question that do probiotics prevent only the early onset NEC in the preterm neonates?

REFERENCES 1.Deshpande G, Rao S, Patole S, Bulsara M. Updated meta-analysis of probiotics for preventing necrotizing enterocolitis in preterm neonates. Pediatrics 2010;125 (5); 921-930 2. Kitajima H, Sumida Y, Tanaka R, Yuki N, Takayama H, Fujimura M. Early administration of Bifidobacterium breve to preterm neonates: randomised control trial. Arch Dis Child Fetal Neonatal Ed. 1997;76(2):F101-F107 3. Dani C, Biadaioli R, Bertini G, Martelli E, Rubaltelli FF. Probiotics feeding in prevention of urinary tract infection, bacterial sepsis and necrotizing enterocolitis in preterm neonates: a prospective double-blind study. Biol Neonate. 2002;82(2): 103-108 4. Costalos C, Skouteri V, Gounaris A, et al. Enteral feeding of premature neonates with Saccharomyces boulardii. Early Hum Dev. 2003;74(2):89-96 5. Bin-Nun A, Bromiker R, Wilschanski M, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight neonates. J Pediatr. 2005;147(2):192-196 6. Lin HC, Su BH, Chen AC, et al. Oral probiotics reduce the incidence and severity of necrotizing enterocolitis in very low birth weight infants. Pediatrics. 2005; 115(1):1-4 7. Manzoni P, Mostert M, Leonessa ML, et al. Oral supplementation with Lacibacillus casei subspecies rhamnosus prevents enteric colonisation by Candida species in preterm neonates: a randomised study. Clin Infect Dis. 2006;42(12):1735-1742 8. Mohan R, Koebnick C, Schildt J, et al. Effects of Bifidobacterium lactis Bb12 supplementation on intestinal microbiota of preterm neonates: a double placebo controlled, randomized study. J Clin Microbiol. 2006;44(11):4025-4031 9. Stratiki Z, Costalos C, Sevastiadou S, et al. The effect of a Bifidobacter supplemented bovine milk on intestinal permeability of preterm infants. Early Hum Dev. 2007;83(9):575-579 10. Lin HC, Hsu CH, Chen HL, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight preterm infants: a multicenter, randomized, controlled trial. Pediatrics. 2008;122(4):693-700 11. Samanta M, Sarkar M, Ghosh P, et al. Prophylactic probiotics for prevention of necrotizing enterocolitis in very low birth weight newborns. J Trop Pediatr. 2009; 55(2):128-131 12. Rouge„S C, Piloquet H, Butel MJ, et al. Oral supplementation with probiotics in very-low-birth-weight preterm infants: a randomized, double- blind, placebo- controlled trial. Am J Clin Nutr. 2009;89(6):1828-1835

Conflict of Interest:

None declared

Dear Editors,

We read with great interest the recent articles published in Pediatrics, relating to probiotic supplementation in preterm infants. [1- 3] Deshpande and colleagues’ updated meta-analysis concludes that there are significant benefits of probiotics in reducing the all-cause mortality and necrotizing enterocolitis [NEC], and coupled with the reduction in the time to establish full enteral feeds, warrants their use as routine therapy in preterm neonates (subject to a suitable probiotic product being available). Moreover they suggest that further placebo controlled trials are unnecessary given the statistical significance, including tight confidence intervals, extremely low p values, and low number needed to treat to prevent a case of NEC.[1] In their commentary on this study, Tarnow-Mordi et al argue passionately that changing practice to supplement premature infants with probiotics would result in thousands more survivors each year. They then question whether placebo-controlled trials should continue. [2]

We, as Soll[3], would suggest a more cautious approach as there are so many unknowns about probiotics. For example, as yet we do not know which probiotic or probiotic combination is the best to use, nor do we know the quality of currently available products (do they contain the numbers of organisms specified on the label? Do they contain extraneous organisms? Are the organisms specified viable?). Importantly, there are no licensed probiotics to prescribe outside clinical trials. Yet to be determined is the appropriate dose, and when to commence or discontinue probiotics. The potential sepsis from probiotics is unknown, as in most diagnostic laboratories blood culture media would not support the fastidious anaerobic probiotic growth. In addition, in the meta-analyses failed to explain reasons for the reduction in mortality.

Whilst the reported findings are very encouraging, we agree with our West Australian colleagues that the selection of a safe and suitable product with documented probiotic properties and close monitoring is essential before probiotics can be provided as a routine therapy. Therefore, we believe that it is extremely important that well designed randomized controlled trials, addressing these outstanding issues, be completed. Indeed, based on encouraging data from the earlier studies of Lin [4] and Bin-Nun,[5] we embarked on and are halfway through recruiting to the ProPrems Study, a multicentre probiotic study, to investigate the efficacy of probiotic supplementation very premature infants (<1500g and <32 weeks gestation) from soon after birth, on the incidence and severity of definite late-onset sepsis. There are multiple secondary outcome factors including NEC, death and longer term morbidity, including allergy. Following two interim analyses the Data Safety Monitoring Committee agreed that the ProPrems trial continue. In addition, and novel to other trials, molecular biology tools are being used to ascertain colonization of the infants intestine by the ingested probiotics, as well as ensuring the integrity of each batch of probiotic utilized.

We were very warmed by the very rational editorial of Roger Soll [3] who concluded that the neonatal community needs to push for further larger probiotic trials such as ProPrems. We continue to welcome Australasian centres to join the ProPrems study. Soll also states that little is known about probiotics in extremely low birth weight infants, nor in infants receiving breast milk. Although ProPrems was not powered to answer these questions, valuable information about these subgroups will be ascertained. Ultimately, we all want any changes in clinical practice to be well informed by the best evidence which will translate to better neonatal outcomes.

Suzanne M. Garland, MBBS MD FRCPA FRANZCOG Ad Eundem FAChSHM1

Susan E Jacobs, MBBS, MD, FRACP Newborn Services, Royal Women’s Hospital, Victoria, Australia

Jacinta M Tobin MBBS PhD FRACP Paediatric Gastroenterologist and Subdean University of Melbourne

On Behalf of the ProPrems study group

1Director, Department of Microbiology and Infectious Diseases, Royal Women's Hospital and Professor, Department of Obstetrics and Gynecology, University of Melbourne, Melbourne, Victoria, Australia.

References

1. Deshpande, G., S. Rao, and S. Patole, Updated Meta-analysis of Probiotics for Preventing Necrotizing Enterocolitis in Preterm Neonates. Pediatrics, 2010. 125(5): p. 921-30.

2. Tarnow-Mordi, W.O., et al., Probiotics Reduce All-Cause Mortality and Necrotizing Enterocolitis: It Is Time to Change Practice. Pediatrics, 2010. 125(5).

3. Soll, R.F., Probiotics: Are We Ready for Routine Use? Pediatrics. 125(5): p. 1071-2.

4. Lin, H.C., B.H. Su, and A.C.e.a. Chen, Oral probiotics reduce the incidence and severity of necrotizing enterocolitis in very low birth weight neonates. Pediatrics, 2005. 115 . p. 1-4.

5. Bin-Nun, A., R. Bromiker, and M.e.a. Wilschanski, Oral probiotics prevent necrotizing enterocolitis in very low birth weight neonates. . J Pediatr 2005 147: p. 192-6.

Conflict of Interest:

No commercial conflicts of interests although I am Chief Investigator of muticentred trial on probiotics in premature babies as a muticentred randomized control trial

We wish to raise several concerns about the Meta-Analysis of Deshpande and Colleagues1 and the accompanying commentary 2 . These authors make several very strong recommendations about the lack of need for additional trials, suggest that it is now “unethical” practice to not use probiotics in these babies and suggest the involvement of parents in recommending use of these agents in premature infants. We base our concerns primarily on flaws in their meta-analysis, but also a lack of information in the original studies about cause of death, increased sepsis seen in certain subgroups in individual studies (e.g., the smallest babies in the Lin, et al multicenter trial3 had 12 babies in the probiotic group and only 1 baby in the control group that developed sepsis), the large number of different probiotics used in the different studies upon which the meta-analysis was based and the lack FDA approval and lack of quality control in the manufacturing sector.

The sequential meta-analysis, although sophisticated, is technically incorrect for several reasons: (A) They used an empirically weighted analysis which leads to bias and underestimation of standard errors. See Shuster 4 and Shuster et. al. 5 for further details; (B) They did not account for rare event methods in their analysis (see Shuster, Jones, and Salmon 6 ; (C) They used a sequential method that does not apply to their analytic situation. They used the method of Reboussin et. al. 7 , which is problematic on two fronts. First, when using empirically weighted methods, the information fraction, needed for the analysis, is completely unknown, even if the maximum horizon of studies is specified. For example, if you are analyzing at study 7, and the maximum horizon is 15, the information fraction is the ratio of the variance of the estimator after seven studies to the unknown variance of the estimator after 15 studies. The unknown weights (sizes) of future studies will affect the present information fraction needed at the time one does the seventh. Generally, it will be below 7/15 because earlier studies tend to be smaller. Another factor that is of concern is the “independent increments of information”, needed for the method to be valid. This is violated, at least to some extent, because the empirical weighting can lead to different relative weights for the existing studies as new studies are added. For example, if the weight ratio for study 1 to study 2 is 75% after seven studies, it may not be 75% after the eighth study is added. This depends upon the impact of study number eight on the change in the overall diversity of the studies after it is added. If it is very different from the other estimates, it will push the old 75% closer to 100%. Keeping relative weights fixed is crucial to the independence of the increments of information.

As it turns out, the biostatistical author was preparing a sequential methods paper for unweighted meta-analysis, based on the O’Brien-Fleming 8 method. This will be submitted in July or August, 2010 but correctness is not an issue. The Deshpende1 data will now be used as an example in the article. As long as a maximum horizon of studies is specified, unweighted meta-analysis has known information fractions (fraction of studies completed) and independent increments of information. Four analyses were done. We used five studies as the time of our first look, both 12 and 15 as our maximum number of studies, and looked at the primary endpoint (NEC) at P=0.05 (two-sided) and a trivariate Bonferroni controlled analysis at P=0.05/3=0.0167 (two-sided) to reflect the three endpoints (NEC, sepsis, and total mortality). We view the results as equivocal as to probiotics. NEC, the primary endpoint, achieved an overall P-value below 0.05, stopping after study #10 when the maximum horizon was 12 studies. But it would not have stopped had the maximum horizon been 15 studies. The trivariate analysis would not have stopped for either maximum number of studies. We assumed a missing at random analysis (as did Deshpende et. al 1), for the two studies missing survival data and the one study missing sepsis data. In short, although our point estimates are similar to those of Deshpende et. al. 1, the true sampling errors are far larger than claimed, and hence, the efficacy of probiotics should not be viewed as evidence-based at this time.

As mentioned in an accompanying commentary by Soll 9, it is also notable that none of the studies in the meta-analysis were done in the US and the conditions could be very different. We therefore suggest a more cautious approach in order to avoid another “misadventure” in neonatal intensive care with the routine use of these agents in all very low birthweight babies.

References.

1. Deshpande G, Rao S, Patole S, Bulsara M. Updated Meta-analysis of Probiotics for Preventing Necrotizing Enterocolitis in Preterm Neonates. Pediatrics 2010.

2. Tarnow-Mordi WO, Wilkinson D, Trived A, Brok J. Probiotics reduce all-cause mortality in necrotizing enterocolitis: It is time to change practice. Pediatrics 2010; epublished ahead of print.

3. Lin HC, Hsu CH, Chen HL, et al. Oral probiotics prevent necrotizing enterocolitis in very low birth weight preterm infants: a multicenter, randomized, controlled trial. Pediatrics 2008;122(4):693-700.

4. Shuster JJ. Empirical vs. natural weighting in random effects meta -analysis. Statistics in Medicine, in press PMID: 19475538 (E-Publication; Will appear 5/30/2010)

5. Jonathan J Shuster JJ, Hatton RC, Hendeles L. and Winterstein AG. Reply to Discussion of “Empirical vs. natural weighting in random effects meta-analysis”Statistics in Medicine, in press. (Will appear 5/30/2010)

6. Shuster JJ, Jones LS, Salmon DA. Fixed vs. Random effects meta- analysis in rare event studies: the rosiglitazone link with myocardial infarction and cardiac death. Statistics in Medicine 2007; 26(24):4375- 4385.

7. Reboussin DM, DeMets DL, Kyungmann K, and Lan KKG. Computations for group sequential boundaries using the Lan-DeMets spending function method. Controlled Clinical Trials 2000; 21(3): 190-207.

8. O’Brien PC and Fleming TR. A multiple testing procedure for clinical trials. Biometrics 1979; 35: 549-556.

9. R. F. Soll Probiotics: Are We Ready for Routine Use? Pediatrics, May 1, 2010; 125(5): 1071 - 1072.

Conflict of Interest:

None declared

An updated meta-analysis of the effects of probiotics for the prevention of necrotizing enterocolitis (NEC) in preterm infants by Deshpande et al. (1) published in Pediatrics concluded that there are significant benefits of probiotic supplements in reducing death and disease in preterm neonates. Given these results, in the same issue of the journal, Tarnow-Mordi et al. (2) stated that when it comes to probiotics and NEC, ‘it is time to change practice.’ In our opinion, possible methodological limitations require rephrasing the statement to ‘Is it time to change practice?’

The major concern with regard to the meta-analysis by Deshpande et al. is whether it is appropriate to pool data on different probiotic microorganisms. It is tempting to produce a single estimate of the treatment effect. However, the results of a meta-analysis of all probiotics, regardless of the microorganisms used, may be misleading if appropriate consideration is not given to the interpretation of the pooled results.

The value of performing a meta-analysis is that by combining trials, the sample size is increased and, thus, the power. In the case of probiotics, pooled data on different microorganisms allow one to: 1) establish whether there is evidence of an effect of probiotics; 2) determine the direction of the effect; 3) determine the size of the effect (and the 95% CI around the effect); 4) assess the consistency of the effect across studies; and 5) identify the most promising probiotic(s). If many trials exist involving the administration of different probiotics to different participants with similar results consistently being seen in the various trials, the effect of the probiotic(s) has some generalizability. In addition, pooled data on different probiotics are important for demonstrating whether further research on these probiotics is substantiated. If so, this pooled data potentially may help to identify the most promising microorganisms as well as the research questions to be addressed in future studies.

Still, there are a number of arguments against pooling data on different probiotics. First, there is evidence that the beneficial effects of probiotics, particularly the immunomodulatory effects of individual probiotics observed in the host, differ greatly and are strain specific (3,4). Second, probiotics vary by organism. In addition to the most commonly used lactic acid bacteria (e.g., lactobacilli, bifidobacteria), the yeast Saccharomyces boulardii is often used. All of these microorganisms have different properties and antipathogenic mechanisms. Consequently, their efficacy, perhaps also safety, may vary (5). Third, the dose of probiotics may be important, as documented in a number of studies (6-9). Finally, the results of some studies do suggest a different response to probiotics in various populations (10-12). Collectively, these data suggest that is difficult to consider probiotic supplementation as a homogeneous intervention. Pooling data from different genera, species, strains, and doses of probiotics obtained in different settings and/or populations, presumably with variations in their native intestinal microbiota, may result in misleading conclusions. The risk is that the results could be erroneously extrapolated to other probiotics. This concern is shared by many experts in the field of probiotics and/or neonatal medicine (13-14).

What could be the solution? Given these concerns, in our opinion, the best approach would be to perform a meta-analysis that evaluates the effect of administering a clearly defined, single-organism, probiotic preparation or an equally well-defined combination of probiotic microorganisms for treatment of a specific disease or condition. However, a lack of data often makes this infeasible. With few exceptions, only seldom are there data from more than single studies on a given probiotic microorganism(s). However, results from 2 studies by Kalliomaki et al. (15) and by Kopp et al. (16) that assessed the effect of Lactobacillus GG use for the primary prevention of atopic dermatitis, which were carried out based on an almost identical study protocol but produced conflicting results, demonstrate the importance of such repeat studies. With a lack of repeat studies, can we be sure that all/any probiotic(s) studied for the prevention of NEC are really effective?

In conclusion, we believe that the strong statement in favor of the routine use of probiotics without clearly stating which probiotic(s), at what dose, for how long, etc. goes beyond the evidence. Also, the readers should be aware that such a statement is in contrast to the recent recommendation made by the Committee on Nutrition of the European Society for Paediatric Gastroenterology, Hepatology and Nutrition.13 The Committee concluded that the presently available data do not permit recommending the routine use of prebiotics or probiotics as food supplements in preterm infants; the Committee also recommended that each probiotic strain and potential combinations need to be characterized separately for each product.

Prof. Hania Szajewska(1), Prof. Raanan Shamir(2)

1. Department of Paediatrics, The Medical University of Warsaw; 2. Institute of Gastroenterology, Nutrition and Liver Diseases Schneider Children's Medical Center, Sackler Faculty of Medicine, Tel-Aviv University, Israel

REFERENCES

1.Deshpande G, Rao S, Patole S, Bulsara M. Updated Meta-analysis of Probiotics for Preventing Necrotizing Enterocolitis in Preterm Neonates. Pediatrics. 2010 Apr 19. [Epub ahead of print] PubMed PMID: 20403939. 2.Tarnow-Mordi WO, Wilkinson D, Trivedi A, Brok J. Probiotics Reduce All- Cause Mortality and Necrotizing Enterocolitis: It Is Time to Change Practice. Pediatrics. 2010 Apr 19. [Epub ahead of print] PubMed PMID: 20403934. 3. O'Mahony L, McCarthy J, Kelly P, et al. Lactobacillus and bifidobacterium in irritable bowel syndrome: symptom responses and relationship to cytokine profiles. Gastroenterology 2005;128:541-51. 4. Wickens K, Black PN, Stanley TV, Mitchell E, Fitzharris P, Tannock GW, Purdie G, Crane J; Probiotic Study Group. A differential effect of 2 probiotics in the prevention of eczema and atopy: a double-blind, randomized, placebo-controlled trial. J Allergy Clin Immunol 2008;122:788- 94. 5.Canani RB, Cirillo P, Terrin G, Cesarano L, Spagnuolo MI, De Vincenzo A, Albano F, Passariello A, De Marco G, Manguso F, Guarino A. Probiotics for treatment of acute diarrhoea in children: randomised clinical trial of five different preparations. BMJ 2007;335:340. 6. Whorwell PJ, Altringer L, Morel J, et al. Efficacy of an encapsulated probiotic Bifidobacterium infantis 35624 in women with irritable bowel syndrome. Am J Gastroenterol 2006;101:1581-90. 7. Basu S, Chatterjee M, Ganguly S, Chandra PK. Efficacy of Lactobacillus rhamnosus GG in acute watery diarrhoea of Indian children: a randomized controlled trial. J Paediatr Child Health 2007;43:837-42. 8. Basu S, Paul DK, Ganguly S, Chatterjee M, Chandra PK. Efficacy of high- dose Lactobacillus rhamnosus GG in controlling acute watery diarrhea in Indian children: a randomized controlled trial. J Clin Gastroenterol 2009;43:208-13. 9. Misra S, Sabui TK, Pal NK. A randomized controlled trial to evaluate the efficacy of lactobacillus GG in infantile diarrhea. J Pediatr 2009;155:129-32. 10. Vanderhoof JA, Whitney DB, Antonson DL, Hanner TL, Lupo JV, Young RJ. Lactobacillus GG in the prevention of antibiotic-associated diarrhea in children. J Pediatr 1999;135:564-8. 11. Arvola T, Laiho K, Torkkeli S, et al. Prophylactic Lactobacillus GG reduces antibiotic-associated diarrhea in children with respiratory infections: a randomized study. Pediatrics 1999;104:e64. 12. Thomas MR, Litin SC, Osmon DR, Corr AP, Weaver AL, Lohse CM. Lack of effect of Lactobacillus GG on antibiotic-associated diarrhea: a randomized, placebo controlled trial. Mayo Clinic Proc 2001;76:883-9. 13. Rijkers GT, Bengmark S, Enck P, Haller D, Herz U, Kalliomaki M, Kudo S, Lenoir-Wijnkoop I, Mercenier A, Myllyluoma E, Rabot S, Rafter J, Szajewska H, Watzl B, Wells J, Wolvers D, Antoine JM. Guidance for substantiating the evidence for beneficial effects of probiotics: current status and recommendations for future research. J Nutr 2010;140:671S-6S. 14. ESPGHAN Committee on Nutrition: Agostoni C, Buonocore G, Carnielli VP, De Curtis M, Darmaun D, Decsi T,Domellöf M, Embleton ND, Fusch C, Genzel- Boroviczeny O, Goulet O, Kalhan SC, Kolacek S, Koletzko B, Lapillonne A, Mihatsch W, Moreno L, Neu J, Poindexter B, Puntis J, Putet G, Rigo J, Riskin A, Salle B, Sauer P, Shamir R, Szajewska H, Thureen P, Turck D, van Goudoever JB, Ziegler EE; ESPGHAN Committee on Nutrition. Enteral nutrient supply for preterm infants: commentary from the European Society of Paediatric Gastroenterology, Hepatology and Nutrition Committee on Nutrition. J Pediatr Gastroenterol Nutr 2010;50:85-91. 15. Kalliomäki M, Salminen S, Arvilommi H, Kero P, Koskinen P, Isolauri E. Probiotics in primary prevention of atopic disease: a randomized placebo- controlled trial. Lancet 2001;357:1076-9. 16. Kopp MV, Hennemuth I, Heinzmann A, Urbanek R. Randomized, double- blind, placebo-controlled trial of probiotics for primary prevention: no clinical effects of Lactobacillus GG supplementation. Pediatrics 2008;121:e850-6.

Conflict of Interest:

None declared