Here is the other study of saccharomyces that I mentioned previously on this blog (Demirel G, Erdeve O, Celik IH, Dilmen U: Saccharomyces boulardii for prevention of necrotizing enterocolitis in preterm infants: A randomized, controlled study. Acta Paediatrica 2013) recently published on-line 271 VLBW babies were randomized to receive the probiotic yeast or control. Infants were enrolled when they were ready to start enteral feeds. There was no effect on NEC. Secondary outcomes showed that feeding intolerance might have been reduced, and clinical sepsis, but not definite culture proven sepsis.

They had no invasive fungal sepsis in either group. A reduction of which might be one reason for including saccharomyces in future multi-species probiotic mixtures. But further work will be required before doing that.

The same group has already published a randomized comparison of the same fungus to Demirel G, et al: Prophylactic saccharomyces boulardii versus nystatin for the prevention of fungal colonization and invasive fungal infection in premature infants. Eur J Pediatr 2013:1-6. This was a comparative RCT in 181 VLBW infants who received either Saccharomyces or Nystatin to prevent colonization and infection with Candida. Infants were randomized at 72 hours of age. There were no significant differences between the groups’ outcomes, 2 nystatin and 0 probiotic babies developed invasive candida.

Both of these studies have the same important error, which probably doesn’t affect the conclusions but should have been picked up by the referees: several infants died after enrollment in the trial, and they are then eliminated from the data, as if they had been ineligible. They are not even included in the numbers of deaths. This means that it is not an intention to treat analysis, any event occurring after enrollment must be included, 9 deaths among 190 subjects is a lot of deaths, and if unevenly distributed might mean that there is a difference between the groups. No justification is given for eliminating these babies from analysis. It is certainly possible that probiotics might have effects within a few hours of being given (positive or negative) so all babies randomized must be included.

There is also something strange though that I cannot understand. Both of these studies were performed in the same mega-hospital, which has an enormous NICU and 4000 admissions per year. The first was performed between March and November 2011 during which time they had a total of 365 VLBW infants admitted, 86 were excluded and the remainder were randomly assigned to saccharomyces or control. The second study was performed between May 2011 and November 2011, there were 240 admissions of VLBW infants, and 50 were excluded, leaving the 190 who were randomized, 9 of whom died later and were excluded.

It looks like from these publications that the studies were performed at the same time in the same NICU, but the total numbers don’t add up, as there aren’t enough total babies to give the numbers in the CONSORT flow diagrams. Is it possible that there were babies in both studies? If so the controls in the first study were actually getting nystatin, and their data have been duplicated. If not there seems to be another error of some sort.

Finally a study of a lactobacillus: Oncel MY, Sari FN, Arayici S, Guzoglu N, Erdeve O, Uras N, Oguz SS, Dilmen U: Lactobacillus reuteri for the prevention of necrotising enterocolitis in very low birthweight infants: A randomised controlled trial. Archives of Disease in Childhood – Fetal and Neonatal Edition 2014, 99(2):F110-F115. This is again from the same 150 bed NICU, but from a different time period. 424 VLBW infants were randomized if they survived to start enteral feeds, to either Lactobacillus reuteri or placebo. Unfortunately the authors do the same thing with the early deaths in this study as well, infants who died within the first week of life were eliminated post-enrollment. 16 deaths are therefore not analyzed (7 probiotic vs 9 placebo). Unlike the majority of other studies, the authors showed no effect on NEC, which was 5% among placebo babies, and 4% with probiotics. Nor on teh combined outcome of survival without NEC, which was 90% with probiotics and 97% with placebo (to invert their numbers).

Even if you add back in the babies eliminated from consideration as they died too soon, (which I have done) the chi-square is nowhere near significance, p=0.37 with Yates correction. So this is a negative study, is it the wrong probiotic? The wrong dose? Inactive organisms? Just the expected random variation?

This particular organism, (it seems to be exactly the same, with the same DSM number) has been used before, in a retrospective cohort study fewer of the 79 babies under 1 kg who received L reuteri developed NEC than those admitted before routine use. This was a much higher risk group, with a much higher rate of NEC before initiation of probiotic prophylaxis than the control group rate in the new study. Maybe that is the difference, a lack of power of the new study, which was prospective and randomized. Oncel et al did have fewer episodes of sepsis, and like most of the studies that have reported their findings, less feeding intolerance, and shorter duration of TPN as a result.