- A prospective, blinded, randomized, multi-center controlled trial was performed in 7 NICUs across Taiwan from April 2005- May 2007 to investigate the effect of probiotic therapy on the incidence of NEC (≥ Stage 2) or death in VLBW, premature infants. Methods and results include:
- Inclusion/Exclusion Criteria: VLBW (<1500g), premature (<34 week) infants who survived to feed enterally were eligible; however, of those infants, any with severe asphyxia (Stage III), fetal chromosomal anomalies, cyanotic congenital heart disease, congenital intestinal atresia, or gastroschisis were excluded as were those who were fed exclusively with formula or were kept NPO for >3 weeks.
- Intervention and duration: After randomization by sequential, computer-generated numbering, study group was given Infloran (L acidophilus + B bifidum, 109 CFU each) at 125 mg/kg/dose BID mixed with breast milk or formula (mixed feeders) for six weeks from the initiation of enteral feeding. Control group was given breast milk or formula (mixed feeds) alone. Breast feeding team prepared breast milk and formula for both groups. A strict feeding protocol was followed at all sites.
- Outcomes/results: Primary outcomes included death or NEC ≥ Stage 2. NEC was diagnosed and classified by two independent attendings who did not know the group assignment of the infants. If there was not agreement, a senior neonatologist was consulted to establish the final diagnosis. Secondary outcomes included: culture-proven sepsis without NEC, CLD, IVH, PVL, feeding amount per week, days to full enteral feeding and weight gain per week. Consensus was established at all centers regarding the definition of outcomes and all clinical variables. Adverse effects including culture-proven sepsis attributable to L acidophilus or B bifidum, flatulence and diarrhea were recorded. The study group had 4 cases of death or NEC of 217 participants; the control group had 20 cases of NEC or death of 217 (p= 0.002). NNT = 14. Study group had 4/217 cases of NEC compared with 14/217 cases in the control group (p=0.02). NNT = 22. For secondary outcomes, there was no statistically significant difference in these measures between the groups with multivariate analysis, although sepsis trended toward a higher frequency in the study group.
- Validity: Patients were randomized and this was concealed from all caregivers. (Breast Feeding Team in charge of mixing all formula and breast milk, with or without probiotics, was obligated to know assignments in order to mix these appropriately; however, Team otherwise not involved in patient care.) Principle Investigators at each center also knew group designation; article does not state explicitly that these individuals were not involved in patient care. Patients were analyzed in the group to which they were assigned. Control and study groups were largely similar; however, post-study analysis of demographic and clinical variables showed the study group to have a statistically significant lower average birth weight and lower first arterial pH. These differences make the study group appear “less well,” and should have skewed any potential effect of the intervention toward no effect or even harm. This was not the case for primary outcomes.