I thought I would discuss the new study by O’Connor and colleagues in a little more detail (O’Connor DL, et al. Nutrient enrichment of human milk with human and bovine milk-based fortifiers for infants born weighing <1250 g: a randomized clinical trial. The American journal of clinical nutrition. 2018). They randomized babies who were under 1250 g birth weight whose mothers were planning to breast feed. There were about 250 eligible babies, and about half of them (127) were enrolled and randomized. The primary outcome variable of this new study from the OptiMom program was feeding intolerance, defined as having feeds stopped for more than 12 hours, or having a >50% reduction in feed volume. That might not be the best outcome, in terms of clinical significance, but choosing most aspects of trial design requires compromise, between what is important, what is essential, and what is possible.

The authors note in the introduction that a trial with NEC as the primary outcome would have needed nearly 1200 babies. I think that is a study that now really needs to be done. But, Prolacta supplied the human milk based fortifier for the study at the cost of manufacturing, the trial was otherwise funded by the CIHR.

I think trying to finance a trial similar to this new one, but much bigger at 1200 babies, might be impossible. According to a study of the cost effectiveness of an “exclusive human milk diet” (compared to a diet using bovine milk supplements and mother’s milk with a bovine fortifier) by Ganapathy et al in 2011, the fortifier cost about $10,000 per baby. Just the cost of the fortifier for a 1200 patient study would be 6 million dollars, if the 600 human milk fortifier babies received about the same amount of fortifier as in that analysis, and it was supplied at market rates. Even if it was supplied at cost it might be prohibitive. I think the next best thing we can do is to perform epidemiologic studies, preferably with propensity score matching, to see, in babies who all receive mothers milk supplemented with pasteurized donor milk, if human milk based fortifier really does lead to less NEC than bovine protein based milk fortifier.

In the absence of any such data I certainly can’t see my health care system funding the human milk based fortifier! I think that the O’Connor study might lead to some insurers in the USA wanting to stop re-imbursing this cost also, if there are many that do so.

The 95% confidence intervals for the difference in stage 2 NEC between the 2 groups in O’Connor’s study was about -8% to +8%. In the Ganapathy analysis that I just mentioned they calculated the cost-effectiveness of a mother’s milk with human milk based fortifier diet based on a reduction from 16% to 6% in NEC. With such a reduction in NEC the hospital costs of the fortifier were calculated to be less than the hospital costs of medical and surgical NEC.

In the two studies from the OptiMom project that I have discussed here recently (the study above and the comparison between artificial formula and donor human milk as supplements), the frequency of NEC in the groups that received mother’s milk, with a donor human milk supplement as necessary, and bovine fortifier, was very similar at just under 5%.

If we calculate the cost-effectiveness of an exclusive human milk based diet by assuming that the human milk diet completely eliminated NEC (which isn’t likely to be true, see below), then we can redo the analysis of Ganapathy.

Using a 5% frequency of NEC you can use the same calculations as Ganapathy did, to calculate the additional costs of NEC, basically, at an incidence of 16%, the costs of NEC per baby were about $24,000, so at 5% the additional costs per baby average somewhere around $8000. Human milk based fortifier is then not quite cost-effective even if you completely eliminated NEC! Of course cost-effectiveness is not the only concern: if you can truly prevent NEC then that is worth something! It certainly seems highly unlikely that a new large trial will show NEC to be completely eliminated by human-milk based fortifier, decreasing from 5% to 0%.

In the new study babies randomized to the human-milk based fortifier had an incidence of NEC of 5%; in the older study Sullivan S, et al. An Exclusively Human Milk-Based Diet Is Associated with a Lower Rate of Necrotizing Enterocolitis than a Diet of Human Milk and Bovine Milk-Based Products. The Journal of pediatrics. 2010;156(4):562-7.e1. The exclusive human milk group had an incidence of NEC of 6%.

That means that in total there are about 29+138+64 (=231) babies from the randomized trials who have received an exclusive human milk based diet, and in those 3 groups there were 12 cases of NEC stage 2 or more. For an incidence of just under 5%, and 95% confidence intervals of just under 3% to just over 9%, 99% confidence intervals are 2.2% to 10%

So it seems unlikely that a larger study would show a complete elimination of NEC, and cost-efficacy will depend on whether NEC really is decreased by replacing bovine-protein based fortification with human-milk based fortification.

It seems from the available studies that supplementing mother’s own milk with a pasteurized donor milk when there is not enough mother’s milk, decreases NEC, even when the milk is fortified with bovine milk products. In contrast there are no reliable data that going the next step and replacing the fortifier with a human-milk based product has any influence on the primary outcome of O’Connor, which was major feeding interruptions, or any of the other indicators of tolerance, which were very similar between the groups. It may be that diluting the bovine proteins in human milk, rather than giving a full feed of bovine-based milk, has very different impacts on gut inflammation. If there is an impact on NEC it is probably quite small. Growth variables were also no different between groups, partly because, probably, the human milk based group were immediately fortified to a higher assumed calorie density than the bovine based fortification group. With this difference in feeding approach, the weight, length, and head circumference remained very similar.

Late onset sepsis was numerically lower in the human milk based group, but may have been due to random error, Intriguingly though, severe retinopathy was much lower in the human milk based group (RoP needing intervention or stage 4 or 5) 6 cases (bovine fortifier) vs 1 case (human fortifier). Retinopathy of all grades was very similar in Sullivan’s study, and Cristofalo’s by diet group; so this might just be a chance, finding, but needs to be pursued in other epidemiologic work.