Dear all:

I'm thinking about what future studies on NEC and donor milk should include the lessons learned 
from the Schanler study and improve the study design accordingly.

The drop in NEC suggests that any additional that any additional studies should be done with a 
larger sample size that reflects the incidence you can expect with the current level of care.  This 
should be established prior to conducting the study using RECENT records from the hospital 
where the study is to take place rather than extrapolated from other studies where the standards 
of care may differ in subtle, but important ways.  Since NEC is such a serious outcome, I think the 
expected differences between the group should be much smaller than the 25 percentage points 
chosen by the authors - at most a difference of 5 to 10 percentage points.  

The high percentage of infants in the donor milk group who were given formula is something that 
can't be ignored and is just as important.  Without verification that the actual nutrient content is 
the same in each treatment group one could easily be comparing apples to oranges and you would 
have know way of knowing whether or not that was the case.  Individual testing of every milk 
sample would not be feasible either logistically or economically.  So, several approaches come to 
mind:

1)  Use a pooled analysis technique such as the ones devised by the statisticians at Cornell.  This 
would drastically reduce costs and the statistical procedures involved in pooling are well worked 
out for how you determine an appropriate sample size that is much lower than is needed for 
individual samples.  (The China study used samples from hundreds of thousands of people).  The 
only danger of this technique is if you get "outliers".  That is values that are way off the charts.  If 
one sample happened to be an outlier, it could dramatically shift the results of a whole pooled 
sample.  I'm sure the statisticians have worked out the probabilities of this happening and 
included this in their tehcniques.

2) Adopt a modified quality assurance technique. Laboratories set aside a certain number of 
samples for "quality assurance" testing either in a repeat analysis or with a more costly "gold 
standard" technique that would be unaffordable to use on all the samples.  So for a study of this 
nature you would choose a subsamples from the donor milk and the mother's own milk and 
determine the "average" fat and protein content and selected important nutrients.    In this way the 
mother's own milk and the donor milk could be calibrated in such a way that the "average" 
nutrient content of the milks was similar.  

3).  Now here is where those of you who work in NICUs would have a much clearer picture on this 
possibility than I would.  Is there any reason why the donor milk could not be pooled so that it is 
fairly consistent?  Will this introduce loss of fat when this milk is transferred from one container to 
another?  Introduce too much potential for contamination?  Or would it be possible to have a 
pretty homgenous sample and do it safely?

If 3) were possible that would make it impossible to really compare to the mother's own milk 
group because the milk would be highly variable.  However, comparing the mother's own milk 
group to the other two groups is ALWAYS going to be comparing apples to oranges because you 
can't ethically randomize that group when it is known that mother's own milk confers benefits.  
This group should never be compared in any sort of "probability" analysis for that reason.

I'm also thinking that future studies could also include subgroups that looked at the interactive 
effects of kangaroo care within the pooled donor and formula groups. 

Susan E. Burger, MHS, PhD, IBCLC



 

             ***********************************************

To temporarily stop your subscription: set lactnet nomail
To start it again: set lactnet mail (or digest)
To unsubscribe: unsubscribe lactnet
All commands go to [log in to unmask]

The LACTNET mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software together with L-Soft's LSMTP(R)
mailer for lightning fast mail delivery. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html