Dear all:

Johns Hopkins taught me to deconstruct, Cornell taught me to reconstruct --- so 
therefore, I'm sending the final piece --- the recontruction of how one might design a 
study about hormones in milk.

My dissertation advisor was a brilliant man and he pointed out, counter to what many 
think that you don't actually need to know why an intervention works to know that it does 
work.  The example is embedded in the word "Limey" that Americans sometimes use to 
describe those from Great Britain.  This term comes from the fact that British seafarers 
discovered that you could prevent scurvy long before anyone isolated vitamin C.  You do 
NOT need to know the cause if the intervention works.

So, in thinking about lactation consultants and our often too short of a window for 
interventions --- the intervention we would implement if growth hormone in milk were 
causing an oversupply, would be to eliminate the growth hormone in milk.  For those who 
were already drinking milk without growth hormone, we would not need to intervene so 
they would not be our study group.

There is a simple design to achieve this.  Select a group of women who are drinking 
cow's milk with growth hormone.  Provide them with milk in identical containers. Half the 
group would be randomly assigned to receive milk with growth hormone, half the group 
would be randomly assigned to drink the milk with growth hormone that they WERE 
ALREADY DRINKING.  In this fashion, you would not be violating human ethics standards 
in ways that you would if you were giving growth hormone milk to those who weren't 
already drinking it.  Moreover, it targets the potential group for intervention. 

You could start the intervention at various time points --- during pregnancy, after delivery 
etc --- and randomize within  those groups to isolate when you might have the biggest 
impact. 

The trickiest part would be to come up with a really tight definition of oversupply.  If you 
do not use very tight definitions, you might lose the "effect" because you might actually 
be including symptoms that are not truly oversupply. You would also need to standardize 
the observers to make sure that they were "diagnosing" oversupply the same way. There 
are methods to assess interobserver variation and minimize it.

In this fashion, you could test the "efficacy" of the short run intervention of removing 
milk with hormones from a woman's diet.  You would also need to indentify subgroups 
within your sample in case one subgroup responds and another subgroup does not.  If you 
do not collect data on potential subgroups you may miss a real effect that is washed out 
in the overall population impact.

The second step would be to assess the "effectiveness" of interventions to convince 
mothers to remove milk with hormones from their diet.  The "effectiveness" is always 
less than the "efficacy" because some women will not want to follow through on the 
intervention and inevitably some women in your nonintervention group will start to do the 
intervention on their own.

This does not get at the impact of generational effects --- that would take a much 
LONGER period of observation in order to really determine if what we are seeing is an 
"ecologic fallacy" or Nora Ephron effect or something that is likely to be causal.

Best, Susan Burger

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

Archives: http://community.lsoft.com/archives/LACTNET.html
To reach list owners: [log in to unmask]
Mail all list management commands to: [log in to unmask]
COMMANDS:
1. To temporarily stop your subscription write in the body of an email: set lactnet nomail
2. To start it again: set lactnet mail
3. To unsubscribe: unsubscribe lactnet
4. To get a comprehensive list of rules and directions: get lactnet welcome