Hi Katie
I just spotted your email among a myriad of mailbase messages that I hadn't
read for a while. I get my email package to sort mailbase stuff from more
urgent messages into a separate folder so that I can read it at "leisure".
You have probably had an answer to this by now but if you haven't, the milk
study is described by Gore and Altman, Collecting and Screening Data. In
Statistics and Ethics in Medical Research. BMJ Publications (1988).
The study was done in 1930 in schools in Lanarkshire and involved 20,000
children: 10,000 given milk and 10,000 controls. Height and weight
measurements were taken at the beginning of a six-month period starting in
Winter and ending in Summer.
The main feature of the study were that it was biased because
1) randomisation broke down: teachers gave milk to poorer children. The two
groups were not similar at the beginning of the study, the milk group having
lower mean weight and height than the control group.
2) the children were weighed with their clothes on at the start (Winter) and
end of the study (Summer) which messed up any growth estimates.
As a result, controls appeared to do no worse than children receiving milk.
The impact of the study was that it encouraged Thatcher "the Milk Snatcher"
as she was called at the time, to stop free school milk in the 70s.
Hope this is helpful
Good luck in your new job, Katie!
----- Original Message -----
From: Alison Hill <[log in to unmask]>
To: Evidence-based mailbase <[log in to unmask]>
Sent: 01 November 1999 23:46
Subject: School milk
> Request from Katie Enock ([log in to unmask]) Please reply to
> Katie, and she will post up responses.
>
> Please can anyone help. There was a study conducted in the 1950s looking
at
> the benefits of milk on primary school children. The school nurse (I
think)
> was randomizing the children and she put all the needy children in the
> treatment group. Needless to say - there was an amazing treatment
effect -
> which was due to appalling randomization.
>
> I would be grateful if you could give me the formal reference please.
>
> Thanks in anticipation.
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|