Intended for healthcare professionals

Rapid response to:

Primary Care

Preventing childhood obesity by reducing consumption of carbonated drinks: cluster randomised controlled trial

BMJ 2004; 328 doi: (Published 20 May 2004) Cite this as: BMJ 2004;328:1237

Rapid Response:

My take on the studies limitations.

The first study, ‘Preventing obesity by reducing consumption of
carbonated drinks: cluster randomised controlled trial’ (James, Thomas,
Cavan and Kerr, 2004) aims to reduce the consumption of carbonated drinks
in 615 children aged 7-11 years old via the delivery of a focused
educational programme on nutrition in schools. This study utilises a
cluster, randomised controlled experimental design. The actual procedure
is really rather complicated and lacks detail as in it leaves the reader
with more questions than answers and because of this, accurate replication
would likely be difficult. For example, it is not clear precisely how
much time and method of delivery was devoted to each component;
discouragement of ‘fizzy’ drinks, affirmation of a balanced healthy diet,
drinking water, presenting art, writing songs/raps outlining healthy
messages. And as a result of this it is impossible to identify which
aspects were actually effective and which were unnecessary. There is no
effort to detail exactly how these messages were delivered to the
participants apart from the considerations they made with regards to the
young age of the participants. The researchers tailored the study in many
ways to try to make the programme accessible to children. Firstly they
did this by keeping the message simple; ‘by decreasing sugar consumption
they would improve overall well-being’, to presumably help the children
understand the message though this is not explained and there is no
reference to previous research to justify this. Secondly, the study
incorporates a music competition for the participants to write a song or
rap with a healthy message and also produce and present pieces of their
own art. Once again there is no rationale for this but is probably used
to get the participants involved and thinking about the message and to
hold their attention. Clearly this would not be likely to be utilised in
a study using adults as participants and is more appropriate to childhood
intervention. The song writing and art presentations sound like very
appropriate methods for this intervention and demonstrate the importance
of the Lifespan approach when designing interventions to reach children
but unfortunately there is absolutely no indication of how these were
implemented or more importantly the impact that they might have had. It
would have been very interesting to read feedback of a qualitative nature
about any difficulties they had implementing this and whether the children
really became involved or not. In a perfect world one could imagine all
the children in a class competitively and enthusiastically painting and
singing and drinking water in their element but is that a realistic
approximation? After all, the interpretation of the results hinges on
these details.

On the positive side, the participants in the sample were
representative of the target population as this is aimed at
reducing/preventing childhood obesity and the size of the sample was large
enough for the number of measures and the effect size. The clusters (in
this case class) were randomly associated to either intervention or
control group and written consent was sought from both parents and
participants. It is claimed that each group were similar and equally
distributed in terms of age, sex and consumption of sweetened carbonated
drinks but there is no statistical test to confirm this. One problem with
the sampling, which the researchers point out, is that school s contained
classes both in the experimental and the control group and therefore it is
possible that ‘transfer of knowledge may have taken place outside the
classroom’ with participants discussing the different conditions amongst

The measures taken are very empirically dubious. The participants
were asked to keep a three day diary both at the beginning and at the end
of the intervention (over one school year) and keep record of the drinks
that they consumed. It is doubtful whether this could be regarded an
appropriate method of collecting data considering the sample used. Is it
feasible to ask a 7 year old to keep an accurate diary indicative of the
complete beverages they have consumed? This is a poor method of
measurement with participants as young as this and indeed this was
reflected in the low number of completed diaries they received both at
baseline and the climax of the intervention. This resulted in a biased
result as the only data they received was from children motivated to keep
the diary and it is questionable how accurate these will have been in any
case. The anthropometric measurements were however a more appropriate
method of measurement. After the year long intervention the percentage of
obese/overweight participants had reduced in the intervention group
whereas the percentage had increased in the control group. This does
indeed indicate that the intervention did have a positive effect on the
participants’ health; it is just unclear precisely which elements are
responsible for this as the intervention aimed at healthy eating as well
and reducing carbonated drinks. Also, teachers in the intervention group
were asked to ‘encourage and reiterate the message in lessons’. This
alone could be responsible for the difference and is completely out of the
researchers’ control. Some teachers may encourage the health promoting
message frequently and others less so if at all. This study could be
described as being an amalgamation of multiple attempts to deliver the
health promotion message of reducing carbonated drinks and healthy eating
to children in various forms with flimsy description of precise procedures
but actually in the end it is reasonably successful. It has some
interesting methods of engaging the children and because it is multi-
faceted it seems to bombard the messages and the results seem to suggest
that the do influence the participants’ eating and drinking behaviour.

In line with the Lifespan approach, this study is important because
it is targeting children at a time when the instance of childhood obesity
is on the increase. By successfully equipping children with the ability
to avoid becoming obese, you automatically reduce incidence of future
adults becoming obese and techniques that will make a difference over the
lifespan. Magarey et al (2003) found that over 70% of obese children went
on to become obese adults. Therefore it is important to target children
with obesity prevention messages in order to prevent the next generation
being obese. This research targeted reduction of carbonated drinks as its
main message. This is interesting because children have fairly limited
control over what they eat and drink. This is because they are not old
enough to shop and cook independently and are therefore likely to be
eating similarly to their families or what is available at school. This
intervention target carbonated drinks as hey are arguably within a child’s
control of consumption. This is therefore highly appropriate along with
the general healthy eating message. This study does contribute to
existing research as it is set in the school setting and thus reaches its
targeted population. There are very few other school based obesity
interventions (Ebbeling, 2006)

Competing interests:
None declared

Competing interests: No competing interests

12 January 2007
Stephen D East
MSc Health Psychology student + Dual Diagnosis Support Worker
TS10 1RD