A population
based study demonstrating an increase in the number of overweight and obese
children between 1989 and 1998
Authors response
The Editor
British Medical Journal
BMA House
Tavistock Sq
London WC1H 9JR
Dear Sir,
Thank you for accepting the manuscript entitled "A population-based
series of cross sectional studies demonstrating an increase in the number
of overweight and obese children between 1989 and 1998" for publication
in the British Medical Journal. A copy has also been sent by email to your
address. The document is in Word format, apart from figures 5 and 6, which
are in Excel format.
I have enclosed a separate sheet with our responses to all the revisions
that you have requested.
We are happy for a shortened version to be published in the BMJ.
Yours sincerely
Dr Peter Bundred MD
Reader in Primary Care
Editorial comments
-
We have considered the reviewer's comments.
-
We do have some information on the socio-economic status of children included
in the study, but have not yet had an opportunity to analyse the data.
It would not be possible to include it in this paper.
-
We have examined the incidence of obesity and overweight as measured by
BMI in boys and girls, and results are included in the paper.
-
The study was population-based and did not include subject identifiers;
it was therefore not possible to link the infant group to the same children
measured three to four years later.
Referee’s comments
-
No action
-
No action
-
Addressed in discussion (pg 9)
-
All information on missing and inaccurate data has been included in methods,
and details of the reasons for exclusion are included in the first of the
tables submitted not for inclusion in the paper, but submitted as additional
information, for publication on the web as appropriate. Table of Study
group (infants and children) with details of exclusion. (pg 21): the criteria
for exclusion of outliers was greater or less than 5 standard deviations
from the median as shown in the table.
-
No comment necessary
-
The decision to use nonparametric statistical methods was based upon the
data being a censored sample, which is implicitly not from a normal distribution.
The referee seems to be confusing the distribution of uncensored standard
deviation scores (SDS) in the reference population with the distribution
of censored SDS in our study population. If parametric analysis
(see enclosed appendix I) is used, then the inferences drawn are almost
identical to those from nonparametric analysis; this is not surprising
given the force of large numbers. We regard the nonparametric approach
as statistically more cautious and robust.
-
Our choice of three decimal places for P values is based upon the classical
description of "statistically highly significant" terminating below a threshold
value of 0.001.
-
Trend in the central location (median) is accompanied by equivalent shifts,
and not by a change in the size of the spread of the data. This reflects
consistency in the data over time, which pertains to the referee's previous
statement on data quality.
-
In our opinion, quadratic or other additive regression modelling of these
data does not add any clinically useful information to the paper.
-
We observed the real differences on which the referee has commented. We
do not have an explanation for this. We are confident, however, that the
differences are not an artefact of the reference method used to calculate
SDS; British Growth Reference Charts that we used for infants and pre-school
children were derived from the same general population sampled at the same
time, thus they provide a fixed reference point for both infants and pre-school
children.
-
We agree with these points.
-
We agree that these figures should only be included in the electronic version.
-
Addressed