Background:

Early life exposures have been hypothesised to affect subsequent levels of glucose and insulin. We aim to draw together all of the research done to date looking at this possible relationship in all normal populations in both children and adults.

Scientific hypothesis:

That there is an association between birthweight and subsequent levels of glucose and insulin.

Objectives:

To establish the difference in mean levels of fasting and post load glucose and insulin per one Kg increase in birthweight. We also wish to examine the effect of potential modifiers and confounders.

Methodology and planned statistical analyses:

The first stage is to identify all studies with published reports on the relationship between birthweight and subsequent levels of glucose and insulin in general populations of both children and adults. Investigators will be asked to provide data (preferably) or to carry out pre-defined analyses. Individual study data will be analysed with multiple logistic regression to obtain estimates for the differences in glucose and insulin that are associated with a 1 kg increase in birthweight, based on the full range of birthweights for each study. Models will be fitted without adjustment and then with additional adjustments for

* age and (if appropriate) gender

* for social class

* current body mass index

Analyses will also examine the effect of excluding subjects with

* birthweights greater than 4kg

* maternal diabetes.

The estimates will then be included in meta-analyses, carried out using fixed-effects modelling if there is not appreciable evidence of heterogeneity.