Use of multiple linear regression and logistic regression models to investigate changes in birthweight for term singleton infants in Scotland

Aim. To illustrate the use of regression and logistic regression models to investigate changes over time in size of babies particularly in relation to social deprivation, age of the mother and smoking. Background. Mean birthweight has been found to be increasing in many countries in recent years, but there are still a group of babies who are born with low birthweights. Design. Population‐based retrospective cohort study. Method. Multiple linear regression and logistic regression models are used to analyse data on term ‘singleton births’ from Scottish hospitals between 1994–2003. Results. Mothers who smoke are shown to give birth to lighter babies on average, a difference of approximately 0·57 Standard deviations lower (95% confidence interval. 0·55–0·58) when adjusted for sex and parity. These mothers are also more likely to have babies that are low birthweight (odds ratio 3·46, 95% confidence interval 3·30–3·63) compared with non‐smokers. Low birthweight is 30% more likely where the mother lives in the most deprived areas compared with the least deprived, (odds ratio 1·30, 95% confidence interval 1·21–1·40). Conclusions. Smoking during pregnancy is shown to have a detrimental effect on the size of infants at birth. This effect explains some, though not all, of the observed socioeconomic birthweight. It also explains much of the observed birthweight differences by the age of the mother. Relevance to clinical practice. Identifying mothers at greater risk of having a low birthweight baby as important implications for the care and advice this group receives.