South West local group meeting: Modelling numbers of births

The South West local group held a meeting on Wednesday, 20th February 2019 at the University of Plymouth. The speaker was Mario Cortina Borja, Professor of Biostatistics of the UCL Institute of Child Health.

Mario described that, in England and Wales, there are several information systems which contain routinely collected data on newborn babies. In England, these include the Office for National Statistics’s (ONS) birth registration records, NHS Numbers for Babies (NN4B) and Hospital Episode Statistics (HES), and for Wales, the Patient Episode Database for Wales (PEDW) and the National Community Child Health Database (NCCHD). 

Mario then explained the data linkage methods used to match records across these databases. This procedure led to a data set of over five millions singleton births with linked information on birth data and morbidity and mortality in the first year of life from infants born between January 2005 and December 2014.

Based on this national data set, Mario then showed the data visualisations and analysis used to identify the daily, weekly and yearly cycles and pattern of number of births. By plotting the number of live births against the timing of births, line graphs illustrated the changing number of live births with the year, day of month and hour of day of births. The immediate information from the line graphs include in the cyclic patterns of number of births over years, the most popular birthdays as well as the relatively low number of births during some specific periods.

Bar charts illustrated variations in the number of live births by seven days of a week as well as public holidays. Line charts showed the cyclical patterns of frequencies of live births along different types of day over years.

Mario then discussed regression modelling of variations in time and day of births, with independent variable including onset of labour, mode of birth, gestational age in singleton births delivered in NHS maternity units in England between 2005 and 2014.

Negative binomial regression models were fitted to the number of births on days of the year, to account for the great variability in the data. The regression models allowed estimating the incidence rate ratios of the expected number of births on a type of day adjusting for long-term trends, yearly cycles, seasonal variations and clock changes. The interactions between gestational age and types of day were estimated for eight care pathways defined as combinations of type of onset of labour and mode of birth. These include spontaneous and induced mode of labour combined with spontaneous, emergency caesarean and instrumental modes of birth, as well as elective and emergency caesareans with no labour. The estimated incidence rate ratios and their 99% confidence interval were plotted as error bars by different types of day and gestational ages for each of these care pathways. The line graphs of average number of births per hour illustrate in what hours within a day specific types of onset of labour and mode of delivery are concentrated. 

These models show that timing of births varies considerably by care pathway and gestational age, and these patterns have implications for midwifery and medical staffing in NHS maternity units. Mario concluded that further research is needed to fully understand the processes behind these findings.

This project was funded by the National Institute of Health Research (NIHR) and was led by Professor Alison Macfarlane (City University) in collaboration with Nirupa Dattani, Dr Miranda Dodwell, Rod Gibson, Dr Gill Harper, Dr Peter Martin and Dr Mary Newburn.

Load more