OP92 Neonatal mortality in NHS maternity units by timing of birth and mode of birth: a retrospective linked cohort study

Aim

Potential ‘weekend effects’ in healthcare prompt concerns that care could be of lower quality during non-working hours, but may reflect differences in case mix or other factors. To investigate this we compared neonatal mortality in English hospitals from 2005 to 2014 by time of day and day of the week.

Methods

We analysed data from a retrospective cohort of 6,054,536 singleton live births in England 2005—2014, created by linking Office for National Statistics’ birth and death registration and birth notification data with Hospital Episode Statistics. Working hours were defined as 07:00—19:00 on weekdays, and non-working hours were all other times on weekdays and all weekends and public holidays. The primary outcome was neonatal mortality attributed to asphyxia, anoxia or trauma. On advice through our public involvement and engagement strategy, analysis was stratified by mode of onset of labour and mode of birth. We fitted logistic regression models in R to estimate odds ratios for mortality, adjusting for covariates.

Results

After adjustment, the odds of neonatal mortality outside of working hours were similar to those during working hours for spontaneous births (OR: 1.09, 95% CI: 0.87—1.35), instrumental births (0.96, 0.70—1.32) and emergency caesareans following spontaneous (1.06, 0.83—1.34) or induced (1.14, 0.74—1.76) onset of labour. Emergency caesarean births without labour had a higher risk of neonatal mortality attributed to anoxia, asphyxia or trauma during non-working hours compared with working hours. Further stratification showed that this higher risk was at night-time (weeknight: 1.56, 1.15—2.11; weekend/holiday: 1.75, 1.24—2.47) and that risk during the daytime on weekends and holidays was not significantly higher than on weekdays (0.95, 0.63—1.46). Even then, the number needed to harm was 1258 for weeknights and 933 for weekend/holiday nights, indicating very low absolute risk.

Conclusion

We found no significant association between time of birth and neonatal mortality. The apparent ‘weekend effect’ may be caused by deaths among the relatively small number of babies who were born by caesarean section apparently without labour outside normal working hours. Obstetric staffing should be planned to allow for these relatively unusual emergencies, although reorganisation has opportunity costs and these findings would not warrant major changes. Continuous updating of the data linkage would facilitate ongoing monitoring of these findings.