ISQUA16-1955USING ROUTINELY COLLECTED ADMINISTRATIVE DATA TO FURTHER EXAMNE VARIATION IN OBSTETRIC OUTCOMES BY DAY OF THE WEEK

Abstract

Objectives: Recent published work demonstrated variation in birth outcomes by day of the week in English Public Hospitals. [1] Over the seven indicators examined, we found a statistically significant association consistent with a lower standard of care for babies born and women admitted over the weekend. The so called "weekend effect". Perinatal death, injury to the neonate, puerperal infection and three day neonatal readmissions were all higher at weekends compared with weekdays. Our objective for this analysis was to examine whether previously unaccounted for case-mix factors could account for these variations. Methods: We identified maternal and neonatal records in English public hospitals between 1st April 2010 and 31st April 2012 using a national administrative dataset. Using previously published case-mix adjustment variables [1] including gestational age, birth weight, maternal age, delivery method, ethnicity and socio-economic deprivation, we calculated the odds of three maternal outcomes (perineal tear, puerperal infections and three day maternal emergency readmissions), and four outcomes for babies (in-hospital perinatal mortality, injury to neonate, selected neonatal infections and three day neonatal emergency admissions) for weekend versus weekdays. We explored the effect of specifically excluding elective caesareans and antenatal still births and adjusted for induction of labour. Results: We identified 1,332,835 deliveries and 1,349,599 births. After exclusion of elective caesareans, we found a persistently high risk at weekends of puerperal infections (OR 1.05, 1.00-1.09), in-hospital perinatal mortality (OR 1.07, 1.02-1.12) and of injury to the neonate (OR 1.05, 1.02-1.08) compared with weekdays. For perinatal mortality, the higher weekend risk persisted even after exclusion of antenatal stillbirths (OR 1.09, 1.02-1.16). After adjusting for induction of labour, puerperal infection rates remained higher at weekends (OR 1.07, 1.02-1.12) Conclusion: The "weekend effect" for birth outcomes seems to persist, even after accounting for potential selection bias or mis-classification of stillbirth date. Administrative data remain a useful starting point for examining variations in the quality and safety of health care. Further work is required to understand the potential reasons for these apparent disparities.