Assessing the Effect of Risk Factors on Rates of Obstetric Transfusion over Time Using Two Methodological Approaches

Background: While red blood cell transfusion rates have declined in most Australian medical specialties, obstetric transfusion rates have instead been increasing. Obstetric transfusions are mostly linked to postpartum haemorrhage, the rates of which have also increased over time. This study used two methodological approaches to investigate recent trends in obstetric transfusion in New South Wales (NSW) and the extent to which this was influenced by changing maternal and pregnancy characteristics.

Methods: Linked birth and hospital records were used to examine rates of red blood cell transfusion in the postpartum period for mothers giving birth in NSW hospitals from 2005 to 2015. Logistic regression models were run to examine the contribution of maternal and pregnancy risk factors to changing rates of transfusion. Risk factors were divided into "pre-pregnancy" and "pregnancy related". Crude and adjusted estimates of the effect of year of birth on obstetric transfusion rates were compared to assess the effect of risk factors on rates over time using two approaches. The first compared actual and predicted odds ratios of transfusion for each year. The second compared the observed increase in transfusion rate with that predicted after controlling for the risk factors.

Results: Among 935,659 births, the rate of obstetric transfusion rose from 13 per 1000 births in 2005 to 17 in 2011, and remained stable until 2015. From 2005 to 2015, postpartum haemorrhage increased from 74 to 114 per 1000 births. Compared with the rate in 2005, the available maternal and pregnancy characteristics only partially explained the change in rate of transfusion by 2015 (Method 1, crude odds ratio 1.39 (95% CI 1.25, 1.56); adjusted odds ratio 1.29 (95% CI 1.15, 1.45)). After adjustment for maternal and pregnancy characteristics, obstetric transfusion incidence was predicted to increase by 10.3%, but a 38.7% increase was observed (Method 2).

Conclusion: Rates of obstetric transfusion have stabilised after a period of increase. The trend could not be fully explained by measured maternal and pregnancy characteristics with either of the two approaches. Further investigation of rates and maternal and clinical risk factors will help to inform and improve obstetric blood product use.