Temporal trends of postpartum haemorrhage in Switzerland : a 22-year retrospective population-based cohort study

Kaelin Agten Andreaa, Passweg Danielb, von Orelli Stephanieb, Ringel Nancyc, Tschudi Ruedid, Tutschek Borise a Fetal Medicine Unit, St George’s University Hospital NHS, London, United Kingdom b Department of Obstetrics and Gynaecology, City Hospital Triemli Zurich, Switzerland c Department of Obstetrics and Gynecology, New York University School of Medicine, USA d Sevisa AG, Ermatingen, Switzerland e Pränatal Zürich, Zurich, Switzerland, and Medical Faculty, Heinrich Heine University, Düsseldorf, Germany


Introduction
Postpartum haemorrhage (PPH) is a leading cause of maternal morbidity and mortality worldwide [1,2].The World Health Organization (WHO) definition of postpartum haemorrhage encompasses all blood losses over 500 ml within 24 hours after birth [3].The most common causes associated with postpartum haemorrhage are uterine atony, retained placenta and postpartum coagulation defects [4,5].Known risk factors include advanced maternal age, hypertensive disorders, chorioamnionitis, placenta praevia, polyhydramnios, labour induction, prolonged and augmented labour, and vaginal operative delivery [4,[6][7][8].Population-based studies from Ireland, Canada, Australia and the USA have demonstrated an increase in the incidence of postpartum haemorrhage during the past decade [9][10][11][12].Changes in maternal demographics and obstetric practice may have contributed to the rising numbers of postpartum haemorrhages, such as older maternal age or the rising numbers of caesarean deliveries [5].The objective of this study was two-fold: to investigate the incidence of postpartum haemorrhage and to assess changes in risk factors in Switzerland during the last two decades.

Materials and methods
This was a retrospective population-based cohort study using de-identified data from the national Swiss Hospital inpatient database for obstetric and gynaecological hospital admissions -"Arbeitsgemeinschaft Schweizer Frauenkliniken" (ASF) database -from January 1993 to December 2014.This database contains detailed individual patient demographics, hospital admission data, and discharge and delivery data of approximately 40% of all deliveries in Switzerland, including rural and urban hospitals as well as teaching hospitals.The database is primarily used for clinical services auditing by the participating hos-pitals.The Swiss Society of Obstetrics and Gynaecology is responsible for the data, which is maintained by Sevisa AG.Data for this database are collected by the responsible physicians and then anonymously submitted to Sevisa AG.Between January 1993 and December 2004, all diagnoses and procedures were coded according to an ASF-database specific code.Beginning in January 2005, the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10), was also added for most items in the database.ICD-10 codes are provided in the text in brackets after the corresponding items if applicable.We extracted all patients from the ASF database who delivered between January 1993 and December 2014, including preterm deliveries (>24+0) and multiple pregnancies.The sample comprised all modes of delivery (vaginal delivery, instrumental delivery and caesarean section).Individual patient data sets were assessed.We identified patients with postpartum haemorrhage using the ASF data and/ or ICD-10 codes as follows: Overall postpartum haemorrhage (O72), postpartum haemorrhage resulting from uterine atony (O72.1 ± O43.2), retained, trapped or adherent placenta (O72.0),delayed secondary postpartum haemorrhage (O72.2), and postpartum haemorrhage associated with postpartum coagulation defects (O72.3).In the ASF database, "postpartum haemorrhage" is defined as an estimated postpartum blood loss over 500 ml as defined by the WHO [3].In addition, for the purposes of our study we defined "severe postpartum haemorrhage" as postpartum haemorrhage requiring blood transfusion and/or emergency hysterectomy.The dataset provided information on whether a patient required packed red blood cell (RBC) transfusion.We only used data on postpartum anaemia defined as haemoglobin over 80 g/dl collected from 2005 to 2014 because of a change in the definition of postpartum anaemia in the ASF database.Before 2005 postpartum anaemia in the ASF database was defined as haemoglobin below 100 g/dl.

Maternal factors
We derived maternal demographics, such as age, parity and history of caesarean section form the ASF data set.

Statistical analysis
Data were analysed after collecting all maternal factors, pregnancy-related factors and delivery-related factors of all extracted patients.Postpartum haemorrhage and related factors were reported per 100 deliveries and analysed over the 22-year period from 1993 until 2014.Temporal trends were assessed using the Linear-by-Linear Association (LLA; Mantel-Haenszel test for trend).The LLA test for trend equals (N-1)*r^2, with N as sample size and r being the Pearson correlation between the two variables.A p-value <0.05 for the test is consistent with the presence of a statistically significant trend.Year of delivery was used as the independent variable.Risk factors for overall and severe postpartum haemorrhage were analysed using cross-tabulations and logistic regression analysis to estimate unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs).Covariates adjusted for in the multivariable logistic regression models included maternal age (≤19, 20-34, 35-39, and ≥40 years) and other maternal factors, and the pregnancy-related and delivery-related factors derived from the ASF database as mentioned above.
The covariates are known or potential risk factors for PPH, based on published literature [6][7][8][9].All analyses were conducted using SPSS Statistics Version 22 (IBM Corp., Armonk, NY).The study was exempt from institutional board review because only de-identified, anonymised data were used.

Maternal factors
Maternal, pregnancy and delivery-related factors for 1993 and 2004 are shown in table 1. Maternal age during the study period showed a steady increase (p <0.001), especially the proportion of women older than 35 years.In 1993 only 1.3% of all mothers were 40 years or older; this number rose to 3.3% in 2014.The incidence of primiparous women of 40 years and older increased from 0.3 to 1.7% (p <0.001).The numbers of women with history of caesarean section almost doubled in the same time period, from 7.4 to 14.0% (p <0.001).Maternal age at parturition was not associated with a risk of PPH.

Delivery-related factors
Delivery-related factors also changed over time (table 1).Induction of labour was more frequently applied over the years, in 15.4% of deliveries in 1993 to 19.0% in 2014 (trend p = 0.002).Prolonged first and second stages of labour was more frequent from 2.1 to 2.5% (trend p = 0.004), and 4.5 to 8.4% (trend p <0.001), respectively.Oxytocin use for augmentation of contractions remained stable (24.6 vs 26.2%, trend p = 0.277).The proportion of deliveries with episiotomies decreased from 59.9 to 13.1% during the study period (trend p <0.001).The use of epidural anaesthesia increased greatly from 9.9 to 27.3%   (trend p <0.001).In 1993, around three quarters (77%) of women had a spontaneous vaginal delivery.In 2014 only 58.6% of women delivered spontaneously.Caesarean deliveries showed a substantial increase from 13.5% in 1993 to 30.9% in 2014 with increasing rates of both elective and emergency caesarean sections (table 1).Operative vaginal deliveries also showed an increase from 9.5 to 10.5% over time.All changes in delivery modes over time showed statistically significant trends (all p <0.001).
The incidence of postpartum haemorrhage increased over all delivery modes during the study period (fig.3).Operative vaginal delivery was associated with the highest rate of postpartum haemorrhage, followed by spontaneous vaginal delivery, emergency caesarean delivery and elective caesarean delivery.

Discussion
The incidence of postpartum haemorrhage in Switzerland almost doubled from 1993 to 2014.This substantial increase was paralleled with increasing numbers of uterine atony.Seventy-seven percent of all postpartum haemorrhages were associated with uterine atony.Although abnormally invasive placentation and placenta praevia accounted for the highest risk of postpartum haemorrhage, their overall numbers in the population remained relatively low, suggesting that these factors cannot be made accountable for the increase in postpartum haemorrhage.The use of blood transfusions paralleled the increasing incidence of postpartum haemorrhage.Interestingly, the number of patients with postpartum anaemia remained stable over the years, which indicates a timely diagnosis of postpartum haemorrhage and adequate intervention.Furthermore, changes in maternal demographics or delivery mode could not sufficiently explain the increase in postpartum haemorrhage.
During the study period, progressively longer durations of labour were accepted by healthcare professionals, both in the first and in the second stage of labour.The wide availability of epidural anaesthesia paralleled longer delivery durations.Perinatal interventions to shorten the duration of the second stage, such as labour augmentation with oxytocin and operative vaginal delivery were associated with higher risks of postpartum haemorrhage.Our results underline that postpartum haemorrhage is a relatively common obstetric complication and the documented increase in incidence over time may be partially attributed to a more liberal approach regarding the duration of labour, meaning longer time intervals until intervention such as operative vaginal interventions or caesarean section are generally accepted.
The strengths of our study were the long timespan of 22 years, whereas most previous studies on this topic have been restricted to only a few years of observation, and the population-based design that included approximately 40% of all hospital deliveries in Switzerland [5,10].Furthermore, our data source included deliveries from rural hospitals, urban teaching hospitals, and urban non-teaching hospitals, whereas other studies were limited to tertiary care centres.
Similar to other studies, we showed an increasing incidence of risk factors for postpartum haemorrhage, such as placenta praevia and abnormally invasive placenta, as well as an increase in maternal age [5,[10][11][12][13][14][15][16][17][18][19][20][21].Changes in perinatal care issues with a more liberal approach to the duration of labour, especially the second stage, and the rising numbers of labour inductions may be associated with the rising numbers of postpartum haemorrhage.Similarly, studies from France, Israel, Australia and the USA showed the significance of these factors for the development of postpartum haemorrhage [6][7][8][22][23][24][25][26].A Norwegian study "blamed" the increasing numbers of caesare- an deliveries for the rise in postpartum haemorrhage [27].Combs et al. demonstrated that risk factors for haemorrhage during caesarean delivery were a protracted active phase of labour and second-stage arrest [24].In contrast, in our data caesarean sections (elective and emergency) showed odds ratios smaller than 1, suggesting some protective potential for postpartum haemorrhage compared to spontaneous vaginal delivery.This result is most likely skewed and might reflect at least to some degree low quality in the data acquisition in the first place.Visual estimation of blood loss in vaginal deliveries is difficult and not objective.For caesarean deliveries, postpartum haemorrhage is sometimes defined as blood loss of greater than 1000 ml, making a direct comparison between delivery modes difficult [28].
A major limitation of our study was the underlying blood loss estimation in the participating hospitals, which then translated into some uncertainty in the ASF database.
There might be under-or over-diagnosing of postpartum haemorrhage because of clinical diagnostic subjectivity.Nevertheless, the results of our study are consistent with previous reports and clinical expectations, suggesting that our findings are likely to be valid [5,10].Lutomski et al. highlighted the need for further research focusing on aetiological factors of postpartum haemorrhage and preventative measures [12].Contributing factors for the increased number of prolonged second stage of labour could be the more frequent use of epidural analgesia and the increased proportion of primiparous women.Nevertheless, prolonged second stage was an independent risk factor for postpartum haemorrhage, even after correction for both of these factors.Another limitation was the change of the questionnaire in 2004/2005, which is a potential source of bias.Since the definition of postpartum haemorrhage did not change, we believe that our results are not substantially affected by that change.
In conclusion, postpartum haemorrhage is a relatively common and potentially dangerous obstetrical complication with increasing incidence over the last two decades in Switzerland.Its increase over time may be partially attributed to a more liberal approach regarding longer durations of labour and subsequent increase in uterine atony.

Figure 1 :
Figure 1: Incidence of postpartum haemorrhage over the 22-year period.Trends in the rate of postpartum haemorrhage (PPH) and haemorrhage by underlying aetiology from 1993 to 2014.The increase in all PPH was paralleled by a rise in uterine atony.Y-axis indicates the incidence (in %) and x-axis shows the corresponding year.

Figure 2 :
Figure 2: Incidence of significant pregnancy-related risk factors for PPH over a 22-year period.Graphs show temporal trends of significant pregnancy-related risk factors for postpartum haemorrhage (PPH) based on logistic regression analysis.There was an increase in incidence of abnormally invasive placenta (AIP) (above, blue line) and placenta praevia (above, green line), the two risk factors with the highest odds-ratios.Y-axis indicates the incidence (in %) and x-axis shows the corresponding year.

Figure 3 :
Figure 3: Postpartum haemorrhage (PPH) and delivery mode.Temporal trends in postpartum haemorrhage (PPH) rates in relation to delivery mode from 1993 to 2014.Instrumental (operative) vaginal deliveries showed the highest association with PPH throughout the entire study period.CD = caesarean delivery.Y-axis indicates the incidence (in %) and x-axis shows the corresponding year.

Table 1 :
Risk factors associated with postpartum haemorrhage.