Endometritis risk factors after arterial embolisation for postpartum haemorrhage

Abstract Various complications of arterial embolisation (AE) for postpartum haemorrhage (PPH) are reported. Endometritis (EM) frequently causes abscesses, increasing hysterectomy risk. However, risk factors for EM after AE for PPH are unclear. We explored these risk factors. We included patients who underwent AE for PPH in our hospital from 2005 to 2020 and compared those who did (EM group) and did not develop EM after AE (non-EM group) in a case-control study. Twenty patients met the study criteria; eight patients (40%) had EM. There were no differences in risk factors between groups involved in infection, such as premature rupture of membranes. However, the contrast medium extravasation rate on computed tomography scans before the AE procedure was significantly higher in the EM group (p=.019) compared to the non-EM group. The greatest EM risk factor was contrast medium extravasation before AE for PPH, determined by classification and regression tree modelling (relative risk: 4.5). Impact Statement What is already known on this subject? Reportedly, the clinical success rate of arterial embolisation (AE) for critical haemorrhage in obstetrics is high, around 90%. However, information regarding AE complications is limited. Endometritis is one of these complications, which not only causes prolonged hospitalisation but may also require further treatment, such as hysterectomy. However, the incidence rate and risk factors for EM remain unknown. What do the results of this study add? In this study, 40.0% of patients developed EM after AE for PPH. Extravasation of contrast medium was the top risk factor (relative risk: 4.5 compared to those without EM, p=.019). The second-leading risk factor was a bleeding volume greater than 2500 mL (relative risk: 4.5 compared to those without EM, p=.019). What are the implications of these findings for future clinical practice and/or future research? We created an EM prediction model using extravasation and a bleeding volume greater than 2500 mL. The model was 87.5% sensitive and 66.7% specific. This prediction model allows for the early detection and treatment of EM by recognising high-risk patients and providing intensive postpartum management.


Introduction
Postpartum haemorrhage (PPH) occurs in approximately 2% of all deliveries, accounting for one-fourth of maternal deaths (World Health Organization 2012).Brown et al. (1979) reported arterial embolisation (AE) as a treatment strategy for uncontrollable PPH, which has become one of the general strategies for controlling PPH.
The 2015 Japanese Society of Interventional Radiology (JSIR) procedural guidelines state that the clinical success rate of interventional radiology for critical haemorrhage in obstetrics is high, around 90%, and the complication rate is between 6 and 7% (Sone et al. 2015).The references in this guideline mostly feature clinical successes (Zwart et al. 2010), such as haemostasis.However, references regarding AE complications are few, and the definitions of AE-related complications are unclear.Therefore, the incidence rate of AE complications must be reconsidered.
Endometritis (EM) can cause prolonged hospitalisation (Van Otterloo et al. 2018), and some patients with EM suffer from abscesses, develop septic shock (Breathnach et al. 2007), or require a hysterectomy to control the infection site (Milasinovi� c et al. 1996).However, the incidence rate and risk factors for EM remain unknown.Therefore, this study investigated the EM incidence rate and risk factors after AE for PPH in a case-control study.

Materials and methods
Patients who underwent AE for PPH in our hospital from 2005 to 2020 were included in the study.We compared patients who did and did not develop EM after AE for PPH.
PPH was defined as bleeding greater than 1000 mL within 24 hours of delivery, based on the American College of Obstetricians and Gynecologists Practice Bulletin No. 183: Postpartum Hemorrhage guideline (Committee on Practice Bulletins-Obstetrics 2017).To distinguish EM from postembolisation syndrome, we only diagnosed EM if the patient had at least two of the following symptoms three or more days after AE: (1) a body temperature of 38 � C or higher, (2) abdominal or lower abdominal tenderness and (3) purulent discharge from the cervix based on the National Healthcare Safety Network Patient Safety Component Manual (Centers for Disease Control and Prevention 2021).All cases were chart reviewed and confirmed.EM is related to AE of the internal iliac artery (Woodhams 2016).Therefore, cases not involving the internal iliac artery were excluded.Patients receiving immune suppressants or those with infections before delivery or intrauterine foetal death were also excluded.
We collected continuous variable data regarding age, body mass index, parity, gestational week, infant birth weight, caesarean section (CS) operative time, labour duration for vaginal deliveries (VDs), bleeding volume, blood transfusion, length of hospital stay, and time to AE from delivery.Bleeding volume was defined as the amount of bleeding occurring after delivery until AE.We also collected nominal variable data regarding the pregnancy type (singleton or not), conception method (assisted reproductive treatment or not), smoking history (yes or no), group B streptococcus (carrier or not), gestational diabetes mellitus (GDM; yes or no), hypertension disorder of pregnancy (yes or no), preterm rupture of membranes (PROM; yes or no) and mode of delivery (VD, CS or operative VD).Contrast medium extravasation was defined as an extravascular area of hyper attenuation between 80 and 150 Hounsfield units (Hallinan et al. 2014).The computed tomography examinations were prepared as one set containing the venous phase only (90 seconds) and another set containing both arterial (40 seconds) and venous phases (Godt et al. 2021).Extravasation was diagnosed by a radiologist using computed tomography before the AE procedure.Embolised blood vessels were also identified by a radiologist (e.g. the bilateral uterine artery, the bilateral internal iliac artery and the unilateral artery).
Data were analysed by the Mann-Whitney U-test and Fisher's exact test (univariate analyses) and the classification and regression tree (CART) model (multivariate analyses).All analyses were performed in R (version 3.5.0,R Foundation for Statistical Computing, Vienna, Austria) with EZR (Kanda 2013).p Values <.05 were considered statistically significant.The primary outcomes were the EM incidence rate and risk factors.The secondary outcome was an EM prediction model.This study was approved by the Institutional Ethics Committee (registration number, 21-R190) and complied with the requirements under the Declaration of Helsinki.Informed consent was obtained in the form of a website opt-out.

Results
In total, 24 patients underwent AE for PPH from 2005 to 2020, and 20 patients met the eligibility criteria.Eight patients (40%) developed EM (EM group), and twelve patients did not (non-EM group).Tables 1 and 2 detail the patient information.
The general EM risk factors (Faro 2005) did not differ between EM group and non-EM group in the univariate analyses, including the number of gestational weeks, bleeding volume, CS operative time, VD labour duration, mode of delivery, GDM and PROM.However, the extravasation (p ¼ .019,non-EM: 16.7%; EM: 75.0%) and length of hospital stay (p ¼ .013,median stay, non-EM: 11 days; EM: 25 days) significantly differed between the EM group and non-EM group.
Factors associated with causing EM were ranked using the CART model.Extravasation of contrast medium was the top risk factor (relative risk: 4.5 compared to those without EM); the second-leading risk factor was a bleeding volume greater than 2500 mL (relative risk: 4.5 compared to those without EM).Furthermore, we identified a significant difference between the EM group and non-EM group after changing the bleeding volume classification from a continuous variable to a nominal variable (p ¼ .019,i.e. bleeding volumes greater than 2500 mL; Figure 1).We did not include the length of hospital stay in the multivariate analysis because there was multicollinearity between EM and the length of hospital stay.
We created the EM prediction model using extravasation and a bleeding volume greater than 2500 mL (i.e. the nominal variable).The mode was 87.5% sensitive and 66.7% specific (Figure 2; area under the receiver operating characteristic curve: 0.771, 95% confidence interval: 0.585-0.956).

Discussion
Even though AE is the most common procedure for PPH in many institutions, uncertainty remains regarding complications.The effects of EM after AE are severe.However, few reports have addressed the incidence rate and risk factors; this study aimed to clarify these points.
In this study, 40.0% of patients developed EM after AE for PPH.However, the JSIR guideline only reports AE complications in 6-7% of cases (Sone et al. 2015).These guidelines consist of only observational studies and do not strictly define AE complications.Furthermore, some patients included in the studies did not have PPH.JSIR recommends grade C for the rate of AE complications, but the evidence suggests there is more to consider regarding EM after AE.Thus, more in-depth evaluations are necessary, such as systematic reviews and meta-analyses for each complication.
Notable EM risk factors after AE for PPH in this study are contrast medium extravasation and the bleeding volume greater than 2500 mL.Extravasation of contrast medium depicts a bleeding rate greater than 0.5 cm 3 /min, reflecting e Birth weight of infant: as for twin pregnancy, the sum of birth weight of infant was showed.f Extravasation of contrast medium was diagnosed by radiologist using computer tomography before the AE procedure.an increased total blood loss (Hamilton et al. 2008, Lee et al. 2012).The endometrium becomes strongly ischaemic from the bleeding and embolisation procedure, promoting microaerophilic and anaerobic bacterial growths.Furthermore, ischaemic tissue decreased the oxygen-dependent bactericidal capacity of leukocytes, the local immune response and antimicrobial agent absorption (Gupta et al. 2007).Thus, endometrial tissue may become easily infected, leading to EM.
Currently, there are few evidence-based procedures to prevent EM.However, recognising high-risk patients and intensive postpartum management allows for the early detection and treatment of this disease.
We performed a case-control study in a single institution.Therefore, confounding factors are likely.For example, the patients' mean age in this study (39.0 years) is higher than the mean age in other studies (Yoon et al. 2006, Zwart et al. 2010).Furthermore, we did not examine the kinds of embolisation materials, but postoperative tissue ischaemia changes depending on whether reversible or irreversible embolisation material substances are used.Finally, the number of cases was small because this was a single-centre case-control study.

Figure 1 .
Figure1.Multivariate analysis results using the classification and regression tree model.Contrast medium extravasation is the highest risk factor for EM development (relative risk: 2.5 compared to those without EM).The second leading risk factor is a bleeding volume greater than 2500 mL. a Extravasation of contrast medium in a computer tomography scan.
A: artery; atonic: atonic bleeding; B: bilateral; CS: caesarean section; extra: the presence of extravasation of contrast medium on computer tomography scan;OVD: operative vaginal delivery; P: parous; previa: previa placenta; U: unilateral; VD: vaginal delivery.a Non-Em group: the group which was not diagnosed as endometritis.b Em group: the group which was diagnosed as endometritis.

Table 2 .
Patient's background.Non-Em group: the group which was not diagnosed as endometritis.b Em group: the group which was diagnosed as endometritis.
a c Data are presented as median (range) for continuous variables.d Data are presented as number (%) for nominal variables.