Interpretation
Suspicion of a placental insertion abnormality is identified by national15 and international6,7guidelines as an indication for implementing ICS. Our results support the idea that an optimal transfusion strategy based on ICS should be set up during cesarean section in women with placenta accreta.
In our study, placenta previa was not associated with autotransfusion. It is likely that ICS is not effective in this indication because a part of the bleeding is externalized through the vagina. Moreover, when trans-placental incision is performed, a significant proportion of bleeding is removed with amniotic fluid. In these cases, using a single-aspiration strategy may optimize the collection of the bleeding and the efficiency of ICS. This strategy is supported by the work of Sullivan et al.16 and is being developed in English centers17. As proposed in the guidelines of the Royal College of Obstetricians and Gynaecologists (RCOG)18, in placenta previa situations, it may be appropriate to reserve ICS for patients with prepartum anemia, in particular in women who would decline blood products (grade D).
The presence of uterine fibroid(s) was also identified as a risk factor for autotransfusion with ICS. These patients are considered to be at high risk of PPH. The size and location of the fibroid(s) appear to correlate with the occurrence of PPH19. From the pathophysiological viewpoint, the existence of a fibroid in the cavity may lead to incomplete uterine retraction and a reduced response to oxytocics, which may lead to uterine atonia20. However, no studies have evaluated the use of ICS for this indication.
We found a strong association between postoperative bleeding and autotransfusion (OR 10.15, 95% CI 2.6–38.53). The good performance of autotransfusion in this situation can be explained by the large volume of blood contained in a hemoperitoneum, which can be easily collected during revision surgery. In our cohort, 8 patients who did not receive autotransfusion during their cesarean section received a homologous transfusion for postoperative bleeding. ICS had not been set up in these patients even though 5 had undergone revision surgery. Our results suggest that the use of ICS would have allowed autotransfusion in 77% of these patients, thus reducing the need for homologous transfusion in 3–4 women.
The existence of a uterine scar did not appear to influence the need for autotransfusion. However, in patients with multiple scars, it is essential not to ignore the possibility of an associated placenta accreta, which should be suspected when a uterine scar is associated with placenta previa. In 2006, a large prospective cohort found placenta accreta spectrum rates of 40%, 60%, and 67% in patients with placenta previa and 3, 4, and ≥5 anterior cesarean section scars, respectively21.
Despite uterine overdistension associated with a multiple pregnancy, this indication was not associated with autotransfusion in our cohort. We did not study the factors influencing uterine distension, such as fetal weight and quantity of amniotic fluid.
The group of patients with transfusion difficulties (e.g., rare blood groups, Jehovah’s Witness, and sickle cell disease) had a lower risk of autotransfusion in our series. This result was expected given that ICS is implemented in women with variable bleeding risk. In these patients, access to blood products is limited by the scarcity of compatible homologous RBCs or transfusion may be refused because of personal beliefs. ICS makes it possible for these women to access transfusion in the event of haemorrhage. Despite the lack of association, it seems justified to continue using ICS in these women.
In our study, 47% of the women who received autologous transfusion avoided homologous transfusion. This result is consistent with the literature13,17. The rate of autotransfusion in our cohort (33.1%) is lower than that in other studies of populations at risk of bleeding. This may be explained by the heterogeneity of our indications, which argue in favor of our main objective. To help decide more easily when implementing ICS, we studied the association between the bleeding volume and autotransfusion. We found a predictive autotransfusion threshold at 936 mL of bleeding, with high PPV and NPV (77.4% and 95.5%, respectively). This finding is consistent with the 1000 mL threshold suggested in the review by Grainger and al.22. However, this volume is less than the 1500 mL threshold proposed by the French guidelines for the management of placenta accreta (grade C)15.

Clinical implications

Our study confirms, through a large cohort, the effectiveness of ICS in patients with placenta accreta spectrum. We also provide a new point of view about the transfusion management of women with uterine fibroma. In these patients, it seems justified to plan cesarean section in an obstetric department equipped with an ICS device.
Postoperative bleeding cannot be anticipated and is likely to occur in any maternity ward. ICS devices are used mainly in the USA and the UK, where 49% of obstetric departments were equipped with an ICS device in 201123. In France, several factors are slowing the uptake of ICS devices, starting with the economic cost. Regular use of ICS is essential for assessing the abilities of anesthetic and obstetric teams. However, it is not suitable within the practice of maternity wards with a smaller volume of deliveries.
In Lille University Hospital’s obstetric department, the cost-effectiveness of the ICS system is ensured because of a large number of cesarean sections in women at high risk of bleeding. The CATS device is provided by Fresenius Health Care Group, which ensures its maintenance. Each collection kit costs 34 \euro, and the treatment kit is invoiced at 63 \euro. The average cost for a homologous RBC unit in our hospital is 193 \euro. The equivalent of 170 autologous RBC units was transfused during our study. Although we did not design this study to be a medicoeconomic analysis, our work seems to support the idea that ICS is cost-effective for cesarean sections in women at high risk of bleeding, as reported by international studies24,25. Team training was fast and our hospital ensures access to ICS 24 h a day. The estimated assembly time is 3 min.
The implementation of ICS in our department has revolutionized the management of women suspected of having placenta accreta spectrum. Before the implementation of ICS, our transfusion strategy for these patients involved ordering a large number of RBC units from the blood bank. This led to the loss of vital resources.

Strengths and limitations

This work provides new data in a field with few published studies. As a clinical study, our findings may provide reassurance about the safety of ICS in terms of serious adverse effects of autotransfusion, in particular amniotic fluid embolism. Our study included 5 women with sickle cell disease (4 homozygous SS and 1 compound heterozygote SC) who were reinfused safely after ICS. Hemoglobinopathy is listed in some reports as a contraindication to ICS26,27 but the data on these patients remain limited. Given the transfusion difficulties encountered by these patients, most of whom were severely anemic, we considered it reasonable to offer them ICS. The volume of reinfusion was often low because of hemolysis of sickle RBCs, but the transfusion efficiency remained good. Despite the small number of patients in our study, the absence of autotransfusion side effects is reassuring and should encourage further study on ICS in this population.
The small number of patients in some of the groups made some of the analyses impossible. For example, we could not analyze the data for women with a retroplacental hematoma or hemostatic disorder. In terms of adverse effects of autotransfusion, we do not conclude on concerns about alloimmunization because we were enable to collect the necessary data.