A systematic review and meta‐analysis of randomized controlled trials on the effects of internal iliac artery balloon occlusion in placenta previa and placenta accreta spectrum in reducing perioperative bleeding

Abstract

Background

Preoperative prophylactic balloon occlusion (PPBO) of internal iliac arteries (IIAs) in placenta previa and placenta accreta spectrum (PAS) has been widely used as a preventive adjuvant technique to reduce blood loss and to preserve fertility. Nevertheless, the efficacy of PPBO is based primarily on retrospective studies.

Objective

To examine the effectiveness of PPBO of the IIAs in placenta previa and PAS in reducing perioperative bleeding.

Search strategy

MEDLINE, EMBASE, Web‐of‐Science, Scopus, the Cochrane Library, and ClinicalTrials.gov were searched from their inception to July 2023.

Selection criteria

Randomized controlled trial (RCT) in which the use of PPBO on the IIAs was compared to non‐use.

Data collection and analysis

The primary outcome was number of packed red blood cell (RBC) units transfused. We executed meta‐analysis using a random‐effects model. Quality of the studies was assessed using the Cochrane Risk‐of‐Bias tool. The GRADE criteria were used to assess evidence certainty.

Main results

Of 164 reports identified, three RCTs representing 167 women (83 and 84 in the intervention and control groups, respectively) were eligible. Mean number of RBC units transfused was 4.52 ± 1.70 and 3.70 ± 1.88 in the intervention and control groups, respectively (weighted mean difference, 1.04; 95% CI: 0.52–1.55; P = 0.001; I ², 0%; low certainty evidence). Transfusion of other blood products was comparable. Incidence of postoperative fever was higher (P =  0.024) and hospitalization duration was longer (P = 0.001) in the intervention group compared to the control group.

Conclusions

Use of PPBO of the IIAs in placenta previa and PAS was associated with increase in number of RBC units transfused.

1. INTRODUCTION

Placenta accreta spectrum (PAS) is a pathologic adherence of the placenta to the myometrium of the uterine wall and is associated with increased maternal morbidity and mortality that are primarily due to severe peripartum bleeding. ¹ , ² , ³ , ⁴ The incidence of PAS has risen over the previous two decades, largely due to the increasing rates of cesarean delivery and placenta previa; and a prevalence between 0.3% and 2% for PAS has recently been reported. ⁵

Placenta previa—even in the absence of PAS—is also considered a major cause of substantial peripartum hemorrhage and maternal mortality worldwide, with the prevalence of placenta previa at nearly 0.6%. ³ In fact, the risk of lifesaving hysterectomy due to major bleeding after a cesarean delivery for placenta previa remains 30 times higher when compared to deliveries without previa. ⁶ , ⁷

Preoperative prophylactic balloon occlusion (PPBO) of the internal iliac arteries (IIAs) has been used widely as a preventive adjuvant technique prior to surgery to reduce operative maternal morbidity and mortality and also to preserve fertility. The surgical concept is one of lessening the uterine perfusion pressure, which leads to attenuated hemorrhaging; this results in enhancing the likelihood of preserving fertility thus reducing surgical complication rates. In addition, once hysterectomy is required, this intervention may allow for improved hemostatic control. ¹ , ² , ³ , ⁴

The literature upon which this concept is based comes primarily from retrospective and case–control studies that exhibit mixed results, ranging from uncertain to favorable. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ A large meta‐analysis that included retrospective and prospective studies revealed that PPBO of the IIAs constitutes an effective means to minimize blood volume loss and the need for blood transfusion and to reduce hysterectomy rate in women with PAS. ¹⁸ There is a paucity of randomized controlled trials (RCTs) in which investigators examined maternal outcomes in cases of placenta previa and PAS with and without PPBO of the IIAs, and these only involved a small number of participants, came from a single medical center, or comprised results that did not match the results of this recent meta‐analysis or other several retrospective studies. ⁷ , ¹⁹ , ²⁰

Considering the importance of reducing bleeding from placenta previa and PAS in order to lower maternal morbidity and mortality, the inconsistent results derived from the available extant data, the small number of participants included in each RCT, and the adverse events related to balloon catheter use, we posited that a meta‐analysis of all RCTs might lead to more established evidence. For these reasons, we wished to determine the efficacy of PPBO of the IIAs in reducing perioperative bleeding among women with placenta previa and PAS.

2. MATERIALS AND METHODS

This systematic review and meta‐analysis were conducted based on principles of the Preferred Reporting Items for Systemic Reviews and Meta‐analysis (PRISMA) statement. ²¹ A prospectively prepared protocol was also registered with Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PROSPERO), registration no. CRD42022311879.

2.1. Search strategy

An extensive electronic database search was performed, utilizing MEDLINE, EMBASE, Web of Science, Scopus, the Cochrane Library, and ClinicalTrials.gov. We identified published articles from inception to March 2022 in which the efficacy of PPBO of the IIAs for reducing operative bleeding in women with PAS and placenta previa was examined. Later we extended the range until July 2023 and registered this update on PROSPERO. We did not apply any language restriction.

Search terms were used as free text and MeSH expressions as follows: placenta previa OR placenta accreta OR placenta accrete OR placenta increta OR placenta percreta OR placental abnormalities OR placenta accreta spectrum OR PAS; prophylactic balloon OR prophylactic catheter OR prophylactic occlusion OR endovascular intervention OR interventional radiology; iliac artery OR uterine artery; and randomized controlled trial OR randomized controlled study OR RCT OR controlled clinical trial. These terms were also combined with the Boolean operators AND/OR to complete the search. Three authors (MM, RI, and RS) independently conducted the search twice. All relevant articles were retrieved, and references of eligible articles were reviewed to procure additional studies.

2.2. Selection, eligibility, and quality of the studies

All randomized peer‐reviewed trials in which PPBO of the IIAs was compared with no‐PPBO use were included in this systematic review and meta‐analysis. RCTs only were chosen because available data is based mainly on large retrospective studies and their weight and contribution within published meta‐analysis is greater and the meta‐estimates weighted constantly towards the large non‐RCTs estimates. We posited that a meta‐analysis of RCTs alone (level 1 evidence) might lead to more robust evidence.

Abstract‐only, editorial, and review articles were excluded. Additionally, retrospective case/control or cohort studies were also excluded. Data selection was performed by two authors (MM and RI). Any disagreement regarding the inclusion or exclusion of studies was resolved through discussion with a third author (RS) that considered an expert physician in the field.

2.3. Data extraction

Data from full‐text studies were extracted independently by two authors (MM and RI) to limit error and to verify the accuracy of data collection. We then developed tables to include the name of the first author and the year of publication, the country of origin, methods, the intervention group (PPBO insertion) versus the control group, sample size, mean age of the women in each group, mean gestational age of each group, and the primary outcome measured in each trial. Furthermore, we recorded in each group the basic demographic and obstetric variables for all participants.

2.4. Outcome measures

Our primary outcome was the number of packed red blood cell (RBC) units transfused in the PPBO (intervention) group compared to the control group. Secondary outcomes comprised fresh frozen plasma (FFP), platelets, and cryoprecipitate transfusion, surgical and postoperative complications such as estimation of blood loss, operating time, cesarean hysterectomy rate, bladder and bowel injury, re‐laparotomy, fever, scar infection, pneumonia, urinary tract infection, postoperative maternal length of hospital stay, and readmission. Neonatal outcomes that were recorded included Apgar scores and admission to neonatal intensive care units.

2.5. Risk of bias

Bias risk was assessed by two authors (RI and MM) independently utilizing the updated Cochrane Risk‐of‐Bias tool for RCTs. ²² Disagreements were resolved by discussion with a third reviewer (RS) until consensus was achieved. A number of aspects of bias were evaluated in all of the trials, including bias arising from the randomization process, deviation from the intended interventions, missing outcome data, measurements of the outcome, selection of the reported results and other sources of bias.

2.6. Quality of the body of evidence

Overall quality of evidence for the significant outcomes was evaluated by using the GRADE criteria which are based on an assessment of limitations in study design or execution, inconsistency of results, indirectness of evidence, imprecision and publication bias. ²³

2.7. Data analysis

We compared the use of PPBO of the IIAs with no‐PPBO use. The quantitative results for all continuous outcomes are presented using means and standard deviations (SDs), and the pooled effect was calculated using the weighted mean difference (WMD). Outcomes presented as medians and interquartile ranges (ranges between the 25th [Q1] and 75th percentiles [Q3]) were converted to approximated means and SDs by applying the following formula. ²⁴

$$\begin{array}{c}\text{Approximated mean}=\left(\text{Q1}+\text{median}+Q3\right)/4,\hfill \end{array}$$

$$\text{Approximated}\mathrm{SD}=\left(\mathrm{Q}3–\mathrm{Q}1\right)/1.35.$$

Categorical outcomes were measured using relative risk (RRs). The collective difference in the mean RR was calculated by combining the results of all the included studies while considering the appropriate weight of each study. We conducted statistical analysis and graphical displays using the open‐source software OpenMeta [Analyst].

Statistical heterogeneity was assessed with the Cochrane's Q test of heterogeneity (P < 0.1 was considered to be statistically significant), and heterogeneity was evaluated with the I ² statistic (a value >50% was judged as showing significant heterogeneity between studies). We executed meta‐analysis using a random‐effects model based on the DerSimonian and Laird method.

3. RESULTS

3.1. Study selection

The PRISMA flow chart summarizing study selection is presented in Figure 1. Of the 164 identified reports, three RCTs ⁷ , ¹⁹ , ²⁰ were considered eligible and were included in the meta‐analysis (File S1); the three eligible studies included a total of 167 women (83 in the intervention group and 84 in the control group). In the study by Yu et al., ⁷ one woman from the intervention group was excluded from the analysis since placenta previa was excluded at 36 weeks. In the three included trials, ⁷ , ¹⁹ , ²⁰ the PPBO placement was performed before surgery, on the day of the planned delivery, in the radiology suite by an experienced interventional radiologist. The PPBOs were placed in the anterior division of the IIAs in two trials. ¹⁹ , ²⁰ In the third, the PPBOs were positioned in the proximal portion of the IIAs. ⁷ Afterward, women were immediately transferred to the operating room for cesarean delivery.

FIGURE 1.

PRISMA flow chart outlining study selection.

All studies included were registered in the clinical trials registry; one was registered at the National Institute of Health, ¹⁹ and the other two were registered at the Chinese Clinical Trial Registry. ⁷ , ²⁰

3.2. Risk of bias

One trial was considered to be at low risk of bias. ⁷ Two trials, ¹⁹ , ²⁰ scored “some concerns” for the domain “deviations from intended intervention” since it was not reported whether an intention to treat analysis was employed, and “some concerns” for the domain “other sources of bias” since not all cases with PAS were diagnosed pathologically and there are some concerns that in the cases managed without a pathologic confirmation not all had PAS (Figure 2).

FIGURE 2.

Risk of bias summary and graph.

3.3. Study characteristics

In the study by Salim et al., 25 (93%) of their 27 participants manifested placenta previa, 13 (48%) were confirmed to have PAS by pathology, and 24/27 (89%) were diagnosed with PAS at surgery. ¹⁹ In the study by Chen et al., 100 patients (100%) showed placenta previa, 70 (70%) had a pathologic confirmation of PAS, and 98 (98%) had a surgical diagnosis of PAS. ²⁰ In the study by Yu et al., 15 (75%) of the intervention group and 20 (70%) of the control group had a placenta previa while five (25%) and six (30%) in the intervention and control groups respectively had a lower placental edge located at a distance of zero to 5 cm from the internal os. There were no cases of PAS. ⁷ The basic descriptive measures of the three trials are presented in Table 1. All basic demographics and obstetric variables of the participants were similar, except for maternal age (Table 2): the mean maternal age in the intervention group was 33.37 ± 1.33 years compared to 34.65 ± 2.04 years in the study group (P = 0.003).

TABLE 1.

Baseline descriptive data of the three enrolled trials.

Study	Country	Study type	Intervention; controls	Sample size Intervention/control	Placenta previa Intervention/controls	PAS confirmed pathologically Intervention/controls	PAS confirmed surgically only Intervention/controls	Maternal age Intervention; controls a	Gestational age at delivery, weeks Intervention/controls a	Primary outcome	Funding b
Salim et al. (2015) 19	Israel	Randomized controlled trial	Balloon occlusion/control	13/14	13/12	6/7	6/5	34.4 ± 4.3; 37.2 ± 5.6	35.1 ± 1.3; 34.8 ± 1.5	Number of packed red blood cell units transfused	No
Chen et al. (2020) 20	China	Randomized controlled trial	Balloon occlusion/control	50/50	50/50	38/32	10/18	32.5 ± 4.6; 33.3 ± 4	36.4 ± 1; 36.1 ± 1.1	Number of packed red blood cell units transfused	Yes
Yu et al. (2020) 7	China	Randomized controlled trial	Balloon occlusion/control	20/20	19/20 c	−/−	−/−	34.9 ± 1.5; 36.3 ± 1.6	36.4 ± 0.6; 36.2 ± 0.7	Reduction of postpartum hemorrhage	Yes

Abbreviations: PAS, placenta accreta spectrum; SD, standard deviation.

^a Maternal and gestational ages are presented as means ± SD.

^b Chen et al. trial, ²⁰ was funded by a grant from the National Key R&D Program of China (no. 16YFC1000406). Yu et al. trial, ⁷ was funded by the Health and Medical Research Fund, Food and Health Bureau of the Hong Kong SAR government for project funding (reference no: 03140236).

^c One woman from the intervention group was excluded.

TABLE 2.

Basic demographic and obstetric variables of the women included in the meta‐analysis.

Variables	No. of studies analyzed	Intervention group N = 82	Control group N = 84	P value
Maternal age, years	3	33.37 ± 1.33	34.65 ± 2.04	0.003
Gravidity	2	4.72 ± 0.09	4.89 ± 0.38	0.788
Parity	2	1.76 ± 1.40	1.68 ± 1.14	0.860
BMI, kg/m2	3	26.75 ± 0.37	26.80 ± 0.30	0.820
Gestational age at delivery, weeks	3	36.19 ± 0.58	35.90 ± 0.60	0.066
Cesarean delivery
Previous cesarean delivery	3	67/82 (81.71)	61/84 (72.61)	0.082
One previous cesarean delivery	2	46/63 (73.02)	49/64 (76.56)	0.690
Two previous cesarean deliveries	2	13/63 (20.63)	7/64 (10.94)	0.099
Diabetes mellitus in pregnancy	3	18/82 (21.95)	24/84 (28.57)	0.40
Hypertension	3	4/82 (4.88)	4/84 (4.76)	0.849
Other maternal disease	2	26/63 (41.27)	25/64 (39.06)	0.741
Placenta previa	2	25/32 (78.13)	23/34 (67.65)	0.343
Low‐lying placenta	2	8/32 (25)	9/34 (26.47)	0.787
Anterior placental site	2	53/69 (76.81)	48/70 (68.57)	0.204

Note: BMI, calculated as weight in kilograms divided by the square of height in meters. Data are means ± SD or N‐intervention/N‐control (%).

Abbreviations: BMI, body mass index; SD, standard deviation.

3.4. Synthesis of results

The primary outcome, that is, the mean number of transfused RBC units, was 4.52 ± 1.70 units in the intervention group and 3.70 ± 1.88 units in the control group (Table 3). This increase in the mean number of transfused RBC units among the intervention patients compared to the controls was significant (Figure 3), with an estimated WMD of 1.04 units (95% confidence interval [CI]: 0.52–1.55; P = 0.001; I ², 0%; low certainty evidence as assessed by the GRADE criteria because it was downgraded by two levels due to imprecision and indirectness of evidence). We did not observe any significant differences in other blood products in terms of units transfused, including FFP, platelets, or cryoprecipitate. Additionally, there was no significant difference in the mean estimated blood loss during the operation, with an estimated WMD of 81.67 mL (95% CI: −115.26–278.61; P = 0.416). The mean operating time (min) was significantly longer in the intervention group compared to the control (WMD 8.65; 95% CI: 4.51–12.80; P = 0.001; I ², 0%; low certainty evidence as assessed by the GRADE criteria because it was downgraded by two levels due to imprecision and indirectness of evidence), and the incidence of postoperative fever was statistically higher in the intervention group relative to controls (RR 6.63; 95% CI: 1.21–36.27; P = 0.024; I ², 0%; low certainty evidence as assessed by the GRADE criteria because it was downgraded by two levels due to imprecision and indirectness of evidence). We noted no differences between the groups in other perioperative complications that included the rates of cesarean hysterectomy, bowel or bladder injury, re‐laparotomy, or postoperative infectious morbidity (Table 3). There were no cases of sepsis, or maternal deaths in either group. The mean length of maternal hospital stay was, however, different between groups: 9.35 ± 3.31 days in the intervention group and 8.80 ± 3.44 days in the control group (WMD 0.97; 95% CI: 0.62–1.32; P = 0.001; I ², 0%; low certainty evidence as assessed by the GRADE criteria because it was downgraded by 2 levels due to imprecision and indirectness of evidence). The rates of readmission were similar (Table 3) between groups, and neonatal outcomes were comparable (Table 4).

TABLE 3.

Primary and secondary outcomes.

Outcome	No. of studies	Salim et al. (2015) 19	Chen et al. (2020) 20	Yu et al. (2020) 7	Totals a Intervention group‐82 Control group‐84	RR or WMD b (95% CI)	P value	I 2 c
Packed red blood cell units transfused	3	5.2 ± 6.2 (0–23) vs. 4.1 ± 3.8 (0–11)	5.3 ± 5.3 (0–18) vs. 4.7 ± 5.4 (0–21)	2.0 ± 1.0 (0–4) vs. 0.9 ± 0.7 (0–2.8)	4.52 ± 1.70 vs. 3.70 ± 1.88	1.04 (0.52, 1.55)	0.001	0
FFP units transfused	2	2.8 ± 5.4 (0–19) vs. 1.7 ± 2.7 (0–8)	‐	0 (0–0) vs. 0 (0–0)	1.14 ± 1.05 vs. 0.70 ± 0.65	0 (0, 0)	>0.99	0
Platelets transfused	3	0.5 ± 1.7 (0–6) vs. 0 (0–0)	0 (0–0) vs. 0 (0–0)	1.0 ± 1.0 (0–4) vs. 0.8 ± 0.8 (0–3)	0.80 ± 0.19 vs. 0.44 ± 0.29	0.32 (−0.16, 0.79)	0.193	0
Cryoprecipitate transfused	2		7/50 (14) vs. 6/50 (12)	2/19 (10.52) vs. 4/20 (20)	9/69 (13.04) vs. 10/70 (14.29)	0.92 (0.39, 2.17)	0.860	0
Estimated blood loss, mL	3	1600 ± 994 (800–4500) vs. 1614 ± 727 (600–3500)	2630 ± 1585 (527–6133) vs. 2221 ± 1694 (691–7006)	1578.5 ± 341 (1024–2388) vs. 1524 ± 353 (888–2300)	2223.07 ± 489.64 vs. 1953.88 ± 310.09	81.67 (−115.26, 278.61)	0.416	0
Operating time, min	3	84.8 ± 41.9 (35–168) vs. 84.4 ± 37.9 (32–144)	123.6 ± 51.6 (42–235) vs. 113.3 ± 56.7 (36–286)	47.5 ± 8 (30–62) vs. 38.8 ± 5.3 (30–51)	99.82 ± 31.86 vs. 90.73 ± 31.37	8.65 (4.51, 12.80)	0.001	0
Cesarean hysterectomy	3	6/13 (46.2) vs. 7/14 (50.0)	22/50 (44) vs. 16/50 (32)	0/19 (0.0) vs. 0/20 (0.0)	28/82 (34.15) vs. 23/84 (27.38)	1.22 (0.80, 1.87)	0.360	0
Fever	2	1/13 (7.7) vs. 0/14 (0.0)	9/50 (18) vs. 1/50 (2)	‐	10/63 (12.20) vs. 1/64 (1.19)	6.63 (1.21, 36.27)	0.024	0
Scar infection	2	0/13 (0.0) vs. 0/14 (0.0)	0/50 (0.0) vs. 0/50 (0.0)	‐	0/63 (0.00) vs. 0/64 (0.00)	‐	‐	0
Urinary tract infection	2	2/13 (15.4) vs. 2/14 (14.3)	1/50 (2) vs. 0/50 (0.0)	‐	3/63 (3.66) vs. 2/64 (2.38)	1.39 (0.29, 6.66)	0.659	0
Pneumonia	2	0/13 (0.0) vs. 0/14 (0.0)	2/50 (4) vs. 0/50 (0.0)	‐	2/63 (2.44) vs. 0/64 (0.00)	2.79 (0.26, 29.89)	0.390	0
Maternal length of hospital stay, days	3	6.6 ± 3.0 (4–15) vs. 7.1 ± 4.4 (5–22)	12.0 ± 7.6 (4–53) vs. 11.5 ± 9.2 (5–63)	4.3 ± 0.8 (3–6) vs. 3.3 ± 0.3 (3–4)	9.35 ± 3.31 vs. 8.80 ± 3.44	0.97 (0.62, 1.32)	0.001	0
Re‐admission	2	2/13 (15.4) vs. 0/14 (0.0)	0/50 (0.0) vs. 0/50 (0.0)	‐	2/63 (2.44) vs. 0/64 (0.00)	2.91 (0.28, 30.57)	0.367	0
Placenta left in situ	2	6/13 (46.2) vs. 5/14 (35.7)	0/50 (0.0) vs. 0/50 (0.0)	‐	6/63 (7.32) vs. 5/64 (5.95)	1.28 (0.52, 3.11)	0.608	0
Re‐laparotomy	3	0/13 (0.0) vs. 1/14 (7.1)	0/50 (0.0) vs. 0/50 (0.0)	0/19 (0.0) vs. 0/20 (0.0)	0/82 (0.00) vs. 1/84 (1.19)	0.65 (0.08, 5.08)	0.680	0
Bladder injury	2	2/13 (15.4) vs. 2/14 (14.3)	2/50 (4) vs. 2/50 (4)	‐	4/63 (4.88) vs. 4/64 (4.76)	1.04 (0.28, 3.88)	0.956	0
Bowel entry	2	0/13 (0.0) vs. 1/14 (7.1)	0/50 (0.0) vs. 0/50 (0.0)	‐	0/63 (0.00) vs. 1/64 (1.19)	0.53 (0.05, 6.09)	0.615	0

Abbreviations: CI, confidence interval; FFP, fresh frozen plasma; RR, relative risks; SD, standard deviation; WMD, weighted mean difference.

^a Data are means ± SD (range; salim ¹⁹ and Chen et al. ²⁰ or interquartile range; Yu et al. ⁷ ), or N‐intervention/N‐control (%).

^b RR with 95% of CI was used as the pooled effect for categorical outcomes, and WMD with 95% of CI was used as the pooled effect for continuous outcomes.

^c I ² was used to identify heterogeneity (a value >50% was judged as showing significant heterogeneity between studies).

FIGURE 3.

Forest plot of packed red blood cell (RBC) units transfused in the intervention and control groups.

TABLE 4.

Neonatal outcomes.

Outcomes	No. of studies analyzed	Intervention group N = 82	Control group N = 84	P value	I 2 a
Birth weight, g	3	2706 ± 138	2746 ± 127	0.324	0
Male neonate	3	43/82 (52.44)	46/84 (54.76)	0.961	0
Neonatal intensive care unit admission	2	11/69 (15.94)	12/70 (17.14)	0.825	0
Apgar score < 7 at 5 min	3	2/82 (2.44)	2/84 (2.38)	0.940	0

Note: Data are means ± SD or N‐intervention/N‐control (%).

Abbreviation: SD, standard deviation.

^a I ² was used to identify heterogeneity (a value >50% was judged as showing significant heterogeneity between studies).

Complications related to PPBO use were reported in two (15.4%) women in one trial. ¹⁹ One woman had right leg pain and weakness without swelling or ischemia. The second woman was readmitted due to buttock claudication. Both complaints completely resolved spontaneously. In the other two trials, ⁷ , ²⁰ no complications related to the use of PPBO were reported. The mean hospitalization costs were 7456 ± 1799 ($) compared to 4803 ± 1620 ($) in the intervention and control groups, respectively (P < 0.01). This result was reported by only one trial. ²⁰

4. DISCUSSION

4.1. Main findings

The current meta‐analysis of RCTs showed that PPBO of the IIAs in patients with PAS and placenta previa was associated with an increase in the number of packed RBC units transfused, operative time, rate of postoperative fever, and prolonged length of maternal hospital stay. The findings were judged at a low certainty of evidence. Other maternal perioperative and immediate neonatal outcomes were not affected.

4.2. Comparison with existing literature

PAS is a life‐threatening pregnancy complication as it can cause catastrophic bleeding, and it is strongly associated with previous cesarean delivery and placenta previa in the index pregnancy. ¹ Placenta previa is a pregnancy complication that may also lead to severe intraoperative bleeding and to increased hysterectomy rates and admission to maternal intensive care units. ⁷ , ²⁵ As in PAS, previous cesarean delivery augments the occurrence of placenta previa, and for this reason, the incidence of both has risen considerably in recent decades. ²⁵ , ²⁶

Surgical management of these life‐threatening complications remains challenging. Endovascular radiology performed prior to surgery to lessen intraoperative bleeding and in anticipation of a potential hysterectomy is used broadly. The rationale behind the use of PPBO of the IIAs is to reduce perfusion pressure and thereby minimize blood loss, enabling a clearer intraoperative surgical field. Several retrospective studies have also established the advantages of PPBO of the IIAs in reducing blood loss and preventing hysterectomy in cases of PAS. ⁹ , ²² , ²⁷ , ²⁸ , ²⁹ A Bayesian network meta‐analysis that included 59 retrospective and prospective studies with a total sample size of 5150 patients showed that performing PPBO of the IIAs was associated with a significant reduction in blood loss and the need for blood transfusion and that the procedure lowered the hysterectomy rate. ¹⁸ Another recent meta‐analysis also revealed a diminution in blood loss and a reduction in hysterectomy rate when PPBO was used in patients with PAS and placenta previa. However, the studies included were chiefly retrospective, had non‐uniform reporting, lacked consistency as to data analysis, and some were limited by incomplete data according to the authors. ³⁰ In contradistinction, data from individual RCTs showed that the use of PPBO in cases of PAS or placenta previa did not affect intraoperative blood loss or the rate of hysterectomy. ⁷ , ¹⁹ , ²⁰ However, compared with retrospective studies and meta‐analyses, extant RCTs reflected a relatively small number of participants, and each trial was conducted at only a single center, hence the impetus to conduct this current meta‐analysis. When combining these small RCTs, the results contradict the previous two meta‐analyses, ¹⁸ , ³⁰ and this was reflected in the number of packed RBC units transfused, operative time, and rate of postoperative fever all of which increased in the intervention group compared to the control group. It should be noted that the increase of almost 9 min in the operative time is perhaps clinically insignificant. In addition, maternal length of hospital stay was also prolonged. Furthermore, the procedure was not associated with a lower hysterectomy rate.

The rationale for the inadequate performance of PPBO of the IIAs as observed in the present meta‐analysis of RCTs when compared to data based on level‐II evidence may involve a number of factors. While uterine blood flow is extensive and wide vascular anastomoses exist in the gravid pelvis with PAS, perfusion pressure to the uterine arteries typically falls following balloon inflation, and collateral circulation from other pelvic origins such as the external iliac artery or the ovarian artery may actually exacerbate blood flow and intensify the overall blood loss. ³¹ , ³² Although hysterectomy rates were absent or low among several retrospective studies, ¹⁸ , ³³ a low to no‐event rate for hysterectomies casts doubts regarding the accuracy of PAS diagnosis due to the absence of pathologic confirmation. In the absence of PAS and as a consequence of the absence of extensive blood flow and collateral circulations, the performance of PPBO of the IIAs may vary. In the present meta‐analysis, 83 (67.5%) of 123 women with preoperative PAS, had pathologic confirmation. In the remaining cases where the uterus was preserved, strict surgical criteria to confirm PAS were employed.

In addition, complications, radiation exposure, costs and logistical hassles associated with the use of PPBO are issues that should also consider before using the balloon catheters.

4.3. Limitations and strengths

This meta‐analysis encompassed only three relatively small RCTs. The trials were not double‐blinded likely due to the nature of the interventions; and for the three trials included, we did not conduct a subanalysis of the results according to the area or degree of invasion. For this reason, it was inappropriate to conclude that the use of PPBO was ineffective for all stages of penetration or for all areas of uterine invasion. Additionally, several women included had placental previa only without PAS. Moreover, the lack of pathologic conformation in all cases of PAS may lead to some concerns (risk of bias) that not all had PAS. Furthermore, the cited trials did not all clearly state how surgical time was defined. From the caption in the text, it can be concluded that the time of inserting the catheters was not counted in. Finally, our results were limited to the use of PPBO of the IIAs. The effect of PPBO performance with balloons inserted into arteries other than the IIAs was thus beyond the scope of this study. It is also noteworthy that in the current meta‐analysis, we examined the effectiveness of PPBO in PAS and placenta previa as a prophylactic procedure and not as a treatment. Importantly, intravascular catheter use and arterial embolization have been reported to constitute an effective treatment in approximately 95% of cases of severe postpartum hemorrhage. ³⁴

The principal strength of the current study was the inclusion of only RCTs. Furthermore, the combined sample size of the included studies overcame the limitations of the relatively small sample size of each separate trial. Additionally, the results reported in Tables 3 and 4, indicate a very low heterogeneity across all studied parameters among the three studies, suggesting minimal impact on the overall findings. Nevertheless, these results are based on only three trials, and in the future the findings may be updated through additional published trials.

4.4. Implications

Use of PPBO of the IIAs in women with placenta previa or PAS was associated with an increase in the number of RBC units transfused and to prolongations of the operating time and maternal length of hospital stay. As the associated incidence of catheter‐related complications, routine preoperative radiologic intervention of the IIAs in women with PAS and placenta previa is not supported by the results of the present meta‐analysis.

AUTHOR CONTRIBUTIONS

RI and MM: Protocol registration, data collection, article writing and participation in bias and GRADE assessment. CR: Protocol registration and data collection. II: Data analysis. RS: Conceiving the research question and developing the study plan, data collection, data analysis, article writing and participation in bias and GRADE assessment. All authors approved the final version of the manuscript.

FUNDING INFORMATION

This study received no funding.

CONFLICT OF INTEREST STATEMENT

The authors report that they do not have any conflicts of interest to declare.

Supporting information

File S1.

Data S2.

Iskander R, Massalha M, Remer C, Izhaki I, Salim R. A systematic review and meta‐analysis of randomized controlled trials on the effects of internal iliac artery balloon occlusion in placenta previa and placenta accreta spectrum in reducing perioperative bleeding. Int J Gynecol Obstet. 2025;170:1061‐1070. doi: 10.1002/ijgo.70118

DATA AVAILABILITY STATEMENT

Data available on request from the authors after publication with a justified proposal for further research.