Intraluminal pressure readings during the establishment of a positive ‘tamponade test’ in the management of postpartum haemorrhage


  • C Georgiou

    1. Illawarra Health and Medical Research Institute/Graduate School of Medicine, University of Wollongong, and Wollongong Hospital, Department of Obstetrics and Gynaecology, Illawarra, NSW, Australia
    Search for more papers by this author

Dr C Georgiou, The Wollongong Hospital Academic Suite, Wollongong Hospital, Block C, Level 8, Crown Street, Wollongong, New South Wales, Australia. Email


Please cite this paper as: Georgiou C. Intraluminal pressure readings during the establishment of a positive ‘tamponade test’ in the management of postpartum haemorrhage. BJOG 2010;117:295–303.

Objective  To investigate the proposed mechanism by which intrauterine balloons achieve their tamponade effect of creating an ‘intrauterine pressure that is greater than the systemic arterial pressure’.

Design  To determine the intraluminal pressures within a Bakri balloon during the establishment of a positive ‘tamponade test’ in the management of postpartum haemorrhage. To correlate these intraluminal pressures with contemporaneous readings of blood pressure recordings as documented from the operating theatre anaesthetic charts.

Setting  An obstetric unit (approximately 2400 births) in Wollongong, New South Wales, Australia.

Sample  Two women in whom first-line uterotonics were unsuccessful and who required a Bakri balloon to control postpartum haemorrhage secondary to an atonic uterus.

Methods  A DigiMano (Netech Corporation, Farmingdale, NY, USA) pressure recorder was attached via a three-way tap to a Bakri balloon. Anaesthetic charts of the two cases were reviewed retrospectively.

Main outcome measures  Intraluminal pressure readings were recorded after each 50-ml aliquot of normal saline had been insufflated into the balloon whilst the next aliquot was being prepared.

Results  There is a curvilinear relationship between the intraluminal pressure and the balloon volume. The pressure does not exceed the systolic blood pressure of the patient at the time of establishment of a positive tamponade test.

Conclusions  The intraluminal pressure within the tamponade balloon does not exceed the systolic blood pressure of the patient when a positive tamponade test is established.


Postpartum haemorrhage (PPH) is a significant cause of worldwide maternal morbidity and mortality.1,2 There are recommended guidelines for the management of PPH that involve an escalation of interventions from rubbing the uterine fundus to peripartum hysterectomy.3,4 In the case of an atonic uterus, which is the most common cause of PPH, there are various physical, surgical and pharmacological methods that can be used to stop uterine bleeding.4,5 Although evidence-based data in the management of PPH are focused on clinical effectiveness and not on the mode of action, first-line treatment is predominantly based on the achievement of contraction of the uterus using pharmacological agents, e.g. oxytocin, ergometrine, prostaglandin F2α and misoprostol.6,7

Uterine tamponade techniques for the management of PPH were reported as early as 1856.8 They usually used cotton gauze to pack the uterus. Although considered to be effective by those who used them regularly, an argument against their use was their ‘unphysiological’ nature of expanding the uterus.9 The uterus was expected to ‘contract down’.10 This concept, together with the possibilities of trauma, infection, ineffective packing and the development of effective pharmacological uterotonic agents, such as ergometrine and syntocinon, resulted in a gradual decrease in their use.11,12

More recently, a resurgence in the use of uterine tamponade in the management of PPH has occurred using balloon technology.13 A variety of such balloons are available. These include the purpose-designed Bakri balloon, as well as the relatively inexpensive Foley and Condom catheters.14–19 Others balloons that have been described previously in other cavity locations in which bleeding has been problematic have also been used. For example, the Sengstaken–Blakemore tube for the oesophagus and the Rusch balloon for the bladder.20,21

One proposed mechanism by which these balloons achieve their tamponade effect is by creating an ‘intrauterine pressure that is greater than the systemic arterial pressure’.22,23 In practice, the balloon is filled incrementally with fluid until the bleeding stops. Thus, a ‘tamponade test’ is used to determine the clinical effectiveness.24 However, this does not involve the measurement of the intrauterine pressure. Indeed, the in vivo pressures generated by the various balloons do not appear to have been reported.

In this article, we report the measurement of the intrauterine intraluminal pressure (ILP) within a Bakri balloon whilst achieving a positive tamponade test in two women.25

In vitro ILPs in the Bakri balloon

The ILPs were recorded for the Bakri balloon in the laboratory setting. These readings were obtained using a DigiMano (Netech Corporation, Farmingdale, NY, USA) pressure recorder (Figure 1A). The ILP was recorded after 50-ml aliquots of normal saline had been used to insufflate the Bakri balloon until a final volume of 500 ml was reached. In order to ensure consistent recordings from each balloon, three separate series (0–500 ml in 50-ml aliquots) were recorded from three different Bakri balloons. This resulted in a total of nine series of recordings (Figure 2). The readings demonstrate that, in the absence of any external restrictions, ILP reaches a peak of 60–80 mmHg at 50 ml insufflation (‘*’ in Figure 2). This pressure does not vary by more than 10–20 mmHg despite the balloon being incrementally filled to a volume of 500 ml (vertical arrow bar in Figure 2). There is no reduction or fluctuations in ILP when the balloon is allowed to stand for 2 hours after being filled to 500 ml of normal saline. In addition, when an external compressive force is applied, such as squeezing the balloon by hand, the ILP increases as expected (Figure 1B, C).

Figure 1.

 Intraluminal pressure recording (in vitro). (A) The Bakri balloon is connected via a three-way tap to a pressure meter (DigiMano, Netech Corporation, Farmingdale, NY, USA). This enables the intraluminal pressures to be measured independently of insufflation of the balloon with normal saline. By clasping the balloon and using a fixed volume of fluid (400 ml), the pressure meter records the effect of external compression (B, C). Bar, 4 cm.

Figure 2.

 Intraluminal pressure readings (in vitro). Fifty millilitre aliquots of normal saline were used to insufflate the Bakri balloon to 500 ml. The results from three balloons taking three sets of readings at each aliquot are shown. Note that the intraluminal pressure, without any external compression, will reach a peak of 60–80 mmHg at 50 ml (*). The pressure will then not vary by more than 10–20 mmHg up to a volume of 500 ml (inline image).

Case 1

A 31-year-old woman in her second pregnancy presented in labour at 41 weeks and 1 day of gestation. She had previously delivered twins by an elective caesarean section at 38 weeks.

Following oxytocin augmentation (syntocinon; Novartis Pharmaceuticals Australia Pty Ltd, North Ryde, NSW, Australia), she required a caesarean section for failure to progress in the second stage of labour under epidural anaesthesia.

At caesarean section, omental adhesiolysis demonstrated a 3-cm transverse rupture of the uterus within the previous uterine scar. There was minimal bleeding from this site. This site was extended to allow for delivery of the baby. Following the delivery of the placenta, haemorrhage ensued from an atonic uterus. Despite the administration of oxytocin, 10 iu, intravenously, an oxytocin intravenous infusion (40 iu/l running at 250 ml/hour), 5 iu intravenous oxytocin, 500 μg ergometrine intramuscularly (syntometrine; Novartis Pharmaceuticals Australia Pty Ltd) and 250 μg of intramyometrial prostaglandin F2α (Pfizer Australia Pty Ltd, West Ryde, NSW, Australia), the bleeding continued. Although no blood was visible from the peritoneal aspect of the two-layer closure of the lower segment site, the uterus remained atonic.

Vaginal bivalve speculum examination demonstrated ongoing bleeding via the cervical canal. The estimated blood loss at this stage was 2.8 l. A decision was made to insert a Bakri balloon (Cook Medical, Bloomington, IN, USA) and to perform a ‘tamponade test’ before the laparotomy site was closed.

The Bakri balloon was inserted vaginally into the uterine cavity via the cervical canal. A Rampley’s forceps was used to gently maintain the balloon at the uterine fundus as it was being insufflated with normal saline. This prevented the balloon from migrating into the vagina. As the Bakri balloon was filled, the ILP within the Bakri balloon was recorded using the DigiMano pressure meter connected via a three-way tap to the Bakri balloon (Figures 1 and 3) after each 50-ml aliquot.

Figure 3.

 Intraluminal pressure readings of Cases 1 and 2. The intraluminal pressures of Case 1 (full curve) and Case 2 (broken curve) are plotted against the infusion volume of normal saline. A positive tamponade test was achieved at 360 and 350 ml (*) for Cases 1 and 2, respectively. Note that the shape of the two graphs is curvilinear and that they are distinct from each other.

The tamponade test was positive at 360 ml of normal saline (a 60-ml aliquot was inadvertently used during the filling process). The corresponding ILP was 83 mmHg. The systolic and diastolic blood pressure of the patient, as recorded on the anaesthetic charts, varied from 80 to 110 mmHg and 40 to 65 mmHg, respectively. The average blood pressure was 92/46 mmHg with an average mean arterial pressure of 61 mmHg (Table 1).

Table 1.   The intraluminal and blood pressure readings obtained in Cases 1 and 2
CasePrimary cause of PPH (normal histopathology of placenta)EBL (l)Volume in balloon at positive tamponade test (ml)Intraluminal pressure (mmHg)Range of patient’s blood pressure* (systolic/diastolic) (mmHg)Average blood pressure* (mmHg)Average mean arterial pressure* (mmHg)
  1. EBL, estimated blood loss; PPH, postpartum haemorrhage.

  2. *Data obtained retrospectively from ‘anaesthetic charts’ documented during operating theatre sessions.

1Uterine atony2.83608380–110/40–6592/4661
2Uterine atony2.53504390–140/50–60110/5573

In view of the cervix being at full dilatation prior to delivery, the Bakri balloon was placed on a gentle vacuum via a Varivac wound drainage system (National Surgical Corporation, West Leederville, WA, Australia). This negated the need to pack the vagina, maintaining the balloon in the uterus. The laparotomy site was then closed and the stem of the Bakri balloon was secured to the inner thigh of the patient.

Case 2

A 28-year-old nulliparous woman with pregnancy-induced hypertension requiring pharmacological treatment (labetolol, 100 mg, twice daily) was admitted in early labour at 37 weeks and 5 days of gestation. Following the birth of the baby, despite 10 iu of oxytocin given intramuscularly and controlled cord traction, the placenta could not be delivered. A 40-iu oxytocin intravenous infusion (over 4 hours) was commenced, but the placenta failed to be delivered over the next hour. In view of a steady loss of vaginal blood estimated at 1 l, manual removal of the placenta under general anaesthetic was arranged.

Manual removal proved challenging and the placenta was removed in a piecemeal manner. The placenta was ‘reassembled’ and considered to be ‘complete’ following the procedure.

Despite the continuation of the original intravenous oxytocin infusion, rectal administration of 800 μg of misoprostol (Cytotec; Pfizer Australia Pty Ltd) and repeated bimanual compression, the uterus was intermittently atonic and bleeding ensued. At this stage, a total estimated blood loss of 2.5 l was calculated. Following the exclusion of genital tract trauma contributing to the ongoing bleeding, a decision was made to insert a Bakri balloon.

The Bakri balloon was inserted vaginally and intrauterine pressure measurements were obtained as described previously (see ‘Case 1’ above). A positive tamponade test was achieved at 350 ml (Figure 3). The corresponding ILP was 43 mmHg. The systolic and diastolic blood pressure of the patient, as recorded on the anaesthetic charts, varied from 90 to 140 mmHg and 50 to 60 mmHg, respectively. The average blood pressure was 110/55 mmHg with an average mean arterial pressure of 73 mmHg (Table 1).

Following placement, the balloon was again placed on vacuum to maintain the device in the uterus and thereby prevent caudal migration into the vagina. In this second case, ILP recordings continued as the patient’s second-degree perineal tear was being repaired and whilst the patient was transferred from the operating theatre into the recovery bay (Figure 4).

Figure 4.

 Intraluminal pressure (ILP) readings following the establishment of a positive tamponade test in Case 2. The ILP recordings of Case 2 were continued at minute intervals after the establishment of a positive tamponade test (*). The initial 5 min of ILP were not recorded as the perineum and vagina were being assessed for the subsequent perineal repair. The time interval from the beginning (*) to point 1 represents the second-degree perineal repair. Point 2 represents the point of extubation. Point 3 represents a short interval of no recording because of blood sampling.

During this time, the systolic and diastolic blood pressure of the patient, as recorded on the anaesthetic/recovery bay charts, varied from 130 to 135 mmHg and 85 to 90 mmHg, respectively. The average blood pressure was 132/86 mmHg with an average mean arterial pressure of 101 mmHg.

Bakri balloon removal and discharge

Both women were transferred to a High Dependency Unit. Each Bakri balloon was deflated by 50% at 12 hours and the remainder at 24 hours. Case 1 was eventually transfused with 4 units of packed cells (blood), 3 units of fresh frozen plasma and 1 unit of platelets. Case 2 required 2 units of packed cells (blood).

Both women received continuous syntocinon for 24 hours. The Na+ concentration was determined at 12 and 18 hours. These were within the normal range (135–145 mmol/l). Prophylactic intravenous (24 hours)/oral (4 days) antibiotics were administered (1 g cephazolin twice daily/500 mg cephalexin thrice daily and 500/400 mg metronidazole thrice daily).

Cases 1 and 2 were discharged on day 6 and day 4, respectively.

Placental histopathology

Histopathology reports for both placentae were reported as normal and without evidence of placenta accreta.


The management of PPH involves a stepwise series of physical, pharmacological and, eventually, surgical procedures to stop uterine bleeding.26 Once the retained products have been removed and any genital tract trauma repaired, ongoing bleeding is assumed to be from an atonic uterus.

Previously, the uterus was rarely described as ‘atonic’, but frequently described as ‘hypotonic’, implying some residual ability to contract.27 Thus, the current commonly used term ‘atonic’ implies that the uterus is unable to initiate or maintain contractions in order to achieve haemostasis. However, in the majority of cases of PPH secondary to uterine ‘atony’, when uterotonic agents are used they are clinically successful.6,7

Recently, uterine tamponade using balloon technology has been employed in the management of PPH.13 This involves the insertion of a balloon made of rubber or silicone into the uterine cavity and incremental inflation of the balloon with normal saline. This ‘tamponade test’ is considered to be ‘positive’ if bleeding ceases.24

Currently, ‘the principle of balloon tamponade therapy is to fill the uterine cavity to control bleeding with pressure’.28 This is achieved as a result of ‘uniform pressure over the open sinuses of the uterus’.29 An analogy to the ‘first aid technique to stop a vessel from bleeding’ has been made: ‘Applying sufficient pressure to compress the blood vessel often brings resolution. The technique works because the pressure on the blood vessel is greater than the pressure within the vessel. If pressure is applied for long enough, the blood will clot and form a permanent seal’.30

In non-uterine systems in which bleeding is successfully counteracted by balloon tamponade, relatively low ILPs are used. For example, only 25–30 mmHg is required when using the Sengstaken–Blakemore tube in the oesophagus.31 These values are based on portal hypertension readings, pressures required to collapse the coronary–oesophageal circuit and experimental confirmation that the oesophagus can withstand such pressures.30

In the bladder, a pressure ‘equal to the diastolic arterial pressure’, approximately 75–80 mmHg, is required.32 This ‘increased intravascular pressure induces a reduction of blood circulation’.33

In the case of the postpartum uterus, the creation of an ‘intrauterine pressure that is greater than the systemic arterial pressure’ has been proposed.22,23 However, there does not appear to be any published experimental recordings of postpartum uterine pressures or data on the actual pressures required to stop uterine bleeding from an atonic uterus. Although the original description of the Bakri balloon stated that it ‘could withstand 300 mmHg’in vitro, this article appears to be the first report of ILP recordings from a balloon within a postpartum uterus.15

The two cases reported here challenge the proposal that the intrauterine pressure should exceed the patient’s systolic pressure. The uterine bleeding was controlled with ILP of 83 mmHg (Case 1) and 43 mmHg (Case 2) (Table 1). In both cases, ILP was actually lower than the systolic blood pressure (Table 1). With respect to the diastolic blood pressure, the final ILP was greater than (Case 1) and less than (Case 2) of these documented values, respectively (Table 1).

These cases demonstrate that, as the balloon is insufflated to achieve a clinically effective ‘positive’ tamponade test, the numerical relationship between this volume and the resulting ILP is curvilinear (Figure 3). In both cases, as the volume of fluid increases within the balloon, the ILP gradually increases. This pressure rises and falls during the establishment of a positive tamponade test (Figure 3). When the tamponade test is positive, the ILP is lower than the previously recorded readings when bleeding was ongoing (Figure 3).


The observed ILP readings (Figure 3) have a number of limitations. The major limitation is that there are only two cases. More cases need to be obtained to provide reproducibility of these results, in particular the curvilinear relationship between ILP and balloon volume during the attainment of a positive tamponade test. The second limitation is being able to demonstrate uterine activity simultaneously and independently from ILP recordings. Otherwise, other explanations, such as the transmission of intra-abdominal pressure secondary to respiration, may account for the ILP fluctuations observed post-tamponade (Figure 4).

As the relationship between ILP, intraluminal volume, systolic/diastolic blood pressures and bleeding cessation appears to be inconsistent, alternative mechanisms have been proposed to explain the attainment of a positive tamponade test. These include uterine wall compliance/structure, the balloon–uterine interface, secondary uterine activity and distal effects on the uterine arteries.

Uterine wall compliance/structure

If, during PPH, the uterus is completely ‘atonic’, it may be considered to be unable to counteract an increase in uterine cavity volume. Provided that the structure of the uterine wall does not limit this expansion, the volume:ILP pattern would be expected to be similar to in vitro experiments in which there are no external restrictions to expansion of the balloon (Figure 2).

If, during PPH, the uterus is again considered to be ‘atonic’, but is able to withstand a defined increase in uterine cavity volume, ILP would again be expected to be similar to Figure 2, as the balloon volume approximates that of the uterine cavity (Figure 2). However, as the enlarging balloon becomes restricted by the non-compliant uterus, the pressure would be expected to rise again in a linear fashion.

Although neither of these patterns was observed in the two cases described, a positive tamponade test was achieved.

Helie described a curvilinear arrangement of muscle fibres that form a figure of eight around the uterine vasculature. This structural arrangement of the uterine muscle fibres results in haemostasis when the ‘physiological contracting-down’ postpartum uterus contracts, an effect referred to as ‘living sutures’.34

In the ‘unphysiological’ expansion of the uterine cavity caused by the uterine balloons, the distended uterine wall (Figure 5) may equally result in a conformational alteration of the myometrial vasculature that also contributes to haemostasis. Thus, the clinical endpoint of a positive tamponade test may not only be a function of the pressure exerted by the balloon, but also of the resulting uterine structure of the distended uterus.

Figure 5.

 The thinning out of the uterine wall during Bakri balloon insufflation. Ultrasound images of before (A) and after (B) insufflation of a Bakri balloon in a case of secondary postpartum haemorrhage. ‘C’ represents the uterine cavity with the tip of the Bakri balloon seen prior to insufflation with normal saline. The dotted sphere represents the insufflated Bakri balloon. The arrowed bars represent the thickness of the uterine wall of 3.4 cm and 1.7 cm in (A) and (B), respectively.

Balloon–uterine interface

Studies using condom-catheters also result in a positive tamponade test at similar volumes to other balloons.18,19 However, the pressure required to insufflate the thin-walled condom-catheter in vitro is significantly less than that of the other balloons (approximately 12–15 mmHg, data not shown). If such low pressures are sufficient to generate a positive tamponade test in vivo, it may be that it is not the magnitude of the pressure exerted which is important, but the effect of contact by the balloon at the uterine surface that elicits a clinical response.

A similar observation has been made in the use of an ‘OAT patch’ when employed to stop the bleeding from a vascular bed or a large vessel, such as the aorta or inferior vena cava.35,36 Here, the approximation of a piece of autologous tissue, such as a piece of rectus sheath, prevents ongoing bleeding. Again minimal external pressures are required to stop bleeding. The authors of this method have proposed mechanisms including: (1) lamina flow within the damaged vessel creates suction on the overlying patch (Venturi effect); (2) resistance to flow between the large patch and the vessel wall beyond the defect may be sufficient impedance to stop flow completely; and (3) the patch provides a framework for the deposition of fibrin and platelets.35

Secondary uterine activity

The curvilinear pattern of the intrauterine pressures during the establishment of a positive tamponade test may be interpreted as the uterus exhibiting an inherent contractile ability. Thus, the rising and falling pressures may represent phases of uterine contractile activity and relaxation, respectively. Clearly, the initial phases of uterine activity are insufficient to result in haemostasis, as the tamponade test is still negative at this time (Figure 3). However, as the uterus distends to accommodate the increasing intraluminal volume, haemostasis is achieved (‘*’ in Figure 3).

Therefore, regardless of the pressure exerted in the uterine cavity, the haemostatic effect may in fact be achieved by stimulating a contractile response of the uterine musculature secondary to insufflation of the balloon. It is noteworthy that, in addition to the known association of polyhydramnios and uterine activity, in vitro stretching of human myometrial cells results in an up-regulation of mRNA expression of the oxytocin receptor.37

Furthermore, the intrauterine pressure readings obtained from ongoing measurements, following the establishment of the positive tamponade test, suggest that the uterus is not an inert component, as the term ‘atonic’ implies. This is demonstrated by the cyclical variation in ILP during this time (Figure 4). Of note is that there were no fluctuations in ILP when the balloon was left to stand for 2 hours in vitro (see ‘In vitro ILPs in the Bakri balloon’ above).

It is uncertain whether this assumed ‘contractile response’ is a result of the stretching of the myometrium secondary to the presence of the Bakri balloon, or to the ongoing oxytocin infusion. The literature demonstrates the use of oxytocin in the majority of cases in which a balloon is used. However, oxytocin usage in these situations is not empirical, and some studies do not comment on the use of syntocinon when balloon tamponade technologies are used.13

Distal effects on the uterine arteries

One study has proposed that the effect of the tamponade is not at the balloon–uterine cavity interface, but at the uterine arteries by direct compression ‘akin to non-surgical uterine artery ligation’.38

Studies involving actual ligation of the internal iliac artery suggest that the conversion of an arterial into a venous-like system would enable a clot to form and remain in place.39 Of note, however, is the ability of balloon tamponade to be used in patients who have impaired coagulopathy.40

Furthermore, in the cases described by Burchell,39 reverse flow in the demonstrated collateral circulation may explain why surgical internal iliac ligation is not 100% effective at achieving haemostasis. Therefore, although probably significant, mechanisms other than lowering the internal iliac or uterine artery pressure must be involved in achieving a positive tamponade test.


These two cases suggest that, although a positive tamponade test is correlated with a rise in ILP, the uterus does not appear to be entirely ‘atonic’. The actual ILPs are not necessarily reflective of the clinical outcome and do not have to exceed the patient’s systolic blood pressure to achieve a positive tamponade test.

The resulting tamponade effect when distending a balloon in the uterine cavity may actually be caused by a number of mechanisms, including uterine shape changes, secondary uterine activity, balloon–endometrial interactions and distal effects on the flow within the uterine arteries.

Disclosure of interests


Contribution to authorship

CG performed the in vitro experiments, was the primary surgeon in both cases, obtained Ethics Committee approval and wrote the manuscript.

Details of ethics approval

Ethics committee approval was granted retrospectively from the University of Wollongong/South Eastern Sydney Ilawarra Area Health Service and Medical Human Resources Ethics Committee (HE09/240).




The author would like to acknowledge the excellent and efficient library staff at Wollongong Hospital (Christine Monie, Sharon Hay, Vivienne Caldwell and Jill Brady). In addition, the author would like to thank the Department of Biomedical Engineering at Wollongong Hospital (Gary De Lucia and Marcus Poplawski) for providing the ‘Digimano’ pressure transducer.