| |

Received: 29 December 2018    Revised: 29 April 2019    Accepted: 14 June 2019

DOI: 10.1002/ijgo.12889

CLINICAL ARTICLE
Obstetrics

Randomized feasibility study of suction-­tube uterine

tamponade for postpartum hemorrhage

G. Justus Hofmeyr1,* | Katrin Middleton2 | Mandisa Singata-Madliki1

1
Effective Care Research Unit, Universities
of the Witwatersrand, Fort Hare, and Walter
Sisulu, and Eastern Cape Department of
Health, East London, South Africa
2
Department of Obstetrics and
Gynaecology, Frere Hospital, and Walter
Sisulu University, East London, South Africa
*Correspondence
G. Justus Hofmeyr, Frere Hospital, East
London, South Africa.
Email: justhof@gmail.com
Funding Information
National Research Foundation
Abstract
Objective: To identify an inexpensive catheter suitable for uterine suction tamponade
(UST) for postpartum hemorrhage and assess its functionality.
Methods: Randomized, single-­center, double-­blind feasibility study in East London,
South Africa, among 45 women undergoing cesarean delivery between October and
November 2018. A search of medical supply websites for catheters with predefined
characteristics (inexpensive, flexible, wide-­bore, circumferentially-­arranged side aper-
tures, rounded tip) identified the FG36 Levin stomach tube. During cesarean, the tube
was placed in the uterus and connected transvaginally to a suction unit. Participants
were randomized via a computer-­generated random sequence to early (after uterine
closure; n=24) or delayed (after skin closure; n=21) UST.
Results: The tube functioned well with respect to stability in the uterus and aspirating
blood from the uterine cavity without blockage. Blood loss was similar between the
groups (mean difference, 7.3 mL; 95% confidence interval, −61 to 75; P=0.433), as were
secondary outcomes. There were no complications. Absolute effectiveness was not
tested because there was no non-­suction group.
Conclusion: The FG36 Levin tube was found to be a suitable device for “suction-­
tube uterine tamponade”. There was no difference in functionality between early
and late UST. Future trials should assess the effectiveness of this approach for
postpartum hemorrhage.
Clinical trial registration: Registered in the Pan African Clinical Trial Registry as
PACTR201809584199573.

KEYWORDS
Cesarean delivery; Feasibility study; Improvised suction tube; Postpartum hemorrhage;
Randomized; Uterine suction tamponade

1 |  INTRODUCTION
Uterine balloon tamponade is widely promoted to treat postpartum
hemorrhage (PPH),1 but one randomized trial has suggested harm.2
Two recent commentaries have called for further objective research
on uterine tamponade methods.3,4 Uterine suction tamponade (UST)
is an alternative method aligned with the physiologic mechanism of
placental site hemostasis (uterine contraction). The application of
UST during the cesarean procedure might reduce blood loss without
systemic effects and improve visualization during closure of the
hysterotomy wound.
In terms of UST devices, a flexible disposable5 and two rigid
stainless-­steel6,7 cannulas have shown promising results when
used therapeutically and prophylactically after vaginal and cesarean
delivery. In addition, a standard 6-­mm wound suction drain has been
used for UST.4

Int J Gynecol Obstet 2019; 1–5 © 2019 International Federation of |  1

wileyonlinelibrary.com/journal/ijgo  
Gynecology and Obstetrics
 |
2       Hofmeyr ET AL.
A limiting factor for implementation of purpose-­designed UST
devices is affordability and availability when needed. The aim of the
present study was therefore to identify a widely available, inexpensive
catheter or tube that might be used for UST, to assess the feasibility
and functionality of the chosen device, and to conduct a pilot compar-
ison of the relative effect of early versus delayed UST on uterine blood
loss during cesarean delivery.
If feasibility and/or effectiveness were demonstrated, this would
justify use of this simple method in larger trials for prevention of blood
loss at CS and treatment of PPH following CS and vaginal birth.
2 |  MATERIALS AND METHODS
The present exploratory, 1:1 randomized, single-­center, double-­blind
feasibility study of a suitable suction tube for UST was conducted
among women undergoing cesarean delivery at Frere Hospital, a ter-
tiary referral hospital in East London, South Africa, between October
15 and November 23, 2018. The protocol was approved by the Human
Research Ethics Committee of the University of the Witwatersrand
(reference M180370; September 10, 2018). All participants provided
written informed consent.
Prior to the feasibility trial, the FG36 Levin stomach washout
tube (Akacia Medical, Cape Town, South Africa) was chosen for the
following attributes: readily available and inexpensive; manufactured
from soft PVC, making uterine perforation unlikely; sufficiently rigid
for easy insertion without an instrument and unlikely to block; a pro-
file (12-­mm diameter and rounded end) similar to those of Panicker 6
7
and Ram stainless steel catheters, for which there is considerable
reported observational evidence, but with the safety advantage of
flexibility and wider side ports (10 mm for FG36 vs 4–5 mm for Ram);
and radial arrangement of four side ports so that they cannot all be
obstructed by application to the anterior and posterior uterine walls.
The feasibility of the FG36 Levin tube for UST was tested among
women undergoing cesarean delivery. A convenience sample size of
45 participants was chosen for the exploratory pilot study. No formal
sample size calculation was performed. The protocol conformed to
8
requirements laid out in the CONSORT statement. The analysis plan
was finalized before recruitment began. Good clinical practice (GCP)
procedures were followed. Adverse events and serious adverse events
were to be recorded on standard adverse events forms and submitted
to the Witwatersrand University ethics committee.
The inclusion criteria were low-­risk pregnancy; age 18–45 years;
planned cesarean; and informed consent. The exclusion criteria were
participation in another clinical trial and presence of serious medi-
cal complications. Women being treated for HIV infection with no
immune compromise were not excluded.
Women who were booked for elective cesarean or underwent
cesarean in early labor, and who met the inclusion criteria were asked
to participate. An information form described the procedures, and
emphasized the voluntary nature of participation (which might not
have any benefit) and the freedom to not participate or to withdraw
consent at any point. A GCP-­trained research assistant was available
to explain any uncertainties. Prospective participants were entered in
a screening log, and a screening form was completed.
Participants were randomized to either early UST (applied after
closure of the uterus) or delayed UST (applied after closure of the skin
wound). Allocation cards marked “A” or “B” via a computer-­generated
random sequence without blocks were placed in consecutively num-
bered opaque sealed envelopes by a member of staff who was not
involved in the study. The participants, surgeon, and those assessing
outcomes were all blind to the group allocations.
All cesarean procedures were performed by a single surgeon (GJH)
following a standard method. After initiation of spinal analgesia, the
vagina was cleansed with chlorhexidine and the cervix examined digitally.
A low-­profile plastic “fracture” bedpan was slipped under the participant's
thighs. The participant's legs were positioned slightly apart. The abdomen
was opened transversely by the Joel Cohen/Misgav-­Ladach method using
mainly blunt dissection. Participants with previous surgery frequently
required additional sharp dissection of adhesions. The myometrium was
opened transversely in the midline with a scalpel and the uterotomy was
extended laterally with finger traction before the membranes were rup-
tured. The neonate was delivered with fundal pressure and obstetric for-
ceps if required. Cord clamping was delayed for 30–60 seconds while the
newborn was shown to the mother. Uterotonics were administered by
the anesthetist according to routine practice (most commonly oxytocin
2.5 IU by slow intravenous bolus plus 7.5 IU by intravenous infusion). The
placenta was delivered with cord traction.
After confirmation that there were no anticipated complications,
a research assistant entered the participant's name on the trial reg-
ister and opened the next in the numbered series of sealed opaque
allocation envelopes. The open end of the FG36 Levin catheter was
passed downward through the cervix and vagina, and the rounded
tip was inserted in the uterine fundus. The open end of the cathe-
ter was retrieved under the drapes by a floor nurse and connected to
10-­mm inner diameter silicone suction tubing, which was connected
to an electric surgical suction unit in an adjoining room. Suction pres-
sures as low as 70 mm Hg (Inpress device) 5 and as high as 650 mm Hg
(Ram device) 7 have been reported for UST. It was estimated that 100–
150 mm Hg would provide adequate suction to collapse the uterus,
while being low enough to be achievable with simple suction sources;
thus, the suction unit was pre-­set at approximately 120 mm Hg. The
suction tubing was connected via a T-­connection to a three-­way tap
concealed in a box with a sliding lid and settings “A,” “B,” and “C”.
Setting C was a closed position (suction on), whereas A and B were
an open position (i.e., vented/no suction) and another closed position
(suction on). The researchers were blind to which setting was which. The
research assistant set the suction vent to setting A or B in accordance
with the randomized card. The uterus was closed with interrupted reverse
figure-­of-­eight PGA sutures on a blunt needle. As soon as the uterus was
closed, the assistant switched on the suction unit. If tubal ligation had
been requested, it was carried out by using a modified Pomeroy method
with 2/0 chromic catgut sutures. The sheath was closed with a continu-
ous PGA blunt needle suture. The skin was closed with 2/0 interrupted
nylon mattress sutures. As soon as the skin was closed, the research assis-
tant switched the suction apparatus vent to setting C (closed/suction on).
Hofmeyr ET AL.
After cleaning and dressing the wound, the position and stability of
the catheter in the uterus was noted; the catheter was then removed.
The uterus was massaged to determine whether there was any resid-
ual blood that had not been aspirated by the catheter. The catheter
was inspected for blockages, and blood remaining in the catheter and
tubing was shaken into the suction unit bottle. The blood collected in
the bedpan and that collected in the suction bottle were measured
with a graduated measuring cylinder. Froth on top of the blood from
the suction bottle was ignored in the measurement.
All times and measurements were recorded on a checklist, and par-
ticipants and their neonates were followed until discharge from hos-
pital. Data were transferred from the checklist and hospital notes to
a case record form. The confidentiality of participants was maintained
by use of participant study ID numbers on all study documents.
Core outcome sets for prevention and treatment of PPH are not
yet available on the CROWN database (http://www.crown-­initiative.
Assessed for eligibility (n=57)
Randomized (n=45)
Allocated to early UST (n=24)
♦ Received allocated intervention (n=24)
♦ Did not receive allocated intervention (n=0)
Lost to follow-up (n=0)
Discontinued intervention (n=0)
Analysed (n=24)
♦ Excluded from analysis (n=0)
F I G U R E   1   CONSORT flow diagram showing recruitment and analysis of the study participants.
|
      3
org/core-­outcome-­sets/ongoing-­core-­outcome-­sets/). Therefore, the
primary outcome was uterine blood loss measured during the cesar-
ean procedure and until removal of the catheter and massage of the
uterus. Secondary outcomes were duration of surgery; need for blood
transfusion; postoperative hemoglobin level; highest postoperative
temperature; incidence of wound infection, endometritis, and postna-
tal complications; and time from surgery to discharge (days).
The results were reported in accordance with CONSORT guide-
lines.8 The data were transferred from the case record forms to Excel
(Microsoft, Redmond, WA, USA) and analyzed by using Epi Info version
7.2.1.0 (CDC, Atlanta, GA, USA). Data entry and cleaning were conducted
blind to group allocation. Owing to non-­normal distributions and small
sample numbers, continuous variables were reported as median (range)
with comparison by non-­parametric statistical tests (Mann-­Whitney or
Wilcoxon two-­sample), or as mean difference (95% confidence interval
[CI]). Categoric variables were reported as number (percentage) or risk
Excluded (n=12)
♦ Not meeting inclusion criteria (n=4)
♦ Declined to participate (n=1)
♦ Other reasons (n=7)
Allocated to delayed UST (n=21)
♦ Received allocated intervention (n=21)
♦ Did not receive allocated intervention (n=0)
Lost to follow-up (n=0)
Discontinued intervention (n=0)
Analysed (n=21)
♦ Excluded from analysis (n=0)
 |
4       Hofmeyr ET AL.

ratio (95% CI) and compared by Fisher exact test owing to small num- T A B L E   1   Baseline characteristics of the study participants.a
bers. A P value of less than 0.05 was taken as statistically significant.
Delayed UST
Characteristic Early UST (n=24) (n=21)

Age, y 32 (19–43) 32 (19–43)

3 |  RESULTS
Nullliparous 1 (4) 1 (5)
Previous cesarean
Among 243 women undergoing cesarean delivery at Frere Hospital
during the study period, 57 women were screened and 45 were 1 16 (67) 12 (57)

enrolled in the pilot trial (Fig. 1). The exclusions included ineligibility, 2 6 (25) 6 (29)

non-­consent, pre-­enrolment complications, and assessment of the Weight, kg 87 (58–143) 81 (54–131)

uterine cervix as too tightly closed for passage of the suction tube. BMIb 35 (24–54) 32 (21–58)
All randomized participants received the allocated treatment. There Polyhydramnios 1 (4) 1 (5)
were no changes to the protocol after the trial commenced and there HIV infected on treatment 13 (54) 9 (47)
was no interim analysis. For the cesarean procedure, one woman in the Other pregnancy 4 4
early UST group received general anesthesia due to failure to achieve complicationsc
subdural needle placement. Only one neonate was delivered with the Reason for cesarean
use of obstetric forceps. For two women, the cervix was insufficiently Previous cesarean 22 (92) 19 (90)
dilated and an FG26 catheter was used instead of the FG36 catheter. Non-­reassuring fetal 1 (4.2) 1 (5)
The baseline characteristics of the 45 study women are summa- heart rate
rized in Table 1, and were well balanced between the UST groups Fetal growth impairment 0 (0) 1 (5)
except for a difference in the percentage of women who received a Estimated macrosomia 1 (4) 0 (0)
prophylactic oxytocin bolus exceeding 2.5 IU (25% in the early vs 57%
Pre-­operative hemoglobin, 12.4 (7.6–14.1) 12.3 (8.9–13.3)
in the delayed UST group). Because randomization took place after g/dL
delivery of the neonate and placenta, neonatal outcomes are summa- Birthweight, g 3325 (2200–4480) 3160 (2170–4580)
rized with the baseline data (Table 1). Apgar score at 5 min <7 0 (0) 0 (0)
The catheter seemed to be an effective device with respect to aspi- d
Fetal anomaly 1 (4) 2 (10)
rating all blood from the uterine cavity. While suction was maintained,
Neonatal hypoglycemia 0 (0) 1 (5)
the catheter was fixed within the uterus and could not be dislodged
General anesthesia 1 (4) 0 (0)
with light traction. In two cases (one in each group), the catheter had
Post enrolment
moved down in the uterus. This was thought to be related to the study
Tubal ligation 9 (38) 7 (33)
design, which included a delay between insertion of the catheter and
turning on the suction, during which time the catheter might be pulled Routine uterotonic

down by the weight of the suction tubing. In a non-­research setting, Oxytocin slow bolus 6 (25) 12 (57)
>2.5 IU
the catheter would be fixed in place by application of suction immedi-
ately after insertion. Oxytocin infusion >10 IU 8 (33) 8 (38)

After removal of the catheter, the uterus was massaged to expel any
residual blood. In 36 cases, there was no blood at all; in 8 cases, less than
5 mL; and in one case, a blood clot of approximately 20 mL. In no cases
were any of the apertures or the catheter itself obstructed by blood clots.
The outcomes of the study women are summarized in Table 2.
There was no difference in the measured blood loss or in any of the sec-
ondary outcomes between the two study groups. There were no cases
of postoperative complications, blood transfusion, wound infection, or
endometritis. Forty-­three women were discharged on postoperative
day 2 (n=25) or day 3 (n=18). One participant with twin pregnancy was
discharged on day 4; and one was discharged on day 5 after investiga-
tion of mild unexplained tachycardia, which resolved.
Oxytocin 5 IU/ 1 (4) 1 (5)
ergometrine 0.5 mg i.v.
Abbreviations: BMI, body mass index (calculated as weight in kilograms
divided by the square of height in meters); i.v., intravenous; UST, uterine
suction tamponade.
a
Values are given as median (range) or number (percentage).
b
Data missing for 2 women in the early UST group and 1 woman in the
delayed UST group.
c
Uncomplicated variable fetal heart rate decelerations, gestational diabe-
tes on diet only, mild gestational hypertension, iron deficiency anemia, and
treated tuberculosis (early UST group); chronic hypertension; diabetes/
chronic hypertension; early fetal heart rate decelerations; twin pregnancy
with fetal growth restriction twin 2 (delayed UST group).
d
Trisomy 21 (early UST group); transposition of great vessels, necrotizing
enterocolitis, neonatal death; trisomy 13, neonatal death (delayed UST group).

4 |  DISCUSSION pilot trial, it remained in place as long as suction was maintained and
it seemed to be an efficient method of evacuating all blood from the
The FG36 Levin catheter examined in the present study met the cri- uterine cavity. Narrower Levin tubes might also be effective for UST
teria for an inexpensive ready-­made catheter in general use. In the and are even less expensive (e.g. FG24).
Hofmeyr ET AL.       5|
T A B L E   2   Outcomes of the study participants by treatment group.a

Outcome Early UST (n=24) Delayed UST (n=21) MD (95% CI) RR (95% CI) P valueb

Operative
Total blood loss, mL 149 (13–466) 126 (24–462) 7.3 (–61 to 75) 0.433
Blood in receiver, mL 76 (4–355) 57 (7–398) –13 (–70 to 43) 0.900
Blood in suction bottle, mL 55 (8–276) 43 (11–130) 16 (–12 to 44) 0.375
Duration of surgery, min 25 (17–39) 23 (16–50) 1.6 (12.4–5.6) 0.310
Duration of suction, min 14 (10–24) 4 (3–5) 11.1 (9.6–12.7) <0.001
Postoperative
Hemoglobin, g/dL 11.2 (6.9–13.2) 11.0 (7.8–12.7) 0.1 (–0.8 to 0.9) 0.891
Highest temperature 37.1 (36.1–37.8) 37.1 (36.3–39.9) –0.2 (–0.5 to 0.1) 0.505
Hospital stay >3 d 1 (4) 1 (5) 0.88 (0.05–13) >0.99
Hospital stay 2 d 9 (38) 11 (52) 0.72 (0.37–1.4) 0.383

Abbreviations: CI, confidence interval; MD, mean difference; RR, relative risk; UST, uterine suction tamponade.
a
Values are given as median (range) or number (percentage) unless stated otherwise.
b
By Mann-­Whitney or Fisher exact test. Intention to treat analysis.

The study has some limitations. Although it provided useful
observational information on the functionality of the chosen cath-
eter, there was no non-­suction control group to assess its absolute
effectiveness. In addition, the potential to show a difference between
the early and delayed UST groups was limited by the small sample
size, wide variation in measurements, and the fact that both groups
received suction tamponade. There were also limitations inher-
ent in the blood measurement methods. The volume measured in
the fracture bedpan was compromised by the possibility of leakage
of amniotic fluid below the drapes into the receiver. No correction
was attempted because there was no way of accurately determin-
ing the volume of contamination, which was assumed to be balanced
because of the randomized trial design.
Uterine suction tamponade is intuitively an attractive option for
management of unresponsive PPH with encouraging results from
previous observational studies using various devices.6–8 Currently,
however, there is no objective evidence of its effectiveness in
randomized trials and the present study was not designed to provide
such evidence.
In conclusion, an inexpensive stomach suction tube in general use
has been demonstrated to be functional in evacuating all blood from
the uterine cavity without blockage. If UST is shown to be an effec-
tive method for treating or preventing PPH in the future, this catheter
has the important public health advantage of being widely available at
minimal cost (less than US $2 for FG36; less than US $0.30 for FG24).
In the absence of electrical suction apparatus, a vacuum extraction
hand pump or a manual vacuum aspiration suction syringe might be
used. Use of the “suction-­tube uterine tamponade” method should be
considered in future trials of UST for PPH.
AUT HOR CONTRI BUTI O N S
GJH conceived the study, drafted the protocol and manuscript, and
performed the trial procedures. KM conducted clinical follow-­up
of the participants. MS supervised the random allocation and GCP
training. All authors revised and approved the final manuscript.
AC KNOW L ED G M ENTS
The study was funded from a National Research Foundation “NRF rating
grant” to G.J.H that was not specific to the study. The study was not
reviewed by the funder.
CO NFL I C TS O F I NT ER ES T
The authors have no conflicts of interest.
REFERENCES
1. Tindell K, Garfinkel R, Abu-Haydar E, et al. Uterine balloon tamponade
for the treatment of postpartum hemorrhage in resource-­poor settings:
A systematic review. BJOG. 2013;120:5–14.
2. Dumont A, Bodin C, Hounkpatin B, et  al. Uterine balloon tamponade
as an adjunct to misoprostol for the treatment of uncontrolled postpar-
tum hemorrhage: A randomized controlled trial in Benin and Mali. BMJ
Open. 2017;7:e016590.
3. Viteri OA, Sibai BM. Uterine balloon tamponade for the management
of postpartum hemorrhage: A challenge and an opportunity for better
evidence. BJOG. 2018;125:540.
4. Hofmeyr GJ. Time to test tamponade. BJOG. 2018;125:538–539.
5. Purwosum Y, Sarkoen W, Arulkumaran S, Segnitz J. Control of post-
partum hemorrhage using vacuum-­induced uterine tamponade. Obstet
Gynecol. 2016;128:33–36.
6. Panicker TN. Panicker's vacuum suction haemostatic device for treating
post-­partum hemorrhage. J Obstet Gynaecol India. 2017;67:150–151.
7. Ram H, Ram HS, Ram S, Panicker V. Vacuum retraction of uterus for the
management of atonic postpartum hemorrhage. IOSR J Dent Med Sci.
2014;13:15–19.
8. Boutron I, Altman DG, Moher D, Schulz KF, Ravaud P; CONSORT NPT
Group. CONSORT statement for randomized trials of nonpharmaco-
logic treatments: A 2017 update and a CONSORT extension for non-
pharmacologic trial abstracts. Ann Intern Med. 2017;167:40–47.