Risk factors for preterm delivery in
women with placenta praevia and
antepartum haemorrhage: retrospective
Objective. To identify risk factors for preterm delivery in women with placenta
praevia and antepartum haemorrhage.
Design. Retrospective study.
Setting. Regional obstetric unit, Hong Kong.
Subjects and methods. Women delivered at Princess Margaret Hospital between
1 January 1990 and 31 December 1997. Possible risk factors for preterm
delivery among women with placenta praevia and antepartum haemorrhage
including onset, pattern, and severity of vaginal bleeding; presence of uterine
contractions on admission; and type of placenta were assessed.
Results. Three risk factors for preterm delivery were identified from univariate
analysis. These included second trimester vaginal bleeding (odds ratio=4.19; 95%
confidence interval, 1.29-13.66), the presence of uterine contractions on admis-
sion (odds ratio=4.00; 95% confidence interval, 1.57-10.19), and a haemoglobin
decrease of more than 20 g/L (odds ratio=3.00; 95% confidence interval, 1.00-
9.04). Using the logistic regression model, second trimester vaginal bleeding
and the presence of uterine contractions were found to be independent risk
factors for delivery before 36 weeks.
Conclusion. Preterm delivery is increased in women with placenta praevia
and antepartum haemorrhage who have second trimester vaginal bleeding or the
presence of uterine contractions. This high-risk group may benefit from close
in-patient monitoring and more aggressive management.
!"#1990 1 1
!"#$%& '()*+,-./ "01234
!"#$%& 20 g/L(
Department of Obstetrics and Gynaecology,
Princess Margaret Hospital, 2-10 Princess
Margaret Hospital Road, Hong Kong
CM Lam, MB, ChB, MRCOG
SF Wong, FRCOG, FHKAM (Obstetrics and
Women with placenta praevia are at increased risk of antepartum haemorrhage
and preterm delivery.1 Perinatal outcome is poorer and morbidity increased
Correspondence to: Dr CM Lam
in women with antepartum haemorrhage.2 The increased adverse perinatal
HKMJ Vol 8 No 3 June 2002 163
Lam et al
outcome seen was related to the risk of prematurity.3
of vaginal bleeding was stratified into mid-trimester and
The management of a woman with bleeding due to
third-trimester bleeding (ie <28 weeks and <32 weeks,
placenta praevia depends on two main factors—the
respectively). The pattern of bleeding was classified as
degree of haemorrhage, and foetal maturity at the time of
persistent or recurrent vaginal bleeding. Each episode of
haemorrhage. Absolute indications for delivery include
bleeding was defined as one occurring after more than a
bleeding after 36 weeks of gestation, foetal distress at
period of 48 hours free from bleeding. As this was a retro-
a viable gestation, and persistent haemorrhage causing
spective study, it was difficult to quantify the amount of
maternal haemodynamic instability at any stage in the
bleeding accurately. Vaginal bleeding was classified into
three grades: spotting, mild-to-moderate, and severe.
Severe bleeding was defined as bleeding resulting in
The purpose of this study was to identify clinical risk
maternal haemodynamic instability (hypotension and
factors associated with preterm delivery in women with
tachycardia). Women with more than vaginal spotting
antepartum haemorrhage and placenta praevia. Such
but not severe bleeding were classified as having mild-to-
infomation might assist in the future management of
moderate bleeding. More objective assessments such as a
patients with antepartum haemorrhage and placenta
drop in haemoglobin of more than 20 g/L 24 hours after
praevia, in particular with regard to selecting those women
the onset of bleeding, and the need of blood transfusion
who might benefit from close surveillance in hospital and
were also used as a guideline. The presence of uterine
more aggressive treatment, such as tocolytics or repeated
contractions was defined as more than one uterine contrac-
tion in 10 minutes on admission tococardiograph.
Subjects and methods
Demographic data and potential clinical risk factors
for the two groups of women were compared. Statistical
The records of all women with placenta praevia at the time
analyses were performed using Statistical Package for
of delivery at Princess Margaret Hospital, a regional obstet-
Social Science (Windows version 9.0; SPSS Inc., Chicago,
ric unit in Hong Kong, between 1990 and 1997 were
US). Chi squared test or the Fisher’s exact test was used,
reviewed. Multiple pregnancies and women with abruptio
where appropriate, to compare categorical variables. The
placentae were excluded. Transabdominal ultrasound
unpaired Student’s t test was used to compare continuous
scanning was used for the diagnosis of placenta praevia. In
variables with a normal distribution. For continuous vari-
doubtful cases, transvaginal sonography had been used. The
ables with non-parametric distribution, the Mann-Whitney
final diagnosis was based on the operative findings.
U test was used. P values of less than 0.05 on two-tailed
Dewhurst’s classification for placenta praevia was used in
analyses were considered statistically significant. Stepwise
all cases.4 Women with asymptomatic placenta praevia had
binary logistic regression was performed to identify the
been admitted at about 32 weeks for observation. Women
independent predictor(s) for preterm delivery. Results of
presenting with antepartum haemorrhage had also been
the analysis were reported as adjusted odds ratios (OR)
admitted for further management. Indications for immedi-
with a 95% confidence interval (CI).
ate delivery were vaginal bleeding after 36 weeks, foetal
distress at a viable gestation, persistent haemorrhage
causing maternal haemodynamic instability at any stage in
pregnancy despite blood transfusion, preterm labour, and at
There were 35 931 deliveries at Princess Margaret Hos-
term, or when other obstetric complications arose.
pital between 1990 and 1997. A total of 305 women with
placenta praevia were identified during the study period.
A case-control design was used to assess for potential
The incidence of placenta praevia was 0.85%. A total of
clinical risk factors associated with preterm delivery. Cases
142 women with placenta praevia experienced ante-
were defined as women who had delivered before 36 weeks’
partum haemorrhage before 36 weeks’ gestation. Of
gestation, given that women presenting with antepartum
these 142 women, 25 were excluded from the study
haemorrhage after this gestation would normally be
because of twin pregnancy, abruptio placentae, missing
delivered. Women who delivered after 36 weeks comprised
records, or uncertain gestation. Fifty-nine women were
the control group.
delivered at or before 36 weeks and 58 women after 36
Socio-demographic characteristics including maternal
age, gravidity, parity, education level, working status, smok-
The demographic data of the two groups are listed in
ing habit, type of placenta praevia, and relevant obstetric
Table 1. There was no difference in age, parity, working
history such as previous miscarriage, termination of
status, educational level attained, marital status, smoking
pregnancy, dilatation and curettage, Caesarean section, and
status, history of drug abuse, or obstetric history. The
previous placenta praevia were retrieved. Data on potential
indications for delivery in each case are summarised in
predictors for preterm delivery were also retrieved—the
Table 2. The majority of women in the preterm group
onset, pattern, and severity of vaginal bleeding; presence of
were delivered because of antepartum haemorrhage (61%).
uterine contractions; and the type of placenta. The onset
The three most common indications for delivery included
164 HKMJ Vol 8 No 3 June 2002
Placenta praevia and antepartum haemorrhage
Table 1. Demographic data comparing women with placenta praevia and antepartum haemorrhage who delivered before
versus after 36 weeks of gestation
Delivery at or before 36 weeks
Delivery after 36 weeks
No. of patients
Age (mean, [SD]) [years]
Tertiary education (%)
Termination of pregnancy (%)
Dilatation and curettage (%)
Previous placenta praevia (%)
Previous Caesarean section (%)
Haemoglobin level at booking (g/L)
† Unpaired Student’s t test
‡ Fisher’s exact test
§ Mann-Whitney U test
Table 2. Indications for delivery for women with placenta
placenta praevia mainly reflected prematurity. Brenner et
praevia and antepartum haemorrhage (n=117)
al6 found that approximately 40% of women with placenta
praevia also experienced premature rupture of membranes,
Massive antepartum haemorrhage
spontaneous labour, haemorrhage, or other problems that
resulted in delivery before 37 weeks’ gestation. Women
with placenta praevia without antepartum haemorrhage
Persistent antepartum haemorrhage
Premature prelabour rupture of membranes
however, do not appear to have an increased risk of preterm
delivery.2 We had previously shown that risk of preterm
birth was mainly associated with antepartum haemorrhage
massive antepartum haemorrhage, labour, and mature
and threatened preterm delivery.2 In the current study, two
foetus. Other indications included persistent vaginal
risk factors for preterm delivery were identified in women
bleeding, and premature prelabour rupture of membranes.
presenting with antepartum haemorrhage—second trimes-
ter bleeding and the presence of uterine contractions on
Univariate analysis identified three risk factors that were
significantly associated with preterm delivery (Table 3).
These included second trimester vaginal bleeding, presence
This was a retrospective case-control study and one
of uterine contractions on admission, and a haemoglobin
might argue that the observed risk factors simply reflected
decrease of >20 g/L. Women presenting with vaginal spot-
the indications for delivery. This might be true for women
ting only were at lower risk of being delivered prematurely.
presenting with uterine contractions, although it was
Type of placenta did not influence the risk of preterm
difficult to differentiate a cause-effect relationship. On the
other hand, this study failed to demonstrate that the pattern
and severity of vaginal bleeding had any influence on
Two independent risk factors were identified after
preterm delivery. Women with placenta praevia presenting
stepwise logistic regression analysis. These were second
with mid-trimester vaginal bleeding had a higher risk of
trimester vaginal bleeding (OR=4.80; 95% CI, 1.09-21.10)
preterm delivery. This appears to be an unequivocal risk
and the presence of uterine contractions on admission
factor, as most obstetricians were reluctant to deliver these
(OR=9.54; 95% CI, 3.34-27.20).
women until a more advanced gestation.
In the US, many obstetricians have adopted a policy
of permitting selected women with placenta praevia and
The use of expectant management was first advocated by
antepartum haemorrhage to return home as part of
Macafee5 in an attempt to reduce the number of premature
expectant management.7,8 Most of the studies supporting this
births and allow the pregnancy to continue until the baby
approach have shown a reduction in hospitalisation and
had grown to a size and age that would give a reasonable
cost.9,10 However, these small studies had insufficient
chance of survival. According to the initial Macafee regimen,
power to assess the safety issue. The group at high risk
women were confined to a fully equipped and staffed
for preterm deliveries identified in this study, namely those
maternity hospital from the time of the initial diagnosis of
with second trimester bleeding or the presence of uterine
placenta praevia until delivery. Preterm birth continued to
contractions, should not be offered such a choice. Moreover,
be a major problem even when expectant management was
in this high-risk group more aggressive treatment may
employed. The poor perinatal outcome seen in women with
reduce the risk of preterm delivery.
HKMJ Vol 8 No 3 June 2002 165
Lam et al
Table 3. Risk factors associated with delivery before 36 weeks in women with antepartum haemorrhage
Delivery at or
before 36 weeks (%)
36 weeks (%)
(95% confidence interval)
Severity of bleeding
Haemoglobin drop (>20 g/L)*
Presence of uterine contractions*
Major placenta praevia
Onset of bleeding
Need for blood transfusion
* Significant risk factors
Various reports have shown that aggressive management
Lam CM, Wong SF, Chow KM, Ho LC. Women with placenta praevia
of placenta praevia, including tocolytic therapy, repeated
and antepartum haemorrhage have a worse outcome than those who
blood transfusion, and prolonged hospital stay leads to
do not bleed before delivery. J Obstet Gynaecol 2000;20:27-31.
improved outcomes.7,11-18 Besinger et al12 demonstrated that
Mabie WC. Placenta previa. Clin Perinatol 1992;19:425-35.
Neilson JP. Antepartum haemorrhage. In: Whitfield CR, editor.
tocolytic use delayed delivery and was associated with an
Dewhurst’s textbook of obstetrics and gynaecology for postgraduates.
increase in birth weight. Towers et al13 have shown that there
5th ed. Oxford: Blackwell Science Ltd; 1995:164-74.
was no increase in morbidity and mortality associated with
Macafee CH. Placenta praevia—a study of 174 cases. J Obstet
an aggressive management approach in a tertiary setting.
Gynaecol Br Emp 1945;52:313-24.
Cervical cerclage has been shown to prolong pregnancy in
Brenner WE, Edelman DA, Hendricks CH. Characteristics of patients
with placenta previa and results of “expectant management”. Am J
women with second trimester vaginal bleeding but there is
Obstet Gynecol 1978;132:180-91.
currently insufficient evidence to recommend this practice
Cotton DB, Read JA, Paul RH, Quilligan EJ. The conservative
outside of a clinical trial.17,18 Results from retrospective
aggressive management of placenta previa. Am J Obstet Gynecol 1980;
studies to date indicate that women with clinical risk
factors of second trimester bleeding or the presence of
Wing DA, Paul RH, Millar LK. Management of the symptomatic
placenta previa: a randomized, controlled trial of inpatient versus
uterine contractions may benefit from close surveillance in
outpatient expectant management. Am J Obstet Gynecol 1996;175:
hospital, tocolytic therapy, and repeated blood transfusion.
Further studies in this group of patients, particularly those
Droste S, Keil K. Expectant management of placenta previa: cost-
with second trimester bleeding, will help to define the role
benefit analysis of outpatient treatment. Am J Obstet Gynecol 1994;
of such treatment in decreasing preterm delivery and
10. Mouer JR. Placenta previa: antepartum conservative management,
improving clinical outcome. In addition, knowing that these
inpatient versus outpatient. Am J Obstet Gynecol 1994;170:1683-6.
women are at increased risk of preterm delivery, the option
11. Silver R, Depp R, Sabbagha RE, Dooley SL, Socol ML, Tamura RK.
of more aggressive treatment should be discussed and
Placenta previa: aggressive expectant management. Am J Obstet
offered to this patient group.
12. Besinger RE, Moniak CW, Paskiewicz LS, Fisher SG, Tomich PG.
The effect of tocolytic use in the management of symptomatic
placenta previa. Am J Obstet Gynecol 1995;172:1770-5.
13. Towers CV, Pircon RA, Heppard M. Is tocolysis safe in the managment
Women with placenta praevia and antepartum haemorrhage
of third-trimester bleeding? Am J Obstet Gynecol 1999;180:1572-8.
who have second trimester vaginal bleeding or the presence
14. Watson WJ, Cefalo RC. Magnesium sulfate tocolysis in selected
of uterine contractions were shown to be at higher risk of
patients with symptomatic placenta previa. Am J Perinatol 1990;7:
preterm delivery in this study. These women should receive
15. Saller DN Jr, Nagey DA, Pupkin MJ, Crenshaw MC Jr. Tocolysis
close in-patient monitoring. Further studies investigating
in the management of third trimester bleeding. J Perinatol 1990;10:
the effect of more aggressive treatment, such as the use of
tocolytics and repeated blood transfusion, are indicated.
16. Sampson MB, Lastres O, Tomasi AM, Thomason JL, Work BA Jr.
Tocolysis with terbutaline sulfate in patients with placenta previa
complicated by premature labor. J Reprod Med 1984;29:248-50.
17. Arias F. Cervical cerclage for the temporary treatment of patients with
placenta previa. Obstet Gynecol 1988;71:545-8.
Rosen DM, Peek MJ. Do women with placenta praevia without
18. Tessarolo M, Bellino R, Arduino S, Leo L, Wierdis T, Lanza A. Cervi-
antepartum haemorrhage require hospitalization? Aust NZ J Obstet
cal cerclage for the treatment of patients with placenta previa. Clin
Exp Obstet Gynecol 1996;23:184-7.
166 HKMJ Vol 8 No 3 June 2002