International Journal of Gynecology and Obstetrics 81 (2003) 191–198

Article
The relationship of placenta previa and history of induced abortion
L.G. Johnsona,*, B.A. Muellera,b, J.R. Dalinga,b
a
Fred Hutchinson Cancer Research Center, Division of Public Health Sciences, Seattle, WA, USA
b
Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA

Received 19 August 2002; received in revised form 12 December 2002; accepted 20 December 2002

Abstract

Objectives: We evaluated the risk of placenta previa being associated with a history of induced abortion by different
surgical procedures. Methods: Cases (ns192) were women who had a singleton delivery complicated by placenta
previa at a major obstetric care hospital in western Washington state between April 1, 1990 and December 31, 1992.
Controls (ns622) were women with singleton deliveries not complicated by placenta previa or abruption. Odds
ratios, determined by logistic regression, approximate the relative risks. Results: Vacuum aspiration abortion was not
associated with an increased risk of placenta previa (OR 0.9, 95% CI 0.6–1.5). However, the risk of placenta previa
increased with the number of sharp curettage abortions (OR 2.9, 95% CI 1.0–8.5 for G3). Conclusions: Risk of
placenta previa may be increased in a dose response fashion by multiple sharp curettage abortions. However, vacuum
aspiration does not confer an increased risk, and may be a better alternative.
䊚 2003 International Federation of Gynecology and Obstetrics. Published by Elsevier Science Ireland Ltd. All rights
reserved.

Keywords: Placenta previa; Induced abortion; Dilation and curettage; Vacuum curettage

1. Introduction the potential for unexpected hemorrhaging that

may become life-threatening in as little as 15 min
Placenta previa (PP) complicates approximately w1,7x.
one in every 200 births in the US. It typically is It has been suggested that surgical abortion,
accompanied by voluminous third trimester bleed- such as those by vacuum aspiration (VA) or
ing and pre-term cesarean delivery w1–3x. Pre-term dilation and sharp curettage (D&C) may cause
delivery resulting from PP is associated with scarring and adhesions to the uterus, which then
increased rates of perinatal morbidity and mortality impede proper placentation in subsequent pregnan-
w1,4–6x. Although the maternal mortality due to cies w1x. The risk of PP in women with one or
blood loss found with this condition is rare in the more prior induced abortions is reportedly 1.3–2.7
presence of modern obstetric practices, PP confers times that of women reporting no prior abortion
w8–12x. Few studies, however, had sufficient pow-
*Corresponding author. Tel.: q1-206-667-4630; fax: q1-
206-667-5948. er to examine the possible effect of multiple
E-mail address: lgodefro@fhcrc.org (L.G. Johnson). induced abortions. None have reported results

0020-7292/03/$30.00 䊚 2003 International Federation of Gynecology and Obstetrics. Published by Elsevier Science Ireland Ltd.
All rights reserved.
doi: 10.1016/S0020-7292Ž03.00004-3
192 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198

related to analyses of different abortion procedures, Approval for this study was granted from the
which may differentially be associated with the institutional review boards for human subjects’
occurrence of subsequent damage to the uterus. protection of the participating hospitals prior to
Early induced abortion by VA has been the most the conduct of the study. In-person interviews were
commonly used method for the last decade w13,14x, conducted with the majority (97%) of participants;
but sharp curettage for induced abortion continues the remainder of the interviews were administered
to be practiced in some settings in the US. We over the telephone. The structured interview
evaluated the association of PP with self-reported included questions about lifestyle, demographics,
history of induced abortion, and examined the and reproductive and medical histories. To obtain
effects of multiple procedures and of type of information about the number and type(s) of
abortion method used. previous pregnancies a woman had had, the num-
ber of each type (miscarriage, ectopicytubal preg-
2. Materials and methods nancy, induced abortion, stillbirth, single live birth,
multiple live birth, and other) was specifically
queried. Pregnancy information sheets were filled
All women whose pregnancies or deliveries out for each of the previous pregnancies, which
were complicated by PP (ICD-9 codes 641.0 and included information regarding: the outcome and
641.1) at the five major obstetric care hospitals in date, length of gestation, presence of complica-
King County, Washington between April 1, 1990 tions, and characteristics of the infant such as
and December 31, 1992 were identified as cases. gender and birthweight. For pregnancies ending in
This study was part of a larger study to evaluate induced abortion, information about the procedure
risk factors for placental abnormalities (including (VA, D&C, or other method), gestational length,
placental abruption as well). Potential cases were and presence of any complications was obtained.
excluded if the index pregnancy resulted in a Since it was possible there would be confusion
multiple birth, was -20 weeks gestation, or if the among the participants regarding the type of abor-
infant birthweight was -500 g. Controls were tion procedure that had been used (since both VA
selected from women with singleton births not and D&C may be referred to as ‘dilation and
complicated by either PP or abruptio placenta. curettage’), a list of procedures was included on
They were frequency matched to cases on month, the pregnancy information sheet, and VA was listed
year, and hospital of delivery. They were also as the first method, so that women would be more
matched on weekday vs. weekend delivery to likely to accurately report a ‘Suction (Vacuum
control for possible differences in diagnostic pro- aspiration)’ (wording used in questionnaire) abor-
cedures utilized during these times. Permission to tion, rather than erroneously reporting a D&C.
contact potential study subjects was requested from To assess the risk of PP associated with a history
the patients’ physicians. We identified 327 poten- of prior induced abortion, odds ratios (OR) and
tial PP cases, of whom 41 were subsequently 95% confidence intervals (CI) were used to esti-
found to be ineligible because they did not meet mate relative risks. ORs were calculated using
study criteria. Of the remaining 286, permission unconditional multivariate logistic regression. All
from physicians for patient contact was not calculations were performed using the STATA sta-
obtained for 35 women. After contact, 57 women tistical package (Stata Corporation, College Sta-
refused to participate, two were lost to follow-up, tion, TX). Factors that may have affected the
and 192 (67%) were interviewed. Of the 1026 occurrence of PP were evaluated for their potential
potential control subjects identified, 116 were effect on the relationships of interest, including:
found to be ineligible. Of the remaining 910, age at reference, the number of prior spontaneous
physician permission for contact was denied for abortions, parity, previous cesarean sections, pre-
100, 183 women refused to participate, five were vious dilation and curettage (for reasons other than
lost to follow-up, and 622 were interviewed induced abortion), previous PP, use of an intrau-
(67%). terine device (IUD), history of pelvic inflamma-
 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198 193

tory disease (PID), history of uterine fibroids, Table 1

smoking habits, and race. Factors that altered the Risk factors for placenta previa and their distribution, among
women with a singleton delivery between 1990 and 1992 in
age-adjusted risk estimates by at least 10% were western Washington state with and without placenta previa
adjusted for in the final risk estimates, these
included: age at reference, maternal smoking dur- Risk factor Cases Controls
ing the pregnancy, and parity. Analyses were com- (Ns192) (Ns622)
pleted both using the entire study population n % n %
(Table 2), and excluding women without a previ-
Age at reference
ous pregnancy (Table 3), since women without a -20 years 8 4.2 29 4.7
prior pregnancy could not have the exposure of 20–29 years 61 31.8 271 43.6
interest, induced abortion. 30q years 123 64.1 322 51.8
Had a prior spontaneous abortiona 63 39.1 160 34.6
3. Results Had a prior live birtha 126 78.3 329 71.1
Had a prior cesarean deliveryb 36 28.6 89 27.1
Ever used an IUD 33 17.2 60 9.7
Cases were more likely than controls to be older Ever had pelvic inflammatory disease 12 6.3 41 6.6
than 30 years of age at the reference birth (64% Ever had uterine fibroids 13 6.9 32 5.1
vs. 52%; chi-squared P for difference in age Had a prior dilation and curettagec 8 4.2 19 3.1
distribution by case-control statuss0.01) (Table Smoking
None while pregnant 138 73.8 500 80.9
1). Among women who had been pregnant prior 1st trimester only 9 4.8 32 5.2
to the index pregnancy, women with PP were 1st & 2nd trimesters 10 5.3 7 1.1
somewhat more likely to have had a prior live Entire pregnancy 30 16.0 74 12.0
birth; among parous women, cases were somewhat Other patternd 0 0 5 0.8
more likely to have had a prior cesarean section Race
White 153 79.7 497 79.9
delivery. Cases were significantly more likely than Black 8 4.2 34 5.5
controls to have used an IUD for contraception Asian 24 12.5 55 8.9
(Ps0.004), and were more likely to have smoked Other 7 3.7 36 5.9
after the first trimester of their index pregnancy Household income
(Fisher’s Exact P for difference in smoking pattern -$15,000yyear 26 13.5 91 14.6
$15,000 to -$30,000 33 17.2 113 18.2
during pregnancys0.005). There were no other $30,000 to -$45,000 41 21.4 140 22.5
significant differences for any variables by case- $45,000 to -$60,000 46 24.0 129 20.7
control status. Race and income did not substan- $60,000 or more 46 24.0 147 23.6
tially vary by case-control status, although the case a
Among women with prior pregnancy(ies).
group consisted of slightly more Asian women. b
Among women with prior delivery(ies).
Having one or more prior abortions by any c
Includes only D&Cs performed for purposes other than
procedure (excluding infusionyinduction) was not induced abortions.
d
associated with a significantly increased risk of PP Includes all smoking during pregnancy that was not clas-
sified in any of the previous categories (e.g. smoking only
(OR 1.2, 95% CI 0.8–1.8) (Table 2). The risks of during the first and third trimesters); Fisher’s Exact P-value.
PP associated with 1, 2 or G3 prior abortions by
any procedure were, respectively 1.0 (95% CI
0.7–1.6), 1.4 (95% CI 0.8–2.5) and 1.9 (95% CI 95% CI 0.6–1.4), and there was not a significantly
1.0–3.6) (Cochran–Armitage trend test, exact P- increased risk of PP associated with having 1 (OR
0.001). Thirteen women had post-abortion infec- 0.8, 95% CI 0.5–1.3), 2 (OR 1.0, 95% CI 0.5–
tions (seven cases and six controls), and for these 2.2), or G3 prior VA abortions (OR 1.4, 95% CI
women, the OR of PP, was 3.6 (95% CI 1.1– 0.6–3.1) (Cochran–Armitage trend test, exact Ps
11.5), compared with women who never had an 0.28). A small number of women who had a VA
abortion (data not shown). abortion required a D&C immediately following
Overall, having any prior VA abortions was not the VA procedure (one case and 11 controls).
associated with an increased risk of PP (OR 0.9, Excluding these women from the analysis of VA
194 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198

Table 2
Risk of placenta previa among women with a singleton delivery between 1990 and 1992 in western Washington state by number
and type of prior induced abortions, relative to women who had no prior abortions

Abortion method Cases (Ns192) Controls (Ns621) Multivariate modela

n % n % OR (95% CI)
Any typeb
None 111 57.8 409 65.9 Ref.
Any 81 42.2 212 34.1 1.2 (0.8–1.8)
1 40 20.8 135 21.7 1.0 (0.7–1.6)
2 20 10.4 49 7.9 1.4 (0.8–2.5)
3 or more 21 10.9 28 4.5 1.9 (1.0–3.6)
Vacuum aspirationc,d
Any 46 23.9 151 24.3 0.9 (0.6–1.4)
1 23 12.0 96 15.4 0.8 (0.5–1.3)
2 11 5.7 34 5.5 1.0 (0.5–2.2)
3 or more 12 6.3 21 3.4 1.4 (0.6–3.1)
Dilation and curettagec,d
Any 22 11.5 37 5.9 1.9 (1.0–3.4)
1 13 6.8 25 4.0 1.4 (0.8–2.6)
2 5 2.6 7 1.1 2.0 (1.0–4.0)
3 or more 4 2.1 5 0.8 2.8 (1.0–8.1)
Dilation and evacuationd
1 1 0.5 4 0.7 0.7 (0.1–7.7)
Unknown methods
1 9 4.7 15 2.4 1.9 (1.0–3.5)
2q 4 2.1 4 0.6 3.2 (1.0–10.1)
Infusionyinduction methodsd
None 179 93.2 590 95.0
1 0 0.0 12 1.9 – –
a
Adjusted for age at reference, smoking during the pregnancy, and parity.
b
Surgical procedures only, women reporting unknown type(s) of procedure(s) are counted as surgical.
c
Also adjusted for history of abortions by vacuum aspiration or dilation and curettage.
d
Excludes women reporting one or more abortions of unknown method.

abortions did not have a substantial impact on the tage trend test, exact Ps0.01). Excluding women
risk estimates of PP. After excluding these women, who also had prior D&Cs for reasons other than
the risk estimates for having 1, 2, or G3 were 0.9 pregnancy termination did not substantially change
(95% CI 0.5–1.5), 1.1 (95% CI 0.5–2.4), and 1.7 the estimates. Having had any prior abortions by
(95% CI 0.7–3.9), respectively (data not shown). other known methods (D&E or infusionyinduc-
There was also little effect on the risk estimates tion) was not associated with increased risks of
for VA abortions when women who had also had PP.
another abortion(s) performed by D&C were When analyses were restricted to women who
excluded (three cases and two controls). had at least one pregnancy prior to the index birth,
However, having any prior abortions by D&C the risk of PP associated with having had an
was associated with a risk of 1.9 (95% CI 1.0– abortion of any type was 1.2 (95% CI 0.8–1.8)
3.4). The risk of PP associated with having one relative to women who had never had an abortion
prior D&C abortion was 1.4 (95% CI 0.8–2.6), (Table 3). The risk estimates that excluded nulli-
the risk associated with two prior D&C abortions gravid women generally followed the same pattern
was 2.0 (95% CI 1.0–4.0), and the greatest risk as the estimates within the entire cohort, although
(OR 2.8, 95% CI 1.0–8.1) was associated with the magnitude of the estimates was attenuated. As
having G3 prior D&C abortions (Cochran–Armi- before, VA abortion was not associated with risk
 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198 195

Table 3
Among women with a prior pregnancy, risk of placenta previa among women with a singleton delivery between 1990 and 1992 in
western Washington state by number and type of prior induced abortions, relative to women who had no prior abortions

Abortion method Cases (Ns161) Controls (Ns462) Multivariate modela

n % n % OR (95% CI)
Any typeb
None 80 49.7 250 54.1 Ref.
Any 81 50.3 212 45.9 1.2 (0.8–1.8)
1 40 24.8 135 29.2 1.0 (0.6–1.6)
2 20 12.4 49 10.6 1.3 (0.7–2.4)
3 or more 21 13.0 28 6.1 1.8 (0.9–3.5)
Vacuum aspirationc,d
Any 46 28.6 151 32.7 0.9 (0.5–1.4)
1 23 14.3 96 20.8 0.7 (0.4–1.3)
2 11 6.8 34 7.3 1.0 (0.4–2.1)
3 or more 12 7.5 21 4.5 1.3 (0.6–2.9)
Dilation and curettagec,d
Any 22 13.7 37 8.0 1.8 (0.9–3.3)
1 13 8.1 25 5.4 1.4 (0.7–2.6)
2 5 3.1 7 1.5 1.9 (0.9–3.8)
3 or more 4 2.5 5 1.1 2.6 (0.9–7.5)
Dilation and evacuationd
1 1 0.6 4 0.9 0.8 (0.1–7.9)
Unknown methods
1 8 5.0 11 2.4 2.1 (1.0–4.7)
2q 3 1.9 4 0.9 2.7 (0.7–10.3)
a
Adjusted for age at reference, smoking during the pregnancy, and parity.
b
Surgical procedures only, women reporting unknown type(s) of procedure(s) are counted as surgical.
c
Also adjusted for history of abortions by vacuum aspiration or dilation and curettage.
d
Excludes women reporting one or more abortions of unknown method.

of PP (OR 0.9, 95% CI 0.5–1.4). However, there w2,3x. PP is also associated with increased perinatal
was a suggestion of an increased risk of PP morbidity (e.g. respiratory distress syndrome, low
associated with having had 2 (OR 1.9, 95%CI APGAR score, etc. w5,6x) and mortality w1,4x.
0.9–3.8), or G3 prior D&C abortions (OR 2.6, More recently, Li, et al. reported that babies born
95% CI 0.9–7.5). Using the Cochran–Armitage of pregnancies complicated by PP or abruptio
test, a significant trend in risk existed for any type placenta were at 2.1 greater risk of sudden infant
abortion (Ps0.02) and D&C abortion (Ps0.03), death syndrome (95% CI 1.3–3.1), compared with
but no trend was apparent for VA abortion (Ps babies of uncomplicated pregnancies w15x. Given
0.75). the serious nature of these associated events, it is
important to identify women who may be at
4. Discussion increased risk and thus may benefit from closer
monitoring during their pregnancies.
Although the occurrence of PP is relatively rare, Recent data from the US indicate that by age
serious adverse events are associated with its 45, 43% of women in the US will have had at
occurrence, including: maternal hemorrhage least one abortion, and that two-thirds of women
requiring blood transfusion w2,13x, extended post- who elect to have pregnancy termination still
partum hospitalization, low birth weight (associ- intend to have children later in life w14x. Within
ated with premature delivery) w2,4x, and increased our cohort, 42.2% of cases and 34.1% of controls
need for hysterectomy following cesarean delivery had undergone elective abortion prior to the index
196 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198

pregnancy. VA is generally thought to be safe and Our analysis had several limitations, including
free of long-term sequelae. Thus, our findings of the possibility of misclassification due to reporting
no increased risk of PP associated with this pro- bias. Since induced abortion is a sensitive issue, it
cedure is reassuring. Overall, we observed that is likely that we encountered under-reporting of
having had a prior abortion by any procedure was the procedure, and it is also possible that women
not associated with a significantly increased risk who have a pregnancy complication (such as PP)
of PP. However, our data suggest that having had may be more likely to report previous induced
multiple prior induced abortions, or an abortion abortions. However, it is unlikely that there would
that is complicated by infection may be associated be a differential reporting bias by type of abortion
with an increased risk of PP. Abortions performed procedure, which was the factor which we found
by D&C also may be associated with an increased to be an important modifier of risk in this analysis.
risk of PP, although in all estimates, our confidence Additionally, while it is unlikely that a woman
intervals included one. Our results persisted when would forget that she has had an induced abortion,
excluding nulligravid women from the analysis (a she may forget details, such as the procedure(s)
group whose inclusion could potentially lower the used, particularly after a long time. Although the
baseline risk of the reference group, women with- questionnaire used in our study was designed to
out a previous abortion). minimize these problems by eliciting details of
Previous studies of the association of induced each pregnancy, we did note instances where the
abortion and PP have reported conflicting results. length of gestation for the terminated pregnancy
Several controlled studies reported significantly was inconsistent with the type of procedure report-
increased risks of PP, with ORs ranging from 1.3 ed. For example, three D&Cs and six VAs report-
to 3.0, associated with a history of any induced edly occurred after 14 weeks gestation; exclusion
abortions, and ORs ranging from 1.3 to 2.0 for of these from their respective analyses, however,
two or more abortions, although the type of pro- had no effect, nor did inclusion of these as D&E
cedure performed was not evaluated w8–12x. How- abortions (the more likely procedure for abortions
ever, both Rose et al. and Williams et al. reported performed after 14 weeks) alter the risk estimate
no association of PP with prior induced abortion for that procedure. It is unlikely that our results
w16,17x, but, in addition to limited power, the are biased by the exclusion of illegal abortions,
authors suggested that their observed lack of effect since the majority of all of the participants’ repro-
may be due to the use of suction curettage or VA, ductive life was after 1973, when induced abortion
rather than sharp curettage, which would align was made legal by federal statute. It has been
their results with ours. proposed that accuracy of self-reported abortion
Past studies did not take into account the method information may be improved by introducing abor-
of abortion, an important factor in our analysis. It tion questions with a ‘filter question,’ similar to
is possible that D&C is more likely to cause the approach we used for obtaining information
damage or scarring of the uterus than VA, which on past pregnancies w19x. Despite these efforts,
is reflected in the suggestion of an increased risk there was variability in the recall or reporting of
of PP associated with D&C procedures in the prior abortions by cases and controls; 7% of cases
current analysis. That multiple D&C abortions may had unknown type of abortions, compared with
be associated with even greater scarring (and only 3% of controls (Ps0.02).
subsequently greater risk of PP as suggested by The majority (73%) of induced abortions among
our data) is also plausible. The theory of uterine study participants were performed by VA, which
scarring as the mechanism for the association of was the method used for more than 87% of all
PP and induced abortion is supported by a previ- abortions performed in Washington state in 2000
ously reported dose-response association between w20x. Also, among the study participants, a signif-
cesarean section deliveries (a procedure that causes icant number of D&C abortions were performed
extensive uterine scarring) and PP occurrence (23% of abortions reported; 7% of participants
w1,10,16,18x. reported one or more D&C abortions), although it
 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198 197

is no longer a commonly used procedure for previa accreta compared to placenta previa non-accreta.
induced abortions in the US w20x. A small propor- Acta Obstet Gynecol Scand 1998;77(4):391 –394.
w4 x Crane JM, van den Hof MC, Dodds L, Armson BA,
tion of the women in the current study reported an
Liston R. Neonatal outcomes with placenta previa.
abortion performed by D&E, which made it diffi- Obstet Gynecol 1999;93(4):541 –544.
cult to interpret results for this procedure, although w5 x Lin C, Wang S, Hsu Y, Lin Y, Yeh T, Chang F. Risk
no increased risk was observed. for respiratory distress syndrome in preterm infants born
In 1997, over 1.3 million induced abortions to mothers complicated by placenta previa. Early Hum
were performed in the US, and an estimated 46 Dev 2001;60(3):215 –224.
million abortions are performed annually world- w6 x Bekku S, Mitsuda N, Ogita K, Suehara N, Fujimura M,
wide w14x. We observed no significant increase in Aono T. High incidence of respiratory distress syndrome
PP risk associated with abortions being performed (RDS) in infants born to mothers with placenta previa.
J Matern Fetal Med 2000;9(2):110 –113.
by VA, indicating the relatively safe nature of the
w7 x Crane JM, van den Hof MC, Dodds L, Armson BA,
procedure with respect to future reproductive out-
Liston R. Maternal complications with placenta previa.
comes. Other studies have shown the high level of Am J Perinatol 2000;17(2):101 –105.
peri-operative safety associated with suction curet- w8 x Taylor VM, Kramer MD, Vaughan TL, Peacock S.
tage in performing abortions w21,22x. There is a Placental previa in relation to induced and spontaneous
suggestion that multiple induced abortions, partic- abortion: a population-based study. Obstet Gynecol
ularly those performed by D&C, may be associated 1993;82(1):88 –91.
with an increased risk of PP. Because of this, w9 x Chelmow D, Andrew DE, Baker ER. Maternal cigarette
pregnancies to women with a history of multiple smoking and placenta previa. Obstet Gynecol 1996;87(5
Pt 1):703 –706.
such procedures may merit closer monitoring to
w10x Ananth CV, Savitz DA, Luther ER. Maternal cigarette
help prevent adverse sequelae that may accompany
smoking as a risk factor for placental abruption, placenta
this condition. It is important to consider that only previa, and uterine bleeding in pregnancy. Am J Epi-
3% of abortions performed internationally occur demiol 1996;144(9):881 –889.
in the US, and the implications of our findings w11x Parazzini F, Dindelli M, Luchini L, La Rosa M, Potenza
may be greater for countries that continue to MT, Frigerio L, et al. Risk factors for placenta previa.
regularly use the D&C procedure to perform Placenta 1994;15(3):321 –326.
induced abortions (e.g. parts of Africa and South- w12x Hendricks MS, Chow YH, Bhagavath B, Singh K.
east Asia w23,24x), particularly if there is limited Previous cesarean section and abortion as risk factors
ability manage post-abortion infections or PP for developing placenta previa. J Obstet Gynaecol Res
1999;25(2):137 –142.
complications. w13x Hemlin J, Moller B. Manual vacuum aspiration, a safe
and effective alternative in early pregnancy termination.
Acknowledgments Acta Obstet Gynecol Scand 2001;80(6):563 –567.
w14x The Alan Guttmacher Institute: Accessed 2002 Apr.
This work was funded in part by Grant No. 5- Available at: http:yywww.agi-usa.orgypubsy
fb_induced_abortion.pdf.
RO1-HD25230 from the National Institute of Child
w15x Li DK, Wi S. Maternal placental abnormality and the
Health and Human Development, National Insti- risk of sudden infant death syndrome. Am J Epidemiol
tutes of Health, Bethesda, MD. 1999;149(7):608 –611.
w16x Rose GL, Chapman MG. Etiological factors in placenta
References previa—a case controlled study. Br J Obstet Gynaecol
1986;93(6):586 –588.
w1x Mabie WC. Placenta previa. Clin Perinatol w17x Williams MA, Mittendorf R, Lieberman E, Monson RR,
1992;19(2):425 –435. Schoenbaum SC, Genest DR. Cigarette smoking during
w2x McShane PM, Heyl PS, Epstein MF. Maternal and pregnancy in relation to placenta previa. Am J Obstet
perinatal morbidity resulting from placenta previa. Gynecol 1991;165(1):28 –32.
Obstet Gynecol 1985;65(2):176 –182. w18x McMahon MJ, Li R, Schenck AP, Olshan AF, Royce
w3x Zaki ZM, Bahar AM, Ali ME, Albar HA, Gerais MA. RA. Previous cesarean birth. A risk factor for placenta
Risk factors and morbidity in patients with placenta previa? J Reprod Med 1997;42(7):409 –412.
198 L.G. Johnson et al. / International Journal of Gynecology and Obstetrics 81 (2003) 191–198

w19x Huntington D, Mensch B, Miller VC. Survey questions w22x Legarth J, Peen UB, Michelsen JW. Mifepristone or
for the measurement of induced abortion. Stud Fam vacuum aspiration in termination of early pregnancy.
Plann 1996;27(3):155 –161. Eur J Obstet Gynecol Reprod Biol 1991;41(2):91 –96.
w20x Washington State Department of Health: Accessed 2002 w23x Singh S, Wulf D, Jones H. Health Professionals’ Per-
Feb. Available at: http:yywww.doh.wa.govyEHSPHLy ceptions About Induced Abortion in South Central and
CHSydefault.htm噛Birth. Southeast Asia. Int Fam Plann Perspect 1997;23(2):59 –
w21x Thonneau P, Fougeyrollas B, Ducot B, Boubilley D, Dif 67.
J, Lalande M, et al. Complications of abortion per- w24x Makinwa-Adebusoye P, Singh S, Audam S. Nigerian
formed under local anesthesia. Eur J Obstet Gynecol health professionals’ perceptions about abortion prac-
Reprod Biol 1998;81(1):59 –63. tice. Int Fam Plann Perspect 1997;23(4):155 –161.