AMERICAN JOURNAL OF EPIDEMIOLOGY
Copyright © 1975 by The Johns Hopkins University
Vol. 102, No. 3
Printed in U.S A
LATE SEQUELAE OF INDUCED ABORTION: COMPLICATIONS AND OUTCOME OF PREGNANCY AND LABOR SUSAN HARLAP AND A. MICHAEL DA VIES1 Harlap, S., and A. M. Davies (Hebrew U.-Hadassah Medical School, P. O. Box 1172, Jerusalem, Israel). Late sequelae of induced abortion: complications and outcome of pregnancy and labor. Am J Epidemiol 102: 217-224, 1975 The effects of previous induced abortion on pregnancy, labor and outcome of pregnancy were measured in a prospective study of 11,057 pregnancies to West Jerusalem mothers who were interviewed during pregnancy and who subsequently delivered a single live or stillborn infant. The 752 mothers who reported one or more induced abortions in the past were more likely, at the same interview, to report bleeding in each of the first 3 months of the present pregnancy. They were subsequently less likely to have a normal delivery and more of them needed a manual removal of the placenta or other intervention in the third stage of labor. In births following induced abortions, the relative risk of early neonatal death was doubled, while late neonatal deaths showed a 3- to 4-fold increase. There was a significant increase in the frequency of low birthweight, compared to births in which there was no history of previous abortion. There were increases in major and minor congenital malformations, but no significant changes in stillbirth or post-neonatal death rates, nor in mean birthweight or sex ratio. When the effects of other variables were taken into account, there were no significant changes in frequency following an induced abortion as to: ABO and rhesus isoimmunization, toxemia, hydramnios, premature rupture of membranes, induction of labor, breech or vacuum delivery, cesarean section, breech presentation, placenta previa, placental abruption, cord prolapse, cord anomalies, fetal distress or asphyxia, post-partum hemorrhage. abortion; birth weight; delivery; fetal death; labor; pregnancy INTRODUCTION Liberalization of induced abortions in many countries has inevitably been accompanied by concern about their possibly harmful effects. There is now considerable information on early complications which include bleeding, trauma and infection and a small, but not negligible mortality (1-7).
However, in countries with liberal abortion practices, mortality in women undergoing induced abortion is considerably lower than in those giving birth or having spontan e o u s miscarriages, and other early complications of artificial termination are generall y milder a n d fewer t h a n those exPerienced by women at parturition (4, 5). Late sequelae of induced abortion are Received for publication September 24, 1973, and reported to be uterine synechae (7 8)in final form February 12, 1975. , , c ,.. ,„, . . . ' 'Department of Medical Ecology, Hebrew Univer- r e duced fecundity (6), cervical incompesity - Hadassah Medical School, P. 0. Box 1172, tence (7) with increased risk of spontaneJerusalem, Israel. in the second o u s miscarriage, particularly Supported by grant H9/181/10 from the World . . , ,„ , ', , J , . _ , , ..,, i n . Health Organization. trimester (9, 10), premature birth (11, 12) The authors thank N. Grover, Ph.D., M. Haber, and rhesus isoimmunization (13). T h e maM.Sc, M B R. Prywes RN, Dip. Obst Mrs. H. j o r i t y o f s t u d i e s of late sequelae, however Rossman, RN, and Mrs. N. Samueloff, RN, for their h, a v e b,e e n b ,a s e d ,o n ^• , A assistance, and Prof. W. Z. Polishuk for his invaluable retrospective evidence, advice. and have not taken into account t h e differ217
218
HARLAP AND DA VIES
ing demographic and health characteristics of women who have induced abortions. Furthermore, there are many inconsistencies in their findings. The present study is prospective in that information on abortion was obtained at interviews in pregnancy, usually in the second trimester as part of a routine antenatal interview. The characteristics of Jerusalem women reporting previous induced abortions are described elsewhere (14). Briefly, induced abortion is correlated with smoking, maternal age, increasing birth order up to the fourth, and certain ethnic groups; it is negatively correlated with religiosity and age at marriage. Some of these variables themselves affect the chances of an unfavorable pregnancy outcome and must be taken into account in estimation of the risks associated with previous induced abortion. We have used multiple regression analysis as an approach to this problem. While induced abortion is nominally illegal in Israel, other than in certain defined circumstances, the procedure has in fact been available on request for over 30 years (15). Large numbers of abortions are performed by qualified doctors under "clean" conditions in hospitals, clinics and private offices without interference from lawenforcement agencies. The "back-street" abortionist is unknown. In addition, Israel has absorbed many immigrants from EastEuropean countries where induced abortion is very common (5). In the Jerusalem Perinatal Study, 5 per cent of mothers who have been pregnant before admit to an induced abortion in the last pregnancy, while 9 per cent had had an induced abortion in at least one of their previous pregnancies. This frequency rises to 20 per cent in women who are not religiously observant. While little stigma is attached to induced abortion in our society, there may well be some degree of under-reporting and these figures probably underestimate the prevalence of previous induced abortion among pregnant women. More-
over, since women who have induced abortions may tend to avoid further pregnancies, or at least, not allow them to continue, these figures will fall considerably short of the true frequency of induced abortion in our population. MATERIAL AND METHODS
The design, scope and methodology of the ongoing Jerusalem Perinatal Study have been described in detail in previous publications (16). Briefly, for every baby born in Jerusalem a file is opened, containing information from the birth-certificate and details of the delivery abstracted from labor-ward records. Copies of all death certificates are received from the Ministry of Health and a record-linkage technique continuously updates the file with this information as well as that on congenital malformations and hospital inpatient morbidity collected from obstetric departments, mother and child health clinics and hospital pediatric departments. The study covers all 6000-7000 annual births to residents of West-Jerusalem and its rural hinterland (the pre-1967 Israeli sector). All births take place in hospital and the high rates of utilization of mother and child health clinics (17) and hospital inpatient services (18) make it probable that significant malformations are quickly and accurately recorded. For a limited period in 1966-1968, it was possible to interview pregnant women, usually at their first antenatal visit (19), and this information forms the basis of the present study. The interview cohort is derived from women attending municipal mother and child health stations or hospital clinics for their antenatal care. Interviewing was not carried out concurrently in all stations, and mothers attending certain private clinics were not reached, so that the interviewed cohort has certain biases which are summarized in table 1. Statistical techniques. Because there are so many possibly confounding variables to be taken into account, we have chosen to
LATE SEQUELAE OF INDUCED ABORTION TABLE 1
Proportion of mothers interviewed in pregnancy, by birth order, mother's age, birthweight, infant mortality and malformations % having antenatal Birth order 1 2-3 4-6 7+
66.3 62.7 66.9 61.0
Mother's age < 19 20-29 30-39 > 40 '
71.5 65.6 61.7 59.2
Birthweight < 2.5 kg > 2.5 kg
55.8 65.0
Stillbirths Early neonatal deaths Late neonatal deaths Post-neonatal deaths
62.3 41.5 51.4 67.3
Babies with major malformations Babies with minor malformations
64.8 69.5
Total % of mothers interviewed
64.3
use multiple regression analysis to estimate differences in pregnancy outcome between women with and without previous induced abortion. Despite the absence of certain theoretical requirements, regression methods seem the most appropriate to achieve our objective of an overall survey summarizing possible complications of labor in births following induced abortions. For each outcome of pregnancy (death, birthweight, complication) a stepwise regression analysis was done, forcing the variable for the number of previous abortions into the equation at the first step and continuing to incorporate other predictor variables until none remained with a residual effect significant at the 5 per cent level. From the resulting regression equations, standardized rates were calculated for each outcome, in women with or without previ-
219
ous induced abortions. These rates are standardized to the "average woman" of this study, the mother for whom each of the other predictor variables takes its mean values. The rates are given by the equations: Y = A + BaXa + 2 BtXi for births following induced abortions, and Y = A + 2 BtXi for births without previous induced abortions, where Y = probability of each outcome of pregnancy, A = regression constant, Ba = regression coefficient for induced abortions, Xa = the mean number of previous abortions in women with a positive history of abortions, while Bt and Xt are the regression coefficients and the mean values of each of the significant other variables in the regression equation. For each outcome, crude rates are presented with standardized rates and significance levels which refer to thefc-testsof the regression coefficients (Ba) for induced abortions. A detailed account of how each variable was coded, and the regression equations for each outcome of pregnancy are available from the authors. The following variables were tested in each regression analysis where appropriate: Outcome: Stillbirth; early, late and postneonatal death; major and minor malformations (defined in previous study (20)); birthweight and low birthweights of l No previous induced previous abortions abortions
Stillbirth rate/1000 Early neonatal deaths/1000 Late neonatal deaths/1000 Post-neonatal deaths/1000 Major malformationsVlOOO Minor malformationsVlOOO Birthweight 1 previous abortions
Significance*
3.8
3.6
6.7
3.6
6.5
p . 001
6.6
6.2
11.8
6.2
11.9
p.001
4.2
4.1
5.5
4.0
5.8
p.05
0.1 0.2 1.4 0.8
0.2 0.2 1.5 0.8
0.1 0.4 0.9 1.3
0.2 0.2 1.5 0.8
0.1 0.2 1.0 1.1
ns ns ns ns
11,057
10,305
752
10,305
752
* See material and methods.
count by regression analysis, there is no significant difference in the resulting standardized rates between mothers with and without induced abortions. Sex ratio, which is tested as an indicator of a possible teratogenic process, is unchanged, following induced abortions. However, both major and minor malformations (defined in a previous study (20)) are increased in the abortion group, and the differences are unlikely to be due to chance. The same is true for rates of low birthweight of less than 2.5 kg and less than 2.0 kg. The increase in risk is relatively greater for the very small babies, the great majority of whom are true prematures, i.e. pre-term births. Mean "birthweight is not significantly altered in the abortion group; the small decrease of 10-15 gm is easily explained by the increased proportion of very small births. There is therefore no evidence, in this population, of growth retardation in the birth following induced abortion. Table 3 shows the findings for firsttrimester bleeding and other complications of pregnancy. Women reporting previous induced abortions at their antenatal interview also reported first-trimester bleeding with significantly greater frequency. Other pregnancy complications, subsequently re-
corded in labor-ward books at the time of the birth, were not significantly increased in the abortion group, when other variables were taken into account by regression analysis. Crude rates of rhesus incompatibility and hydramnios were substantially higher in the abortion group, but the differences disappeared after controlling for confounding variables. Among the complications of labor recorded by midwives and obstetricians (table 4), placenta previa was more frequent (crude rate) in births following induced abortions. Again, after standardization for other variables, there were no significant differences between the abortion group and the controls, for this and other complications of labor. Mothers with previous induced abortion tended to have fewer normal deliveries and more interventions in the third stage of labor (table 5). The crude rate of cesarean sections was higher in the abortion group, but not after standardization. DISCUSSION
From the public health point of view, the most significant finding of this prospective study is the excess of neonatal deaths and low birthweight in births following induced abortion. While the probability of death is
222
HARLAP AND DA VIES TABLE 4
Complications of labor, by previous induced abortion Crude rates (%) Complication of labor
Breech Premature rupture of membranes Placenta previa Placental abruption Cord prolapse Other cord anomalies Fetal distress/asphyxia Post-partum hemorrhage Total singleton births
Total No previous interviewed induced cohort abortions
Standardized rates (%)* >1 previous abortions
No previous induced abortions
>1 previous abortions
2.9 2.1 0.4 0.3 0.4
2.9 2.0 0.3 0.4 0.4
2.9 2.5 0.8 0.1 0.1
2.9 2.1 0.4 0.4 0.4
2.8 2.0 0.5 0.1 0.2
12.3
12.3
12.2
12.3
12.3
5.5 1.3
5.5 1.3
5.5 1.1
5.5 1.3
6.2 1.3
11,057
10,305
752
10,305
752
Significance'
ns ns ns ns ns ns ns ns
* See material and methods. TABLE 5
Interventions in labor, by previous induced abortions iCrude rates
Intervention in labor
No intervention (normal delivery) Induction Forceps or vacuum Cesarean section Intervention in 3rd stage (manual removal of placenta) Total singleton births
Standardized rates (%)*
(%)
Total No previous interviewed induced cohort abortions
>1 previous abortions
No previous induced abortions
Significance* >1 previous abortions
82.1
82.3
79.9
p < .05
8.2 5.2 3.4 1.9
79.1 8.6 4.4 5.2 2.9
82.2
8.3 5.1 3.5 2.0
8.2 5.2 3.4 1.9
9.1 4.7 3.8 3.0
ns ns ns
11,057
10,305
752
10,305
752
p
Copyright © 1975 by The Johns Hopkins University
Vol. 102, No. 3
Printed in U.S A
LATE SEQUELAE OF INDUCED ABORTION: COMPLICATIONS AND OUTCOME OF PREGNANCY AND LABOR SUSAN HARLAP AND A. MICHAEL DA VIES1 Harlap, S., and A. M. Davies (Hebrew U.-Hadassah Medical School, P. O. Box 1172, Jerusalem, Israel). Late sequelae of induced abortion: complications and outcome of pregnancy and labor. Am J Epidemiol 102: 217-224, 1975 The effects of previous induced abortion on pregnancy, labor and outcome of pregnancy were measured in a prospective study of 11,057 pregnancies to West Jerusalem mothers who were interviewed during pregnancy and who subsequently delivered a single live or stillborn infant. The 752 mothers who reported one or more induced abortions in the past were more likely, at the same interview, to report bleeding in each of the first 3 months of the present pregnancy. They were subsequently less likely to have a normal delivery and more of them needed a manual removal of the placenta or other intervention in the third stage of labor. In births following induced abortions, the relative risk of early neonatal death was doubled, while late neonatal deaths showed a 3- to 4-fold increase. There was a significant increase in the frequency of low birthweight, compared to births in which there was no history of previous abortion. There were increases in major and minor congenital malformations, but no significant changes in stillbirth or post-neonatal death rates, nor in mean birthweight or sex ratio. When the effects of other variables were taken into account, there were no significant changes in frequency following an induced abortion as to: ABO and rhesus isoimmunization, toxemia, hydramnios, premature rupture of membranes, induction of labor, breech or vacuum delivery, cesarean section, breech presentation, placenta previa, placental abruption, cord prolapse, cord anomalies, fetal distress or asphyxia, post-partum hemorrhage. abortion; birth weight; delivery; fetal death; labor; pregnancy INTRODUCTION Liberalization of induced abortions in many countries has inevitably been accompanied by concern about their possibly harmful effects. There is now considerable information on early complications which include bleeding, trauma and infection and a small, but not negligible mortality (1-7).
However, in countries with liberal abortion practices, mortality in women undergoing induced abortion is considerably lower than in those giving birth or having spontan e o u s miscarriages, and other early complications of artificial termination are generall y milder a n d fewer t h a n those exPerienced by women at parturition (4, 5). Late sequelae of induced abortion are Received for publication September 24, 1973, and reported to be uterine synechae (7 8)in final form February 12, 1975. , , c ,.. ,„, . . . ' 'Department of Medical Ecology, Hebrew Univer- r e duced fecundity (6), cervical incompesity - Hadassah Medical School, P. 0. Box 1172, tence (7) with increased risk of spontaneJerusalem, Israel. in the second o u s miscarriage, particularly Supported by grant H9/181/10 from the World . . , ,„ , ', , J , . _ , , ..,, i n . Health Organization. trimester (9, 10), premature birth (11, 12) The authors thank N. Grover, Ph.D., M. Haber, and rhesus isoimmunization (13). T h e maM.Sc, M B R. Prywes RN, Dip. Obst Mrs. H. j o r i t y o f s t u d i e s of late sequelae, however Rossman, RN, and Mrs. N. Samueloff, RN, for their h, a v e b,e e n b ,a s e d ,o n ^• , A assistance, and Prof. W. Z. Polishuk for his invaluable retrospective evidence, advice. and have not taken into account t h e differ217
218
HARLAP AND DA VIES
ing demographic and health characteristics of women who have induced abortions. Furthermore, there are many inconsistencies in their findings. The present study is prospective in that information on abortion was obtained at interviews in pregnancy, usually in the second trimester as part of a routine antenatal interview. The characteristics of Jerusalem women reporting previous induced abortions are described elsewhere (14). Briefly, induced abortion is correlated with smoking, maternal age, increasing birth order up to the fourth, and certain ethnic groups; it is negatively correlated with religiosity and age at marriage. Some of these variables themselves affect the chances of an unfavorable pregnancy outcome and must be taken into account in estimation of the risks associated with previous induced abortion. We have used multiple regression analysis as an approach to this problem. While induced abortion is nominally illegal in Israel, other than in certain defined circumstances, the procedure has in fact been available on request for over 30 years (15). Large numbers of abortions are performed by qualified doctors under "clean" conditions in hospitals, clinics and private offices without interference from lawenforcement agencies. The "back-street" abortionist is unknown. In addition, Israel has absorbed many immigrants from EastEuropean countries where induced abortion is very common (5). In the Jerusalem Perinatal Study, 5 per cent of mothers who have been pregnant before admit to an induced abortion in the last pregnancy, while 9 per cent had had an induced abortion in at least one of their previous pregnancies. This frequency rises to 20 per cent in women who are not religiously observant. While little stigma is attached to induced abortion in our society, there may well be some degree of under-reporting and these figures probably underestimate the prevalence of previous induced abortion among pregnant women. More-
over, since women who have induced abortions may tend to avoid further pregnancies, or at least, not allow them to continue, these figures will fall considerably short of the true frequency of induced abortion in our population. MATERIAL AND METHODS
The design, scope and methodology of the ongoing Jerusalem Perinatal Study have been described in detail in previous publications (16). Briefly, for every baby born in Jerusalem a file is opened, containing information from the birth-certificate and details of the delivery abstracted from labor-ward records. Copies of all death certificates are received from the Ministry of Health and a record-linkage technique continuously updates the file with this information as well as that on congenital malformations and hospital inpatient morbidity collected from obstetric departments, mother and child health clinics and hospital pediatric departments. The study covers all 6000-7000 annual births to residents of West-Jerusalem and its rural hinterland (the pre-1967 Israeli sector). All births take place in hospital and the high rates of utilization of mother and child health clinics (17) and hospital inpatient services (18) make it probable that significant malformations are quickly and accurately recorded. For a limited period in 1966-1968, it was possible to interview pregnant women, usually at their first antenatal visit (19), and this information forms the basis of the present study. The interview cohort is derived from women attending municipal mother and child health stations or hospital clinics for their antenatal care. Interviewing was not carried out concurrently in all stations, and mothers attending certain private clinics were not reached, so that the interviewed cohort has certain biases which are summarized in table 1. Statistical techniques. Because there are so many possibly confounding variables to be taken into account, we have chosen to
LATE SEQUELAE OF INDUCED ABORTION TABLE 1
Proportion of mothers interviewed in pregnancy, by birth order, mother's age, birthweight, infant mortality and malformations % having antenatal Birth order 1 2-3 4-6 7+
66.3 62.7 66.9 61.0
Mother's age < 19 20-29 30-39 > 40 '
71.5 65.6 61.7 59.2
Birthweight < 2.5 kg > 2.5 kg
55.8 65.0
Stillbirths Early neonatal deaths Late neonatal deaths Post-neonatal deaths
62.3 41.5 51.4 67.3
Babies with major malformations Babies with minor malformations
64.8 69.5
Total % of mothers interviewed
64.3
use multiple regression analysis to estimate differences in pregnancy outcome between women with and without previous induced abortion. Despite the absence of certain theoretical requirements, regression methods seem the most appropriate to achieve our objective of an overall survey summarizing possible complications of labor in births following induced abortions. For each outcome of pregnancy (death, birthweight, complication) a stepwise regression analysis was done, forcing the variable for the number of previous abortions into the equation at the first step and continuing to incorporate other predictor variables until none remained with a residual effect significant at the 5 per cent level. From the resulting regression equations, standardized rates were calculated for each outcome, in women with or without previ-
219
ous induced abortions. These rates are standardized to the "average woman" of this study, the mother for whom each of the other predictor variables takes its mean values. The rates are given by the equations: Y = A + BaXa + 2 BtXi for births following induced abortions, and Y = A + 2 BtXi for births without previous induced abortions, where Y = probability of each outcome of pregnancy, A = regression constant, Ba = regression coefficient for induced abortions, Xa = the mean number of previous abortions in women with a positive history of abortions, while Bt and Xt are the regression coefficients and the mean values of each of the significant other variables in the regression equation. For each outcome, crude rates are presented with standardized rates and significance levels which refer to thefc-testsof the regression coefficients (Ba) for induced abortions. A detailed account of how each variable was coded, and the regression equations for each outcome of pregnancy are available from the authors. The following variables were tested in each regression analysis where appropriate: Outcome: Stillbirth; early, late and postneonatal death; major and minor malformations (defined in previous study (20)); birthweight and low birthweights of l No previous induced previous abortions abortions
Stillbirth rate/1000 Early neonatal deaths/1000 Late neonatal deaths/1000 Post-neonatal deaths/1000 Major malformationsVlOOO Minor malformationsVlOOO Birthweight 1 previous abortions
Significance*
3.8
3.6
6.7
3.6
6.5
p . 001
6.6
6.2
11.8
6.2
11.9
p.001
4.2
4.1
5.5
4.0
5.8
p.05
0.1 0.2 1.4 0.8
0.2 0.2 1.5 0.8
0.1 0.4 0.9 1.3
0.2 0.2 1.5 0.8
0.1 0.2 1.0 1.1
ns ns ns ns
11,057
10,305
752
10,305
752
* See material and methods.
count by regression analysis, there is no significant difference in the resulting standardized rates between mothers with and without induced abortions. Sex ratio, which is tested as an indicator of a possible teratogenic process, is unchanged, following induced abortions. However, both major and minor malformations (defined in a previous study (20)) are increased in the abortion group, and the differences are unlikely to be due to chance. The same is true for rates of low birthweight of less than 2.5 kg and less than 2.0 kg. The increase in risk is relatively greater for the very small babies, the great majority of whom are true prematures, i.e. pre-term births. Mean "birthweight is not significantly altered in the abortion group; the small decrease of 10-15 gm is easily explained by the increased proportion of very small births. There is therefore no evidence, in this population, of growth retardation in the birth following induced abortion. Table 3 shows the findings for firsttrimester bleeding and other complications of pregnancy. Women reporting previous induced abortions at their antenatal interview also reported first-trimester bleeding with significantly greater frequency. Other pregnancy complications, subsequently re-
corded in labor-ward books at the time of the birth, were not significantly increased in the abortion group, when other variables were taken into account by regression analysis. Crude rates of rhesus incompatibility and hydramnios were substantially higher in the abortion group, but the differences disappeared after controlling for confounding variables. Among the complications of labor recorded by midwives and obstetricians (table 4), placenta previa was more frequent (crude rate) in births following induced abortions. Again, after standardization for other variables, there were no significant differences between the abortion group and the controls, for this and other complications of labor. Mothers with previous induced abortion tended to have fewer normal deliveries and more interventions in the third stage of labor (table 5). The crude rate of cesarean sections was higher in the abortion group, but not after standardization. DISCUSSION
From the public health point of view, the most significant finding of this prospective study is the excess of neonatal deaths and low birthweight in births following induced abortion. While the probability of death is
222
HARLAP AND DA VIES TABLE 4
Complications of labor, by previous induced abortion Crude rates (%) Complication of labor
Breech Premature rupture of membranes Placenta previa Placental abruption Cord prolapse Other cord anomalies Fetal distress/asphyxia Post-partum hemorrhage Total singleton births
Total No previous interviewed induced cohort abortions
Standardized rates (%)* >1 previous abortions
No previous induced abortions
>1 previous abortions
2.9 2.1 0.4 0.3 0.4
2.9 2.0 0.3 0.4 0.4
2.9 2.5 0.8 0.1 0.1
2.9 2.1 0.4 0.4 0.4
2.8 2.0 0.5 0.1 0.2
12.3
12.3
12.2
12.3
12.3
5.5 1.3
5.5 1.3
5.5 1.1
5.5 1.3
6.2 1.3
11,057
10,305
752
10,305
752
Significance'
ns ns ns ns ns ns ns ns
* See material and methods. TABLE 5
Interventions in labor, by previous induced abortions iCrude rates
Intervention in labor
No intervention (normal delivery) Induction Forceps or vacuum Cesarean section Intervention in 3rd stage (manual removal of placenta) Total singleton births
Standardized rates (%)*
(%)
Total No previous interviewed induced cohort abortions
>1 previous abortions
No previous induced abortions
Significance* >1 previous abortions
82.1
82.3
79.9
p < .05
8.2 5.2 3.4 1.9
79.1 8.6 4.4 5.2 2.9
82.2
8.3 5.1 3.5 2.0
8.2 5.2 3.4 1.9
9.1 4.7 3.8 3.0
ns ns ns
11,057
10,305
752
10,305
752
p
