AMERICAN JOURNAL OF EPIDEMIOLOGY Vol. 123, No. 4 Copyright C1986 by The Johns Hopkins Univenity School of Hygiene and Public Health Printed in U.SA. All rights reserved

LOW IN FOLLOWING INDUCED ABORTION: NO EVIDENCE FOR AN ASSOCIATION1

MICHAEL B. BRACKEN, KAREN G. HELLENBRAND, THEODORE R. HOLFORD, AND CAROL BRYCE-BUCHANAN

Bracken, M. B. (Dept of Epidemiology and Public Health, Yale U. School of

Medicine, New Haven, CT 06510), K. G. Hellenbrand, T. R. Hotford, and C. Bryce- Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 Buchanan. Low birth weight In pregnancies following induced abortion: no evi- dence for an association. Am J Epidemiol 1986; 123:604-13. Compared with women delivering a first , those delivering a second pregnancy after aborting the first have similar rates of low (<2,500 g) birth weight newboms (relative risk (RR) G2A1/G1 = 0.86, 95% confidence interval (Cl) = 0.49-1.51) and mean birth weight (A = 16.3 g, p = 0.63). Abortion of the first pregnancy prevents the reduction tn low birth weight and increase in mean birth weight in the second pregnancy which delivery of the first pregnancy normally bestows (RR G2P1/G2A1 = 0.48, 95% Cl = 0.25-0.90; A = 135.3 g, p < 0.0001). Two prior induced abortions do not significantly increase risk for low birth weight (RR G3A2/G1 = 1.14, 95% Cl = 0.37-3.56) or decrease mean birth weight (A = 29.0 g), compared with women delivering their first pregnancy. The second of two deliveries has a reduced risk of low birth weight irrespective of whether both deliveries follow an aborted first pregnancy. Adjustment for confounding factors did not materially change these results. Low birth weight rates were higher after abortions performed in hospital compared with elsewhere (p = 0.03), but mean birth weight was not affected. Gestation at abortion, vacuum aspiration or dila- tation and curettage, and abortion complications were unrelated to birth weight of subsequent pregnancies. Pregnancies conceived within six months of a prior abortion or delivery had lower birth weight than if the antecedent pregnancy ended more than six months previously.

abortion, induced; , low birth weight

Low birth weight has been implicated as etiology of low birth weight is generally one of the single most important predictors unknown, although the contributions of of perinatal death (1, 2) and has recently race, young maternal age, primigravidity, been identified as a major research priority and maternal cigarette smoking are reason- by the National Institutes of Health. The ably well established (3). Whether or not prior induced abortion is Received for publication February 6, 1985, and in a riskfacto r forlo w birth weight has been final form June 4,1985. a controversial issue (4-6). If induced abor- ^Z^tT^CZ'ZuctS^ *d tion does increase subsequent risks of low Obstetrics and Gynecology, Yale University School of birth weight, there could be long-term neg- Medicine, New Haven, CT. atiVe effects on the national low birth parEt *£&££& SX'bHc K£N£ weight rate as a result of the increasingly University School of Medicine, 60 College Street, New large number of women delivering at term Haven, CT 06510. after experiencing a prior induced abortion

(HDH357). quent low birth weight would be a consid- 604 ABORTION AND LOW BIRTH WEIGHT 605 eration in the personal decision faced by (n = 5,331) agreed to be contacted. Women many women to seek an induced abortion were considered ineligible for interview if (8). they a) planned to deliver somewhere other Because of the well known increase in than at Yale-New Haven Hospital (n = 83); birth weight in second deliveries (3), it is b) demonstrated poor English language important to control for both gravidity and comprehension (n = 107); c) were not preg- parity when examining effects of induced nant at the time that the study interview abortion. Since control of both parameters was conducted (n = 208); or d) were famil- is not possible within one comparison iar with the study (n = 7). In all, 4,926 eligible subjects were contacted regarding

group, the choice of controls is crucial. The Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 present report addresses four research the research. Of these, a number who had questions, the first three given below pro- initially agreed to be contacted subse- posed by Hogue et al. (6) and the fourth quently refused to be interviewed (n = 473) proposed by Bracken (4). or could not be reached to arrange an in- Hypothesis 1: First pregnancy abortions terview (n = 263). Four women whose re- do not increase risk of low birth weight in sponses were considered by the interviewer the second pregnancy compared with deliv- to be unreliable were not analyzed. Thus, ered first pregnancies. valid interviews were completed on 4,186 Hypothesis 2: Abortion, unlike a first (85.0 per cent) of eligible subjects. birth, does not reduce risk of low birth Information concerning pregnancy his- weight among second deliveries. tory, including prior induced abortions, de- Hypothesis 3: Abortion of the first two mographic characteristics, contraceptive pregnancies will not increase risk of low practice, medical history, and exposure to birth weight in first births any more than possible risk factors was obtained by a abortion of the first pregnancy alone. standardized interview schedule adminis- Hypothesis 4: Abortion of the first preg- tered by trained interviewers. The vast ma- nancy will not increase risk of low birth jority of the interviews were conducted in weight in the third pregnancy if the second the women's homes, and a small number at pregnancy was delivered. That is, if abor- the women's places of employment, their tion increases the risk of low birth weight physician's office, or the research office. in subsequent pregnancies, this risk will be The interviews took place within a few only for the pregnancy immediately follow- weeks of the women's first or second pre- ing the abortion. natal visit. Information about the pregnancy's out- METHODS come was obtained by one of three methods. The daily deliveries at Yale-New Haven All women who made a first prenatal visit Hospital were surveyed for the names of to a private obstetrics or midwifery prac- study participants. Data regarding preg- tice, health maintenance organization, or nancy outcome and the condition of the hospital clinic in the greater New Haven newborn were abstracted from the mothers' area between May 12, 1980 and March 12, and ' medical charts. Periodically, 1982 and who anticipated delivery at Yale- participating obstetric practices were con- New Haven Hospital were invited to par- tacted to obtain information about women ticipate in the study. A total of 6,219 women whose names had not appeared on the daily were introduced to the study by their phy- delivery records within 42 weeks of their sician or midwife (without exception, all date of conception. For 76 women, we could physicians and midwives with admitting not ascertain the outcome of their preg- privileges at Yale-New Haven Hospital co- nancy, and 60 women did not deliver at the operated in the research), and 85.7 per cent Yale-New Haven Hospital as intended. Of 606 BRACKEN ET AL. the total number (n = 4,050) of pregnan- sis, we constructed a general linear model cies, 97.2 per cent were livebirths, and of to estimate the effect of prior abortion his- these, 3,891 (98.9 per cent) were singleton tory on birth weight, while adjusting for deliveries. maternal characteristics. To test the study hypotheses, we cate- Because of the inherent correlation of gorized subjects according to their preg- with birth weight, we next nancy history. Five groups were included evaluated abortion history effects on ges- in the analysis: women pregnant for the tational age for the five hypotheses. This first time (Gl: n = 1,194); women pregnant analysis was restricted to the G2A1, Gl, for the second time whose first pregnancy and G2P1 pregnancy history groups be- Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 was aborted (G2A1: n = 286); women preg- cause of their large number of subjects. An nant for the third time whose first two analysis of variance was used to examine pregnancies were aborted (G3A2: n = 58); the association of gestational age with birth women with one prior livebirth (G2P1: n = weight. The multivariate models were rean- 880); and women for whom the index preg- alyzed to include terms for gestational age. nancy was their third pregnancy, the first Characteristics of the abortion procedure ending in an induced abortion and the sec- itself were examined for the G2A1 group. ond in a livebirth (G3A1P1: n = 108). All prior pregnancies could have been multiple. RESULTS The above grouping excludes all women with a history of , stillbirth, ec- Association of maternal characteristics with topic pregnancy, or more than three prior low birth weight pregnancies. Low birth weight is defined as Women were more likely to deliver low <2,500 g and prematurity as <37 weeks birth weight newborns if they were aged 20 gestational age at delivery dated by last or less; not currently married (and espe- menstrual period. cially were separated); were Hispanic or The association of low birth weight with black; not Catholic; did not complete their selected maternal characteristics was deter- high school education; were themselves ex- mined. Next, using the five defined cate- posed to diethylstilbestrol in utero; used gories of pregnancy history, we examined oral contraceptives in the year before con- the association of low birth weight with ception; or used marijuana or cigarettes pregnancy and abortion history. Relative during pregnancy. Spermicide use at con- risks of low birth weight were calculated to ception and moderate consumption of al- test the four hypotheses of interest. The cohol in pregnancy suggested an associa- hypotheses were also tested by performing tion with modest increases in birth weight. t tests on mean birth weight for the five The following variables were not associated categories of pregnancy history. The refer- with low birth weight: average daily caf- ence group for both of these tests was feine use in pregnancy; whether or not cur- women with one previous abortion (G2A1). rently employed; history of infertility in Using the same reference group, we exam- year before pregnancy; ever any obstetric ined the association of pregnancy and abor- or gynecologic surgery; or any injury since tion history with selected maternal vari- becoming pregnant. Also unrelated to low ables. birth weight were use of the intrauterine To test the effect of prior induced abor- device in year before pregnancy and use of tion on the risk for delivering a low birth rhythm. Women interviewed in the first weight newborn, we fit a log-linear model trimester (which, because of the study de- to the four hypotheses, adjusting simulta- sign, reflects being a private rather than a neously the more important maternal char- clinic patient) were less likely to deliver low acteristics. In a second multivariate analy- birth weight newborns. ABORTION AND LOW BIRTH WEIGHT 607 Association of pregnancy and abortion having delivered their first have a signifi- history with low birth weight cantly decreased risk of low birth weight when compared with women in their second Table 1 shows the low birth weight rates pregnancy who aborted their first (RR = in the five pregnancy groups of interest, 0.48,95 per cent CI = 0.25-0.90). The mean and table 2 shows their mean birth weight birth weight of the G2A1 group is 135.3 g comparisons. The following observations lighter than that of the G2P1 group (table can be made concerning the risk of low 2). As expected, however, women with one birth weight after induced abortion. prior delivery (G2P1) have a substantially

Hypothesis 1, G2A1 vs. Gl. Women deliv- reduced risk of a low birth weight newborn Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 ering their first pregnancy have a risk of in their second pregnancy than women de- low birth weight similar to that of women livering their first (Gl) pregnancy (RR = delivering their second pregnancy after 0.55 for G2P1/G1, 95 per cent CI = 0.34- having aborted their first (relative risk 0.89). Correspondingly, the Gl newboms (RR) = 0.86,95 per cent confidence interval are, on average, 119 g lighter than the G2P1 (CI) = 0.49-1.51). In addition, the mean newborns (t = 5.31, df = 2,072, p < 0.0001). birth weight for these two groups differs by Thus, the initial bivariate analysis of hy- only 16.3 g. potheses 1 and 2 indicates that abortion of Hypothesis 2, G2A1 vs. G2P1. Women the first pregnancy does not directly in- delivering their second pregnancy after crease risk of low birth weight in the first

TABLE 1 Description of pregnancy and abortion history and association with low birth weight*

Low birth Relative rule 95% Pregnancy-abortion history Labelt Total no. weight (%) for G2AU confidence interval

1 prior pregnancy—aborted G2A1 286 5.24 No prior pregnancies Gl 1,194 4.52 0.86 0.49-1.51 1 prior pregnancy—delivered G2P1 880 2.50 0.48 0.25-0.90 2 prior pregnancies—both aborted G3A2 58 5.17 0.99 0.29-3.30 2 prior pregnancies—first aborted, second delivered G3A1P1 108 2.78 0.53 0.16-1.75 All others 1,331 5.03

Total 3,857 4.44 * Birth weight not ascertained for 34 newborns. t The label takes into account the present (index) pregnancy. t Using different reference groups.

TABLE 2 Mean birth weight by abortion and pregnancy history* Birth weight (g) Label Mean Standard deviation G2A1 286 3,314.2 539.5 t t Gl 1,194 3,330.5 510.6 0.480 0.632 G2P1 880 3,449.5 496.1 3.922 0.0OO1 G3A2 58 3,359.0 495.7 0.585 0.559 G3A1P1 108 3,428.8 474.9 1.941 0.053 • F = 8.40, df = 4, p = 0.0001. t Reference category. 608 BRACKEN ET AL. delivery. Abortion of the first pregnancy, sociated with low birth weight, must be however, appears to prevent the normal adjusted to ensure that it does not account reduction in risk of low birth weight to the for the higher low birth weight rate in the second pregnancy that delivery of the first G2A1 group versus the G2P1 group. pregnancy usually bestows. Hypothesis 3, G2A1 vs. G3A2. Low birth Multivariate analysis weight in the third pregnancy, after abor- The results of the multivariate analysis tion of the first two, is almost identical to are shown in table 3. Adjustment for ma- that in women who aborted their first preg- ternal variables does not influence the con- nancy and are delivering their second (RR clusions drawn from the bivariate analysis. Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 = 0.99). Additionally, the mean birth The Gl newborns are 22.8 g heavier than weight of the G3A2 group is only 44.8 g the G2A1 newborns, which is not a signifi- heavier than that of the G2A1 group (table cant difference. As before, the G2P1 new- 2). Thus, two prior induced abortions do borns are heavier than the G2A1 newborns not appear to increase the risk of low birth (by 129.1 g) and have a 0.57 decreased risk weight any more than does one prior abor- of low birth weight. Women with two prior tion. Compared with primigravidas, the risk abortions have somewhat heavier newborns of low birth weight after two abortions in- (39.6 g) than women with a single prior creased by only 14 per cent (RR = 1.14 for abortion. An intervening delivery after in- G3A2/G1, 95 per cent CI = 0.37-3.56), and duced abortion (G3A1P1) leads to a further mean birth weight decreased by 29.0 g (t = increase in mean birth weight (72.0 g). 0.42, df = 1,250, p = 0.68). Hypothesis 4, G2A1 vs. G3A1P1. An in- Influence of gestational age tervening delivery after induced abortion The association of gestational age with almost halves the risk of subsequent low birth weight for each of the three pregnancy birth weight and results in a mean birth history groups was examined. The expected weight increase of 114.6 g (table 2). The increase in birth weight for longer gesta- rate of low birth weight and the mean birth tional ages is found for each pregnancy weight of the G3A1P1 group are similar to group (F = 128.7, p < 0.0001). A similar those of the G2P1 group. Abortion of the distribution of birth weight for the G2A1 first pregnancy, therefore, appears to have and Gl subjects in the more mature new- no effect on birth weight of the third preg- borns was noteworthy; for example, there nancy if an intervening delivery has oc- was a difference of only 8.8 g in the gesta- curred. tional age group 39+ weeks but a much larger difference of 192.5 g in the gesta- Association of pregnancy and abortion tional age group 30-36 weeks. These differ- history with maternal characteristics ences were further increased for the G2A1 Some maternal characteristics may have versus G2P1 comparisons (117.7 g and influenced the association of pregnancy 477.1 g, respectively). There was no inter- and abortion history with low birth weight. action of pregnancy history with gesta- For example, the Gl group was signifi- tional age on mean birth weight (F = 1.72, cantly more likely than the G2A1 group to p = 0.142). be aged less than 20 years. Since younger The multivariate models shown in table age is itself associated with low birth 3 were reanalyzed to include terms for ges- weight, any effect of the prior induced abor- tational age, as well as an interaction term tion on reducing birth weight might be for the association of birth weight and ges- missed if age were not adjusted. Alterna- tation with pregnancy history. This analy- tively, the G2A1 group contains a larger sis did not materially influence the mean proportion of cigarette smokers than the birth weight results already described G2P1 group. Cigarette smoking, itself as- (G2A1 was -8.3 g compared with Gl and ABORTION AND LOW BIRTH WEIGHT 609

TABLE 3 Effect of selected variables on estimated mean birth weight gain or loss and on risk for low birth weight* Standard Estimate Adjusted relative 95* Variable error p valuet risk of low birth confidence (8) estimate weight interval} Intercept 3,498.6 160.4 0.0001 Maternal age (<20 years) -33.4 43.1 0.4376 1.00 0.47-2.11 Unmarried +2.5 42.4 0.9530 1.83 0.83-4.00 Black (vs. other) -270.5 44.8 0.0001 6.16 2.82-13.52 Catholic (vs. other) -8.3 24.9 0.7387 0.63 0.34-1.10

Education («12 years) -20.1 28.3 0.4772 1.04 0.58-1.86 Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 Exposed to diethylstilbestrol -391.8 134.1 0.0035 6.61 1.38-31.75 Oral contraceptives within 12 months -8.7 27.0 0.7466 1.48 0.90-2.45 Spermicide at conception +3.0 32.2 0.9256 0.77 0.39-1.53 Cigarettes in pregnancy -139.0 27.0 0.0001 1.69 1.01-2.82 Marijuana in pregnancy -61.1 41.2 0.1380 1.22 0.60-2.47 Alcohol in pregnancy +26.5 27.3 0.3308 0.93 0.54-1.61 >150mg of +6.9 32.2 0.9256 0.95 0.63-1.78 Interviewed at >12 weeks gestation +8.5 25.8 0.7401 1.22 0.70-2.10 G2A1 vs. Gl +22.8 40.4 0.5728 0.94 0.45-1.20 vs. G2P1 +129.1 41.6 0.0020 0.57 0.25-1.30 vs. G3A2 +39.6 81.3 0.6260 1.03 0.26-4.08 vs. G3A1P1 +72.0 64.2 0.2618 0.45 0.09-2.16 * Low birth weight = <2,500 g. t Using type III sums of squares.

-116.1 g compared with G2P1). There con- TABLE 4 tinued to be no interaction of pregnancy Mean birth weight by type of procedure for G2A1 history and gestational age on decreasing pregnancy group mean birth weight after adjustment for ma- Birth weight (g) ternal factors (F = 1.99, p = 0.09). Risk of Abortion procedure n* Mean Standard low birth weight, when adjusted for gesta- deviation tional age, also remained essentially un- Menstrual extraction 5 3,534.0 1,205.1 changed (G2A1 vs. Gl, RR = 1.00; G2A1 Vacuum aspiration 201 3,315.8 504.0 vs. G2P1, RR = 0.54, 95 per cent CI = 0.20- Dilatation—curretage 42 3,336.4 615.9 1.44). The effect of prior induced abortion Dilatation—evacuation 2 3,015.0 134.4 on prematurity itself was examined in a Instillation—saline 8 3,386.2 676.5 logistic regression analysis. The indepen- Instillation—unknown agent 5 2,712.0 293.7 dent variables were those shown in table 3. * Type of procedure unknown by 23 respondents. The relative risk of prematurity in the G2A1 group compared with the Gl group by dilatation and curettage (16.0 per cent) was 0.95 (95 per cent CI = 0.53-1.71) and, and saline instillation (3.0 per cent). Be- compared with the G2P1 group, 1.05 (95 cause of small numbers, we can reliably per cent CI = 0.57-1.93). compare only vacuum aspiration with di- latation and curettage. The rate of subse- Characteristics of the abortion procedure quent low birth weight is almost identical To examine characteristics of the abor- (5.0 and 4.8 per cent, respectively), as is tion procedure on subsequent birth weight, mean birth weight (3,315.8 g and 3,336.4 g, we made comparisons within the G2A1 respectively, t = 0.232, df = 241, p = 0.817). group (table 4). Most abortions were by Only eight women had saline instillation; vacuum aspiration (76.4 per cent), followed none of these had a low birth weight child, 610 BRACKEN ET AL. and their mean birth weight of 3,386.2 g 284, p = 0.42). Ten women (3.5 per cent) was somewhat heavier than that following suffered a pelvic or genital tract infection first-trimester abortions. after their abortion. None delivered a low The abortion facility (private clinic, hos- birth weight newborn, however, and no sig- pital, or physician's office) did not influ- nificant differences in mean birth weight ence subsequent mean birth weight (F = were observed compared with women with- 0.84, df = 2, p = 0.43), but the rate of low out infections (3,283.0 g and 3,315.3 g, re- birth weight after hospital abortion (8.6 per spectively, t = 0.19, df = 284, p = 0.85). cent) was greater than that following other Five women required dilatation and curet- abortions (2.9 per cent). Adjustment for tage to treat an incomplete abortion; none Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 maternal race (more blacks had hospital had a low birth weight newborn, and their abortions) did not materially change this mean birth weight was relatively high finding (adjusted RR = 2.66, p = 0.03). A (3,662.0 g). large number of variables were examined to try to identify the characteristics of hos- pital abortion which might be associated Interpregnancy interval with the subsequent increase in low birth Women who conceived their second preg- weight. In all respects but one, the hospital nancy within six months of either deliver- abortions could not be distinguished from ing or aborting the first were at increased abortions performed elsewhere: Hospital risk of delivering low birth weight new- abortions had a higher rate of subsequent borns (table 5). When conception occurred infection (eight of the 10 immediate post- within six months of delivery, there was a abortion infections occurred in the 40.7 per significant drop in mean birth weight cent of abortions performed in hospital). (171.9 g, p = 0.05) compared with longer As shown below, however, infections are interpregnancy intervals. A 145.3-g decline not associated with low birth weight. The in mean birth weight followed conception type of anesthesia given (none, sedatives within six months of an abortion. only, or local) was also unrelated to mean birth weight (F = 1.39, df = 2, p = 0.25). Women aborted under local anesthesia had DISCUSSION a somewhat higher rate of low birth weight The present analysis suggests that in- (9.22 per cent) than women aborted with- duced abortion of the first pregnancy does out anesthesia or using only sedatives (4.0 not increase risk of low birth weight, nor per cent, RR = 2.30, 95 per cent CI = 0.80- does it significantly affect mean birth 6.60). weight of the subsequent pregnancy when Of the G2A1 group, 92.1 per cent aborted compared with first-delivered pregnancies. before 13 weeks gestational age (57.8 per Induced abortion of the first pregnancy, cent before nine weeks and 26.0 per cent however, appears to postpone the improve- before seven weeks). Gestational age at ment in birth weight that would normally abortion was unrelated to risk of either low occur in the second pregnancy after a pre- (xl = 0.21, p = 0.976) or mean birth vious delivery. Figure 1 plots the cumula- weight (F = 0.27, df = 3, p = 0.85). tive birth weight distributions for the Gl, Thirteen of the 286 women in the G2A1 G2A1, and G2P1 pregnancy history groups. group (4.5 per cent) reported hemorrhaging The G2P1 birth weight distributions are either at the time, or within a month, of heavier across the entire weight distribu- their abortion. None of these women had a tion, whereas the Gl and G2A1 birth weight low birth weight newborn, and their mean distributions are virtually indistinguish- birth weights were not different from those able. Figure 1 also shows how selection of not suffering this complication (3,426.9 g the Gl or G2P1 control groups will influ- and 3,308.8 g, respectively, t = 0.77, df = ence conclusions about the impact of a prior ABORTION AND LOW BIRTH WEIGHT 611

TABLE 5 Association of mean birth weight and low birth weight with interpregnancy interval

Birth weight 95% Prior pregnancy Time Low birth Relative risk of n confidence outcome (months) Mean Standard weight (%) low birth weight deviation interval Delivery (G2P1) >6 847 3,456.0 493.6 2.4 * 33 3,284.1t 539.7 6.1 2.57 (0.64-10.25)

Abortion (G2A1) >6 274 3,320.3 539.2 4.7 • <6 12 3,175.0J 552.7 16.7 3.51 (0.90-13.69) Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 * Reference category. 11 •= 1.96, df = 878, p = 0.05. 11 = 0.91, df = 284, p = 0.36.

posed women. Moreover, pregnancies con-

90 ceived within six months of delivery were associated with similar reductions in birth Gl // 8 weight, suggesting that any effect of the £ ° \// short interpregnancy interval is due to a H 70 jji biologic impact of the interval itself rather CC jjl £ 60 than to the induced abortion procedure per l'i UJ II se. Dilatation and curettage did not differ- > 50 Sj entially affect birth weight compared with 5 II ' 3 40 if 1 the more widely performed vacuum aspi- G2A1 jlj ration, a finding which conflicts with some 3 30 earlier Eastern European reports but which /// \ 20 is in agreement with others, primarily from //1 G2PI Japan (see reference 6 for review). Hospi- 10 //' tal-induced abortions were followed by a higher rate of low birth weight, but this 2000 3000 4000 could not be attributed to any differences BIRTH WEIGHT (gms) in abortion technique, complication rate, gestational age at abortion, or maternal FIGURE 1. Cumulative birth weight distributions characteristics. This finding is deserving of for the Gl, G2A1, and G2P1 pregnancy histories. more study. It is possible that women who were terminated in hospital may have dif- induced abortion on subsequent birth fered on variables we did not measure—for weight. Termination of the first two preg- example, preexisting genital infection— thereby putting them at higher risk for nancies did not lead to any increased risk subsequent low birth weight. of low birth weight in the first delivery (the third pregnancy). Abortion of the first preg- The proportion of low birth weight new- nancy had almost no effect on birth weight borns in this study (4.44 per cent for sin- of the third pregnancy when an intervening gletons, 5.48 per cent overall) compares delivery occurred. Abortions complicated with a rate of 6.8 per cent found for the by hemorrhage and infection did not appear United States in 1982 (9). The sociodemo- to influence birth weight in the subsequent graphic characteristics of this population pregnancy. are also similar to those reported for the Conceptions occurring within six months United States (9). From the present data, of abortion had lower birth weight new- we can estimate the proportion of low birth borns, but this was based on only 12 ex- weight first singleton deliveries which can 612 BRACKEN ET AL. be attributed to prior induced abortion. REFERENCES (The actual comparison is for G2A1 + 1. Bakketeig LS, Hoffman HJ, Oakley ART. Peri- G3A2/G1 + G2P1.) Using the formula of natal mortality. In: Bracken MB, ed. Perinatal Miettinen (10) as programmed by Rothman epidemiology. New York: Oxford University and Boice (11), we calculate a rate differ- Press, 1984:99-151. 2. Hoffman HJ, Meirik O, Bakketeig LS. Methodo- ence of 0.016 (95 per cent CI = 0.006- logical considerations in the analysis of perinatal 0.038), indicating that only 1.6 per cent of mortality rates. In: Bracken MB, ed. Perinatal newborns weighing <2,500 g can be attrib- epidemiology. New York: Oxford University uted to the effect of prior induced abortion. Press, 1984:491-530. 3. Albennan E. Low birthweight In: Bracken MB, This low rate is expected in light of the ed. Perinatal epidemiology. New York: Oxford very small relative risks found for low birth University Press, 1984:86-98. Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021 weight after induced abortion and the low 4. Bracken MB. Induced abortion as a risk factor for perinatal complications: a review. Yale J Biol Med prevalence of low birth weight itself. More- 1978;51:539-48. over, our rate difference estimate probably 5. Cates W Jr. Late effects of induced abortion: inflates the total effect of induced abortion hypothesis or knowledge? J Reprod Med 1979; 22:207-12. on all pregnancies since 1) our analysis 6. Hogue CJR, Cates W Jr, Tietze C. The effects of indicates that second deliveries are even induced abortion on subsequent reproduction. Ep- less affected by prior induced abortion, 2) idemiol Rev 1982;4:66-94. 7. Centers for Control. Abortion surveil- the analysis excludes women with multiple lance, 1979-80. Atlanta, GA, 1983. pregnancies and a history of spontaneous 8. Bracken MB, Klennan LV, Bracken M. Abortion, abortion who are at higher risk for low birth adoption or motherhood: an empirical study of decision-making during pregnancy. Am J Obstet weight, and 3) the calculation of a rate Gynecol 1978;130:251-€2. difference does not account for the fact that 9. National Center for Health Statistics. Advance maternal factors shown in our multivaria- report of final natality statistics, 1982. Monthly ble analysis are associated with low birth Vital Stat Rep 1984;23:no. 6. 10. Miettinen OS. Proportion of disease caused or weight and would further reduce the rela- prevented by a given exposure, trait or interven- tive risks due to abortion. It can reasonably tion. Am J Epidemiol 1974;99-.326-32. be concluded, therefore, that prior induced 11. Rothman K, Boice JD. Epidemiologic analysis with a programmable calculator. Washington, DC: abortion had no meaningful effect on the NIH publication no. 79-1649, 1979. low birth weight rate in our study popula- 12. World Health Organization Task Force on the tion. Sequelae of Abortion. Gestation, birthweight, and spontaneous abortion in pregnancy after induced The birth weight and prematurity results abortion. Lancet 1979;l:142-5. 13. Hern WM. Long-term risks of induced abortion. of this study are in close agreement with In: Sciarra JJ, ed. Gynecology and obstetrics. Vol reports from Eastern Europe (12-14), 6. Hagerstown, MD: Harper & Row, 1982;63:l-8. Scandinavia (15-17), Japan (18, 19), and, 14. Schott G, Ehrig E, Wulff V. Prospective studies into pregnancies of primiparae with record of ther- more recently, the United States (20-22). apeutic termination of previous pregnancies or of This growing body of evidence strongly sug- spontaneous abortion and assessment of fertility. gests that first-trimester vacuum aspiration I. First communication. Zentralbl Gynakol 1980;102:932-8. and dilatation and curettage abortions, as 15. Obel E. Pregnancy complications following legally presently performed, do not increase risk induced abortion. Acta Obstet Gynecol Scand for subsequent low birth weight. Multiple 1979^58:485-90. abortion has been less frequently studied 16. Meirik O, Bergstrom R. Outcome of delivery sub- sequent to vacuum aspiration abortion in nullip- (6). Our results indicate that two prior arous women. Acta Obstet Gynecol Scand abortions also do not elevate subsequent 1983;62:499-509. low birth weight risk. Risks due to second- 17. Meirik O, Nygren K-G, Bergstrom R, et al. Out- come of delivery subsequent to induced vacuum- trimester procedures are not well eluci- aspiration abortion in parous women. Am J Epi- dated, although the proportion of abortions demiol 1982;116:415-29. performed in the second trimester (10.0 per 18. Furusawa Y, Koya Y. The influence of artificial abortion on delivery. In: Koya Y, ed. Harmful cent in 1980 (7)) is increasingly rare. effects of induced abortion. Tokyo: Family Plan- ABORTION AND LOW BIRTH WEIGHT 613

ning Federation of Japan, 1966:74-83. 21. Schoenbaum SC, Monson RR, Stubblefield PG, 19. Roht LH, Aoyama H, Leinen GE, et al. The etal. Outcome of the delivery following an induced association of multiple induced abortions with or spontaneous abortion. Am J Obstet Gynecol subsequent prematurity and spontaneous abor- 1980; 136:19-24. tion. Acta Obstet Gynaecol Jpn 1976;23:140-5. 22. Chung CS, Steinhoff PG, Smith RG, et al. The 20. Daling JR, Emanuel I. Induced abortion and sub- effects of induced abortion on subsequent repro- sequent outcome of pregnancy in a series of Amer- ductive function on pregnancy outcome. Hono- ican women. N Engl J Med 1977;297:1241-5. lulu, HI: Hawaii East-West Center, 1983. Downloaded from https://academic.oup.com/aje/article/123/4/604/72671 by guest on 01 October 2021