GESTATION, BIRTH-WEIGHT, AND SPONTANEOUS ABORTION IN PREGNANCY AFTER INDUCED ABORTION

142 INTRODUCTION

Hospital

THE

Practice

concern

GESTATION, BIRTH-WEIGHT, AND SPONTANEOUS ABORTION IN PREGNANCY AFTER INDUCED ABORTION

Report of Collaborative Study by W.H.O. Sequelæ of Abortion*

Task Force

on

of pregnancy in 7228 from eight European cities was studied. In two of the three city clusters, there was a significantly higher risk of adverse outcome in terms of either mid-trimester spontaneous abortion, preterm delivery, or low-birth-weight infant among women whose only previous pregnancy had been surgically terminated than among primigravidæ or women whose only previous pregnancy had ended in live birth. In one city cluster in which surgical termination was accomplished both by conventional dilatation and curettage and by vacuum aspiration (V.A.), an increased risk of short gestation was noted for V.A., but the overall total risk of adverse outcome was not significantly increased. In the third city cluster, in which surgical termination was nearly entirely by V.A., induced abortion was not associated with any increased risk of adverse pregnancy outcome. The effects of spontaneous abortion on the subsequent pregnancy are similar to those of induced abortion.

Summary

The

of legalised abortion has led to increasing about its possible long-term effect. Reports from Hungaryl2 suggested that widespread induced abortion had reduced the mean birth-weight, the mechanism being incompetence of the cervix and premature onset of use

outcome

women

labour. Studies in Britain3 and Israel4 confirmed the Hungarian findings but those in Japan,Taiwan,6 Yugoslavia,’ U.S.A.,8 and Holland9 failed to do so, perhaps because of differences in the characteristics of women who seek abortion-age at childbearing, parity, attitudes towards contraception, and duration of pregnancy at the time of termination-and in the operative technique. It has also been suggested 10 that induced abortion may increase the risk of spontaneous abortion in the succeeding pregnancy. In an attempt to overcome some of the difficulties and the deficiencies of earlier studies the W.H.O. Task Force on the Sequelæ of Abortion designed a collaborative study between nine cities in eight countries. The data in this report are from the eight European cities only.

PATIENTS

A history was taken from 33 188 patients at their first prenatal attendance in accordance with an approved protocol. The data were transferred to the International Computing Centre, W.H.O., Geneva, for consistency checks. The study population was group 1 (the index group, the last pregnancy ended in an induced abortion); group 2 (the last pregnancy ended in a spontaneous abortion or stillbirth); group 3 (the last and group 4 (first pregpregnancy ended in a live nancy). Groups 3 and 4 were stratified samples matched with group 1 for age, gravidity, and social class. The 12 813 women in the study population were interviewed a second time at about 32 weeks’ gestation: samples were taken (derivation in the figure and in table i) to reduce the extent of the interviewing and data processing for the control groups 3 and 4 but matched pairs were not constructed. The techniques used for the original induced abortion for all patients in the index group (obtained from the clinical records or by postal questionnaire) are shown in table i. This paper deals only with vacuum aspiration (v.A.) and conventional dilatation and curettage (D&C), the two most widely used methods for first-trimester abortion. In Lodz and Warsaw information on technique was available only for 8.2% and 91%

birth);

working group for this investigation was: Investigators: (Copenhagen, Denmark) DR M. MADSEN, DR E. OBEL, DR E. OSTERGAARD, DR J. PHILIP; (Helsinki, Finland) DR O. KARJALAINEN, DR M. MANDELIN; (Debrecen, Hungary) DR L. LAMPE; (Jerusalem, Israel) DR A. M. DAVIES†; (Warsaw, Poland) DR F. SZCZOTKA, DR H. WIOR; (Seoul, Republic of Korea) DR S.-B. HONG; (Stockholm, Sweden) DR G. AF GEIJERSTAM; (Newcastle-upon-Tyne, U.K.) DR S. L. BARRON,‡ DR J. K. RUSSELL; (New York, U.S.A.) DR C. TIETZE,† DR A. O. VARMA,† (Ljubljana, Yugoslavia) DR L. ANDOLSEK, MRS M. OGRINC-OVEN, DR M. POMPE; (W.H.O., Geneva, Switzerland) DR M. *The

A. BELSEY†, DR K. Y. RUSSELL.

EDSTROM, MRS P. HEINER, MISS K. KINNEAR, MRS

†Working group for this publication. ‡Working group drafter.

TABLE I-DATA FRAME

O.P.P.:

only previous pregnancy.

143

tiple or complicated by diabetes or by an intrauterine contraceptive device in situ, if gestation was longer than 46 completed weeks, or if gestation at recruitment was longer than 42 completed weeks. This analysis was restricted to cases where menstrual dates were reliable (89.7% of the total) and, for birth-weight, where the pregnancy resulted in a live birth. It was not possible to study the frequency of spontaneous abortion in the first 3

Derivation of study groups. o.p.p. only previous pregnancy;

N.P.P. no

previous pregnancy.

of cases, respectively, but D&c is almost the only method used in these cities and has been assigned to all cases for which the method could not be ascertained. Legal limitations and local customs determined the choice and method used and only in Helsinki, Ljubljana, and Stockholm were both v.A. and D&c used in adequate numbers. For the remaining five cities, the predilection for one method is limited to Debrecen, Lodz, means that evaluation of D&c and Warsaw and for v.A. to Copenhagen and Newcastle. The outcome of pregnancy for all patients was transcribed from the hospital records on to precoded forms. The results of the validation procedures in Copenhagen, Debrecen, and Stockholm have already been described.’1

METHODS

This paper, the first of a series, examines only the hypothesis that termination of a first pregnancy affects the outcome of the succeeding pregnancy by increasing the frequency of low birthweight, short-gestation infants, and mid-trimester spontaneous abortion. Although mean birth-weight and gestation have been examined,12 small changes in the mean can overshadow changes in the frequency of low birth-weight and we believe that the use of a dichotomy which examines the rate of lowbirth weight for the live infant at risk is more useful to the clinician. The choice of suitable control groups posed problems. For comparison we needed not only women pregnant for the first time but also women whose previous pregnancy ended in a live birth. A third group, whose last pregnancy ended in spontaneous abortion, was added in order to compare the effect of spontaneous with induced abortion. Five groups, three of them controls, were constructed according to the previous pregnancy experience as follows:

(1) only previous pregnancy (o.P.P.) was an induced abortion by D&C (1088); (2) o.p.p. was an induced abortion by v.A. (555); (3) o.p.p. was a spontaneous abortion (1372); (4) o.p.p. was a live birth (2186); (5) no previous pregnancy (2027). Cases were excluded if the current pregnancy ended in an induced abortion or maternal death, if the pregnancy was mul-

months of pregnancy because of the wide variation in the gestation at the time of recruitment. By 14 weeks, however, about half our patients had been recruited and spontaneous abortion between 14 and 27 weeks gestation could be examined. We originally intended ’to pool the results from the eight European cities but differences in smoking habits, gestation at recruitment, and distribution of birth-weight made total pooling unacceptable. Some pooling was necessary to obtain adequate numbers for analysis. The cities were therefore grouped in three clusters determined by their predilection for either D&C or V.A. as the method of induced abortion (table II): in cluster A (Debrecen, Lodz, and Warsaw) and in cluster B (Copenhagen and Newcastle) D&c and v.A., respectively, were used in almost all cases; in cluster C (Helsinki, Ljubljana, and Stockholm) both methods were used approximately equally. To take account of the multiple factors known to affect birth-weight and gestation, we used the Mantel-Haenzel summary chi-square and a general linear interaction model which provided a series of values of X2 with one degree of freedom. (Further details of the statistical analysis may be obtained from the authors). The results from the general linear interaction model were similar to those using Mantel-Haenzel summary chi-square, but only the latter are presented in this report. RESULTS

preliminary analysis showed that the reduction in birth-weight associated with cigarette smoking in pregnancy varied from 120 to 147 g according to group. There was a corresponding increase in the frequency of low birth-weight. Low birth-weight (<2501 g) within the three city clusters (without taking into account any of the background variables known to affect birthweight) was more frequent in city cluster A than in B and C, and more frequent in all three clusters (except The

mean

TABLE

II-FREQUENCY

OF LOW BIRTH WEIGHT

SINGLETON LIVE BIRTHS

(<2500g)

FOR

144 for o.p.p. D&c in cluster C) after induced abortion than in primigravidae (table n). Once allowance had been made for variables known to affect birth-weight (maternal age and height and smoking at 32 weeks gestation) and gestation (smoking and maternal age as before and gestation at booking), in the city cluster A, the 9.5% rate of low birth-weight in the induced abortion group and the 9.9% rate in the o.p.p. spontaneous abortion group were significantly higher than the rate in the o.p.P. live birth (5-5%) At least part of this difference can be explained by the effect of birth rank. The differences were not significant in the other two city clusters. TABLE

III-FREQUENCY

OF SHORT GESTATION

(<258

TABLE

V-FREQUENCY OF ADVERSE OUTCOME ABORTION, SHORT GESTATION,

SPONTANEOUS

WEIGHT WITH NORMAL

(MID-TRIMESTER AND LOW BIRTH

GESTATION).

DAYS) FOR

SINGLETON LIVE BIRTHS

TABLE

IV-FREQUENCY

OF MID-TRIMESTER SPONTANEOUS

ABORTION

Short gestation was more common in the control groups within cluster A (Debrecen, Lodz, and Warsaw) than in controls in clusters B or C (table III). In city cluster C, the short gestation rate of 5.7% in the o.p.p. induced abortion (v.A.) and 64% in the o.p.p. spontaiieous abortion were significantly higher than the rate of 20% in the o.p.p. live birth group. No significant differences were found between the other groups in city cluster C nor in city clusters A and B with respect to rate of short gestation. Spontaneous abortion after induced abortion was more frequent in city cluster A (by implication, after D&c) than in clusters B or C. Within cluster A (East European cities) mid-trimester spontaneous abortion in the o.P.P. D&c group was more common than in the o.p.p. live birth group or in primigravidae (table iv). No differences were noted between groups in the clusters B or C. When the three adverse outcomes were summarised for women who had been registered by the 20th week of gestation, only in city cluster A were the differences still

statistically significant (table v).

DISCUSSION

Women who have had an induced abortion are not a random sample of the population,13 but the degree to which they deviate from the norm varies with the attitude of their society. In general, our findings suggest that they were more likely to smoke, to have had failed contraception, and to be uncertain of menstrual dates when compared with primigravidx of similar age or with women who had had a previous spontaneous abortion. All these social factors tended to vary from city to city, although the trends were consistent. The advantage

145 of using a multicentre study was the rapidity with which data was collected, the wide variety of experience, and the likelihood that its results would prove more acceptable to those who make medical policy. The diversity of social variables within the eight cities was too great to allow all the data to be pooled. We compromised by pooling cities according to the pattern of the method of abortion used. D&C was used almost universally in the three cities in Hungary and Poland (cluster A).. of short gestation and of low birthboth higher in city cluster A than in either weight of the others and this extended to all groups including those previously terminated by D&c. The observation is of considerable importance since much of the early literature which suggested that abortion is followed by premature birth and mid-trimester abortion came from Hungary. Certainly, mid-trimester abortion appears to be more common in pregnancies which follow D&c than The

frequency

were

in primigravidx among patients in city cluster A, even when allowance is made for factors such as smoking, gestation, booking, and maternal stature (the effect of social class was neutralised by the stratifying sampling procedure). It is not possible, without a prospective study, to measure the effect of the abortion technique but simple regression lines for dilatation of the cervix against gestation at the time of abortion 10 suggest that, in the case of D&c, the extent of dilatation does not vary much with the length of gestation and tends to be around 12 mm. Induced abortion by v.A. was associated with an increased risk of a short gestation in city cluster C. The apparent lack of an effect of D&C on short gestation and low birth-weight, in contrast to the demonstrated effect of v.A., might be explained by an increased risk of midtrimester spontaneous abortion in the D&C group. The data suggest that the adverse effect of D&C may be expressed in terms of mid-trimester spontaneous abortion, whereas if there is an adverse effect of vacuum aspiration it may not become manifest until the last trimester of pregnancy as a preterm delivery and lowbirth-weight infant. This discrepancy of the time in gestation for the expression of an effect also may explain the failure to demonstrate an adverse effect of induced abortion in other studies which have examined birth certificates or hospital records of births only, or in which case recruitment was late in the course of the pregnancy. In all three centres, the effect of spontaneous abortion on the next pregnancy appears to parallel those of induced abortion: In summary, it cannot be concluded that either spontaneous or induced abortion is entirely harmless in terms of the effects on the next pregnancy, although at least in one city cluster, where nearly all procedures performed were by v.A., no effect could be observed, There are, of course, many questions yet unanswered and a series of other studies sponsored by W.H.O. will enquire into the sequelae of prostaglandininduced abortion, vacuum aspiration vs D&c, the effect of multiple abortion, and the occurrence of secondary

infertility. Requests for reprints should be addressed W.H.O., 1211 Geneva, Switzerland

to

Dr M. A.

Belsey,

Occasional

Survey

DISCRIMINATION BETWEEN IRON-DEFICIENCY AND

HETEROZYGOUS-THALASSÆMIA SYNDROMES IN DIFFERENTIAL DIAGNOSIS OF MICROCYTOSIS

J. M. ENGLAND Department of Hœmatology, St. Mary’s Hospital Medical School, London W2 PATRICIA FRASER

Department of Medical Statistics and Epidemiology, London School of Hygiene and Tropical Medicine, London WC1 Three formulæ have been advocated for interpreting the blood-count when differentiating between iron deficiency and heterozygous discriminant thalassæmia: the function, M.C.V.-R.B.C.-(5×Hb)-k, the M.C.V/R.B.C. ratio, and the (M.C.V.)2×M.C.H. product. In a comparison of their diagnostic accuracy in microcytic adults from several countries, the discriminant function was the most satisfactory. 417 of 455 uncomplicated cases (91·6%) of iron deficiency, heterozygous &agr;1- and &bgr;- thalassæmia were correctly identified. The function was not able to distinguish heterozygous &agr;2-thalassæmia from iron deficiency. A flow chart illustrates how the discriminant function can be used in hæmatological practice.

Summary

INTRODUCTION

SEVERAL formulae have been proposed for distinguishing the microcytosis of iron deficiency from that of heterozygous thalassaemia. England and Fraser’ introduced a simple linear discriminant function derived from the mean cell volume (M.c.v. in fl), red-blood-cell count (R.B.C.X 1012/1), and haemoglobin concentration (Hb in g/dl). The function was of the form M.c.v.-R.B.c.-(5xHb)-k, where k is a constant determined by the method used to calibrate the Coulter Counter:2 positive values for the discriminant function indicate iron deficiency and negative values heterozygous p-thalassaemia. Mentzer3 suggested that an even simpler index, M.C.v./R.B.C., was equally capable of distinguishing the two conditions, values below 13 indicat-

1. Klinger, A. Ther. Umsch. med. Biblphie, 1970, 27, 681. 2. Czeizel. A., and others, Br. J. prev. soc. Med. 1970, 24, 146. 3. Richardson, J. A., Dixon, G. Br. med. J. 1976, i, 1303. 4. Harlap, S., Davies, A. M. Am. J. Epidemiol. 1975, 102, 217. 5. Muramatsu, M. Bull. Inst. pub. Hlth, 1972, 21, 127. 5. Muramatsu, M. Bull. Inst. pub. Hlth, 1972, 21, 127. 6. Daling, J. R., Emanuel, I. Lancet, 1975, ii, 170. 7. Hogue, C. J. Am. J. Obst. Gynec. 1975, 123, 675. 8. Daling, J. R., Emanuel, I. New Eng. J. Med. 1977, 297, 1241. 9. Van der Slikke, J. R., Treffers, P. E. Br. med. J. 1978, i, 270. 10. Wright, C. S. W., Campbell, S., Beazley, J. Lancet, 1972, i, 1272. 11. W.H.O. Task Force on Sequelæ of Abortion (Belsey, M. A.) in Proceedings of the P.A.R.F.R. International Workshop and Postgraduate Course on Pregnancy Termination: Methods, Safety, and New Developments. Nassau, Bahamas, 23-26 May, 1978 (in the press). 12. W.H.O. Task Force on Sequelæ of Abortion (Barron, S. L.) ibid. (in the

press). 13.

Harlap, S., Davies, A. M. Bull. Wld

Hlth Org.

1975, 52, 49.