Journal of Epidemiology and Community Health 1990; 44: 259-264

REVIEW ARTICLE:

Induced abortion as cancer risk factor: a review ofepidemiological evidence

Larissa I Remennick

In spite of the fact that efficient contraception isavailable in many countries, induced abortion isstill a quite prevalent means of birth control,especially in certain subpopulations of women(young, socially deprived, etc). Judging by thepublished statistics, induced abortion rates varydramatically between countries, with the lowestlevel being in West Germany (5-7 per 1000women aged 15-441) and the highest in theRussian Federation of the USSR (123-6 in themid-80s2).Ever since its legalisation in 1955, induced

abortion has been the primary means to reducefertility in the USSR, resulting in 7 116 000terminations in this country in 1986 (comparedwith only about 5 611 000 livebirths2), or 101 2per 1000 women of reproductive age.2 The realprevalence of induced abortion in the USSR iseven higher than reflected by official statistics,since there is substantial underreporting in manyareas, especially for the youngest and primigravidwomen.3 Results of a few surveys suggest that onaverage a sexually active woman has 2-4 inducedabortions by the end of her reproductive period,and roughly 10-15% ofthe women have their firstpregnancy interrupted.3 Induced abortion ratesare highly differentiated between national,regional, and social subgroups of the population.Thus the range of variation between the nationalrepublics in registered induced abortion rates in1985 was fourfold, ie, 30-8% in Azerbaidjanversus 123-6% in the Russian Federation.2

Sociodemographic and public healthimplications of multiple induced abortion in theUSSR are well known: fertility reduction due tosecondary postabortive sterility/subfecundity,high prevalence of chronic gynaecologicalconditions in young and middle aged women,obstetric complications, and high infantmortality.5 But up to now, clinical (and a fewepidemiological) investigations have been focusedon immediate and short term complications ofinduced abortion and its influence on subsequentreproductive function,5 while the long termcumulative effect of pregnancy termination hashardly been assessed. The issue of possibleoncological implications of induced abortiondiscussed in this paper has traditionally been "noman's land" in a highly specialised medicalscience, as it has been missing from thecurriculum of both oncologists andgynaecologists. Epidemiologically, inducedabortion has also been neglected as a specific riskfactor involved in causation of sex hormone

related cancers in women. To my knowledge, nostudy in this area has examined the potential effectof induced abortion in relation to length ofgestation, termination procedure used, presenceof immediate complications, etc.This lack of attention is not easy to understand,

since involvement of induced abortion in somestage of carcinogenesis in hormone dependentorgans is biologically highly plausible. Someclinical and laboratory studies have shown thatsudden interruption of the endocrine,immunological, neurological and other processeswhich adjust the organism to coexistence with thefetus cause a strong and long lasting impact,described by some investigators as a "hormonalblow".4 In cases of repeated induced abortion,such changes may accumulate, graduallyproducing chronic hormonal changes (such asprogesterone deficiency) and/or immunedisorders, to which psychological depressioncaused by this dramatic life event may alsocontribute.45 These shifts may lead todisturbance of mechanisms of cell differentiationand proliferation, weakening of immunity, andother systemic changes involved in malignanttransformation.468 Therefore consideringinduced abortion in epidemiological analysis asjust another short term pregnancy rather than aspecific biological event may be misleading.

Breast cancerAmong hundreds of epidemiological studies offemale breast cancer conducted all over the world,abortion is mentioned in no more than 20 papers.In many cases the authors do not even distinguishbetween spontaneous abortions, or miscarriages,and induced abortions-obviously quite differentevents, both biologically and socially. The lack ofinterest in induced abortion among Westernepidemiologists probably reflects the relativelysmall perceived social significance of this factor insocieties with developed contraceptive culture.Besides, in many Western countries inducedabortion was illegal until recently, and femalecohorts that have reached their breast cancer agehad spent their reproductive years withoutopportunity of legal pregnancy termination.Hence European and North Americanepidemiologists did not have much chance tostudy induced abortion in terms of inherentcancer risk. In some studies where the authorsintended to study induced abortion among otherreproductive factors in breast cancer, the

All-Union CancerResearch Centre,AMS USSR, 115 478Moscow, Kashirscoyeshosse 24, USSRL I Remennick

Accepted for publicationJune 1990

on Novem

ber 3, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.44.4.259 on 1 D

ecember 1990. D

ownloaded from

Larissa I Remennick

numbers of respondents reporting such abortionswere so small that the outcome statisticalmeasures made no statistical sense. (In a case-control study by Brinton et al,9 only 1-17% ofcases and 1 20% of controls had ever had inducedabortion.)Due to limited number and low statistical

power of the various studies that have been donethe aetiological role of induced abortion and itsrelationships with other reproductive variables(particularly in terms of their relative timing) aredubious. Most of the studies consideredtermination of the first pregnancy as a risk factorfor premenopausal breast cancer. The first ofthese, conducted by Pike et al in Los Angeles andpublished in 1981,10 had a strong impact on theepidemiological community. Comparison ofreproductive histories of 163 women aged lessthan 33 years at diagnosis of breast cancer withtwo groups of age matched controls (neighboursand friends of presumably similar social status)showed that cases had significantly higherprevalence of both first trimester inducedabortion and spontaneous abortion before theirfirst full term pregnancy. The level ofrelative risk(RR) of 2-4, p < 0-005, did not decrease afteradjustment for other reproductive variables, butwas lower for women who subsequently hadlivebirths (1-8). Remarkably, abortions after 3months ofgestation or occurring after the first fullterm pregnancy did not affect the risk of earlybreast cancer.

Interpreting their results, Pike et al refer to thedata from the international case-control study byMacMahon et al," showing that a terminated firstpregnancy produced an increased rather thandecreased breast cancer risk. The underlyinghypothesis was that the long lasting protectiveeffect of the first full term pregnancy was due tocombination of cell differentiation and alteredhormone profile after the first birth (the latterlasting for decades). As proliferation of the breasttissue occurs mainly in the early first pregnancy itis plausible that induced or spontaneous abortionafter the first trimester of the first pregnancy hasno sizeable effect on the risk.6 10

After the study by Pike et al there have been atleast four attempts to reproduce theirresults.9 12-14 In the retrospective cohort study byHadjmichael et al,'2 breast cancer risk wasassessed among 33 115 women who had giventheir first birth between 1946 and 1965 inConnecticut. Reproductive histories of thesewomen were obtained from their obstetricrecords, and breast cancer incidence in the cohortwas followed up to the year 1980 through theConnecticut cancer registry. It was found thatspontaneous abortion before the first full termpregnancy was associated with a 3 5-fold increasein breast cancer risk in comparison with womenwithout history of spontaneous abortion,irrespective of the number of such abortions andother known breast cancer risk factors. Riskincreased with time elapsed since the date ofspontaneous abortion, so that over a 20 yearperiod the overall breast cancer risk increasedmore rapidly in women who had had a

spontaneous abortion. Unfortunately, there wereno data on induced abortion before the first fullterm pregnancy in the obstetric records (since

they were illegal at the time), and interviews werenot performed.

Interpreting these results, the authorsaddressed the experimental data of Russo andRusso,7 who analysed the influence of pregnancyin rats (full term, full term with subsequentlactation, and terminated) on the development ofbenign and malignant lesions in the mammarygland. They showed a protective effect of fullterm pregnancy, with and without lactation,against neoplastic changes in mammary gland viacomplete differentiation of mammary tissues.Hormonal changes of pregnancy affecting growthand differentiation ofmammary cells (increases inoestrogens, progesterone and prolactin levels)were also shown to be protective. Pregnancytermination abruptly halts these processes, whichresults in underdevelopment of the mammarygland, making it more susceptible to malignantchange. The animals with interruptedpregnancies had as high an incidence of tumoursas virgin rats treated with a carcinogen. Thisexperimental evidence may also be true forhumans.8 12

In the case-control study by Brinton et a19(1362 newly diagnosed breast cancer patients and1250 healthy controls selected through the BreastCancer Detection Demonstration Project invarious parts of the USA) induced abortionemerged as a risk factor only for nulliparouswomen. Relative risk in these, as opposed tonullipara without induced abortion, was highenough to be close to statistical significance (5 5,with 95% confidence limits (CL) 0 8-36-8),although it was based on very small numbers.Another American study'3 and one British case-control study'4 failed to identify any riskassociated with abortion before the first full termpregnancy, but in both studies the results werebased on small numbers due to low prevalence offirst pregnancy termination.One earlier American investigation'5 and three

recent European studies'6S8 have also givennegative answers. A case-control study in thecities of Northern Italy'6 (1108 newly diagnosedbreast cancer patients and 1281 hospital non-cancer controls) produced non-significant relativerisk measures: 1 18 for one induced abortion and0 72 for two or more. But in women with one ormore termination before the first full termpregnancy relative risk increased to 1 42 and wasof borderline statistical significance (95%confidence interval 0 91-2 20). In the largescreening programme based cohort study inNorway, women with induced and spontaneousabortions had a slightly decreased breast cancerrisk, although non-significant and not showing adose-response gradient.'7 A cohort study of49 000 women with a history of first trimesterinduced abortion in Sweden'8 followed up in thecancer registry produced a relative risk estimate of0-8 (95% CI 0-58-0 99).However, in a number of studies from all over

the world (in the USA, Canada, France,Denmark, Japan, and Israel) abortions, eithermultiple or occurring before the first full termpregnancy, have been shown to be significantlyassociated with increased breast cancer risk.19-28In most cases, this evidence was related toinduced abortion, though in a few studies it was

260 on N

ovember 3, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.44.4.259 on 1 Decem

ber 1990. Dow

nloaded from

Induced abortion and cancer risk

related to any pregnancy interruption. Relativerisk values for this risk factor (in women withhistory of abortion(s) compared to those without,adjusted for other significant variables) werewithin the range of 1 5-4-0, ie, similar to mostother reproductive characteristics.A recent population based case-control study in

Denmark,26 including almost all the new breastcancer cases in the country for one year, hasshown that induced abortion in the first andsecond trimesters of the first pregnancy wassignificantly associated with breast cancer risk(RR 1-43 with 95% CI 1-101-84). This relativerisk value closely approximated that fornulligravid women (147; 1 14-190). Womenwith two or more induced abortions before theirfirst full term pregnancy had a breast cancer riskof 1 73 (0-76-391) relative to those withoutinduced abortions, all relative risk values adjustedfor age, residence, and age at first birth. Here too,induced abortion after the first full termpregnancy did not significantly influence breastcancer risk (RR 1-35; 0-71-2-56).The latest investigation specially addressing

the issue of abortion and breast cancer was apopulation based record linkage study in upstateNew York.27 It was based on cancer registrationdata and fetal death reports, the latter containingdata on all known spontaneous and inducedabortions of any gestational length. History offetal death was compared in 1451 breast cancercases under the age of 40 years diagnosed during1976-80, and 1451 population controls matchingthe cases by year of birth and residence. Matchedpair analysis showed that termination of earlypregnancy (mean length ofgestation 9-6 weeks forcase series and 11-5 weeks for controls) wassignificantly more frequent in breast cancer cases.Relative risk for induced abortion was 1-9 (95%CI 1-2-3-0) and for spontaneous abortion 1-5, butnon-significant (0-7-3 7). Relative risk was thehighest in women with a history of multipleterminations with no intervening livebirths (40;15-13-6). On the whole, the number ofinterrupted pregnancies was 70% higher amongcases, irrespective of the outcome of previouspregnancies.

In Canada, the 30 year follow up of a cohort ofregistered nurses28 has shown that women withinduced abortions had a significantly higherincidence of benign breast disease as compared towomen with similar reproductive histories butwithout induced abortions.

In the Soviet Union where the potential publichealth impact ofinduced abortion might be muchstronger than in the West, there have also beenvery few epidemiological studies directlyassessing induced abortion as a cancer risk factor.Among other reproductive factors, it wasevaluated in a case-control study in the NorthCaucasus.29 Relative risk of breast cancer inwomen with three or more induced abortions wasshown to be 3 4 times higher, and in those withone or two induced abortions twice as high as inwomen with no such history. Women whointerrupted their first pregnancy when they wereyounger than 25 had a 18 times higher risk ofbreast cancer than those who had their firstinduced abortion later in life. Since theassociation between induced abortion and breast

cancer risk could be confounded by parity, thesecharacteristics were analysed together and alsostratified for two histological types of breastcancer (differentiated and undifferentiated).These stratified data have shown that in allsubgroups by parity (0 1-1-3+), relative riskincreased with number of induced abortions(1-2 3+), and for any given number of inducedabortions the risk was lower for women with morelivebirths. The highest risk of both breast cancertypes was shown for nulliparous women withthree or more induced abortions. The protectiveeffect of higher parity was more clear for womenwith more induced abortions and differentiatedtumours. This study therefore suggested anindependent, and obviously inverse, aetiologicalimpact of parity and induced abortion on breastcancer.

In another case-control study in Moscow withhospital controls similar to cases in terms of age,menopausal status, education and residence30 nooverall trend of breast cancer risk in relation tonumber of induced abortions was shown. The riskwas slightly increased only in women with inducedabortion before the first full term pregnancy. Butthis study also failed to show any role of otherknown reproductive characteristics involved withbreast cancer (parity, lactation) and revealed onlythe age related aspect of reproductive activity asaetiologically important. It may be argued that theMoscow population as a whole is unfavourable forepidemiological examination of reproductive riskfactors because of the homogeneity ofchildbearingpatterns of Muscovites with predominantly lowfertility, the use ofinduced abortion as the means ofbirth control, and short lactational experience.Methodologically speaking, it is difficult to revealaetiological significance in the case of a universallyprevalent exposure.

Cervical cancerSeveral studies have suggested a positiveassociation between induced abortion and cancerof the cervix.23 31-38 Thus a case-control study inParis and seven other French cities23 has shown analmost fivefold increase in relative risk for womenreporting two or more terminations (afteradjustment for sexual and other significantvariables). In the case-control study in Chile, thecountry with endemic cervical cancer rates,women with both induced and spontaneousabortion had significantly increased risk,although for spontaneous abortion it wassomewhat higher (1-94 v 1-38).32

Descriptive epidemiological studies in theUSSR393 have suggested that regional variancein cervical cancer incidence is related more toinduced abortion than to fertility rates. Thus themajority of cervical cancers in Armenia have beenregistered in three major cities-Yerevan,Leninokan and Kirovokan-where inducedabortion rates have been high. In other towns andregions of this republic, with similar reproductive(and presumably sexual) populationcharacteristics but half the induced abortion rate,the incidence of cervical cancer has been threetimes lower. The correlation coefficient betweeninduced abortion rate and cervical cancerincidence in Armenia was 067.39

261 on N

ovember 3, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.44.4.259 on 1 Decem

ber 1990. Dow

nloaded from

Larissa I Remennick

Probably for the same reason, in the CentralAsian Soviet republics, which have multinationalpopulations, cervical cancer incidence inmigrants, mostly Russians, is much higher than inindigenous women. For example, in Kirgizia,cervical cancer incidence in Russian women is fivetimes higher than in Kirgiz women, while theaccumulated numbers ofinduced abortions by theend of reproductive age is 3 3 for Russians v 1 8for Kirgiz women.40 This suggests that higher andearlier fertility in Kirgiz women (their expectednumber of children is about five v about two inRussians2) does not affect their cervical cancer

risk.The same relationship is more obvious for the

USSR overall. A correlation study based on

official abortion statistics and regional cancer

incidence data for the period 1959-8541 showed a

significant contribution ofinduced abortion to thevariance of cervical cancer (and to a lesser extent

also breast cancer) rates between areas of thecountry. The available statistical indicators ofinduced abortion prevalence (rate per 1000women aged 15-49; induced abortion:livebirthratio; proportions of registered out of hospitalterminations and induced abortions inprimigravidas in the total number of inducedabortions have proved to be strong and consistentdeterminants of variation in the incidence of bothfemale cancers (see table). This influence ofinduced abortion rates on regional distribution ofcervical cancer and breast cancer (withadjustments made in correlation and regressionanalysis for the time lag between reproductiveevents and diagnosis) was independent of otherreproductive characteristics studied (earlymarriage rates as a surrogate measure for earlyonset of sexual activity; age at first birth; parity;lactation).4'

Suggested mechanisms of induced abortioninfluence on cervical carcinogenesis may bemultiple. The first mode of action may be viageneral endocrine stress in the reproductivesystem resulting from termination of pregnancyrelated processes. Another is through mechanicaltrauma and possible infection associated withdilatation and curettage or incomplete evacuationof the embryo and placenta. Chronicinflammatory lesions may arise in cervical tissueon the site of this trauma (erosions, endocervicitis,leucoplakia), as well as cell abnormalities(dysplasias). In course of time, the latter mayundergo malignant transformation and/orfacilitate the action of exogenous carcinogenicagents.4 34 42 43 Such a pathogenic chain is in goodagreement with the concept of three stage cervical

Correlation betweenstatistical indicators ofabortion and age adjustedincidence rates of breastcancer and cervical cancer

for women of 70 areas ofRussia

carcinogenesis: from dysplasia (mild, moderate,and severe) to carcinoma in situ and invasivecarcinoma.34 42 It has also been shown thatdysplasias and intraepithelial neoplasms of thecervix are more prevalent in women with thehistory of induced abortion.32 35 42

Ovarian cancer

The possible influence of abortion on ovariancancer development is obscure and has hardlyever been specifically examined. Most of thestudies of ovarian tumours have been directedtowards two protective reproductive factors:pregnancy experience per se, and hormonalcontraception as an imitator of pregnancy

experience, both acting by decreasing the totalperiod ofovarian activity.446 A few studies havesuggested no effect45 46 or a weak and non-

significant protective effect47 48 of both inducedand spontaneous abortion. A recent case-controlstudy in Japan has shown a significant protectiveeffect ofinduced abortion, with their numbers not

specificed, on ovarian cancer risk (RR 0 6, 95h0CI 03-0 9) when a few other reproductivevariables were controlled for.49 No mechanism ofthis protective action is suggested by the authorsother than that short term pregnancies are addedto the total pregnancy experience.

Endometrial cancer

No special evidence is available on the role ofinduced abortion in the causation of endometrialcancer, although here an effect is biologicallyplausible since the inner lining of the uterus is thesite of dilatation and curettage or vacuum

aspiration precedures. As with ovarian tumours,several authors have argued that the more

pregnancies a woman had had, regardless ofwhether they proceed to term or not, the lower herrisk of endometrial cancer,50 although theprotection conferred by an incomplete pregnancy

is much smaller than that of a full term one: as

estimated by Henderson et a15' 5-6 pregnanciesterminated by induced or spontaneous abortionroughly equal one pregnancy resulting in a

livebirth. In a Norwegian cohort study by Kvaleet al,52 there was no definitive trend inendometrial cancer risk according to the numberof induced or spontaneous abortions. Amongparous women, the odds ratio for those with threeor more abortions, as compared to abortion freesubjects, was 1-03 (95% CI 0-58-1 82) in logisticregression analysis with adjustment for parity, ageat first birth, and some demographic variables.

Abortion indicators Time lag Parametric tests Non-parametric rank criteriaand outcome variables (years)

r rp Spearman p Kendall t

Abortion rate, %Breast cancer 10 0-45 0-23 0-25 NS NSCervical cancer 10 0-77 0-63 0-76 0 77 0 57

% of out of hospital abortionsBreast cancer 0 0 80 0 48 0 36 NS NSCervical cancer 0 0 53 NS 0 34 NS NS

% of abortion in primigravid womenBreast cancer 0 0-65 0 22 0-79 045 0 29Cervical cancer 0 0 58 0 33 0-29 NS NS

Cited from Remennick4lCoefficients used: r = linear (Pearson's); r1, = partial; 4 = correlation ratioAbortion/livebirth indicator is not shown since its meaning is close to that of abortion rate, and so are the coefficients

262 on N

ovember 3, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.44.4.259 on 1 Decem

ber 1990. Dow

nloaded from

Induced abortion and cancer risk

No studies could be found relating abortion tothe risk of any non-reproductive cancer.

ConclusionTwo competing approaches can be seen in thestudies assessing induced abortion as a potentialcancer risk factor and as a biological eventgenerally. The first attitude tends to stress thepregnancy per se, short or full term, with all itsprotective changes (differentiation of breasttissues, intermission in ovarian activity, etc).Therefore, an interrupted pregnancy isconsidered as a legitimate supplement to the totalpregnancy experience, and calculations such as

"5-6 terminated pregnancies = 1-0 full termpregnancy'"5' are the logical extension of thisviewpoint. Hence terminated pregnancies are

considered as protective, or at least neutral,reproductive events. "4-7 452The alternative approach is focused not so

much on pregnancy experience but rather on thefact of its sharp termination with concomitanthormonal and immunological stress, incompletedifferentiation and growth of breast tissues, etc.The above mentioned calculation would beirrelevant within this approach, and itsproponents see abortion as a riskfactor.7 8 10 12 23 25-27 29 35 37 41-43 An initialattitude of researchers towards abortion usuallydetermines the way they interpret results, sinceoutcome risk measures are often of moderatevalue and/or borderline statistical significance.There are rational arguments for bothapproaches, but the second seems more plausiblebiologically and is also backed up by certainclinical and experimental data.The role of induced abortion in female cancers

obviously needs further scrutiny, preferably inanalytical studies specially addressing this issueand in populations with high induced abortionrates (so far all but a few studies have beenconducted in the countries with low/moderateinduced abortion rates). One of the questions tobe clarified is the suggested cross over effect in theaction of pregnancy related risk factors for breastcancer.8 15 2153 Is it true that any pregnancy,including one terminated by abortion, confersprotection in the long run (ie, for post-menopausal breast cancer) but increases the riskduring reproductive years? Is the effect ofpregnancy termination separate and opposite to

the effect of pregnancy per se? How does inducedabortion add to a woman's reproductive cancer

risk profile when performed at various ages,

before, after, or in between live births?So far, two categories of women have been

shown to experience increased breast cancer riskdue to induced abortion: those with terminatedfirst pregnancy,8 10-12 25 29 30 41 and nulliparouswomen.8927 29 But both findings have beenchallenged in other studies.'3 14 16 17 Anotheraspect of the problem is possible accumulation ofrisk with multiple terminations, highly prevalentin the USSR and some other countries. Yetanother question to be answered in future studiesis whether the gestational age and type oftermination technique (dilatation and curettage,

vacuum aspiration, prostaglandins, and otherabortifacients) influence subsequent cancer risk,

and in what way. And, finally, what are theintermediate pathogenic stages between inducedabortion and the development of cancer of thecervix uteri (and probably also of the corpus uteri)?As in most other cancer aetiology puzzles, both

epidemiology and other sister disciplines (clinicaland experimental oncology, gynaecology, socialmedicines, etc) can contribute to an improvementof our understanding of abortion as a factor incarcinogenesis.

1 Tietze C, Lewit S. Epidemiology of induced abortion. In:Abortion and sterilization. Medical and social aspects. NewYork: Academic Press, 1981: 39-56.

2 Population of the USSR, 1987. Statistical yearbook.Moscow: Financy i Statistika, 1988 (in Russian).

3 Remennick LI. The role of induced abortion amongmethods of birth control in the USSR and abroad. In:Problems of demographic development of the USSR. Instituteof Sociology, Moscow: USSR Academy of Sciences, 1988:86-89 (in Russian).

4 FrolovIB. Impact of pregnancy termination on the organismand reproductive function of a woman. Akush Gynecol1982; 4: 6-8 (in Russian).

5 Kulakov VI, Zack IR, Kulikova NN. Induced abortion and itscomplications. Moscow: Medicine Publ, 1987 (in Russian).

6 Pike MC, Gerkins VR, Casagrande JT, Brown J, Gray GE.The hormonal basis of breast cancer. Natl Cancer InstMonogr 1979; 53: 187-96.

7 Russo J, Russo IH. Susceptibility of the mammary gland tocarcinogenesis.II. Pregnancy interruption as a risk factorfor tumour incidence. Am J Pathol 1980; 100: 497.

8 Krieger N. Exposure, susceptibility, and breast cancer risk: ahypothesis regarding exogenous carcinogens, breast tissuedevelopment, and social gradients, including black/whitedifferences, in breast cancer incidence. Brast Cancer ResTreat 1989; 13: 205-23.

9 Brinton LA, Hoover R, Fraumeni JF. Reproductive factorsin the aetiology of breast cancer. Br J Cancer 1983; 47:757-62.

10 Pike MC, Henderson BE, Casagrande JT, RosarioI, GrayGE. Oral contraceptive use and early abortion as riskfactors for breast cancer in young women. Brj Cancer 1981;43: 72-6.

11 MacMRhon B. Epidemiology of cancer of the breast. In:Levin DL, ed. Cancer epidemiology in the USA and theUSSR. Bethesda: National Cancer Institute Publications1980: 93-7.

12 Hadjmichael OC, Boyle CA, Meigs JW. Abortion beforefirst livebirth and risk of breast cancer. Br J Cancer 1986;53: 281-4.

13 Rosenberg L, Palmer J, Kaufman DW, et al. The risk ofbreast cancer in relation to induced and spontaneousabortion before first birth. AmJ Epidemiol 1986; 124: 516.

14 Vessey MP, McPherson K, Yeates D, Doll R. Oralcontraceptive use and abortion before first full-termpregnancy in relation to breast cancer risk. Br J Cancer1982; 45: 327-31.

15 Paffenbarger R, Kampert J, Chang H. Characteristics thatpredict the risk of breast cancer before and after themenopause. Am Y Epidemiol 1980; 112: 258-68.

16 La Vecchia C, Decarli A, Parazzini F, et al. Generalepidemiology of breast cancer in Northern Italy.Int JEpidemiol 1987; 16: 347-55.

17 Kvale G, Heuch I, Eide GE. A prospective study ofreproductive factors and breast cancer.I. Parity. Am JEpidemiol 1987; 126: 831-41.

18 Lindefors Harris B-M, Eklund G, Meirik 0, Rutqvist LE,Wiklund K. Risk of cancer of the breast after legal abortionduring first trimester: a Swedish register study. Br MedJ31989;299: 1430-2.

19 Stewart HL, DunhamLJ. Epidemiology of cancer of theuterine cervix and corpus, breast and ovary in Israel andNew York city. J Natl Cancer Inst 1966; 37: 1-95.

20 Choi NW, Howe GR, Miller AB, et al. An epidemiologicstudy of breast cancer. Am J Epidemiol 1978; 107: 510-21.

21 Kelsey JL, Fischer DB, Holford TK, et al. Exogenousestrogens and other factors in the epidemiology of breastcancer. J Natl Cancer Inst 1981; 67: 327-33.

22 Nishiyama F. The epidemiology of breast cancer inTokushirna prefecture, Japan. Shikoku Acta Med 1982; 38:333-43.

23 Le MG, Bachelot A, Doyon F, et al. Oral contraceptive useand breast or cervical cancer: preliminary results of aFrench case-control study. In: Hormones and sexualfactorsin human cancer aetiology. Amsterdam: Exerpta Medica,1984: 139-47.

24 Hirohata T, Shigematsu T, Nomura A, et al. Occurrence ofbreast cancer in relation to diet and reproductive history: acase-control study in Fukuoka, Japan. Natl Cancer InstMonogr 1985; 69: 187-90.

25 White E, Daling J, Norsted TL, Chu J. Rising incidence ofbreast cancer among young women in Washington State. JNatl Cancer Inst 1987; 79: 239-43.

26 Ewertz M, Duffy SW. Risk of breast cancer in relation toreproductive factors in Denmark. Br J Cancer 1988; 58:99-104.

263 on N

ovember 3, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.44.4.259 on 1 Decem

ber 1990. Dow

nloaded from

Larissa I Remennick

27 Howe HL, Senie RT, Bzduch H, Herzfeld P. Early abortionand breast cancer risk among women under 40. IntEpidemiol 1989; 18: 300-4.

28 Hislop TG, Elwood GM. Risk factors for benign breastdisease: a 30-year cohort study. Can Med AssocJ 1981; 124:283.

29 Dvoirin VV, Medvedev AB. The role of reproductivehistory in breast cancer causation. In: Methods and results ofstudies of breast cancer epidemiology. Tallinn 1978: 53-6 (inRussian).

30 Levshin VF, Chepurko AD. Reproductive history andbreast cancer. Sov Med 1986; 6: 15-21 (in Russian).

31 Cascante E, Sagot S. Aspectos epidemiologicos de laneoplasia intraepitelial e invasore del cuello uterino. ActaMed Costa Ricense 1979; 22: 347-60.

32 Molina R, Thomas DB, Dabancens A, et al. Oralcontraceptives and cervical carcinoma in situ in Chile.Cancer Res 1988; 48: 1011-5.

33 Harris RW, Brinton LA, Cowdell RH, et al. Characteristicsof women with displasia or carcinoma in situ of the cervixuteri. Br Cancer 1980; 42: 359-69.

34 Hulka BS. Risk factors for cervical cancer. Chron Dis 1982;35: 3-11.

35 Zaninetti P, Francheschi S, Baccolo M, Bonazzi B, GottardiG, Serraino D. Characteristics of women under 20 withcervical intraepithelial neoplasia. Int Epidemiol 1986; 15:477-82.

36 Fujimoto I, Kurikawa Y, Nomada T, et al. Epidemiologicalstudy of the preinvasive cervical cancer. J'pn Cancer Res1987; 33: 651-60.

37 Parazzini F, La Vecchia C, Negri E, et al. Risk factors foradenocarcinoma of the cervix: A case-control study. BrCancer 1988; 57: 200-4.

38 Groenroos M. Etiology of premalignant cervical lesions inteenagers. Acta Obstet Gynecol Scand 1980; 59: 79.

39 Bazikian KL. On the role of abortion in incidence of uterinecancer in Armenia. In: Aspects of prevention and earlydetection of the cervical cancer. Tbilisi, 1971: 199-200 (inRussian).

40 Kanunnikova LN. Incidence and regional distribution ofcervical cancer in Kirgizia. In: Aspects of clinical andexperimental oncology. Frunze, 1980: 42-6 (in Russian).

41 Remennick LI. Reproductive patterns and cancer incidencein women: a population-based correlation study in theUSSR. IntJ Epidemiol 1989; 18: 498-510.

42 Smachtina OA, Mezinova NN. On some etiologic factors ofcancers of the uterine corpus and cervix. In: Epidemiologicalaspects of prevention and early detection of gynaecologicalmalignancies. Tbilisi, 1977: 73 (in Russian).

43 Charkviani LI. Epidemiological aspects of prevention andearly detection of uterine cancer. Tbilisi, 1977 (in Russian).

44 Beral V, Francheschi S. The prevention of ovarian cancer.In: Hormones and sexual factors in human cancer aetiology.Amseterdam: Exerpta Medica, 1984: 19-24.

45 Wu ML, Whittemore A, Paffenbarger RS, et al. Personaland environmental characteristics related to epithelialovarian cancer. I. Reproductive and menstrual events andoral contraceptive use. AmJ Epidemiol 1988; 128: 1216-27.

46 Harlow BL, Weiss NS, Roth CJ, Chu J, Daling JR.Case-control study of the borderline ovarian tumors:reproductive history and exposure to exogenous femalehormones. Cancer Res 1988; 48: 5849-52.

47 Cramer DW, Hutchinson GB, Welch WR, Scully RE, RyanKJ. Determinants of ovarian cancer risk. Reproductiveexperiences and family history. J7 Natl Cancer Inst 1983; 71:711-6.

48 Risch HA, Weiss NS, Lyon JL, Daling JR, Liff JM. Eventsof reproductive life and the incidence of epithelial ovariancancer. Am J Epidemiol 1983; 117: 128-39.

49 Mori M, Harabuchi I, Miyake H, Casagrande JT,Henderson BE, Ross RK. Reproductive, genetic, anddietary risk factors for ovarian cancer. AmJ Epidemiol 1988;128: 771-7.

50 Kelsey JL, LiVolsi VA, Holford TR, et al. A case-controlstudy of cancer of the endometrium. Am J Epidemiol 1982;117: 333-42.

51 Henderson BE, Casagrande JT, Pike MC, Mack T, RosarioI, Duke A. The epidemiology of endometrial cancer inyoung women. Br 7 Cancer 1983; 47: 749-56.

52 Kvale G, Heuch I, Ursin G. Reproductive factors and risk ofcancer of the uterine corpus: a prospective study. CancerRes 1988; 48: 6217-21.

53 Kelsey JL, Berkowitz GS. Breast cancer epidemiology.Cancer Res 1988; 48: 5615-23.

264 on N

ovember 3, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.44.4.259 on 1 Decem

ber 1990. Dow

nloaded from