Reprinted by the U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE PUBLIC HEALTH SERVICE from AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY Vol. 132, No. 2, September 1978 Legalized abortion: Effect on national trends of maternal and abortion-related mortality (1940 through 1976) WILLARD CATES. Je. M.D.. M.PLIL ROGER W, ROCHAT, M.D. DAVID A, GRIMES, M.D RL W. TYLER, Ju. M.D. Allania, Georgi ‘Both non-aborton related matornal and aborion-related mortality decined prior tothe Supreme. ‘Court docisions 0! 1973. In order to determine the effect of legalized abortion on maternal mortally, Wwe have analyzed the seculr trends in national abortion mortally ratlos for 1940 through 1976, Compared the trends fo those maternal moraliy ratios, and hypothesized reasons for dtlerencas between these trends. Between 1940 and 1950 and after 1965, deaths trom abortion decined moro rapidly than deaths from other causes associated with prognaney and chicbith. However, between 41951 and 1965, maternal morality related to pregnancy or childbirth decined more rapidy than abortion-elated mortalty. Five possible explanations exist fr the more rapid deine in abortion deaths since 1865—selacted underreporing, changes in coding practices, improved safety ol ilegal abortion. inteducton of more ettective contraception, and increased availabilty of legal abortion We consider the last two explanations as the most likely reasons for the accelerated decine in abotion-elated deaths. (AM. J. OBSTET. GYNECOL, 192: 211, 1978) BoTH MATERNAL AND ABORTION mortality had been declining prior to the United States Supreme Court decisions in 1973, which declared restrictive state abortion laws unconstitutional.’ In order to determine the effect of legalized abortion on maternal mortality ratios, we analyzed the secular trends in national abor jon mortality ratios for 1940 through 1976, compared the trends 10 those of maternal mortality ratios, and hypothesized reasons for dillerences between these ends. Methods National mortality statistics in the United States are compiled by the National Center tor Health Statistics; cause of death is based on etiologies listed on the olficial death certificate. Although death certilicatesare From the Abortion Surcillance Branch, Center far Disease Control Bares of Epidemiangs, Family lasing Evaluation Divi, Prrsenied atthe One Hundred Fustythird Annual Merting ofthe American Associaton forthe Advaneement of Scenee, Denerr, Colorado, February 23, 1977, Reprint requests: Dr. Willard Cats, Chie, Abortion Suareilaner Branch, Center jor Dive Conte, dan, Georgie 30333, (002:887878/02182-021 1800.40 © 1978 The C. V. Monby Co, limited in the amount of information they contain, vital statistics provide a consistent data base for monitoring national abortion mortality trends. ‘We will present the vital statistics data in the form of the annual mortality ratio, which is the number of deaths from pregnancy in one year divided by the number of liv births during'the same year. Although customarily referred to as the maternal mortality rat, this term is inappropriate since women who died of ‘causes such as abortion or ectopie pregnancy usually have not been delivered of live fetuses, Because the denominator does not include the numerator, the cor- rect terminology should be mortality rato, rather than rate? ‘We used live births as the denominator because they provide the best available indicator of reproductive ‘outcome during the 37 year period. The ideal denomi= nator would be all pregnant women, with the mortality risk calculated in « life-table fashion.t Alternatively, a denominator of live births plus abortions might pro- vide a better estimate of the total population at risk of death from pregnaney and childbirth. Unfortunately. wwe have accurate national data only on legal abortions since 1969. when the Genter for Disease Control first ant212 Cates etal beyan ity surveillance program Moreovel no relic able statistics on the number of illegal abortions are iwailable for this period: indireet measures, such local surveyst or number uf deaths.” are not con Sstent through the periods studied. ‘Therefore, we Ihave chosen to use live births as the denominator for In our amily. we compared te mortality ratios: (1) the abortion mortality tie or the number of abor= tion-related deaths per one nsillion live hirthss this ratio ine ludes deaths from all citegories of abortion legally and (2) the maiernal mortality ratio evelading abrtion deaths Which is he total mumiber of deaths from pregnancy Induced. illegally induced. aid! spontaneou ths per one mili live births. Abort illion live births Saurce: United States Vital Statistics, Nation mare a September 15, 1978 ‘Am. J Ober. Gynec. equals abortion deaths per one W Cemter fi Health Statsies, 1940 10 and childbirth per one million live births. Our use of the nonabortion maternal mortality ratio for compari son is intended to reflect an estimate of the rate of decline in abortion deaths that might be attributable to improved standards of general maternal health care developed during the period under study. Finally, we used a loygvithmic veale for comparing the wwe ratios to portray more accurately their relative rates of decline Uhronghout the study intervat Results ‘The maternal mortality ratio, exuding abortion, has continvally declined from 1940 to 1976 at an aver rate of 13 per cent before 1956 and 5 petcent therealier (Fig. 1), The abortion niotality ratio passed dau three mia pluses between 1940 and 1976, In the first phase (1910 to 1950) there was steady dectine at ant average annual rate of 1a per cent In the second phase (IYSL us» 19655) the abwnti mortality rte virtually leveled of at at average ate nual rate of decrease of any 17 per cent Therealter the abortion mortality ratio recreasedt at an average annual rate of HT per cent beoseen 1966 and 1970, «nd accelerated t0 25 per cent per yea be: ween 1970 and 1976, Theretore, from IH te 1950 and from 1955 w 1976, deaths tram abortion declined more rapidly than deaths frasn other causes associated ‘ith pregnancy and childbirth, However, from 1950 te 1965. the maternal mortality ratio declined snore rapidly than) abortion montality ratio, Comment ‘The decline in nonabortion maternal mortality prob- ably is due to (1) improvements in obstetric practice (sophistication of blood replacement, antibiotic ther- apy. snd surgical techniques): (2) innovations in train= ing programs tor posigiaduare healt proessionals; (3) the identification of high-risk patient groups and increased eflorts to deliver obstetie services to these groups: and (4) the expansion of fuilities for obstetric care, resulting in beter prenatal evaluation, improved postpartum family planning, and an increasing per- ‘centage of births occurring in hospitals throughout the The introduction of antibioties may be a reason that abortion mortality ratio declined more rapidly than maternal mortality ratio between 1940 and 1950, Con- sistent with this hypothesis is the fact that sepsis ac- counted for neanly 80 per cent of abortion deaths i 1940 but only 30 per cent of al other macersal deaths.” The relatively stable abortion mortility ratio fron 1950 to 1965 may represent an equilibrium between the impact of improved medical «re for women (rep: resented by the declining maternal mortality ratio) and the still siable number of illegal abortions. Tietze" has hypothesized that during the 1960's an increasing percentage of illegal abortions were performed by more-expetienced practitioners, Berween 1965 and 1976 the increased use liveness of contraceptives, as well as the yr bility of legal abortions, may have produced the rel atively greater decline of abortion deaths. Between 1963 and 1970 oral contraceptives and intrauterine devices were used by an inereasing percentage of mar ried women? and this wasassociated with the decline in unwanted fertility during those years.” Assuming that unwanted conceptions followed the same pattern as unwanted births, there probably was a rediwed de- i effe- avail Legalized boron 213 mand for pregiiancy termination and, therefore, a reduced population at risk of dying from abortion, Si- multancously, abortion laws were liberalized, and sub- sequently, the number of reported abortions hi ‘creased each year since 1969." The largest increases in the number of legal abortions were reported in 1970 and 1971, concurrent with the accelerated fall in abor- tion mortality ratio, ‘The increased availability of legal abortion would re- duce the ‘otal number of deaths resulting from abor: tion, primarily from substituting the risks of the safer legal abortion procedures for these of the more dan- xerous illegal methods of pregnaney cer Concurrently, the trend could underscore the velative safety of the legal abortions that were actually substi tuted for term births after 1970." ‘The accelerated decline in abortion deaths after 1970 is even more striking if three other factors are considered. First, the average age of women who have died from abortion and from other conditions of preg- nancy has not changed over time. Moreover, the age- specific mortality ratios show an increasing differential (favoring abortion) between abortion and nonabortion ‘maternal deaths as age increases."" Because women who obtain legal abortions are generally older than those who give birth at term, adjusting the ratios to a standard population of women who obtained abortions Further widens the gap between abortion and nonab- ortion maternal deaths." Second, if the growing: number of legal abortions were substituted solely for live births then (I) the number of women at risk of death from abortion would increase, while (2) the number at risk of death from other causes of pregnaney and childbirth would decrease. If this were to occur, the abortion mortality ratio should fall as rapidly than the nonabortion ma- ternal mortality ratio: the opposite appears to have oc curred. Third, by 1975, illegal abortions had nearly been climinated in the United States. We estimated that fewer than 20,000 clandestine procedures were per- formed in the United States in 1974,* compared with 200,000 to 1,200,000 per year during the 1960's. The small number of illegal abortions after 1974 would limit the relative ‘of substituting legil for legal abortions. Thus, it may be that such factors as operator skill, abortions at ear~ lier gestations. and better procedures, which affect the safety of legal abortions, have improved over the last several years, However, other sources of bias in the data must be considered before accepting the hypothesis thatthe i creasing availability of legal abortion contributed in part to the decline in abortion deaths, First. ascertai fluence on abortion mortality ratio