somatic complaint score between smokers and nonsmokers. However, when smokers were compared with nonsmokers of the same parity, education, work history, and psychosomatic complaint seore. cigarette smokers still had a significantly higher proportion of smal] infants than did nonsmokers. .\s previously mentioned, whenever other factors known or suspected to influence birth weight have been controlled, cigarette smoking has always been demonstrated to have an inde- pendent and significant effect. Ounsted (49) offered evidence that the best predictor of the birth weight of a mother’s future offspring was the birth weight of her previous children. Herriott, et al. (35) found prematurity rates for previous pregnancies among smokers to be markedly higher than among nonsmokers, independent of parity, height, and social class. Evidently a woman whose previous infants have been small tends to continue to have relatively smaller than average infants in subsequent pregnancies. The question is. will those infants be even smaller than expected tf she smokes? Goldstein, et al. (28). in a comprehensive review, proposed a research design in which a woman would serve as her own contro] to compare outcomes of pregnancies during which she smoked with those during which she did not with consideration of the effect of parity on the outcome. Yerushalmy (772) has recently tested this type of research design. using data from his Oakland Growth Study. With information on the age at which a woman began to smoke cigarettes, her smoking status during the pregnancy actually studied, her prior reproductive experience, and the outcome of her present pregnancy, the author compared the outcomes of pregnancy during periods of smoking and nonsmoking using the woman as her own control. As the author noted, “If smoking causes the increase in Jow-birth-weight infants, then the incidence of low birth weight for infants born to smoking mothers during the period before they acquired the smoking habit. should be relatively low. If, on the other hand, the high incidence of low birth weight is due to the smoker, then it should be high for infants of future smokers also when they were born before their mothers started to smoke.” Yerushalmy then proceeded to compare the reproductive experiences of four groups of women: (a) Those who smoked in none of their ‘pregnancies, (b) those who smoked in all of their pregnancies, (c) those who were smoking now but previously had not smoked during some pregnancies (future smokers), and (d) those who were ex- smokers now but had previously smoked during some pregnancies. These outcomes are shown in figure 5. The incidence of low-birth- weight infants in the pregnancies of the future smokers. before they started to smoke, was similar to that for women who smoked in every pregnancy, which was significantly higher than that of infants from 426 mothers who had never smoked. Ie also noted that ex-smokers, during the period before they quit, gave birth to relatively few low-birth- weight infants; the imcidence was significantly lower than for mothers who smoked during all of their pregnanctes. He concluded that the findings cannot be eastly reconciled with a cause-ctfect basis for smok- ing and birth weight. He said, “Rather the evidence appears to support the hypothesis that the higher incidence of low-birth-weight infants is due to the smoker, not the smoking.” There are several considerations which limit the interpretations which can be drawn from this study. The information on smoking behavior of the women during past pregnancies was apparently de- rived from the woman's age when she began to smoke, her smoking behavior early in the study pregnancy, and the age at which she had her prior pregnancies. Thus, if the woman reported that she began smoking at a certain age, and that she was still smoking at the time of the study. it was apparently inferred that she had smoked during all of her pregnancies. Since no questions were specifically asked about actual smoking behavior during each previous pregnancy, it is possible that the woman indeed had not smoked during every pregnancy or that the amount or way she smoked had differed from current smoking Figure 5.—Percent of tow birth weight white infants by smoking status of their mothers. Gravidas’ smoking habits Percent low birth weight infants in previous pregnancies Nonsmoker 7 (during alt pregnancies) | 2,529 Nonsmoker |° (future smoker) 9.5 210 Smoker a |" + (during all pregnancies) a 2,076 Smoker ee (future ex-smoker) 6.0 | 651 Percent “Difference is statistically significant (P <0.01). **Difference is Statistically significant (P <0.02). SOURCE: Adapted from Yerushalmy, 3. (112). 427 habits. This would be important to know given the strong dose- response relationship which has been established between cigarette smoking and low birth weight, and would tend to make the reproduc- tive outcomes for ex-smokers similar to those of nonsmokers, and diferent from those of women who smoked in all pregnancies. For ex-smokers, the age at which smoking began was not elicited. Hence, some of the infants of ex-smokers may have been born before their mothers acquired the smoking habit. This would also tend to make the reproductive experiences of ex-smokers more like those of nonsmokers and different from those of women who smoked in all pregnancies. No direct adjustment for age, parity, and other variables was reported, although Yerushalmy stated that the study population was limited to the births that occurred to women at age 25 years or less. He noted that, “In order to adjust for parity, the same comparisons were performed for firstborn infants only. The numbers were reduced considerably, but the same tendencies as found above were noted.” However, no data were presented. Primiparous births and births in teenagers are strongly associated with the delivery of low-birth-weight infants. If the pregnancies which occurred among future smokers included a predominance of very young women and primiparous births, the reproductive experiences of future smokers would tend to be similar to those of women who smoked during all pregnancies, and diiferent from those of nonsmokers. In the absence of more precise information on actual smoking behavior during pregnancy and more rigorous adjustment for maternal age, this study does not provide a critical test of the hypothesis that it is the smoking during pregnancy which is responsible for the high proportion of small-for-dates in- fants born to women who smoke. Experimental Studies Sropres IN ANIMALS Tobacco Smoke Several investigators have demonstrated that exposure of pregnant rats or rabbits to tobacco smoke leads to a reduction of birth weight in the offspring, as compared to controls (23, 87, 117). Younoszai, et al. (117) reported data from studies in rats which indicated that some agent present in cigarette smoke other than nicotine was responsible for the reduction in birth weight observed. The authors suggested that carbon monoxide might also not be responsible for the retardation of 428 fetal growth; however, the evidence presented was inadequate to support a firm conclusion. Haworth and Ford (33) recently extended the experiments of Younoszai. A group of pregnant rats was exposed to cigarette tobacco smoke for 6 to 8 minutes, five times a day, from days 3 to 20 of ges- tation. These rats were compared with another group whose food intake was restricted to the amount actually consumed by the tobacco- exposed rats, and both were compared to a well-fed control group. The animals in both experiments were killed on the 2ist day of gestation, and weights of the entire body, the liver, and the kidney of each fetus were recorded. The total average fetal weight of the group exposed to tobacco smoke was significantly lower than that of both the food-restricted and control groups. The fetal weights of the latter two groups were quite similar. Protein and DNA analyses were performed separately on the entire forebrains and hindbrains of the fetuses and on the entire carcass. Both DNA and protein were sig- nificantly and proportionately reduced in the carcass and hindbrains of the animals exposed to tobacco smoke. This implies that cell number was reduced and cell size was normal, and suggests that the exposure to tobaceo smoke either inhibited cellular proliferation or accelerated cellular destruction. Nicotine Several workers have demonstrated that chronic injections of large doses of nicotine into pregnant rats resulted in a reduction of birth weight of the offspring (7, 8, 9, 23, 40). Other investigators have de- termined that tritium-labelled nicotine injected into pregnant rabbits and C"-labelled nicotine injected into pregnant mice crossed the placenta to the developing embryo and fetus (89, 98). Kirschbaum, et al. (47) found no significant acute effects of small doses of nicotine, injected intravenously into near-term sheep, on blood gas composition, pH, blood pressure, or heart rate in either the ewes or their fetuses. The authors concluded that the influence of maternal smoking upon. the fetus must result from chronic effects or through the effects “of other variables which they did not study. Recently, Suzuki, et al. (94) evaluated the short-term effects of in- jected nicotine on the cardiovascular performance, acid-base status, and oxygenation of pregnant female Rhesus monkeys and their infants during the second half of gestation using the mothers as their own controls. Nicotine was administered either as a single intravenous dose of 0.5 to 1.0 mg. or as a continuous infusion of 100 pg./kg. over 495-028 O--73-——-9 429 a 20-minute period. The injection of nicotine in the larger, single dose into the mother produced a rise in maternal blood pressure and a fall in maternal heart rate, and an immediate fall in both fetal blood pressure and fetal heart rate followed by persistent hypotension and tachycardia in the fetus. Subsequent to the injection of 1.0 mg./kg. of nicotine into pregnant monkeys, in a single dose, significant changes in the arterial blood of the older fetuses included a fall in pH, a rise in base deficit, and a fall in oxygen tension. Carbon dioxide tension remained unchanged. Nicotine injected directly into the fetus prompted an immediate rise in fetal blood pressure and a fall in fetal heart rate. These responses were similar to those previously seen in the mothers following a direct injection of nicotine. The changes were more prominent in older rather than in younger fetuses. The authors sum- marized their findings by stating that: (a) fetuses in different ges- tational stages are differentially responsive to a given dose of nico- tine, probably because of the different stages of development of the autonomic nervous system; (b) diminished intervillous space per- fusion resulting from vasoconstriction in the uterine circulation ap- pears to be mainly responsible for the fetal asphyxta following the injection into the mother, because fetal hypotension and bradycardia were not preceded by the transient hypertension seen following the direct administration of nicotine to the fetus; (c) the differences be- tween the results obtained by Kirschbaum and by Suzuki, et al. may reflect either the considerable dosage differences or species differences; and (d) the doses which the authors employed were much larger than those which a human mother would absorb from usual cigarette smok- ing, but that differences in tolerance to nicotine between the Rhesus monkey and humans would imply that the dosages were, in fact, com- parable and that, “Hence, it can be envisaged that the concentration of nicotine which could be reached in the organism of a smoking mother would reduce oxygen availability to the fetus.” Carbon Monoxide Longo (45) has reviewed the work of several investigators which has demonstrated the transplacental passage of carbon monoxide from mother to fetus in animals. A recent study which related CO to birth weight was published by Astrup (2). He found that continuous ex- posure throughout gestation of pregnant rabbits to different levels of ambient carbon monoxide resulted in a statistically significant dose- related reduction in birth weight (table 2). The actual significance level was not reported. 430 TABLE 2.—Effect of carbon monoride erposure of pregnant rabbits on birth weight Group 1, Group 2, Group 3, 0 percent § to 10 percent 16 10 14 percent COHb COokib Colib Number of pregnant rabbits__________ 17 14 17 Total number of babies_._.-.... 116 S1 123 Average weight of babies in grams_____ 53.7 51.0 44.7 Source: Astrup, P. (£). Polycyclic Hydrocarbons Polycyclic aromatic hydrocarbons (PAH) such as benzo(a)pyrene (BAP) are constituents of cigarette smoke which have been impli- cated in the generation of cancers in many animal species (711). No studies presently available relate benzo(a)pyrene to a reduction in birth weight of exposed offspring. Evidence suggests, however, that BAP does reach and cross the placenta. Aryl hydrocarbon hydroxylase (AHH) is a part of the cytochrome P-450- containing microsomal enzyme system, present in many tissues of different species. This enzyme system is induced to hydroxylate polycyclic aromatic hydro- carbons after exposure of cells to PAH. Several investigators have utilized the inducibility of the enzyme system to demonstrate indirectly that benzo(a)pyrene and other polycyclic hydrocarbons reach the placenta and fetus. Welch, et al. (208) extended this work by administering the poly- cyclic hydrocarbon, 3-methylcholanthrene (8-MC), to rats during late gestation. The metabolism of benzo(a) pyrene was studied in vivo (1s- ing tritium-labelled benzo(a)pyrene) and in vitro. AHH activity was increased in fetal livers to adult levels by pretreatment with 3-MC Since a relatively high dose of polycyclic hydrocarbon was required to stimulate enzyme activity in the fetus, compared to the dose which stimulated placental enzyme activity, the authors suggested that the placenta may protect the fetus from exposure to polycyclic hydro- carbons. However, immaturity of the fetal enzyme system might also account for its apparent relative insensitivity to polycyclic hydro- carbons. Therefore, an exposure of the fetus to levels of poly- cyclic hydrocarbon similar to those experienced by the mother cannot be ruled out by the available data. 431 Schlede and Merker (86) have studied the effect of benzo(a) pyrene administration on aryl hydrocarbon hydroxylase activity in Che mater- nal liver, placenta, and fetus of the rat during the latter half of gestation. The pregnant animals were treated with large oral doses of benzo(a)pyrene 24 hours prior to sacrifice. Control rats had no detectable levels of aryl hydrocarbon hydroxylase in their placentas. Treatment with benzo(a) pyrene resulted in barely detectable placental levels on gestation day 13, but steadily rising values until day 15, and then constant levels thereafter. No activity was detected in the fetuses of untreated controls. In the treated animals, the fetal enzyme activity rose steadily from the 13th to the 18th day of gestation. The authors concluded that the stimulatory effect of benzo(a)pyrene treatment on aryl hydrocarbon hydroxylase activity in the fetus demonstrates that benzo(a)pyrene readily crosses the rat placenta. Srupres 1 Huxrans Carbon Monoxide Smokers and their newborn infants have significantly elevated levels of carbon monoxide as compared with nonsmokers and their infants (31, 34, 88, 116). Recently, Baribaud, et al. (5) studied 50 nonsmokers and 27 cigarette smokers and their newborns. All smokers inhaled. The authors found that the mean level of CO content in the blood of non- smokers was 0.211 volumes percent compared with 0.672 volumes per- cent in the blood of smokers. The values for blood samples from the umbilical cords of their newborns were 0.352 and 0.949 volumes per- cent, respectively. Moreover, a definite dose relationship was found between CO levels and number of cigarettes smoked. Younoszai, et al. (716) found, in addition to elevated carboxyhemo- globin levels among the infants of smoking mothers, significant elevation of mean capillary hemotocrits and significant reduction of standard bicarbonate levels, as compared to the infants of nonsmoking mothers. Since no evidence for nicotine effects upon blood glucose, serum FFA levels, or urinary catecholamines, or for hypoxia was present, they concluded that the higher hematocrit levels in the infants of smoking mothers may have represented a compensatory response to the decreased oxygen-carrying capacity of the blood due to the presence of carboxyhemoglobin. Longo (44) pointed out that a level of 9 percent carboxyhemoglobin in the fetus is the equivalent of a 41 percent decrease in fetal blood flow or fetal hemoglobin concentration. In reviewing the studies of CO levels in human mothers and their newborns, he made the follow- 432 ing comments: “These samples were obtained at the time of vaginal delivery or Cesarean section and may not accurately reflect the normal values of (COHb), for several reasons. The number of cigarettes smoked by the mothers during labor may be less than their normal consumption and was not specified in these studies. The blood sam- ples were collected at varying time periods following the cessation of smoking. In addition, many of the samples were probably taken early in the day before COHb levels had built up to the levels reached after prolonged periods of smoking. Thus actual levels of (COHb) » and (COHb)r may be higher than the reported values.” Polycyclic Hydrocarbons The results of several studies concur that cigarette smoking is strongly associated with the induction of aryl hydrocarbon hydrox- ylase in the human placenta (18, 38, 61, 99, 109). This finding implies that benzo(a)pyrene or other polycyclic hydrocarbons reach the placenta. To date, evidence to support the passage of polycyclic hydro- carbons through the placenta to the human fetus has not been published. Vitamin B,. and Cyanide Detoxification McGarry and Andrews (48) determined serum vitamin B,, levels in 826 women at their first prenatal clinic visit. They found that the serum levels for smokers were significantly lower than for nonsmokers. After adjustment for gestational age, parity, social class, hemoglobin level, hypertension, and maternal weight, smokers stil] had signifi- cantly lower levels of Bie. They also found a direct, statistically sig- nificant dose-response relationship between cigarettes smoked and serum vitamin By,» level. They again confirmed the relationship be- tween smoking and low birth weight. The authors suggested that the lowered vitamin B,, levels reflect a disorder of cyanide detoxification. Cyanide is a demonstrable ingredient in cigarette smoke (29, 60, 62, 64,68, 74,91). Vitamin C Venulet (105, 106, 107) has demonstrated that the vitamin C level is significantly lower in the serum of women who smoke cigarettes during pregnancy, compared to values for their nonsmoking counter- parts. Possible Mechanisms The following mechanisms have been proposed for the production of low birth weight and other unfavorable outcomes of pregnancy following exposure to cigarette smoke: 433 1, A direct toxic influence of constituents of cigarette smoke upon the fetus (2, 5, 50, 51, 117). - Decreased placental perfusion (94). 3. Decreased maternal appetite and diminished maternal weight gain with secondary effects upon the fetus (6, 33, 36, 65, 75, 99- 17). . A direct effect upon the placenta (36, 57, 65, 110). . An oxytocic effect on uterine activity (44). . A disturbance of vitamin B,, metabolism (48). . A disturbance of vitamin C metabolism (703, 106, 107). to “Tm On He Of the potential mechanisms, available evidence suggests that neither decreased maternal appetite and decreased maternal weight gain nor a direct effect upon the placenta are responsible for a sig- nificant reduction in birth weight. Existing evidence does not permit firm conclusions concerning the relative significance of the remaining mechanisms. Timing of the Influence of Cigarette Smoking on Birth Weight Several investigators have published results which bear on the time period during which exposure to cigarette smoke most affects fetal growth. Lowe (46) and Zabriskie (/18) have offered evidence which suggests that cigarette smoking influences fetal growth most during the second half of pregnancy. Butler, et al. (15) found that the birth weights of infants of women who did not smoke after the fourth month of pregnancy were essentially the same as those of the infants of nonsmokers. This implies that the influence is most probably exerted after the fourth month of pregnancy. Herriott, et. al. (35), however, found that women in lower socioeconomic classes who gave up smoking early in pregnancy tended to have intermediate weight babies as com- pared with nonsmokers and persistent smokers, but his numbers of women were small and the results were not statistically significant. Underwood, et al. (100) found that cigarette smoking in any single trimester was associated with a lower birth weight of the infant, although the difference between the birth weights of infants of women who smoked only during a single trimester and infants of non- smokers was not statistically significant because of small numbers. Several investigators have detected a nearly constant difference be- tween the birth weights of the infants of smokers and nonsmokers, delivered during the last month of pregnancy, following gestations of comparable length {fig. 1, (7/)]. Although this observation is 434 compatible with the suggestion that the influence of cigarette smoking upon the fetus occurs prior to the last month of pregnancy, it is based upon data derived from cross-sectional rather than longitudinal studies. The results of many human epidemiological studies sugeest that maternal smoking prior to pregnancy does not influence fetal weight gain (15, 25, 46, 49, 113). Site of Action at the Tissue and Cellular Level The use of labelled nicotine (98) and the preparations of autoradio- grams have permitted the localization of nicotine within the tissues of the fetus and mother. Tjalve, et al. (98) found high levels of nico- tine in the respiratory tract, adrenal, kidney, and intestine of 16- to 18- day mice fetuses. The use of other labelled constituents during various parts of gestation might further the understanding of how certain ingredients in cigarette smoke produce an impact upon birth weight. Haworth and Ford (33) have reported data which suggest that the reduction of birth weight of rat fetuses caused by the action of the ingredient.(s) of tobacco smoke results from a reduction in cell number, but not in cell size. Significance of the Association Among all women in the United States, cigarette smokers are nearly twice as likely to deliver low-birth-weight infants as are non- smokers. Assuming that 20 percent of pregnant women in the United States smoked cigarettes through the entire pregnancy (extrapolated from data on changes in smoking behavior during pregnancy collected for the British Perinatal Mortality Study), taking into account the apparently different risks of delivering a small-for-dates infant for— . Caucasian and non-Caucasian women who smoke during pregnancy, and considering the number of infants with a birth weight less than 2,500 grams born to Caucasian and non-Caucasian women, an excess of nearly 43,000 occurred in the 286,000 low-birth-weight infants among the 3,500,000 infants born in the United States in 1968, because of the increased risk among women who smoke of having small-for- dates infants. Since neonatal mortality is higher for low-birth-weigth infants, with gestational age held constant, the excess of small-for-dates infants among smoking mothers would imply a significant excess mortality risk as well. 435 Birth Weight Summary A causal association between cigarette smoking and fetal growth retardation is supported by the following evidence: 1. The resulte of all 42 studies in which the relationship between smoking and birth weight was examined have demonstrated a strong association between cigarette smoking and delivery of small-for-dates infants. On the average, the smoker has nearly twice the risk of delivering a low-birth-weight infant as tpat of a nonsmoker. _ This association has been confirmed by both retrospective and prospective study designs. A strong dose-response relationship has been established between cigarette smoking and the incidence of low-birth-weight infants. Available evidence suggests that the effect of smoking upon fetal growth reflects the number of cigarettes smoked daily during a pregnancy, and not the cumulative effect of cigarette smoking which occurred before the pregnancy began. . When a variety of known or suspected factors which also exert an influence upon birth weight have been controlled for, cigarette smoking has consistently been shown to be independently related to low birth weight. _ The association has been found in many different countries, among different populations, and in a variety of geographical settings. . New evidence suggests that if a woman gives up smoking by the fourth month of pregnancy, her risk of delivering a low-birth- weight infant is similar to that of a nonsmoker. _ The infants of smokers experience a transient acceleration of growth rate during the first 6 months after delivery, compared to infants of nonsmokers. This finding is compatible with viewing birth as the removal of the smoker’s infant from a toxic influence. . The results of experiments in animals have shown that exposure to tobacco smoke or some of its ingred{ents results in the delivery of low-birth-weight offspring. New evidence demonstrates that chronic exposure of rabbits to carbon monoxide during gestation results in a dose-related reduction in the birth weight of their offspring. . Data from studies in humans have demonstrated that smokers’ fetuses are exposed directly to agents within tobacco smoke, such as carbon monoxide, at levels comparable to those which have been shown to produce low-birth-weight offspring in animals. 436 Cigarette Smoking and Fetal and Infant Mortality Introduction Several previous studies ofthe relationship between cigarette smok- ing and higher fetal and infant mortality among the infants of smokers have been reviewed in the 1971 and 1972 reports on the health con- sequences of smoking (/0/, 102). In many of these studies, the authors combined two or more categories of fetal and infant mortality. Differ- ent mortality outcomes, such as spontaneous abortion, stillbirth, and neonatal death, are influenced by different sets of factors. Among other factors, the frequency of abortion is influenced by congenital infections, hormonal deficiencies, and cervical incompetency. In addi- tion to other factors, the frequency of stillbirth is influenced by pre- mature separation of the placenta, uterine inertia, and dystocia. Along with other factors, the frequency of neonatal death is influenced by gestational maturity, birth injucies, and delivery room and nursery care. Separate analysis of the relationship of cigarette smoking to each different mortality outcome, with control of the unique set of factors which influences it, may facilitate understanding of the relationship. Spontaneous Abortion Previous epidemiological and experimental studies of the relation- ship between spontaneous abortion and cigarette smoking reviewed in the 1971 and 1972 reports on the health consequences of smoking (J01, 102) form the basis of the following statements: The results of several studies, both retrospective and prospective, have demonstrated a statistically significant association between ma- ternal cigarette smoking and spontaneous abortion (43, 65, 70, 99, 118). Data from some of these studies have documented a strong dose- response relationship between the number of cigarettes smoked and_ the incidence of spontaneous abortions (70, 99, 178). In general, vari- ables other than cigarette smoking (e.g., maternal age, parity, health, desire for the pregnancy, and use of medication), which may influence the incidence of spontaneous abortions, have not been controlled. The results of the one study, in which adjustment for the woman’s desire for the pregnancy was performed, indicated that after such adjust- ment cigarette smoking during the pregnancy retained an association with spontaneous abortion of borderline significance (43). The time period during which cigarette smoking might exert an influence on the incidence of spontaneous abortions has not been determined. Abor- 437 tions have been produced in animals only with large doses of nicotine (23, 96, 104); the relevance of these studies for humans is uncertain. SpoxtTaNEocs ABORTION SUMMARY Although several investigators have found a significantly higher, dose-related incidence of spontaneous abortion among cigarette smokers as compared to nonsmokers, the lack of control of significant variables other than cigarette smoking does not permit a firm con- clusion to be drawn about the nature of the relationship. Stillbirth Epidemiological studies of the association between cigarette smok- ing and stillbirth previously reviewed in the 1971 and 1972 reports on the health consequences of smoking (101, 102) form the basis for the following statements: In one group of retrospective and prospective studies, a higher still- birth rate was found for the infants of smokers as compared ta those of nonsmokers (14, 25, $3). In another group of retrospective and prospective studies, no significant difference was detected in the still- birth rate among the infants of smokers and nonsmokers (/6, 20, 85, 99, - 100). Differences in study size, numbers of cigarettes smoked, or the presence or absence of control of variables, such as age and parity, which may influence stillbirth rates, were probably not sufficient to explain the differences in results obtained. Several recent epidemiological studies have added to our under- standing of the relationship between cigarette smoking and stillbirth. Niswander and Gordon (63) have reported data from 39,215 preg- nancies followed prospectively and collected between 1959 and 1966 at 12 university hospitals in the United States. A random sample of women who presented to hospital prenatal clinics were enrolled in the study. The authors reported no increase in stillbirths among white smokers as compared with white nonsmokers. A higher incidence of stillbirths was found among black women who smoked than among nonsmoking black women. and a dose-response relationship with 5 suggested, although the findings did not attain statistical significance. The results were not adjusted for other vari- ables. Rush and Kass (82) found, in a prospective study of 3,296 City Hospital, a nonsignificant Increase in cigarettes smoked wa pregnancies at Boston 438 stillbirths among white women who smoked, but a statistically signifi- cant increase in stillbirths among black women who smoked (P<0.02). These findings are consistent with those previously outlned by Frazier, et al. (25) and Underwood, et al. (99). Rumeau-Roquette (8/), in a prospective study of 4,824 pregnancies in Paris, demonstrated that the risk of stillbirth was significantly higher for cigarette smokers than for nonsmokers (P<0.001). The authors also presented evidence that a woman with either a previous stillbirth or at least one prior infant weighing less than 2,500 grams at birth was significantly more likely to have a future stillborn infant than a woman without such an obstetrical history. After previous obstetrical history was controlled, smokers still retained a statistically significant increased risk of subsequent stillbirth as compared to non- smokers (P<0.01). Of further interest was the finding that among women who previously had delivered only living infants, weighing over 2,500 grams, cigarette smoking had no influence on the stillbirth rate. Previous experimental studies were reviewed in the 1971 and 1972 reports on the health consequences of smoking (7/01, 102). The authors demonstrated that exposure of pregnant rabbits to tobacco smoke and pregnant rats te large doses of injected nicotine resulted in a signifi- cant increase in stillbirths (7, 8, 23, 87). STILLBIRTH SUMMARY 1. The results of recent studies suggest that cigarette smoking is most strongly associated with a higher stillbirth rate among women who possess less favorable socioeconomic surroundings or an unfavorable previous obstetrical history. In the United States, black women have higher stillbirth rates than white women. The finding that cigarette smoking is associated with an even greater difference between the stillbirth rates of the two groups merits special attention. These findings may provide at least a partial | explanation for the lack of a significant difference in stillbirth rates between smokers and nonsmokers, which some investigators have found. 2. The results of experiments in animals demonstrate that exposure to tobacco smoke and some of its ingredients, such as nicotine, can result ina significant increase in stillbirth rate. 439 Late Fetal and Neonatal Deaths Considerable variation has occurred in the definition of the study population among the studies in which the relationship of cigarette smoking to fetal mortality (other than abortion) and early infant mortality was examined. The most commonly identified study popula- tions have been perinatal deaths, neonatal deaths, and late fetal plus neonatal deaths. Perinatal deaths are a combination of late fetal deaths (ie., stillborn infants) and deaths occurring within the first week of life. Neonatal deaths include all deaths of liveborn infants within the first 28 days of life. EpPipeMIOLOGICAL STUDIES Most of the earlier epidemiological studies of the association between cigarette smoking and late fetal plus neonatal mortality were reviewed in the 1971 and 1972 reports on the health consequences of smoking (101, 102). A review of previously unreported studies (67, 76), as well as reexamination of previously cited studies, forms the basis of the following statements: The results of several prospective and retrospective studies indicate a statistically significant higher late fetal and/or neonatal mortality for the infants of smokers compared to those of nonsmokers (/4. 17, 25, 43). The results of other prospective and retrospective studies iden- tified no significant difference in the mortality rates between the in- fants of smokers and nonsmokers (20, 65, 72, 85, 100, 115). If mortality rates were compared for those infants of smokers and nonsmokers weighing less than 2.500 grams, the infants of nonsmokers apparently had a considerably higher risk than did those of smokers. The results of recent studies, coupled with a critical review of the design and analysis of previous studies, and a reexamination of exist- ing data, may provide at least a partial explanation of discrepancies between the results of previous studies. Comparisons of the Mortality Risks of Low-Birth-Weight Infants Born to Smokers and Nonsmokers The perinatal mortality risk for infants weighing Jess than 2.500 grams appears to be lower for those infants born to women who smoke during pregnancy than for those born to nonsmokers (table 440 3). However, available evidence shows that cigarette smokers’ infants tend to be small-for-gestational age rather than gestationally pre- mature. Hence, within a given birth weight group, the infants of smokers are, on the average, gestationally more mature than those of nonsmokers. Data collected by the National Center for Health Sta- tistics (403) demonstrate that within a given birth weight group, the more gestationally mature an infant, the lower is its mortality risk (fig. 6). Thus, the difference in perinatal mortality risks experienced by the infants of cigarette smokers and nonsmokers, within comparable birth weight classes, reflects the facts that the two sets of infants are not of the same average gestational age, and that gestational age is a major factor influencing late fetal and neonatal mortality. An accu- rate estimate of comparative mortality risks for the infants of cig- arette smokers and nonsmokers requires adjustment for gestational age. For infants of comparable gestational age, lower birth weight is as- sociated with higher mortality (fig. 6). Since infants of cigarette smokers have, on the average, lower birth weights than the infants of nonsmokers, within groups of comparable gestational age, cigarette smokers’ infants should experience higher mortality rates than non- smokers’ infants of similar gestational ages. In a recent review, Meyer and Comstock (45/) provided a more extensive discussion of these points. TaBLe 3.—Compartson of the perinatal mortality for infants weighing less than 2,500 grams, of smokers and nonsmokers Perinatal mortality rate (deaths per 1,000 live births) Auther, reference Smokers Nonsmokers Underwood, et al. (100)_._.._.-.2-- 2-2-2 ++ 187 269 Ontario Department of Health (67)_....---_--- 232 300 Kullander and Kallen (43)_.......-..--------- 129 139 Rantakallio (76)_..-..---. 2-2 eee ee 288 344 Yerushalmy ! (112): Black women_______.._.---------------- 114 202 White women__.__..__-..--------------- 114 218 Butler and Alberman (14)______.-_.---------- 269 284 1 Reported neonatal mortality rates only. 441 Figure 6.—Neonatal mortality rates among single white births in nospitals (by detailed birth weight and specified gestation groups: United States). JANUARY 1 TO MARCH 31, 1950 400 ~ ~~ 200 \ © 28-31 weeks % Ne “ ~A 2? 100 . aa “ 32-35 x 80 }-—_ * “XN oa * weeks o *. 2 60 . = . °o ° “ ° 40 *. ct ~ . o + a . oc 20 *, . . * . ‘* J . . ‘. 10 “s . . . ee tf 4 1 1 _ti. i 4 lo ol ho to do to A ob DS Om HO OF HO os Hy RO ON NH MRE RO on an AN NA NN NN ONO mo BIRTH WEIGHT (in grams) SOURCE: U.S. Public Health Service, National Center for Health Statistics (103). Recent Studies The Ontario Perinatal Mortality Study (66, 67) was conducted among 10 teaching hospitals during 1960 and 1961. In this retrospec- tive study of 51.490 pregnancies, a statistically significant increase in the perinatal mortality rate was demonstrated for smokers’ in- fants as compared with those of nonsmokers; the infants of smokers experienced an overall relative risk of 1.27 (P?<0.001). Moreover, the investigators found a statistically significant dose-response relationship between the amount of cigarettes smoked and the perinatal mortality rate (P<0.001) (fig. 7). 442 Figure 7.—Perinatal mortality rate per 1,000 total! births by cigarette smoking category. wn = 40} = 2 33.4 2 cH oO 6 307 27.7 CL) 5 23.2 0 Q 2 7 C4) C4) 2 99 |b CH C4) 2 Cd CL) a & Cd) = i1o0f i. cy 4 a £ 5 Lo C4 Oy oO. OL, 0 LZ) 4 4 Nonsmoker <1 pack of =21 pack of cigarettes cigarettes per day per day Number of perinatal deaths: 659 425 220 Total births: 28,358 15,328 6,581 (P <0.001) SOURCE: Ontarfo Department of Health (66). Recently Butler, et al. (75) further analyzed the British Perinatal Mortality Study. They found a highly significant association between maternal smoking after the fourth month of pregnancy and both late fetal and neonatal deaths. Infants of smokers had an increase in the late fetal mortality rate of 30 percent, and an increase in the neo- natal mortality rate of 26 percent, compared to the infants of non- smokers. The overall mortality ratio of late fetal plus neonatal deaths was 1.28 (P<0.001). Given the large number of women in the study, and the significant changes in smoking behavior which occurred, they found it possible to consider the effect of a change in smoking 443 behavior between the beginning of pregnancy and the fourth monn on late fetal and neonatal mortality. A statistically significant and dose-related increase in mortality occurred among the infants of mothers who continued to smoke after the fourth month of pregnancy, as compared with the infants of nonsmokers and those of women who smoked prior to the pregnancy but gave up smoking by the fourth month of gestation. Niswander and Gordon (63) reported data from the prospective Collaborative Perinatal Study of the National Institute of Neurologi- cal Disease and Stroke. The 39,215 pregnancies registered at 12 uni- versity hospitals in the United States were almost equally divided between black and white women. They found a nonsignificant increase in perinatal mortality among the infants of white smokers as compared to those of white nonsmokers; the overall mortality ratio was 1.13 (P>0.1). The infants of black smokers, however, had a significantly higher mortality risk than did those of black nonsmokers; the mor- tality ratio was 1.18 (P<0.02). Moreover, a definite dose-response re- lationship between cigarettes smoked by pregnant mothers and mortality risk was shown for black infants. Black women were noted to smoke significantly fewer cigarettes, on the average, than white women. Rush and Kass (82) found, in a prospective study of 3.276 preg- nancies followed at Boston City Hospital, a nonsignificant increase in late fetal plus neonatal mortality rate among the infants of white women who smoked as compared to those of white nonsmokers. How- ever, the infants of black women who smoked had a statistically sig- nificant increase in mortality rate compared to the infants of black nonsmokers (P<0.01). The overall mortality ratio for black women who smoked was 1.86. The difference in frequency of stillbirth among the infants of smokers and nonsmokers was the primary factor which contributed to the significance of the difference in mortality rates. Analysis of Previously Reported Studies Previously reported studies can be divided into two groups: A group in which the late fetal plus neonatal mortality rates for infants born to cigarette smokers were significantly higher than those for the infants born to nonsmokers, and a group in which no significant differences were detected in the mortality rates for the infants born to smokers and nonsmokers. The results of several studies (14, 17, 25, 42, 43, 55, 84, 92) yielded mortality ratios ranging from 1.38 to 1.78. The results of other studies (20, 65, 76, 85, 100, 115) yielded mortality ratios ranging from 1.01 to 1.66. Both groups contained retrospective and prospective studies of comparable size. The two groups did differ 44h significantly, however, with regard to control of variables other than cigarette smoking which infuence perinatal mortality. Factors Which Influence Perinatal Mortality Other Than Smoking Butler and Alberman (/3), on data from the British Perinatal Mortality Study, employed a logit transformation analysis of variance, and demonstrated that maternal height, age, parity, social class, and severe preeclampsia all had a significant independent effect on late fetal and neonatal mortality. Rumeau-Roquette (8/) provided evi- dence that a previous stillbirth or low-birth-weight infant significantly increased the risk of a future stillbirth. Meyer and Comstock (5/) provided examples of how the differential distribution of smoking and other factors which are related to perinatal mortality, in a population of women, can bias data (e.g., black women have higher perinatal mortality rates than do white women, but black women smoke less than white women do. Hence, nonsmokers will tend to include more black women. and smokers more white women. This will tend to reduce anv differences between the groups in mortality rates.) Meyer and Comstock concluded, “Comparisons of mortality rates of smokers” and nonsmokers’ babies should be made within subgroups according to parity, socioeconomic status, and other appropriate risk factors, and not separated by birth weight.” In three of the studies in which a significantly higher mortality risk was demonstrated for the infants of smokers, adjustment for other variables was performed. The results indicated that, after such ad- justment, a significant independent association between cigarette smoking and infant mortality persisted (73 and 15, 17, 81). Of the studies which revealed no significant increase in mortality risks for smokers’ infants, one (7/15) controlled for race alone. Hence, at least part of the discrepancy in results between the two groups of studies may be explained by a lack of control of variables other than smoking. Another possible, at least partial, explanation of the discrepancy _ in results obtained by the two sets of studies is that cigarette smoke ~ may be more harmful to the fetuses of certain women than others. Several developing lines of evidence suggest that. this may be the case: 1. Cigarette smoking and socioeconomic background. Butler. et al. (15) noted that when data from the British Perinatal Mortality Study are grouped by social class of the mother’s husband, the late fetal plus neonatal mortality ratio for infants of smokers and nonsmokers in the upper social classes T and II is 1.10; the mortality ratio for the entire sample was 1.28. Rush and Kass (82) reviewed tho British Perinatal Mortality Study, along with several other studies, and noted that all have shown the strongest association between excess infant mortality and cigarette smoking among the infants of those 495-028 O—73——10 445 mothers with lower socioeconomic status. Comstock and Lundin (76) found excess mortality among smokers’ infants almost entirely con- fined to those whose fathers had a grammar school education or less. Several of the studies which revealed no significant difference in mor- tality among the infants of smokers and nonsmokers were conducted in predominately middle class populations (20, 100, 115). 2. Cigarette smoking and previous obstetrical experience. Peterson, et al. (72) had rigid criteria for entry into his study population of 7,740 women. He included only those women who pre- viously had healthy infants with a birth weight greater than 2,500 grams. He found a significant decrease in birth weight among smokers’ infants, but no significant increase in mortality rates. Rumeau- Roquette (87) found that among women who previously had delivered only healthy infants weighing more than 2,500 grams, cigarette smok- ing was not associated with an increased risk of stillbirth ; among those women with a previous stillbirth, smoking was significantly associated with increased risk of a future stillbirth. 3. Cigarette smoking and genetic differences. The consistent finding that the mortality risk for the infants of black smokers is higher than the risk for the infants of white smokers, even when the socioeconomic background for both is ostensibly similar, suggests that genetic factors also may interact with smoking to pro- duce enhanced risk (82, 99, 115). Available evidence suggests that if those women, who are already likely to have small infants for reasons other than smoking, smoke during pregnancy, their infants will be most unfavorably affected. This means that the women in the United States whose infants will be most affected by cigarette smoking are those who have an unfavor- able socioeconomic situation, have a history of previously unsuccessful pregnancies, and are black. EXvERIMENTAL STUDIES Studies in Animals Studies previously reviewed in the 1971 and 1972 reports on the health consequences of smoking (/01, 102) demonstrate that exposure of rabbits and rats to tobacco smoke and to injections of large doses of nicotine resulted in significantly increased late fetal and neonatal mortality. Astrup (2) has recently studied the effect of continuous exposure of pregnant rabbits to carbon monoxide on stillbirth rates. He found a significantly higher. dose-related incidence of stillbirths and deaths within the first 24 hours of life among the offspring of the experimental rabbits (table 4)- 446 TasLe 4.-Eyfect of carbon monoride erposure of pregnant rabbits on birth weight and neonatal mortality Group 1, Group 2, Group 3, O percent 8 to tO percent 16 to 18 percent COUb COlIb Colin Number of pregnant rabbits__.---.----- 17 14 17 Total number of babies_.......--------- 116 8k 123 Stillborn and babies died within first 24 hours____.__-___-------------+------ iy 78 144 (P<0.001) 11 percent. 110 percent. 136 percent. Bource: Astrup, P. (2). Studies in Humans Some investigators have examined the causes of death among the infants of smokers as compared with those of nonsmokers. Comstock, et.al. (77) found that infants of smokers died more frequently of as- phyxia, atelectasis, and immaturity. Kullander and Kallen (43) found abruptio placentae significantly increased as a cause of death among smokers’ infants. Butler and Alberman (74) found httle difference in the death rates for the infants of smokers and nonsmokers from iso- immunization and malformations, but higher rates were found for smokers’ infants in the groups in which death occurred before or dur- ing labor, or in which death resulted from massive pulmonary hemor- rhage, or pulmonary infection. As the authors noted, “The latter three are conditions known to be associated with small-for-dates babies.” They pointed out that distribution of causes of death in the smoking group could be accounted for almost entirely by the excess of low-birth-. weight babies. This supports the conclusion that the mechanism which affects birth weight also influences mortality. SIGNIFICANCE OF THE ASSOCIATION The following calculation is offered to give some idea of the order of magnitude of increased late fetal and neonatal mortality associated with cigarette smoking during pregnancy. If women who smoked dur- 447 ing pregnancy in the United States had an elevation in risk of 28 per- cent for late fetal and neonatal mortality. as demonstrated by Butler. et al. (15) for Britain, Scotland, and Wales, and if 20 percent of pregnant women smoked throughout the prernaney,' the higher risk of stillbirth and neonatal death for the infants of mothers who smoke cigarettes during pregnancy would account for approximately 4,600 of the 87,263 stillbirth and neonatal deaths in the United States in 1963. A Late Ferra, anp Neonata, Deatn Susr1ary strong, probably causal association between cigarette smoking and higher late fetal and infant mortality among smokers’ infants is supported by the following evidence: 1. Twelve retrospective and prospective studies have revealed a sta- tistically significant relationship between cigarette smoking and an elevated mortality risk among the infants of smokers. In three of these studies, of sufficient size to permit adjustment for other risk factors, a highly significant independent association between smokng and mortality was established. Part of the discrepancy in results between these studies and those in which a significant association between smoking and infant mortality was not dem- onstrated may be explained by a lack of adjustment for risk fac- tors other than smoking. - Evidence is converging to suggest that cigarette smoking may be more harmful to the infants of some women than others; this may also, in part, explain the discrepancies between the results of the studies in which a significantly higher mortality risk was shown for the infants of smokers compared to those of nonsmokers and the results of those studies in which significant differences in mortality risk were not found. - Within groups of similar birth weight, the infants of nonsmokers appear to have a higher mortality risk than do the infants of ciga- rette smokers. This results from the fact that the infants of non- smokers within such similar birth weight proups are on the average gestationally less mature than the infants of cigarette smokers. Available evidence indicates that within groups of sim- ilar gestational age, infants of lower birth weight experience a higher mortality risk. Since the infants of cigarette smokers are 2 Based on extrapolation of data on smoking behavior change during pregnancy from the British Pertnatal Mortality Study, which probably yields a conservative estimate. 448 small-for-gestational age, one should expect that if the infants of cigarette smokers and nonsmokers are compared within similar gestational age classes, the infants of cigarette smokers would have the higher mortality rate. The results of recent studies have documented a statistically sig- nificant dose-response relationship between the number or amount of cigarettes smoked and late fetal and neonatal mortality. New data suggest that if a woman gives up smoking by the fourth month of pregnancy, she will have the same risk of incurring & fetal or neonatal loss as a nonsmoker. Available evidence strongly supports cigarette smoking as one cause of fetal growth retardation. The causes of excess deaths among the infants of smokers are those associated with small- for-dates babies. Data from experiments in animals have demonstrated that expo- sure to tobacco smoke or some of its ingredients, such as nicotine or carbon monoxide, results in a significant increase in Jate fetal and or neonatal deaths. The results of studies in humans have shown that the fetus of a smoking mother may be directly exposed to agents such as carbon monoxide within tobacco smoke, at levels comparable to those which have been shown to produce stilJbirth in experimental animals. Sex Ratio Although a number of small studies have found a slight, usually statistically nonsignificant, increase in the proportion of female infants born to smokers, the three largest studies of Underwood, ct al. (48,505 pregnancies), Butler (15,791 pregnancies), and MacMahon (12,155 pregnancies) have found similar infant sex ratios among both smok- ing and nonsmoking mothers, with the expected slight excess of males among each (table 5). Summary Available evidence strongly indicates that maternal cigarette smok- ing dogs not influence the sex ratio of newborn infants. 449 Tasie 5.—Proportion of male infants delivered to smoking and non- smoking mothers Proportion of rosie infaots Author, reference Pregnancies ———--______—_ cputennes Smokers Nom smokers Underwood, et al. (100)__..---.-------- 48, 505 . 518 -519 None. Butler and Alberman (14)_.-__--------- 15, 791 . 518 . 516 Do. MacMahon, et al. (49)_._.-------------- 12, 155 . al3 ~al2 Do. Kulander and Kallen (43)__..--------- 6, 363 ~ 515 501 Do. Reinke and Henderson '(78)___--------- 3, 156 . 498 517 Do. Frazier, et al.' (26)-.....--------------- 2,915 - 472 - 505 Do. (P> 0.05) Kizer (42)._...----------------------- 2, 095 . 502 .493 None. Herriott, et al. (36)__------------------ 2,745 . 492 ~ 517 Do. Ravenholt, et al (77)_--.--------------- 2, 052 . 501 .533 P<0.05 Lowe (46)_._.------------------------ 2, 042 . 532 .529 None. Russell, et al. (83)__.__----------------- 2, G02 . 513 . 512 Do. ’ Black women. Congenital Malformations Previous epidemiological studies which examined the relationship between cigarette smoking and congenital malformations were re- viewed in the 1971 and 1972 reports on the health consequences of smoking (107, 102). Recently, the authors of the Ontario Perinatal Mortality Study (66, 67), a retrospective study of 51,490 births, re- ported no difference in malformation rate for the infants of smokers— and nonsmokers. The various studies of the association between ciga- rette smoking and congenital malformation have differed significantly with regard to study design, the type of population sampled, sample size and number of infants with malformations, the definition of mal- formation, and results (table 6). Previous experimental work was reviewed in the 1971 and 1972 reports on the health consequences of smoking (101, 102). The chick embryo has been employed in recent studies. The direct application of nicotine to the émbryo results in cephalic hematomas (26), malforma- tions of the cervical vertebrae (93), and anomalies of the heart (27), depending upon dose of nicotine and period of incubation in which exposure occurs. Anomalies of the limbs of chicken embryos can also be induced by exposure of the egg to high Jevels of carbon monox- ide (4). 450