These studies suggest unfavorable effects of maternal smoking
during pregnancy on the child's long-term growth, intellectual
development, and behavioral characteristics. Although these changes
are difficult to study because of the vast complexity of possible
antecedent and confounding variables, high priority should be given to
obtaining conclusive answers about the role of fetal exposure to
maternal smoking in these conditions. The fact that the direction of
observed differences in a variety of different studies is the same adds
to the urgency of this question.

Role of Maternal Weight Gain

In the search for mechanisms through which maternal smoking
reduces birth weight, the question has been asked whether it might be
an indirect result of reduced appetite, less intake of food, and lower
maternal weight gain. Several early studies reported no differences
between smoking and nonsmoking women in intake of food or in
weight gain and concluded that the effect of maternal smoking on
birth weight was not mediated in this way (8, 54, 76, 101, 141, 212).
Recently the question has been raised again by Rush in a study of
births to 166 women of whom 41 smoked throughout pregnancy. His
evidence showed that the mean weekly weight gain was reflected in
the infant's weight at birth (162). In a subsequent study, Davies, et al.
examined the interrelationships of cigarette smoking in pregnancy,
maternal weight gain, and fetal growth. By analysis of covariance of
480 mother-infant pairs from the total of 1,159 included in the study,
these authors stated: "Correction of birth weights within smoking
groups to a common mean maternal weight gain appears to remove
most of the differences between infants of nonsmokers and heavy
smokers, although technically these corrected means are still statisti-
cally heterogeneous." That is, the effect of smoking on birth weight
was still observed although diminished by these procedures. From this
the authors concluded that "a large part of the effect of maternal
smoking is mediated through maternal weight gain with only a very
small additional direct effect on the fetus. This suggests that
increasing weight gain in smoking mothers might prevent some of the
harmful effects of smoking on fetal growth." However, the alternative
explanation that lower maternal weight gain and fetal growth
retardation are both independently related to cigarette smoking in
pregnancy is also mentioned (34).

Other studies have not corroborated these findings. Mau reports
results of the German prospective study in which 6,260 pregnant
women were examined every month from the first trimester to
delivery and the children followed for up to three years. Smoking was
classified as none, 1 to 5,6 to 10, or more than 10 cigarettes per day. No
significant association was found between smoking habit and weight
gain. On the other hand, there was a close correlation between the

g--24


number of small-fordates babies and the smoking habit in a subgroup
of women with normal weight gain (10 to 15 kg). The proportions of
babies below the tenth percentile were 7.7 percent for nonsmokers, 8.4
percent at 1 to 5 cigarettes, 12.5 percent at 6 to 10, and 17.6 percent at
over 10 cigarettes per day. These babies had a general retardation of
weight, length, and head circumference rather than appearing
malnourished (IO?`). These findings are in agreement with the studies
of Miller and Hassanein, who found that the effects of smoking on
fetal growth did not appear to be related to poor maternal nutrition.
Mean weight gains during the last two trimesters of pregnancy were
not significantly different in smoking and nonsmoking mothers and
were above the mean weight gains recommended by the National
Research Council (118).

Meyer investigated the relationship of maternal smoking to
maternal weight gain and to birth weight, using data from the 31,733
births b English-speaking Canadian-born women included in the
Ontario Perinatal Mortality Study (113, 14.2, 1.43). As expected, birth
weight distributions shifted downward as maternal smoking level
increased. Maternal weight-gain distributions, on the other hand, were
the same for smokers and nonsmokers. Furthermore, the proportion of
infants weighing less than 2,500 grams increased with each level of
smoking (none, less than a pack, and more than 1 pack per day) within
each maternal weight-gain group from less than 5 pounds to more than
40 pounds. This evidence supports a direct effect of maternal smoking
on birth weight rather than one mediated through eating. Evaluation
of Rush's study (162) is difficult because of small numbers and because
of population-selection factors that led to large differences between
smokers and nonsmokers in age, parity, marital status, and education.
The study population of Davies, et al. (34) is more homogeneous and
contains 450 smokers, but both studies share a common problem in
interpretation. Meyer points out that an inevitable correlation exists
between maternal weight gain and birth weight insofar as both
increase with gestational age, necessitating careful control of this
factor. Furthermore, the fact that fetal weight is an increasingly
important component of maternal weight gain towards term (51
percent between 30 and 40 weeks) and accounts for a larger proportion
of a low-weight gain than of a high-weight gain ensures a considerable
degree of correlation between the two values. The same baby is
weighed twice, once while growing in. utero and contributing to
maternal weight gain, and again at birth. In this way the mother gains
weight because the baby is growing, and not vice versa. Meyer
concludes that efforts to prevent or reduce smoking during pregnancy
should have greater benefits for mother and child than would efforts
to increase food intake among women who smoke (113).

8-25


Evidence for Indirect Associations Between Smoking and Birth
Weight

Yerushalmy has suggested that smoking is an index to a particular
type of reproductive outcome and does not play a causal role in the
production of small-for-dates infants (206-208). The line of reasoning
and evidence presented by Yerushalmy and the responses to it are
discussed in detail in the 1973 report on The Health Consequences of
Smoking (192). The problems inherent in Yerushalmy's study, in which
he found a higher percentage of low birth weights among 210
nonsmokers who later became smokers than among nonsmokers who
did not take it up, have been described. The most serious of these
problems is the bias introduced by the study design resulting in
significantly younger ages for the "future smoker" group (mean age
19.70 20.15) than for his nonsmokers (22.102 0.04); the doubly retro-
spective nature of the information gathered (women being asked about
smoking habits at the time of previous pregnancies); and lack of
control for other important factors influencing birth weight, such as
primiparity and sex of child.

Silverman addressed the question of whether the smoker rather than
the smoking was responsible for increased frequency of low birth
weight by comparing pairs of births to the same woman, using data
from the 1963 private census of the population of Washington County,
Maryland (28). In this census all members of the household were listed
with birth dates, and all members were asked whether and how much
they smoked and when they had started. Using these data, Silverman
constructed a population of pairs of births that occurred during the 17-
year period prior to the census date of July 15,1963. Assuming that the
mothers did not stop smoking during pregnancy and that the age of
starting was accurately reported, she was able to compare birth
weights in first and second births of 143 women who smoked during
the second pregnancy, but not during the first, with corresponding
birth weights from 382 women who smoked during neither pregnancy
and 491 women who smoked during both pregnancies. The many
problems inherent in this study were faced, and adjustments were
made insofar as possible. For example, as in Yerushalmy's study,
significantly more of the future smokers (44.8 percent) were under 20
years of age at the time of the first study birth, compared with 24.5
percent of the continuing nonsmokers. Young, primiparous mothers
are known to have lighter babies than older mothers with higher
parity. When weights were compared specific for maternal age and sex
of child, the mean birth weight for the first member of the birth pair
was lower in four out of six comparisons and higher in two. With
simultaneous adjustment for the effects of infant sex, maternal age,
and birth order, there were no significant differences in mean birth
weight difference among pairs in which the mother smoked during
both pregnancies and pairs in which the mother smoked during the

8-26


second pregnancy of the pair, but not the first. Comparison of the
mean birth weights for the first infants in each pair showed that
future smokers had babies who weighed less than those of women who
did not take up smoking and more than those of women who were
already smokers and continued to smoke. Silverman concluded: "These
findings neither confirm nor deny the hypothesis that the smoker
rather than the smoking per se causes a reduction in birthweight"
(171).

Evidence for a direct effect of maternal smoking on fetal growth as
presented in this chapter is extremely strong. Furthermore, the
biological effects of carbon monoxide, nicotine, and other known
components of cigarette smoke are compatible with the findings from
epidemiologic studies. Therefore, there seems little value in arguing
that this direct effect does not exist. On the other hand, smokers are to
some extent self-selected, and comparisons of "smokers" and "non-
smokers" in a population reveal differences between them. These may
be related to calendar time trends, peer group influence, cultural and
ethnic background, social class, or personality type. Because the
relationship between maternal smoking and birth weight is so strong,
these differences do not obscure it. More problems arise from lack of
adjustment for differences between smokers and nonsmokers in the
distribution of such factors as age, parity, socioeconomic status, and
race when the relationship of maternal smoking to perinatal mortality
is under study; these issues are discussed in detail in another section of
this chapter. In addition, attention should be paid to the possibility that
psychological makeup and strength of addiction to cigarette smoking
may have an independent influence on some of the outcomes being
studied. Future studies should not only adjust for independent factors
that influence whether or not a woman becomes a smoker and smokes
during pregnancy but should also distinguish between the effects of a
personality type that adopts smoking and the physical effects of the
smoke on mother, placenta, and fetus.

1. Babies born to women who smoke during pr&nancy are on the
average 200 grams lighter than babies born to comparable women who
do not smoke. The whole distribution of birth weights of smokers'
babies is shifted downward, and twice as many of these babies weigh
less than ~,.XIO grams compared with babies of nonsmokers. There is
abundant evidence that maternal smoking is a direct cause of the
reduction in birth weight.
2. Birth weight is affected by maternal smoking independently and
to a uniform extent, regardless of other determinants of birth weight.
The more the mother smokes, the greater the reduction in birth weight
of the baby.

8-27


3. The ratio of placenta weight to birth weight increases with
increasing levels of maternal smoking., This increase may signify a
response to reduced oxygen availability due to carbon monoxide and
may have some survival value for the fetus.
4. There is no overall reduction in the duration of gestation with
maternal smoking, indicating that the lower birth weight of smoke&
infants is due to retardation of fetal growth.
5. The pattern of fetal growth retardation that occurs with maternal
smoking is a decrease in all dimensions: body length, chest circumfer-
ence, and head circumference are smaller if the mother smokes.
Smokers' babies are short for dates as well as light and do not exhibit
reduction in ponderal index.

6. Studies of long-term growth and development give evidence that
smoking during pregnancy may affect physical growth, mental
development, and behavioral characteristics of children at least up to
the age of 11.

7. Overwhelming evidence indicates that maternal smoking during
pregnancy affects fetal growth rate directly, that fetal growth rate is
not due to characteristics of the smoker rather than to the smoking nor
mediated by reduced maternal appetite, eating, and weight gain.

Cigarette Smoking and Fetal and Infant Mortality
Overview

In contrast with the strong, consistent relationship of maternal
smoking to reduced birth weight, the relationship of maternal smoking
to perinatal mortality has been marked by variation in the level of
increased risk for women who smoke. This has led to controversy as to
whether there truly are lethal effects for the fetus or neonate caused
by maternal smoking.

Earlier epidemiological studies of the association between maternal
cigarette smoking and perinatal mortality (fetal deaths, neonatal
deaths, or perinatal deaths) were reviewed in the 1971,1972, and 1973
reports on The Health Consequences of Smoking (190-192). The 1971
report gave details of 12 studies of maternal smoking and the incidence
of spontaneous abortion, stillbirth, and neonatal death (20, 41, 54, 87,
101, 141,151,164, 166,188, 206, 212). The increased risk of loss among
smokers varied from study to study. Inconsistencies between studies
were described, and it was noted that both smoking habits and
perinatal loss were influenced by such factors as social class, maternal
age, and parity. Rush and Kass reviewed the English language
literature in 1972 and found reports of 12,333 perinatal deaths and
abortions with a mean excess perinatal loss for smokers of 34.4 percent.
Where reported, excess loss was higher among the poor and among
blacks. Their study of black and white women in Boston showed excess

8-B


mortality risks of 36 percent for black smokers and 11 percent for
white smokers (166).
The 1973 report (192) summarized studies that were published up to
that date and contained a critical analysis of known reasons for
variability in the strength of the association between maternal
smoking and increased perinatal loss. Much of the controversy about
whether maternal smoking did or did not cause fetal or neonatal loss
centered around the basically irrelevant issues of whether studies were
"prospective" or "retrospective" (usually referring to the time at which
smoking information was obtained rather than to whether the study
was based on a cohort of births or on a set of cases and controls), and
on whether or not the differences were "statistically significant."
Classification of the studies reviewed in the 19'73 report according to
statistical significance revealed that studies in which the higher rates
of mortality for the infants of smokers compared with nonsmokers
reached a significant level (usually p<O.OS or smaller) (20, 22, 30, 54,
86, 89, 124, 142, 14.9, 165, 180) had mortality ratios (smoker rate:
nonsmoker rate) that ranged from 1.33 to 1.78, whereas studies in
which significant levels were not reached (41, 141, 151, 155, 166, 189,
207) had mortality ratios that ranged from 1.01 to 1.06. Both groups
contained retrospective and prospective studies of comparable size.
Statistical significance obviously depended upon the combined effects
of the risk ratio and the size of the study. A further source of
controversy in this matter was the fact that when one compares
neonatal death rates for low-birth-weight babies only, the low-weight
babies of smokers have lower death rates than those of nonsmokers.
This apparently paradoxical relationship is partly due to the relatively
greater maturity of the under-2,!500-gram smokers' babies. It is also
due to the fact that maternal smoking affects birth weight more
strongly than it does neonatal mortality. Because the denominators of
these rates include only babies under 2,500 grams, the downward shift
of birth weight with maternal smoking inflates the denominators and
lowers neonatal mortality rates for smokers. Numerators include a
majority of low-birth weight babies, whether or not the mother
smokes. This matter is discussed more fully in the 1973 report (192) and
in the commentary by Meyer and Cornstock (114).
In the 1973 report, analysis of reasons for variability between studies
included two important points. First was the observation that other
important variables might influence the results if they were unequally
distributed in comparison groups of smokers and nonsmokers. A
logistic transformation analysis of variance applied to data from the
British Perinatal Mortality Study demonstrated that in addition to
maternal smoking, maternal height, age, parity, social class, and severe
Preeciampsia had significant independent effects on late fetal and
neonatal mortality (Figure 5). Meyer and Comstock (114) provided
examples of how the differential distribution of smoking and other

8-29


factors could bias data. For example, as reported in the data from the
Collaborative Perinatal Study of the NINCDS (1959-X%6), U.S.
mortality rates were higher for black than for white babies, while
white women were more often smokers and smoked more cigarettes
than black women (1%`). Selection of births on the basis of smoking
alone would tend to include more nonsmokers who were black and at
high risk and more smokers who were white and at basically low risk,
thereby minimizing the apparent effects of maternal smoking on
perinatal loss. In three reported studies in which adjustment for other
factors was carried out, a significant independent association between
cigarette smoking and infant mortality persisted (20, 22, 30, 169). Of
the studies that revealed no significant increase in mortality risks for
smokers' infants, one (207) 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'
(292).

The second important point presented in the 1973 report was the
suggestion that cigarette smoking might be more harmful to the
fetuses of certain women than of others. Analysis of data by
socioeconomic status (2, 22, -29), race (137, 163, 188, 206, 207'), previous
obstetrical experience (22, 151, 169), and maternal age (20) indicated
that the increased perinatal mortality risk associated with- maternal
smoking varied considerably with these other factors (192).

Spontaneous Abortion

The results of several past studies have demonstrated a statistically
significant association between maternal cigarette smoking and
spontaneous abortion (74, 89, 141, 147, 188, 212). Data from some of
these studies have documented a strong dose-response relationship
between the number of cigarettes smoked and the incidence of
spontaneous abortion (147, 188, 212). Spontaneous abortions are
difficult to study because of problems in ascertainment. The most
complete ascertainment is possible when the mother's history of past
spontaneous abortions is used, despite problems of recall. Differences
in .rates between smokers and nonsmokers are largest when this
method is used (141, 212). In- prospective studies, many early
spontaneous abortions will be missed, and bias will occur if one group
tends to register earlier than the other. Nevertheless, higher rates of
spontaneous abortion are also reported among smoking mothers in
prospective studies (89). The study by Kullander and Kaellen counted
spontaneous abortions through the eighth month of gestation and
noted -that the largest increase was among smoking women whose
pregnancies were unwanted. Although this was a prospective study,
with smoking data collected repeatedly during prenatal care, the

method of analysis was retrospective. Rearrangement of their table to


FIGURE 5.-Theoretical cumulative mortality risk according to
smoking habit, in mothers of different age, parity, and social class
groups
SOURCE: Butler, N.B. (2~).

obtain incidence rates of spontaneous abortion for subgroups of
smokers and nonsmokers gives rates and relative risks of spontaneous
abortion by desideration of pregnancy (Table 4). More of the smokers'
than nonsmokers' pregnancies were unwanted (19 percent versus 13

8-31


TABLE I.-Spontaneous abortions by maternal smoking habit
     and desideration of pregnancy

Spontaneous abortions
per 100 pregnancies

Smoker:
nonsmoker
Relative risk

Total spontaneous abortions

Pregnancy wanted

Pregnancy unwanted

Smokers        Nonsmoker



9.4           7.2


78           6.6


16.0           11.9

1.61


120


la4

percent), but the increased risk of spontaneous abortion was seen
among smokers whether or not the pregnancy was wanted (89).

The method for studying `spontaneous abortions that may be the
least subject to error if carefully done is the traditional, retrospective,
case-control approach, used recently by Kline and coworkers (87). In
their study a log-linear analysis was used to test the hypothesis that
maternal smoking is associated with spontaneous abortion, controlling
for confounding variables such as age, number of previous spontaneous
abortions, induced abortions, and live births. Of the cases of spontane-
ous abortion, 41 percent were smokers compared with 23 percent of the
controls, giving an odds ratio of 1.8, This leads to the conclusion that
smoking during pregnancy is a risk factor for spontaneous abortion.

Perinatal Mortality

Most of the epidemiological studies about which questions of causality
have arisen have used perinatal death (late fetal and early neonatal),
neonatal death, or combinations of these as their outcome variable.
Ascertainment and recordkeeping may start at 20 weeks, at 23 weeks,
or at the time of registration. These differences in definition and
design affect the study results but are not fundamental to the basic
questions raised in the 1973 report and by other authors.

Progress toward resolving these questions has been made since the
1973 report through new studies and analyses in which attention is
paid not only to differences in the number of cigarettes smoked but
also to other characteristics of the study populations. A table from
Fabia's study of a 10 percent random sample of registered births in
Quebec in 1970-71 illustrates this approach (Table 5). Within subgroups
of the population by maternal age, parity, and years of school, the
relative perinatal mortality risk for smoking versus nonsmoking
mothers varies from 1.00 to 1.81 for categories with at least 10 deaths
(47). Table 6 (117) shows examples of a number of studies in which

8-32


TABLE 5.-Perinatal mortality rates per 1,000 live births to
     smoking and nonsmoking mothers, and relative risks
     for infants of smokers by maternal age, parity, and
     years of school (10 4% random sample of medical
     certificates of births in Quebec in 197&71)

Maternal
characteristica

Total births

Perinatal deaths per
1,609 live births

Smoker:
nonsmoker
Relative risk

Age
<=
534
35+

parity
0
l-3
4-b

Years of school
<a
e-11

Nonsmokers     Smokers

3,143.         12.1         16.1         1.23
3.717         12.6         13.2         1.05
751         23.0         41.7         1.81

2.798         14.2         18.7         1.32
3,959         11.2         11.2         1.00
860         21.8         36.1         1.66

14.5
128
13.5

18.8
19.7
( 8.9)

1.30
1.54
(0.W

Excludes birtha weighing lem than 1,001 pame.
Rnteainparenthaedbuedwfewerthml0deatha
WJBCE: Fabii J. (47).

perinatal mortality rates by maternal smoking are shown within
categories of other relevant factors. These studies show that perinatal
mortality rates vary with maternal smoking level and also with the
other factors shown. The general statement can be made that the
detrimental effect of maternal smoking on fetal survival is greater in
groups of women who already have a higher risk of perinatal loss for
other reasons. Women characterized by low social class, low level of
education, less than optimum maternal age, or being black have higher
risks of perinatal mortality than their counterparts, and their relative
increase in risk due to maternal smoking is enhanced. Studies in which
the population, by design or by chance, includes mainly or only women
without other reproductive risk factors show the smallest differences
between the risks of smokers and nonsmokers (22, SO, 47,137,155,163,
206).

A series of articles by Meyer, et al. reports analyses of data from the
Ontario Perinatal Mortality Study of all single births in 10 Ontario
teaching hospitals in 1960-61, including 51,490 births, 701 fetal deaths,
and 655 neonatal deaths (115, 116, 117). For the Ontario study,
Wnsored and supported by the Maternal and Child Health Branch of
the Ontario Department of Health (I.&Z, I@), detailed data were

8-33


TABLE O.-Examples of perinatal mortality by maternal smoking
     status related to other subgroup characteristics

Study population

No. of births


Non-
smokem   Smokers

      Perinatal or neonatal

Category   deaths/l,060 births Relative
                    risk*
  s;;;;m Smokers

British Perinatal Mortality
Survey, England, all
births

11,145

4,660

Social class
12 (high)
  s5

25.8     1.3     1.02
33.5     46.6     1.39

Washington Co. Maryland,
white

7.646

4.641

Father's
education
9-b years
<a Y-

14.4t   16.lt    1.12
li'.St   =.ot    2.16

Northern Finland, white



California, middle to upper
middle class

8.898    2346             232    22.4     1.01



              Race
6,667    3,726     White    ll.ot   11.3t     1.03
2219    1,071     Black    17.1t   21.5t     1.26

Boston City Hospital,
Prenatal Clinic

             Race
513     892      white     292    31.4     1.08
1W     636      Black     28.6    54.1     1.89

Collaborative Perinatal
Study, 12 U.S. centers

8,521



9,862

    Bate and cig-
    w&es/day
11,369     White     31.4
        l-10            31.5     1.00
        11+            38.2     1.22
8,166     Black     38.5
        l-10            41.5     1.08
        11+            57.4     1.49

Quebec, 10% sample of
qistered births

3,912

2,967

Maternal age
  <B
  E-34
  35-b

12.1     16.1     1.33
12.6     132     1.05
23.0    41.7     1.81

*Ratio of mortality rate for smokera' to nonsmoken' babies.
tNematal only.
SOURCE: Meyer, M.B. (117).

collected from routine records, and from interviews with mothers,
anesthetists, and attending physicians, and from autopsy records.
Results related perinatal mortality to social, demographic, and physical
maternal factors, prenatal care, histories of prior pregnancies,
complications of pregnancy, details of anesthesia, delivery, hospital
course, and survival of the infant up to 8 days. The interviews of

8-34


mothers included questions on the maximum amount smoked during
pregnancy, expressed as packages per day (142,143). The large size of
this study and the richness of its available information provided a
valuable resource for sorting out complex interrelationships between
maternal smoking, other factors, and perinatal loss. In the first article
of the series, the differential risk of smoking based on maternal
characteristics was demonstrated by extensive cross-tabulation of
perinatal mortality rates for 3 levels of smoking (none, less than a
pack, 1 pack or more per day) within 52 subgroups of other maternal
variables. Risk ratios for light smokers compared with nonsmokers
showed excess death risks of less than 10 percent for women of young
age, low parity, and normal hemoglobin. At the other extreme,
mothers of high parity, public hospital status, with previous premature
infants, or with hemoglobin under 11 grams and who were heavy
smokers (one pack or more per day) had increased perinatal mortality
risks of 70 to 100 percent. Risks for light smokers who had other
antecedent risk factors and for heavy smokers with otherwise good
prognosis fell between these extremes when compared with nonsmok-
ers. These relationships show how selection of a study population from
one end or the other of this spectrum of smoking-associated risk levels
would influence the relative risk found for smoking when no
adjustment is made for these other factors (117). Other studies in
which similar cross-tabulations have been made between maternal
smoking level and socioeconomic level, maternal age, parity, previous
pregnancy history, and other such factors have corroborated these
findings (2,22,29,47,102,169).

Because of possible interactions between maternal smoking and the
other independent variables, Meyer, et al. undertook further analysis
of the Ontario data to define and measure the independent effect of
maternal smoking on the risk of perinatal mortality. For this a
multiple regression analysis was used to compare the relative
importance of smoking and other factors in their influence on perinatal
mortality and on the frequency of low birthweight, of preterm
delivery, and of placental complications (115). When the rates of
Perinatal mortality by smoking were adjusted for the effects of all
other factors, perinatal mortality rates per thousand births were 23.5
for nonsmokers, 23.2 for smokers of less than a pack per day, and 31.8
for smokers of a pack or more per day. In other words, light smoking
increased the risk by 20 percent and heavy smoking increased it by 35
percent. This is a highly significant, dose-related, independent effect,
hut it is less strong than the relationship to perinatal mortality of
hospital pay status (a 55 percent increase for public status mothers),
W-parity differences, or a history of previous pregnancy loss (190
percent greater risk if there is a previous loss compared with
Primiparity or with a previous pregnancy with no fetal or neonatal
1(=4 (115).

8-35


TABLE 7.-Cause of stillbirth related to smoking habit

Cause of stillbirth

Percentage incidence

Nonsmokers

Smokers

Maternal disease
Maternal hypertension
Difficult labour
Antepartwn hemorrhage
Congenital malformation
Haemolytic disease
Infection
Anoxia (without obvious
Other cause stillbii

0.01             -
0.19            0.17
0.09            0.05
0.11            0.39
0.32            027
             0.13
0.01             -
024            OP
-             0.02

Macerated stillbirth (without obvious cause)

029            0.23

Total stillbirths                           1.30            1.54

SOURCE: Andlews. J. (8).

Cause of Death

The weight of evidence presented in this chapter clearly indicates that
maternal smoking does increase the risk of spontaneous abortion, early
and late fetal death, and early neonatal death. This being so, it is
appropriate to attempt to identify mechanisms of action and interme-
diate pathways between the cigarette smoke and the fatal event. Clues
to these mechanisms might be found if certain causes of death showed
an excess among the infants of smoking mothers. Several authors have
reported cause-specific mortality rates for the infants of smokers and
nonsmokers. Andrews and McGarry (2) reported stillbirth rates of 1.30
per 100 births for nonsmokers and 1.54 per 100 for smokers, among
which 0.11 and 0.39 were due to antepartum hemorrhage for
nonsmokers and smokers respectively. For neonatal deaths, causes
showing excess rates for infants of smoking mothers were "immaturi-
ty (no other cause), " "respiratory distress syndrome," and "pneumo-
nia," with overall rates of 1.10 and I.40 for nonsmokers, and smokers,
respectively (Tables 7 and 8). Comstock,`et al. (30) compared observed
neonatal deaths of smokers' babies with numbers of deaths expected at
nonsmoker rates. Out of 100 total observed deaths, smokers' infants
had excesses of 1'7 due to immaturity, 15 due to asphyxia and
atelectasis, and 7 due to birth injuries, with deficiencies of -7 due to
congenital defects and -4 due to "other," leaving a net excess of +28.
In the prospective study of 9,169 pregnancies carried out by Goujard,
et al. (&?), causes of stillbirth that increased significantly with
maternal smoking were "abruptio placentae" (p = .005) and "unknown
cause" (p=O.O005). Overall differences in stillbirth rates showed an
excess for smokers at a significance level of p=O.OOOl (Table 9).

8-36


TABLE &-Cause of neonatal death related to smoking habit

      Cause of neonatal death             Percentage incidence

                              Nonsmokers        Smokers

Immaturity (no other cause)                    0.25            0.36
Congenital malformation                      0.33            0.31
Pneumonia                              0.06            0.19
Asphyxia-atalectasi                         0.11            0.12
Birth injury                             0.03            0.09
lnf&ioa                               0.03             -
IiaemoIytic dii                          0.01            0.03
Respiratory dir&es syndrome                   0.09            0.16
othet                                 0.11            0.12


Total neonatal deaths                        1.10             1.40

6oURcE: Andrews, J. (S).

TABLE 9.-Stillbirths according to cause in relation to maternal
     smoking during pregnancy

Stillbii

Number of         % of
deliveries        smokem

Ci3mparison
with live
biihs t

cause of death:

VascUI~ . . . . . . . . . . . . . . . .
Abruptio pI&z?ntae . . . . . . .
Meehanii cause . . . . . . . . . .
M~lIatlecus (SyphiIis,
Bh, malfonMtions). . . . . . . . .
Unknown cause _ . . . . . . . . . . , . . . .
wailed records not avai-
IahIe. . . . . . . . . . . . . . .
TOW . . . . . . . . . . . . . . . _ . . . . . . .

8           2590
13           469c
13           15%


24           13%
37           35%


5            -
100           36%

p-o.005



p-0.0006


p-O.call

Ihhktbs . . . . . . . . . . . . . . . . 1 . . .      9069           1241

Meyer and Tom&a (116) have analyzed fetal and neonatal deaths
from the Ontario Perinatal Mortality Study (142, 143) to identify
causes of death that show an excess if the mother smokes and to
examine the relationship of these deaths to complications of pregnancy
and labor. Fetal and neonatal deaths by coded cause and maternal
smoking habit are shown in Table 10. For each cause the observed
numbers for smokers were compared with the number expected at
nonsmoker rates. The differences between observed and expected
numbers indicate the number of deaths in each category attributable
tc maternal smoking. Significance levels of the differences between
smoker and nonsmoker rates, based on the null hypothesis of no
difference, are shown for p values of 0.06 or less.

8-37


TABLE IO.-Fetal and neonatal deaths by coded cause and
      maternal smoking habit (En&h swakinp: mothers)

coded cause

    Ob3WWd

Nonsmoker    Smoker

fipec@d
smoker*

ObWWd-       P
expected
difference    value?

Fetal deaths
Unknown
Malformations
Hemolytic dii
Anoxia
Maternal cause
All others

Total

Neonatal deaths
unknown
Malfonnatioos
Hemolytic disease
Respiratory difficulty
Prematurity alone
Maternal cause
All others

52
22
7
46
33
2
16

Total               173        233
Total births         15240     16,549

75        125       81.4       43.6       0.003
32        24       34.7       -10.7       N.S.
11         15        11.9        3.1       N.S.
16        29        17.4       11.6       N.S.
31         45       33.7       11.3       N.S.
8         13        a.7        4.3       N.S.

173

251

187.9

56.5        -5.5       N.S.
23.9        0.1       N.S.
7.6        0.4       N.S.
50.0       13.0       N.S.
35.8       29.2       OX05
2.2        3.8       N.S.
17.4        -1.4       N.S.

63.1

193.3       33.6       0.06

0.003

N.S. - Not seat.
o Bac+edoanommokermte.

tP value derived from chi square bawd on a null bypotbesii of no differems between smokers and nonsmokera.
SOURCE: Meyer, Y.B. (116).

For fetal deaths, the largest category of coded cause was "un-
known," and by far the largest and most significant smoking-related
difference fell in this category (p=O.O03). Smokers also showed more
than expected fetal deaths due to anoxia and maternal causes and
fewer deaths than expected due to malformations. In other categories
only minor mortality rate differences were found between the two
groups. For neonatal deaths the largest cause of death category was
"unknown," but here there was no excess for smokers' infants. Most of
the smoking-related excess of neonatal deaths was among those
attributable to prematurity alone (p=O.O05), with additional numbers
in the related category of "respiratory difficulty." Differences
between observed and expected deaths in other categories were
negligible.

The tentative conclusion to be drawn from these findings is that
many of the excess fetal deaths associated with maternal smoking do
not have any recognizable pathology but occur from otherwise
unknown causes. A significant excess also occurs as a result of
antepartum hemorrhage or abruptio placentae. The excess neonatal
deaths among the infants of smokers appeared to be due to
prematurity and to related respiratory problems. In other words, these

8-38


deaths occurred in babies who were born preterm, but were without
other pathology. There is no convincing evidence that maternal
smoking increases the incidence of congenital malformations. Results
of published studies, reviewed in the 1973 report, show relative risks
for smokers versus nonsmokers ranging from 0.31 to 1.55 (192).

Complications of Pregnancy and Labor

Observations from the Ontario study and other data showed that
women who smoked during pregnancy had excess fetal deaths either
unexplained or attributed to anoxia and excess neonatal deaths due to
premature delivery. These findings suggested that maternal smoking
might increase the risk of certain pregnancy complications that were
related, in turn, to these causes of perinatal loss. A direct relationship
between maternal smoking level and the incidence of placenta previa,
abruptio placentae, bleeding during pregnancy, and premature rupture
of membranes had been reported previously (2, 31, 63, 115, 189).
Underwood, et al., found higher rates for smokers than for nonsmokers
of bleeding, abruptio placentae, and placenta previa combined, and of
premature rupture of membranes in three groups of women with
different socioeconomic and racial backgrounds (188). In a large study
of births to U.S. Navy wives, the same complications increased with
maternal smoking. In the latter study, the incidence of premature
rupture of membranes increased within four levels of maternal
smoking from none to 31+ cigarettes per day (189). Kullander and
Kaellen found a significant increase in the frequency of abruptio
placentae among children dying before the age of 1 week (89).
Andrews and McGarry found increased incidence of abruptio placentae
and other forms of accidental antepartum hemorrhage to be associated
with maternal smoking. They stated that this was thought to be the
cause of premature delivery in 1.2 percent of smokers compared with
only 0.5 percent of nonsmokers. The incidence of accidental hemor-
rhage specific for parity was higher for smokers than for nonsmokers
at all parities, rising to 3.16 percent of smokers who were para 4 or
more (2). Similarly, Russell, et al. found an increase in vaginal bleeding
during early pregnancy among women who smoked (165). In the study
by Goujard, et al., as previously noted, a large proportion of the
increase in stillbirths among smokers was caused by abruptio placentae
(63). Naeye reviewed the clinical and postmortem material from the
3,897 fetal and infant deaths in the Collaborative Perinatal Project of
the NINCDS (137) and reported an association between perinatal
mortality rates caused by abruptio placentae and number of cigarettes
smoked by the mother (1.31). Abruptio placentae was the underlying
cause identified in 11 percent of all the deaths in this large study (129).
The Ontario data corroborated these findings, as shown in Table 11.
Increasing levels of smoking resulted in a highly significant increase in
the risks of placental abruptions, placenta previa, bleeding, and

8-39


TABLE Il.-Perinatal mortality and selected pregnancy
     complications by maternal smoking levels

Outcome

(A58
births)

Smoking level (packs per day)
(rate3 per 1,ooo total bii)
(128   (&l       chi

bii)     biiS)     sq-*

Perinatal mortality
Abruptio placentae
Placenta previa
Bleeding during
pw*w
Rupture of membranes
>48 hours
Rupture of membranes
only at admission

23.3        28.0        33.4      27.8t
16.1       20.6        28.9      47.q
6.4        8.2        13.1      =`Jt
116.5       141.6       189.1      mw


15.8       23.3        36.8      109.9t


30.3       39.3        45.0      45.7t

`Cachmn's chi squaw for trends.
tp<O.ooool.
SOURCE: Meyer, M.B. (116).

prolonged rupture of membranes-all of which carry high risks of
perinatal loss. Fetal and neonatal deaths from the Ontario study were
analyzed (116) to look for smoking-related excesses of various
complications of pregnancy and labor among those coded by the
original Ontario Perinatal Mortality Study (142). Results are shown in
Table 12. Most diagnoses showed no association with excess mortality
for smokers' babies, but a few stood out as highly significant. As shown
in Table 10, the net excess of fetal deaths for smoking mothers was 63.
Table 12 shows that these deaths were strongly associated with
bleeding during pregnancy, either before (p = 0.01) or after (p = 0.0005)
20 weeks' gestation, with 33 percent of the total excess falling in these
categories. In other coded categories, a significant excess of fetal
deaths occurred among smoking mothers with abruptio placentae
(p=O.OOl) or other obstetrical problems. Analysis of coded complica-
tions of labor showed an excess of 32 fetal deaths coded as abruptio
placentae and 8 coded as placenta previa. Fourteen more than expected
had prolonged rupture of membranes.

Similar comparisons were made for neonatal deaths (Table 8). For
these, the net excess among smoking mothers was 40. Among women
who had vaginal bleeding before 20 weeks' gestation, there were 41
more neonatal deaths observed than expected, accounting for the total
difference (p=O.ooOl). Other categories that showed significant
increases of smoking-associated neonatal deaths are the admission
status of rupture of membranes only, other obstetric complications,
and duration of rupture of membranes over 48 hours, with 19 more
neonatal deaths than expected in the latter group (116).

S-40


TABLE 12.-Fetal and neonatal deaths by maternal smoking and
     other coded conditions (Ontario Perinatal Mortality
     Study data. Canadian-born, English-speaking women,
    N=$1,789 births, 411 perinatal deaths)

Coded condition

       Deaths of smokers' babies
     Oteerved-expected differences'

Fetal        Pt       N.ZlXlatal      pt

Admission status
True labor
Toxemia
Abruptio placentae
Elective cesarean section
Induction
Rupture of membranes only
Other obstetric abnormality

Duration of rupture of membranes
< 24 how
2448 hours
43+ houm
In caul
U&own

Bleeding during pregnancy
None
Before al weeks
After2oweeks

Complications of labor
None
kenta previa
Ahptio pkwentae
Abnormal uterine action
`&balopelvic disproportion,
dyacia
hmltuous hb0r
*urn hemorrhage

15.3       N.S.        26.3       N.S.
-0.9      N.S.        0.7       N.S.
48.5       0.001        25       N.S.
-2.3       N.S.        5.9       N.S.
4.9       N.S.        -4.8       N.S.
0.4       N.S.        13.9       0.04
16.8       0.06         6.0       0.01

322       N.S.
2.3       N.S.
14.3       N.S.
8.5       0.02
5.8       N.S.

2.6
23.7
322

19.2      N.S.        2?2
7.6       N.S.        6.6
32.3       0.002        6.2
0.7       NS.        4.9

-24       N.S.
8.4      N.S.
-4.6       N.S.

N.S.
0.01

13.7       N.S.
3.3       N.S.
19.4       0.01
1.7       N.S.
1.7       N.S.

-5.4       N.S.
41.3       O.oool
3.3       N.S.

N.S.
N.S.
N.S.
N.S.

1.8       N.S.
7.1       N.S.
-8.0       0.06

KS.-Not a@,dfia,,t
`BusdonnoMmokRnte.

The conclusion may be drawn that maternal smoking increases the
risk of fetal and neonatal death at least partly by increasing the
incidence of these complications. The mechanisms of action of various
Components of cigarette smoke in bringing about these events are
discussed in another section of this chapter.

Reeclampf5ia

It has been a consistent finding in almost all published studies that the
incidence of pmlampsia and toxemia, however defined, is negatively
Wiat.4 with maternal smoking (2, 10, 31, 42, 74, 89, 101, 146, 164,

841


189, 212). Some of these studies have shown an inverse dose-response
relationship, the incidence of preeclampsia declining as the number of
cigarettes smoked increased (Z&Z, 189). Data from the British Perinatal
Mortality Study were cross-tabulated by parity, severity of preeclamp
sia, and maternal smoking status. Smokers had lower rates of all
grades of preeclampsia than nonsmokers, whether they were primipar-
ae or multiparae (20). Andrews and McGarry showed that the negative
relationship between cigarette smoking and preeclamptic toxemia was
independent of social class, maternal weight before pregnancy, and
maternal weight gain during pregnancy (2). Despite the favorable
effect of smoking on the incidence of hypertension in pregnancy, there
is a greatly increased risk of perinatal mortality if preeclampsia or
hypertension does develop in a smoker (2, 42, 164). Several authors
have suggested that this negative association may be due to the
hypotensive effect of thiocyanate, which is derived from the cyanide
present in cigarette smoke and regularly found in the blood of smokers
(2,146).

Fkterm Delivery

Previous sections of this chapter have indicated that the downward
shift of the distribution of birth weights with maternal smoking is not
accompanied by a similar downward shift of gestational ages. On the
other hand, abundant evidence has been presented that a smoking-
related increase in preterm delivery plays an important role in the
increased risk of neonatal death for the infants of smokers. Explana-
tion of this apparent paradox is found by examination of the
distribution by gestational age of births to nonsmokers, light smokers,
and heavy smokers as shown in Figure 6, plotted on a semilogarithmic
scale to emphasize relative differences in the early weeks. There is
little difference between the means of these curves because the great
majority of births occur around term in all groups. There is, however, a
significant and dose-related increase in the proportions of preterm
babies born to women who smoke. These preterm deliveries account for
a small proportion of total births but for a large proportion of the
deaths (112).

Published studies in which the percent of births occurring before
term has been related to maternal smoking have consistently shown
higher rates for smokers than for nonsmokers. Some examples are
shown in Table 13. In four studies where all births and perinatal deaths
were included, the risk of early delivery increased from 36 to 47
percent if the mother smoked, and 11 to 14 percent of all preterrn
births could be attributed to maternal smoking (2, 20, 47, 20fl. The
lower relative and attributable risks found in Yerushalmy's study (207)
may have resulted from selection of particular births to be studied and
from the exclusion of fetal deaths. Analysis of the Ontario Study data

8-42


LG
:  2.0-
K
w
n  I .o:.

   0.6,.

  0.4,.

0.2..

O.lA
   20 24 26 32 36 40   44+
        GESTATION: WEEKS

FIGURE O.-Percentage distribution by weeks of gestation of
births to nonsmokers, smokers of less than one pack per day, and
smokers of one pack per day or more
3OURCE:Meyer.Y.B.(IIP).

showed rates of delivery before 38 weeks of 77 per 1,990 births for
nonsmokers, 92 per 1,990 for light smokers, and 116 per 1,000 for heavy
smokers, after adjustment for the effects of other maternal factors
(115).

Pregnancy Complications and Perinatal Mortality by Gestation
Meyer and Tonascia (126) have related the excess fetal and neonatal
mortality of smokers' infants and the excess incidence of pregnancy
complications among women who smoke to the gestational age of
occurrence, using a life-table approach. A starting population of all
Pregnancies in utero at 29 weeks was used to calculate the probabilities
of fetal death, live delivery followed by survival or death, or the
occurrence of a complication followed by fetal death or delivery. At 28
weeks (the next point defined by the data), the population at risk
included those remaining in utero at that point. Figure 7 shows the risk
of perinatal death during each period of gestational age starting at 20
weeks. Risks for smokers' infants were significantly greater in the

8-43


TABLE 13.--preterm births by maternal smoking habit, relative
      and attributable risks, derived from published
       studies

                         F'reterm births*       Relative

    Study       Smokem        per loo          risk:     Attributable
            (proportion)       t&d births       Smokers/Non-    risk-

_

Nonsmokers   Smoke=     smokem       w

Cardiff (2)            ,465        6.7       9.2       1.26        14
Great Britain (P)        ,274        4.7        6.9       1.47        11
Montreal (47)          A32        1.7       10.6       1.33        14
h&g..            A35        1.4       10.1       1.26         4
Cdiiomie (zo7)
White              A02        5.9       6.9       1.10        4
Black              333       13.4       16.7       1.26         6

earlier weeks, remaining higher until term. Separate calculations for
fetal and neonatal deaths (not shown) indicated a fetal death pattern
very similar to the one shown for perinatal deaths. Neonatal deaths
appeared to be due solely to an increased risk of early delivery among
smokers' babies, rather than to differences in survival between
smokers' and nonsmokers' babies of the same gestational age.

A similar approach was applied to the risk of abruptio placentae,
placenta previa, and premature rupture of membranes for smokers and
nonsmokers, as shown in Figure 8. All of these complications are more
frequent in smokers than in nonsmokers throughout gestation, but
again the biggest differences occur in the weeks of pregnancy from 20
to 32 or 34 weeks (116). The relationships between maternal smoking,
these complications, early fetal death, and preterm delivery accompa-
nied by neonatal death are apparent from the statistical associations
between them and from the similar time patterns they share.

Sudden Infant Death Syndrome

Maternal smoking habits have been ascertained in several studies of
the sudden infant death syndrome (SIDS). In all of these, a positive
association has been found between maternal smoking during preg-
nancy and the incidence of sudden infant death. Steele and Langworth,
in a study of 86 cases, each with two matched controls, which were
traced back to the Ontario Perinatal Mortality Study population of
1960-61, found that sudden infant deaths were strongly associated with
the frequency of maternal smoking during pregnancy (p<O.OOl) and
also with the level of maternal smoking. Thirty-nine percent of the
cases were nonsmokers versus 66 percent of controls; 36 percent of the

8-44


FIGURE `I.-Probability of perin& death for smoking and
nonsmoking mothers, by period of gestational age. Bars shoti 95
percent confidence intervals
SOURCE: Meyer, M.B. (116).

cases and 27 percent of the controls smoked less than a pack per day; 24
percent of the cases and 10 percent of the controls smoked a pack per
.day or more. The habits of the remaining 1 to 2 percent of mothers
were unknown (180). Bergman and Wiesner noted the effects of
exposure to cigarette smoke (passive smoking) on infants, including
the increased frequency of respiratory infections in the infants of
smoking mothers, and stated their impression that the amount of
smoking seemed unusually heavy at meetings of parents who had lost
children to SIDS. The authors studied 56 families who lost babies to the
sudden infant death syndrome and 86 control families. They reported
that a higher proportion of SIDS mothers smoked during pregnancy
than controls (61 percent versus 42 percent), more smoked after
pregnancy (59 percent versus 42 percent), and SIDS mothers smoked a
significantly greater number of cigarettes than controls. These authors
indicate that exposure to cigarette smoke (passive smoking) appears to
enhance the risk for SIDS for reasons not yet known (15). However,
whether prenatal or postnatal exposure is more important cannot be
determined. Naeye, et al., in their analysis of 125 SIDS victims from
the population of the Collaborative Perinatal Project of the NINCDS,
stated: "The gestations that produced the SIDS victims were
characterized by a greater frequency of mothers who smoked
cigarettes and had anemia" than was true for the whole population of
53,721 infants or for a set of 375 controls matched on important factors

8-45


GESTATON: WEEKS

FIGURE &-Risks of selected pregnancy complications for smok-
ing and nonsmoking mothers, by period of gestational age at delivery.
A-abruptio placentae; B-placenta previa; C-admission diagnosis,
rupture of membranes only
SOURCE: Meyer. M.B. (116).

(130). Rhead, commenting on studies published to date which
demonstrate an increased incidence of maternal cigarette smoking in
SIDS, states: "It is now...clear that maternal cigarette smoking
contributes to an infant's risk of dying from SIDS" (159).

Summary

1. The risk of spontaneous abortion, of fetal death, and of neonatal
death increases directly with increasing levels of maternal smoking
during pregnancy.

2. Published studies of smoking during pregnancy show a range of
perinatal mortality risk ratios (smokers versus nonsmokers) from a low
of 1.01 to a high of 2.42.

3. Causes of variability between risk ratios in different study
populations have been explained by recent analyses. They include:

8-46


(a) Lack of comparability between smokers and nonsmokers with
respect to other important variables that influence perinatal
mortality, such as race, socioeconomic status, age, parity, and
others.

(b) Interaction between the effects of maternal smoking and these
other variables, which makes maternal smoking more dangerous
for the fetus in some pregnancies than in others.
4. Studies failing to take account of these other variables may show

unusually high or unusually low risk ratios.

5. In one large study, the perinatal mortality risk increased by 20
percent for the infants of smokers of less than a pack per day and by 35
percent for smokers of a pack per day or more, compared with
nonsmokers, after simultaneous adjustment to balance the effects of
variables other than smoking. These increases are similar to those of
other large studies with appropriate control of other variables.

6. Excess deaths of smokers' infants are found mainly in the coded
cause categories of "unknown" and "anoxia" for fetal deaths, and in
the categories of "prematurity alone" and "respiratory difficulty" for
neonatal deaths. This finding indicates that the excess deaths result
not from abnormalities of the fetus or neonate, but from problems
related to the pregnancy.

7. Increasing levels of maternal smoking result in a highly
significant increase in the risks of placental abruptions, placenta
previa, bleeding early or late in pregnancy, premature and prolonged
rupture of membranes, and preterm delivery-all of which carry high
risks of perinatal loss.

8. Although there is little effect of maternal smoking on mean
gestation, the proportion of fetal deaths and live births that occur
before term increases directly with maternal smoking level. Up to 14
percent of all preterm deliveries in the United States may be
attributable to maternal smoking.

9. According to the results of one large study, the most significant
difference between smokers' and nonsmokers' risk of perinatal
mortality and pregnancy complications occurs at the gestational ages
from 20 weeks to 32 or 36 weeks.

10. These findings lead to the conclusion that maternal smoking can
be a direct cause of fetal or neonatal death in an otherwise normal
infant. The immediate cause of most smoking-related fetal deaths is
probably anoxia, which can be attributed to placental complications
with antepartum bleeding in 30 percent or more of the cases. In other
cases, the oxygen supply may simply fail from reduced carrying
capacity and reduced unloading pressures for oxygen caused by the
presence of carbon monoxide in maternal and fetal blood. Neonatal
deaths occur as a result of the increased risk of early delivery among
smokers, which may be secondarily related to bleeding early in
pregnancy and premature rupture of membranes.

8-47


Lactation and Breast Feeding

Introduction

In 1902, Ballantyne (9) suggested the possibility of detrimental effects
of breast feeding on babies whose mothers worked in tobacco factories.
In the intervening years, questions have been raised concerning the
interaction between cigarette smoking and lactation, as well as the
relationship of cigarette smoking to the quantity of milk produced, to
the presence of constituents of cigarette smoke within the milk, and to
effects upon the nursing infant mediated through changes in either the
quantity of milk available or the substances within the milk.

Epidemiological Studies

Underwood, et al. (188), in a study of 2,000 women from various social
and economic strata, observed a trend, though statistically insignifi-
cant, toward more frequent inadequacy of breast milk production
among those smoking mothers who attempted to nurse, as compared to
nonsmokers. They concluded that smoking does not interfere with
breast feeding to any significant degree. However, this study, baaed on
interviews of puerperal women, was not designed to analyze the effect
of smoking on breast feeding and presents only percentile results: No
data are provided to permit a reanalysis to determine the validity of
their conclusions.

Perlman, et al. (14.9) also present anecdotal data. They found that in
their postpartum population practically all smoking women started to
consume cigarettes within two days after delivery. Although they
collected milk between the fourth and ninth postpartum days to
determine nicotine content, they do not report and compare actual
amounts of milk secreted by both smokers and nonsmokers. They noted
that of the 55 smoking, lactating mothers, 11 failed to have enough
breast milk for the needs of their babies. No comparative study was
done in a nonsmoking but otherwise equivalent population.

Mills (120) studied the nursing patterns of 520 women giving birth to
their first live-born infant. Among the mothers nursing their babies
for a minimum of 2 months and beyond, the mean nursing period was
significantly shorter for smokers than for nonsmokers. Moreover,
among the 24 mothers who had given up smoking during at least the
final 3 months of their pregnancies, the average length of nursing was
identical to that of the nonsmokers. There was no significant
difference between smokers and nonsmokers with regard to complete
inability to nurse their offspring. This study is difficult to interpret
because the author did not determine the reason(s) for the discontinua-
tion of nursing among the women.

Surveys of larger populations of women, smokers and nonsmokers,
are needed to determine accurately the effect of smoking on milk

8-48