pared with never smokers. All former smokers had a I .6-fold increase. but this increase
was limited to those who had quit within the preceding X years. Longer durations of
abstinence yielded an odds ratio of I .O.

Concerns about the possibility of differences in sexual activity between smokers and
nonsmokers and the occurrence of STDs limit the ability to draw firm conclusions about
the association of smoking with ectopic pregnancy. There is little information about
former smokers, and consequently, no conclusion can be drawn.

Some data suggest an association between smoking and increased risk of spontaneous
abortion (US DHHS 1989). Data on smoking cessation are very sparse. Kline (I 9X4)
noted that the adverse effect of smoking observed in a case-control study of smoking
and spontaneous abortion (Kline et al. 1977) was limited to current, not former.
smokers. Alberman and colleagues (1976) found that the proportion of spontaneous
abortions with abnormal karyotypes decreased with increased smoking but was identi-
cal for never smokers and women who stopped smoking prior to pregnancy (Alberman
et al. 1976). The interpretation of this finding is uncertain.

Fetal. Neonatal, and Perinatal Mortality

Information linking cigarette smoking with an increased risk of the various measures
of mortality used to assess pregnancy outcome has been reviewed in previous reports
of the Surgeon General and other publications (US DHEW 1979: US DHHS 1980: US
DHHS 1986). Table 3 provides data on perinatal and neonatal mortality from the earlier
reports of the Surgeon General (US DHEW 1979; US DHHS 1980) and adds informa-
tion from a more recent publication on the topic (Rush and Cassano 1983). The studies
are consistent in indicating higher mortality in children born to women who smoke.
The high risk of mortality is independent of various factors. such as education and social
class. that are also associated with mortality.

Kleinman and colleagues (19X8) assessed the effect of smoking on fetal and infant
mortality in 363.621 births in Missouri during 1979-1983. Using multivariate statisti-
cal techniques, these investigators estimated the effects of smoking on fetal and infant
mortality among black and white primiparous and multiparous women. After adjust-
ment for marital status. education. and age. fetal plus infant mortality rates were 25 to
56 percent higher in smokers for all categories of maternal race and parity. The
elevations in the estimated risks of fetal plus infant mortality were statistically sig-
nificant in all categories. In further analyses ofdata from the Missouri births and deaths.
Malloy and coworkers (19Xx) showed that the relative risk of fetal plus infant mortality~
among whites was significantI> elevated for the infants of women who smoked in all
categories of low birthweight. even after adjustment for marital status. education. age.
and parity (Table 1). This data set is unique in its sile. consisting of more than 3SO.OfW)
births. The data indicate that even in the nomtal birthweight infants of smokers-those
that wjeiahed 7.500 g or more-mortalitv was significantly elev#ated for infants of
      c
mothers who smoked.

Information on fetal. neonatal. and perinatal mortality in former smokers is sparse
(Table 5). Butler. Goldstein. and Ross (1973) analy,red data from the British Perinatal
Mortality Survey and estimated that perinatal mortality was the same for women who

376


TABLE 3.-Summary of studies of perinatal and neonatal mortality in smokers
     and nonsmokers during pregnancy

Reference

Number of
birth\

       Perinatal mortality"     Uronalal morlalIl~"
Carepq   Smnher\    Non\mohrr\    Smoher\   Non\mohsr\

Com\toch and   12.37
Lundin
t 1967,

Meyer and     5 I .lYO
Tona\cia
llY77,

Rantahalllo    12.06X
(lY7XI

Ru\h and
ca\\ano
lI9X.3)

Butler.       2 I .7X8
Goldstein. Rev
1972)

Whltr\
Blach\

Amount mohrd

<I wd        2X.0      23.0
>I ppd        33.3

Social cl9s.
~.. .______~.
I+11
III+IV
Farmer\
Lnhnown

XI"    22.4"
25.1      14.6
753 <I    3s .I?
29.4"    36.X"

Amount \mohed

4 q/day
S-11 cig/day
>I5 cigda)

IS.0      IX.7
26. I
2x.3

Andrews and
McGarry
t 1972)

IS.63 I

Nihwander    37,912
and Gordon
(1972)

  I4 cig/day
  S-9 c&/day
IObl9cigiday
  t70 ctg/day

  Race and
Amount smoked

White
I-IOciddq
2 I I cig/da)

Black
IplOcigjdq
21 I c&iq

11.1      32.0

2s
20
32
36

25

       31.1
31.5
3x.2

       3x.5
41.5
57.1

17.6     13.7

377


TABLE X-Continued

TABLE L-Estimated relative risk of fetal plus infant mortality for maternal
     smoking in several birthweight groups, adjusting for maternal
      marital status, education, age. and parity

Y.i'i Cl


Perinatal mortality among those who smoked before pregnancy but quit during preg-
nancy (lS.O/l,OOO) was lower than for either nonsmokers during pregnancy ( I8.7/
I .OOO) or smokers of 5 cigarettes or more per day throughout pregnancy (26.9/1.0(K)).

TABLE 5. --Summary of studies of perinatal mortalit? in smokers throughout
      pregnancy, smokers who quit in the early months of pregnancy, and
      nonsmokers during pregnancy

Reference

Fetal, neonatal, and perinatal mortality are rare events. This limits the study of their
association with smoking cessation. Lack of data makes it impossible to draw a firm
conclusion about the association of smoking cessation with the risk of fetal, neonatal.
or perinatal mortality. However, the limited available data are consistent with the
conclusion that perinatal and neonatal mortality are lower among infants of women who
quit smoking than among those women who smoke throughout pregnancy. The
possibility must be considered that differences between women who quit smoking and
those who continue to smoke account for the lower rate of perinatal and neonatal
mortality in the studies in which this has been observed.

Birthweight and Gestational Duration

Fetal. neonatal. and perinatal mortality are the most direct measures of pregnancy
outcome. Mortality is relatively uncommon, and very large samples are needed for
study. This has led to the widespread study of birthweight and the percentage of births
that are low birthweight (<2.500 p) as surrogates for the study of mortality. This
strategy has been justified by the extremely strong association between birthweight and
the percent of low birthweipht and each of the measures ofmortality (Figure I 1. Equally
important is weight at birth as a determinant of infant health (McCormick 1985 J.


W Perinatal mortality =
      fetal deaths and neonatal
      deaths/total births

x-.. .. .,x Neonatal mortality =
     death through 28 days in
      liveborn infants/live births

0 - -0 Fetal mortality =
       stillbirths/total births

500-    lOOO-    1500-    2000-   2500-   3000-   3500-   4000-
749    1249    1749    2249    2749    3249    3749    4249
BIRTHWEIGHT (g)
FIGURE I.-Perinatal, neonatal, and fetal mortality rates by birthweight
      in singleton white males, 1980

SOURCE: Williams and Chen (1982).


Birthueight is. ho&ever. :I result ofgestational age at birth and the rate of fetal yrou th.
Recognition ot the complex relationship5 amon, 0 ft3tationul duration. rate of fetal
growth. birth\\teight. and mortalitv has led to attempt\ to classif\ infant\ according to
cre\tational duration or joint distribution of hirthueight and ~~c\tatioii;il duration.
Eenerallj. births are catefori/ed ;I!, preterm (~37 ueehs gr\tatiotyl and/or 34 \mal I for
gestational age (SGA) (<lOth percentile of weight for ;I gi\,en ge\tational age ). Joint
cla\4fication is thought to pro\ ide a more discriminating hasi\ for the \tucl! ofetiologic
agents.

Preterm dcli\,ery i\ strongly a\>oci;ltcd V, ith incrc;Lht`\ in the risk of fetal. neonatal.
and perinatal mortalit\ and \j ith significant childhood morbidit!.  Both preterm
delivery and SGA increase the rish of cerebral pal\!. although the ri4 i\ much greater
for prcterm delivery (Ellenberg and Nelson Ic)7Y). SC;+. I\ as\ociatcd M ith increased
ri\k of neonatal and perinatal mortalit) at e\`er! c catation;d qe (Koop\. hlorgan.
Battaglia 19X2: Lubchenco. Searl$. Bralie 1071): Lsith SIDS tBuch et al. IWSJ: and
with neurocognitive deficits. 4ort \taturc. :md small head circumference in childhood
(Fitzhardinge and Steven IY72: Hill et al. IYX4: We\thood et al. lYX3; Ounted and
Taylor 197 I; Harvey et al. 19X7: Ounsted. Moar. Scott IYXI. I YXX: Fancourt et al.
1076).

A\ reviewed in previous Surgeon General'\ report\ (US DHEW lY7Y: US DHHS
IYXO) and in other literature (Landesman-Dv. yer and Emanuel 1979: Longo 19X2:
Werler. Pober. Holme\ 19X5: Kramer 1987). cmokin2 during pregnancy decreases
mean birthweight and increase\ the proportion of IOU' birthweight births. Ei;timates
vary among studies. but birthweight is reduced by an average of approximately 100 g.
and the proportion of 1ow birthweight is approximately doubled by cigarette smoking
(Meyer. Jonas. Tonascia 1976: US DHHS 19X0: L'S DHEW 1979: Mclnto\h 19X-I:
Committee to Study the Prevention of Low Birthweight IYXS: Kramer 19X7). Mean
birthweight decrea5e.s and the percent low birthbeight increases with increasing num-
ber of cigarettes smoked daily. The relationship between cigarette smoking and
decreased birthweight is considered to be causal (US DHEW 1979: US DHHS 1980.
I9XY).

Smoking affects birthweight and the percentage of babies who are born of ION
birthweight by retarding fetal growth. A measure of fetal growth retardation is the
probability ofdeliverinp an infant who is in the lee than 10th percentile for gestational
age. The relative risk of SGA is about 3.5. to l.O-fold higher among the infants ot
\mokerc than for the infant4 of nonsmokers (Ounsted. Mnar. Scott IYX5 1. Preterm birth
is also associated with maternal smoking. although not as strongly. Estimates of the
relative risk of delivering before 37 weeks of yestation are typically about I.5 for
smoking during pregnancy (Committee to Study the Pre\zntion of Low Birthueight
19X5: Kramer 19X7: Shiono. Klebanoff. Rhoads 19X6). Mean Fe\tntional duration
among makers i\ not significantly shorter than it i\ among nonsmoher\ (C`S DHEW
1979: US DHHS 19x0). This finding is consistent with the observation that the ri\h ot
delivering early ik greater amon? smokers than nonsmoher\. but the percentay of


preterm deliveries is so small that the mean would not be affected unless the shift were
very large (US DHEW 1979; US DHHS 1980).

Most studies of cigarette smoking and birthweight have failed to separate never
smokers from women who quit smoking prior to conception. MacMahon. Alpert, and
Salber (1966) first examined the association of pre-pregnancy smoking with
birthweipht and found no significant difference in the mean birthweight of infants
whose mothers smoked before but not during pregnancy compared with never smokers.
Subsequent research has confirmed the absence of an association between smoking
prior to conception and reduced birthweight (Table 6). In all of these studies. smokers
who quit before conception had mean birthweight values that were equivalent or higher
than those of never smokers. Other studies in which information on mean birthweight
could not be derived (Kline, Stein. Hutzler 1987: Anderson et al. 1983: Wainright
1983). with the exception of Zabriskie (1963). have also consistently shown no
association between birthweight and smoking that ceased prior to conception. Zabris-
kie (1963) failed, however. to adjust for smoking during pregnancy. and these results
are not directly pertinent in a comparison of birthweight in never smokers and smokers
who quit before conception.

TABLE 6.-Summary of studies of mean birthweight, by smoking status

In interpreting these data. misclaa\itication of exposure needs to be considered.
MacArthur and Knox ( 198X) reported that women who quit smoking during pregnancy.
and possibly those who quit before pregnancy. were more often living with a partner
who smoked. Passive smoke exposure may adversely affect the fetus (Martin and


Bracken 1986). Furthermore. for whatever reason, some women ma) misrepresent
their smoking status, denying that they have continued amohing. thus leading to an
underestimation of the benefit of smoking cessation prior to conception.

More important. women who quit smoking prior to conception differ in other respects
from women who continue to smohe. Women who quit may have smohcd feuer
cigarettes per day prior to quitting. Studies of smoking cessation prior to conception
have not accounted fully for other difference\ between women who quit and those who
continue to smoke.

Birthweight

Table 7 summarires nonexperimental studies in u hich information on mean
birthweight in nonsmokers. smohers throughout pregnancy. and smoker\ u ho quit after
conception could be derived. The data from each of these studies are con\i\tcnt in t\j o
important ways. First. women who smohed throughout pregnancy delivered infant\
who weighed less than the infants of nonsmokers. Second. women who quit smoking
delivered infants who weighed more than the infants of smokers throughout pregnancy.
In most of these studies. mean birthweight values among infants whose mother\ \toppcd
smoking were the same or higher than those of infant\ of nonsmokers.
Table 8 summarizes nonexperimental studie< estimating the relative rish of IOM
birthweight for continuing smokers and quitters some time during pregnancy compared
with nonsmokers during pregnancy. These studies are consistent with those examining
mean birthweight. Compared with nonsmokers. the risk of low birthweight is elevated
among smokers throughout pregnancy. and the risk is about I .O for women M ho quit.
In addition. Kleinman and Madans (1985) reported no association between the rish ot
low birthweight for women who quit smoking during pregnancy compared with those
who had not smoked in the I2 months prior to conception among participants in the
1980 National Natality Survey (NNS ).

An important aspect of smoking cessation and pregnancy outcome is the timing of
cessation during pregnancy and its relation to birthweight. How early in pregancl
cessation must occur to avoid the adverse effects of smoking on birthweight is a ke\
issue with important implications for counseling pregnant smokers.

In most of the studies examining this question, only information on cessation in the
early months of pregnancy is presented. However. Rush and Cassano ( 1983) found that
mean birthweight among women who quit as late as the seventh to eighth month of
pregnancy was higher than for women who smoked throughout pregnant)`. but lower
than for nonsmokers and for women who quit earlier in gestation. MacArthur and Knox
(1988) concluded that quitting any time before the 30th week of gestation increase\
birthweight when compared with continuing to smoke. Cooper (19X9) assessed
patterns of cigarette smoking by trimester of pregnancy. Women who reponed
smoking during the "first trimester of pregnancy only" had a X)-percent increased ri\h
of having a low birthweight baby. while women who reported smohinp during the "first
and second trimester of pregnancy only" had a 7%percent higher rt\h ot. ;I lo\{


TABLE 7.-Summary of nonexperimental studies of smoking cessation after
     conception, mean increase (+I or decrease (-) in birthweight (g)
     according to timing of cessation

Month of ce\\ation

Smohed

Reference

I 2 3 4 5 6 7 x Y      L'tlktlOWl   throughout

Lowe
(IYSY)

Underwood et al.
t 1967)

Butler. Gold\tein.
Ro\\
(1972)

Andrew and
McGarry t I472 1
Pap07 et 31.
(19X?)

+9X

Rush and
Cnssano ( 19x3 1

Pulhkinen
(IYXS)

Councilman and
MacKay t 19x5)

Kline. Stein.
Hutrler
(19X7)

MacArthur     +`17
and Knox
(IYXXI

       +I3                             -IX'



-IOX       -152                         -730



    +Jh                                -160





-x0                                    -170



          + IO                           -70



          +-I.?     c.76 -90                -155



41                                    -22s



10                                   -3s



                            +I?        -202




    -sx                                 -2-t:

birthweight baby.  Women M ho reported mokin, 17 throughout their pregnancy had a
90.percent increased risk. of having a low birthweifht babl in contrat to nonjmohers.

Most fetal gwwth occurs late in pregnancy. and the primary smoke comtituents
considered as candidate\ in mediating the effect of wloking on fetal grow. th (i.e.. CO
and nicotine leading to intrauterine hypoxia) have short-term reversible effects. The
data in Table\ 6 and 7 wpport the conclusion that the adverse effect of making on
birthweight occur5 in the latter part of Fe\tation. primarily during the third trimester.
and that cessation at any time during gestation i4 lihely to mltlzL
                                 `vte the adverse effect
of smoking on fetal growth.

Because it is difficult to perwade ull pregnant \moher\ to quit vnohing entirely. the
benefit of reducing the number of cigarette\ smoked per day becomes a public health
issue. The obhervstion that cigarette mohing retard\ fetal gro\hth in ;I do\e-rehpon\e

3x3


TABLE 8.-Summary of nonexperimental studies of relative risk of low
     birthweight for smoking cessation after conception

Fruier et al. ( IYhl I
Van den Berg (lY771~
Petltti and Coleman on prr\\I

   Vv'hlte\    <I mo
          I-? mo
          2-3 mo
   Blach\    <I mo
         I-2 mo
          7-1 m.

Andrew and
McGarr? ( iY77)

0.S                `7
                -.
I .(I
(1.6
I A               3X
I .o
I.1

1.3               1.0

fashion supports the benefit of reducing the number of cigarettes smoked per day.
Hebel. Fox. and Sexton (19%) used data from their randomized trial of smoking
cessation to examine this question.  These researchers found that the benefit of
decreased smoking for birthweight during pregnancy w'as almost entirely, restricted to
those who achieved total cessation. perhaps because women who reduce the number of
cigarettes smoked compensate by inhaling more deeply. by puffing more frequently.
or by smoking the cigarette to a shorter butt length. Findings from another randomized
trial support the conclusion that abstinence. not reduction. should be the goal in
pregnancy (MacArthur. Newton. Knox 1987). In this latter study, the intervention led
to a considerable reduction in the reported mean number of cigarettes smoked per day
but almost no difference in the percentage of women who quit entirely: there was no
difference in birthweight between the treatment and control groups (MacArthur.
Newton, Knox 1987). Because of the social stigma associated with smoking during
pregnancy, it is possible that some women in this intervention trial falsely reported a
reduction in smoking; if so. this underreporting would lead to an underestimation of
possible benefits of reducing cigarette consumption.

Whether quitting only during the first half of pregnancy will prev'ent a reduction in
birthweight is another important consideration. Most fetal growth takes place in the
last trimester: early quitting virtually eliminates the effect of smoking on birthweight.
Thus. smoking late in pregnancy may have an adverse effect on birthvveight even if
there is abstinence in the first trimester. Lowe (1959) found that the mean birthweight
of infants of smokers who quit early in pregnancy but resumed smoking was between
that of smokers throughout pregnancy and that of never smokers. Infants of women
whogave upcigarettes by the fifth month ofpregnancy and whodid not resume smohin_g


had a mean birthweight identical to that of never smokers. MacArthur and Knox ( 198X)
also found that infants born to women who quit smokin? early in their pregnancy but
started again before delivery had a mean birthweight value between that of smokers
throughout pregnancy and those of both early quitters and never smokers. These data
indicate that abstinence throughout the third trimester of pregnancy is necessary to
realize the full benefit of smoking cessation for birthweight.

Preterm Delivery

The effect of smoking on birthweight is principally due to a reduction in size for a
given gestational ape rather than to a large decrease in gestational duration (US DHEW
1979: US DHHS 1980). Thus. it would be expected that pregnancy outcome in women
who quit would reflect a predominant effect on size for gestational age.

Andrews and McGarry ( lY72) considered preterm delivery as a distinct endpoint in
continuing smokers and quitters: the latter group included a mixture of women who
quit prior to conception and women who quit during their pregnancy. The rate ot
preterm delivery among nonsmokers was 6.7 per 100 compared with 7.5 per 100 for
ex-smokers and 9.1 per 100 for women who continued to smoke throughout pregnancy
(Andrewc and McGun-y 1971).

Berkowitz. Holford. and Berkowitz ( 1981) examined the association between smok-
ing during each trimesterof pregnancy and the risk of preterm delivery in acase-control
study of 175 mothers of singleton. preterm infants and 3 I3 mothers of singleton. term
infants. The risk of preterm delivery was increased among women vvho smoked in the
third trimester of pregnancy. especially if they smoked heavily (>I0 cigarettes per day ).

Using data from a longitudinal study of pregnant women. Van den Berg and Oechsli
( 1984) reported rates of preterm delivery (137 weeks) among never smokers. smokers
who stopped at the beginning of pregnancy. and continuing smokers for 10,937 white
women whose singleton pregnancies progressed beyond 22 weehs. The rate of preterm
delivery was 5.3 percent in nev'er smokers. 6.X percent in quitters. and 7.6 percent in
continuing smokers. The difference in the rate of pretemr delivery between never
smokers and quitters was not statistically significant (pN.05): however. the difference
between never smokers and continuinp smokers was significant.

In a population-based case<ontrol study of white and black women delivering
singleton infants without congenital anomalies in a large urban county. Petitti and
Coleman (in press) reported that the estimated relative risk of very low birthweight
(<I.500 g) or of other preterm births among blach and white women who quit smohing
prior to the fourth month of gestation was not increased in comparison with those of
nonsmokers. The estimated relative risk of very low birthweight (< I.500 g) in continu-
ing smokers M as 2.5 for u hites and 3. I for blacks and that of other preterm births u as
2.0 for wshites and 3.7 for blacks.

MacArthur and Knox (1988) examined gestational duration according to smoking
during pregnancy. Mean gestational length was 1.7 days shorter among continuing
smokers than nonsmohers. Compared with nonsmohers. gestational periods were 0.4
days shorter for women who quit smoking by the 6th week of pregnancy, I.5 days longer


for women who quit between the 6th and 16th weeks of pregnancy. and 0.3 days longer
for women who quit after the 16th week of pregnancy.

Because of the limited data on the risk of preterm delivery among women who quit
smoking after conception. a firm conclusion about benefit. or lack of benefit. at-
tributable to smoking cessation for this pregnancy outcome cannot be drawn.

Women who smoke during pregnancy are at increased ri>k of bleeding during
pregnancy and of placenta previa and abruptio placentae (US DHEW 1979: US DHHS
1980: Naeye 1978: Naeye 1980). These women are probably at decreased risk of
preeclampsia (US DHEW 1979: US DHHS 1980: Marcoux, Bribson. Fabia 1989). Few
data on these pregnancy complications among former smokers are available.

In Naeye's ( 1980) analysis of data from the Collaborative Perinatal Project, smoking
for more than 6 years (but not short-term smoking) was found to be associated with a
relative risk of I.6 to 1.9 for abruptio placentae and a relative risk of 2.4 to 2.8 for
placenta previa. Women who had stopped smoking by their first prenatal visit were not
at increased risk of abruptio placentae, but were still at twofold increased risk of
placenta previa if they were long-term smokers. However, the latter result was based
on only 18 exposed cases.

Marcoux, Brisson, and Fabia (1989) found that. compared with women who had
never smoked, those who smoked at the time of conception were protected from
preeclampsia (estimated relative risk (RR)=OS 1). whereas women who smoked but
quit prior to conception had the same risk of preeclampsia as never smokers (RR=O.97).
Women who smoked at conception but quit prior to 20 weeks' gestation were not as
protected from development of preeclampsia as were continuing smokers. Because of
the otherwise serious adverse effects of smoking on the fetus, this minor "benefit" of
smoking during pregnancy probably has no public health consequence.

Randomized Trials of Smoking Cessation During Pregnancy

Three randomized trials have been conducted on pregnancy outcome in relation to
advice to stop smoking (Donovan 1977; Sexton and Hebel 1984: MacArthur, Newton,
Knox 1987). Table 9 summarizes the studies and birthweight results. Two other
randomized trials have also been conducted on the effect of various programs on
smoking cessation rates among pregnant women (Ershoff, Mullen, Quinn 1989:
Windsor et al. 1985). and other trials are in progress. Information on pregnancy
outcome is not available, and these studies are not reviewed.

Donovan (1977) studied smokers in three maternity units in England. Women aged
35 years or younger at the start of pregnancy. who smoked more than 5 cigarettes per
day, who had less than 30 weeks of gestation at the first prenatal visit. and who had no
prior perinatal deaths, were randomly assigned to a control group that received usual
prenatal care or to a test group that was given intense individual antismoking advice by
a physician at each prenatal care unit. There were 263 women in the test group and 289
in the control group. Mean daily cigarette consumption decreased from 17.1 cigarettes
per day early in pregnancy to 9.2 cigarettes per day late in pregnancy in the intervention


for women who quit between the 6th and 16th weeks of pregnancy, and 0.3 days longer
for women who quit after the 16th week of pregnancy.

Because of the limited data on the risk of preterm delivery among women who quit
smoking after conception. a firm conclusion about benefit. or lack of benefit. at-
tributable to smoking cessation for this pregnancy outcome cannot be drawn.

Women who smoke during pregnancy are at increased risk of bleeding during
pregnancy and of placenta previa and abruptio placentae (US DHEW 1979; US DHHS
1980: Naeye 1978: Naeye 1980). These women are probably at decreased ri\k of
preeclampsia (US DHEW 1979: US DHHS 1980: Marcoux. Brisson. Fabia 1989). Few
data on these pregnancy complications among former smokers are available.

In Naeye's (1980) analysis of data from the Coltabordtive Perinatal Project. smoking
for more than 6 years (but not short-term smoking) was found to be associated with a
relative risk of I .6 to I .9 for abruptio placentae and a relativ*e risk of 2.4 to 2.8 for
placenta previa. Women who had stopped smoking by their first prenatal visit were not
at increased risk of abruptio placentae, but were still at twofold increased risk of
placenta previa if they were long-term smokers. However, the latter result wa$ ba$ed
on only I8 exposed cases.

Marcoux, Brisson. and Fabia (1989) found that, compared with women who had
never smoked, those who smoked at the time of conception were protected from
preeclampsia (estimated relative risk (RR)=O.Sl). whereas women who smoked but
quit prior toconception had the same risk of preeclampsia as never smokers (RR=O.97 ).
Women who smoked at conception but quit prior to 20 weeks' gestation were not as
protected from development of preeclampsia as were continuing smokers. Because of
the otherwise serious adverse effects of smoking on the fetus, this minor "benefit" of
smoking during pregnancy probably has no public health consequence.

Randomized Trials of Smoking Cessation During Pregnancy

Three randomized trials have been conducted on pregnancy outcome in relation to
advice to stop smoking (Donovan 1977: Sexton and Hebel 1984; MacArthur. Newton.
Knox 1987). Table 9 summarizes the studies and birthweight results. Two other
randomized trials have also been conducted on the effect of various programs on
smoking cessation rates among pregnant women (Ershoff. Mullen, Quinn 1989:
Windsor et al. 1985). and other trials are in progress. Information on pregnancy
outcome is not available, and these studies are not reviewed.

Donovan (1977) studied smokers in three maternity units in England. Women aged
3.5 years or younger at the start of pregnancy, who smoked more than 5 cigarettes per
day, who had less than 30 weeks of gestation at the first prenatal visit. and who had no
prior perinatal deaths. were randomly assigned to a control group that received usual
prenatal care or to a test group that was given intense individual antismoking advice by
a physician at each prenatal care unit. There were 263 women in the test group and 289
in the control group. Mean daily cigarette consumption decreased from 17. I cigarettes
per day early in pregnancy to 9.2 cigarettes per day late in pregnancy in the intervention

ix7


TABLE 9.-Summary of birthweight outcome in randomized trials of
      smoking cessation in pregnancy

Reference

Number of     Smohmg aI end
\uhjecr\      of prq"""c)

I       c      I         c

   Binhueiyht rp)

1     c    Difference (g I'

Donovan I lY77)    x?

Sexton and Hebel   463
liYX4

MacArthur. Newton. 4Yi
Knox (IYX7)

group. but increased slightly from 13.7 to 16.4 in the control group. Mean birthweight
was 3. I72 g in the test group and 3. I83 g in the control group. In the test group IO
percent of the infants had low birthweight (4.500 g) compared with 9 percent in the
control group. There were four perinatal deaths in the test group and one in the control
group. None of the differences in birth outcome betueen the test and control groups
were statistically 4gnificant.

Although this trial might be regarded as evidence against a benefit of smoking
cessation during pregnancy. a number of limitations of the study must be considered.
First. no data are presented concerning the percentage of pregnant smohers who quit
smoking entirely. Reducing cigarette consumption almost certainly has a smaller
     c
benefit for pregnancy outcome than complete cessation. Second. the time at uhich
smoking behavior changed during pregnancy is unclear: data on cigarette consumption
for three periods during pregnancy uere obtained postnatally. and may have been
affected by recall bias. Data from ohser\ational studies discu\\ed in the pre\ ious
section strongly suggest that mohing during the last trimester of pregnancy i\ a critical
mediator of reduction in fttal growth among smokers.

Information from another British randomized trial (MacArthur. Neti ton. Knox 19x7)
alsoquestions the benefit ofmohing ce\\ation during pregnancy. In this \tud!. ~\omen
who \mohed at the time thei were scheduled for ;I prenatal 1 isit ;II a large hospital v.crc
assigned randomI> to a control group that received routine care or to an intervention
group that received supplementary health education about mohing during pregnant!
The planned intervention consisted of ad\ ice to 4top \mohin= 11 and information about
the effects of smoking on the fetus. pre\cnted I isuall> h! a bnohlet or \erball! by the
obstetrician. There were 4X9 \tomen in the control group and 193 in the inter\rntion
group. Mean birthweight for infant\ in the control group v.34 3. I30 g compared with
3.163 g for the intervention group. The pcrccntage\ of 1~ birthueight and perinatal
mortality in the tv.o groups \\erc not reported. The difference in mean hirth\\eight was


not statistically significant as determined by the conventional 0.05 probability value
and a two-sided test.
In this trial, only Y percent of the women in the intervention group quit smoking
entirely. compared with 6 percent of the women in the control group. The failure of
the intervention to cause smoking cessation makes this trial essentially uninformative
concerning the benefit. or lack of benefit. of smohing cessation during pregnancy. In
the intervention group. 28 percent of the women reduced the number of cigarettes
smoked per day, compared with IY percent of the women in the control group. The
greater reduction in cigarette consumption in the inter\,ention group. in the absence of
a difference in mean birthweight between the intervention and control groups. suggests
that reducing smoking does not entirely prevent the adverse effects of smohing on
birthweight.
The third randomized trial (Sexton and Hebel IY8-t) recruited women in a large
metropolitan area from various sources. Smokers of at least locigarettes per day at the
beginning of pregnancy, who had not passed the 18th week of gestation. were randomly
assigned to a control group that received routine advice or to a treatment group that
received intensive. ongoing advice throughout pregnancy from specially trained profes-
sional staff. There were 472 women in the control group and 363 women in the
treatment group. The mean birthweight of infants born to women in the control group
was 3.186 g compared with 3,278 g for infants of women in the treatment group. The
percentage of low birthweight infants was 8.9 in the control group and 6.X in the
treatment group. There were I I stillbirths in the control group and Y in the treatment
group. The difference in mean birthweight was statistically significant (~~0.05. two-
tailed test); the differences in the percentages of low birthweight and in fetal mortality
were not statistically significant.
In this trial, 43 percent of the women in the treatment group had ceased smoking
entirely by the eighth month of pregnancy. compared with 20 percent of the women in
the control group. The intervention was, therefore. highly successful in causing
substantial changes in smoking that exceeded changes in the comparison group. The
investigators ruled out concomitant changes in consumption of alcohol and coffee as
explanations for the increase in birthweight. Weight gain was 1 .O kg greater among the
treatment group than the control group. but at least part of the difference in weight gain
was a result of the higher birthweight of the infant (Sexton and Hebel 1984).
Review of these three randomized trials leads to two conclusions. First, to prevent
entirely the adverse consequences of smoking on birthweight, it is necessary for women
to cease smoking completely. Second. intensive interventions spanning the entire
period of gestation may be necessary to effect large changes among the percentage of
women who abstain from smoking entirely.


Prevalence of Smoking and Smoking Cessation During Pregnancy and Time
Trends in Prevalence and Cessation

Introduction

Ideally, conclusions about the prevalence of smoking during pregnancy and trends
in prevalence would be based on representative samples of pregnant women performed
at regular intervals using the same methodology. Assessment of smoking cessation
during pregnancy and time trends in smoking cessation should be based on repre-
sentative samples of women who start pregnancy as smokers and who are monitored
for smoking behavior throughout gestation. Available data fall short of these ideals.

Furthermore, available information on smoking and smoking cessation in pregnancy
is based almost exclusively on self-reported behavior. Few data on the quality of
self-reported smoking specifically in relation to pregnancy have been collected. and it
is possible that the societal pressures against smoking during pregnancy would make
underreporting more problematic than for other populations (Chapter 2). Similarly,
pregnant smokers who admit to smoking might underreport their daily cigarette
consumption. perhaps to a greater extent than nonpregnant smokers. The effect of
underreporting of smoking and overreporting of cessation would make the data from
former smokers more similar to that of continuing smokers with respect to their
reproductive health outcomes. Also. smokers who reduce the amount of nicotine in
their cigarettes by changing brands or those M ho reduce the number of cigarettes they
smoke per day without quittin, ~7 may compensate to maintain the \ame nicotine do\e
(US DHHS lY88).

Prevalence of Smoking and Smoking Cessation

Pertinent data on smoking during pregnancy from the 1985 National Health Interview
Survey (NHIS) (NCHS IYXX) are presented in Table IO. The IYXS survey focused on
health promotion and disease prevention.  The survey involved nearly 35.000
households and more than YO.Ot)O persons. and the response rate was 95.7 percent.
Information concerning smoking during pregnancy' vvas obtained from all female
household members aged IX to 4-I kears ti ho had had a live birth in the 5 years prior to
the survey. The proportion of vvomen u ho had smohed at any time during the year
preceding pregnancy ~`a\ 32 percent overall. Of Momen uith less than I? years of
education. 46 percent smohed in the year preceding pregnancy. compared with I.3
percent of women with I6 or more years ofeducation. Thirty percent of married women
had smoked. compared u ith 10 percent of formerly married u'omen.

Patterns of smohing cessation or reduction uere reported in detail for some
demographic subgroup\. Overall. 2 I percent of women who smoked prior to pregnancy
quit upon learning of their pregnancy. and an addittonal 36 percent reduced the number
of cigarette\ they smohed. Cessation (but not reduction) *as strongly related to
education and family income. Among uomen with less than 12 years of education. I2
years of education. and more than 12 years of education. 15. 10. and 32 percent quit.


TABLE IO.-Smoking and smoking cessation during pregnancy. summary of
      results of two surveys of national probability samples

respectively. The proportions for reduction in smoking were 34. 3X. and 36 percent.
respectively. Younger mothers were slightly more likely to quit than older mothers.
and white mothers quit slightly more often than black mothers (2 I vs. IX percent ). More
married mothers (23 percent) than never married (19 percent) or formerly married ( I3
percent) mothers quit. although the proportions reducing their smoking levels were
similar (36. 37. and 35 percent. respectively).

Fingerhut. Kleinman. and Kendrick (1990) also reported data on smoking in whites
before and during pregnancy based on the Linked Telephone Survey. which reinter-
viewed I.550 women aged 20 to 44 years who were respondents to the 1985 NHIS.
This analysis confirmed the previous findings that smoking prior to pregnancy and
quitting during pregnancy were strongly related to age and educational attainment.
Information on amount smoked prior to pregnancy was obtained in this survey.
Fifty-nine percent of women who smoked less than I pack per day prior to pregnancy
quit smoking. compared with 25 percent of those who smoked I pack or more per day.
Of the white women who smoked prior to pregnancy, 3Y percent quit during pregnancy
(27 percent when they found out they were pregnant and I2 percent later in pregnancy).
This estimate of quitting during pregnancy is higher than the previous estimate of
quitting from whites in this survey because it includes as quitters both women who quit
upon learning that they were pregnant and those who quit later in pregnancy.

Smoking during pregnancy was also assessed in the 1980 NNS (Prager et al. 1983)
(Table IO). Questionnaires were distributed to a national probability sample of married
women who had had live births in 19X0; the response rate was 56 percent. The
restriction to married women severely compromises the generalizability of results.
especially for subgroups such as blacks and youth because smoking during pregnancy~


is consistently more common among unmarried mothers (Schramm 1980; Rush and
Cassano 19x3) and nearly one-half of black infants are born to unmarried mothers
(NCHS IYXZ). The low response rate might have also affected the validity of the study,
Prayer and associates ( 19X4) asked women how many cigarettes they smoked per day
before and after they found out they were pregnant. Among all married respondents.
3 I percent smoked before pregnancy. Whites were more likely to smoke than blacks
(32 vs. 25 percent). These investigators reported a strong association of smoking with
age. with younger mothers more likely to smoke than older mothers. There were even
more pronounced gradients with education. Among women with less than a high school
education. 50 percent smoked before pregnancy, and this percentage diminished
monotonically to IS percent among women with 16 or more years of education.
Amonp the women in the study (PraLger et al. 19x3) who smoked prior to pregnancy.
IX percent quit after realizing they were pregnant. White women were somewhat more
likely to quit than black women ( IX vs. I3 percent). Mothers older than 35 years of age
were markedly less likely to quit: only 7 percent did. Again. education had a strong
association with quitting: IO percent of mothers with less than I2 years of education
quit. and the percentage increased monotonically to 33 percent among mothers with I6
or more years of education. The patterns of cessation by amount of smoking are also
of interest. Women who were smoking I to IO cigarettes per day at the time of
pregnancy recognition were far more likely to quit than women smoking I1 or more
cigarettes per dav (3 I v's. I2 percent). Among the heavier smokers. 27 percent reduced
their consumption to IO or fewer cigarettes per day even though they did not quit.
Williamson and associates ( 19x9) used data from the Behavioral Risk Factor Surveil-
lance System in 1985 and I986 to compare smoking patterns among pregnant and
nonpregnant women. Data were collected through 19.12-t telephone interviews of a
population-based sample of women in 26 States. with ascertainment of current preg-
nancy status. smohing history. and current smoking practices. Women pregnant at the
time of interview were less likely to be current smokers than nonpregnant women (3 I
vs. 30 percent). but had a similar likelihood of ever having smoked (43 vs. 35 percent).
The proportion of former smokers was thus greater among pregnant women (?1 v's;. I5
percent). largely accounting for the difference in current smoking patterns. This study,
(Williamson et al. I 0x4,) suggests that if 30 percent of women pregnant at the time of
the sutvey smoked prior to pregnancy. then 30 percent of smokers would have had to
quit after becoming pregnant to account for the reported smoking rate of 1 I percent.
Among pregnant women w.ho smoked. the mean number of cigarettes consumed per
day was I?. compared M ith 20 cigarettes per day among nonpregnant w otnen who
smohed. These data suggest that smokers who do not quit upon becoming pregnant
tend to reduce their cigarette consumption (Williamson et al. IYXYI.
Patterns ot`smohing were generally similar across demogaphic subgroups. with one
important exception. Among unmarried women. smohin, ~7 was slightly more common
in pregnant than nonpregnant women (36 vs. 3-t percent,. implying no change in
smoking among unmarried pregnant women. The absence ofpregnancy-related reduc-
tion in smoking for unmarried women was due exclusively to a markedly higher
smoking prevalence for white unmarried pregnant women. The results suggest that
data on married mothers cannot be generalized to unmarried mothers.


A number of investigators reported smohing patterns in selected populations. such as
women delivering in a particular hospital or geographic region or those receiving
prenatal care at a specific clinic.  Table I I summarizes se\,eral of these studies.
Although none are true probability samples. these studies provide an indication of the
diversity of smoking and smoking cessation among different populations. The propor-
tion quitting during pregnancy ranges from 6 to 3Y percent.

Time Trends in Smoking and Smoking Cessation

Kleinman and Kopstein ( 19X7) compared the pattern of smohing cessation during
pregnancy from the similarly designed IY67 and 1980 NNS. Although there were some
changes in the proportion of mothers vv ho were married at the time of each of the 1~ o
surveys and the characteristics of nonrespondents might have varied. the surveys
provide a unique opportunity to assess temporal trends in smoking and smohing
cessation during pregnancy. The percentage of mothers u ho smohed prior to pregnancy
decreased markedly during that period. from 55 to 30 percent for white mothers and 40
to 3 percent for black mothers. The percentage of v.hite mothers who quit after
pregnancy rose from I I to I7 percent between the two surveys. whereas the percentage
of black mothers who quit decreased from I7 to I I percent over that interval. During
the interval between the surveys. the diminution of smoking during pregnancy was more
pronounced for highly educated women. increasing the differential exposure to tobacco
by educational status (Kleinman and Kopstein 1587).

Estimates of Attributable Risk Percent

Although several measures of attributable risk are commonly used to describe the
burden of disease associated with an exposure, the most recent report of the Surgeon
General (US DHHS 1989) has focused on attributable risk percent. frequently termed
etiologic fraction, as the most relevant measure of the likely public health impact of
smoking cessation. Calculation of the attributable risk percent uses the formula as
follows:

where p is the proportion of persons with the exposure and RR is an estimate of the
relative risk of the outcome in those who are exposed compared with those unexposed.

At least three different studies (Meyer. Jonas. Tonascia 1976: McIntosh 1983:
Kramer 19X7) estimated the relative risk of several pregnancy outcomes after reviewing
the research literature. Table I2 summarizes these studies and provides estimates of
attributable risk for prevalences of smoking of 20. 30.40. and SO percent based on the
relative risk estimates from the three studies. As noted earlier. demographic subgroups
of women differ markedly in smoking prevalence. Of those women with less than a
high school education, 50 percent smoked during pregnancy: of those women with some
college education, 20 percent smoked during pregnancy (NCHS 19X8). Approximate-
ly 30 percent of married women and 30 percent of unmarried women smoked prior to


`I'AIILE 1 I.--Patterns of smoking cessation during pregnancy among selected populations



K~lclcllc~             I.~~c;lll,lll                 SOlILY             YW\     `4 Snrohing iilltially

I'jSX            43             3)            NK


I `M1 (30          I7             II            IO


I'j5X            3X             IX            YK

NK             so             NK           1-I


I97h- 70          37            4')            NK


I'JXO            22            3x            NK


I'JXO            NK            2X            NK


I'MI  x1          20            22            NK


I')X I -x2          1')             h            I `1

NK             32             I7            NK


TABLE 12.-Summary of studies that estimated relative risk of various
      pregnancy outcomes for smoking based on a "q nthesis" of
      the literature, and attributable risk percent based on several
      estimates of the pre\,alence of smoking during pregnanq

P"

Prcwrm lklI\~r!

RR         .AR'r

Ile>cr. Joru\.   031   I 2ib       1
TOn;i\LLl      0 Xl              h
l19?hl       0.10              X
          0.w             IO

Llclnlo\tl      0.3)    I.25
I IYX1l       o.io
          I).10
          0.50

Kwn1er       0.21)    NK
/19X7)       0.20
          0.10
          0.50

i        I.81

Y
I I

        2 12

I,YY"

pregnancy (NCHS 198X). The most recent estimates suggest that about 25 percent of
U.S. women smoke throughout pregnancy (NCHS 1988).

The relative risk estimates forperinatal mortality and preterm delivery are remarkably
consistent. especially considering that these authors conducted independent syntheses
of the literature. Estimates of the relative risk of low birthweight ranged from 1.81
(McIntosh 1984) to 2.42 (Kramer 1987). probably because ofdifferences in the number
of studies used to derive the estimate. For this reason. attributable risk percent for a
given prevalence of smoking is more variable for low birthweight than for perinatal
mortality and preterm delivery.

Based on data that indicate that about 25 percent of U.S. women smoke throughout
pregnancy. it can be estimated that 5 to 6 percent of perinatal deaths. 17 to 26 percent
of low birthweight births. and 7 to 10 percent of preterm deliveries could be prevented
by elimination of smoking during pregnancy. In groups with a SO-percent prevalence
of smoking. such as women with less than a high school education. approximately IO
to 1 1 percent of perinatal deaths . 29 to 42 percent of low birthweight births. and 13 to
18 percent of preterm deliveries might be prevented by elimination of smoking during
pregnancy. These contributions to adverse pregnancy outcome are sizable. and smoh-
inp is probably the most important modifiable cause of poor pregnancy outcome among
women in the United States (Kramer 1987).


Age at Natural Menopause

Introduction

The significance of menopause extends beyond marking the end of female reproduc-
tive potential. The age at which menopause occurs also may have implications for the
risks ofosteoporotic fractures. irchemic heart disease. and cancers of the reproductive
system. Thus. the effect of smoking on the age of menopause could have potentially
broad health implications.

In fact. an early natural menopause has been observed consistently among women
who smoke cigarette\. As summarized in Table 13. the major studies addressing this
topic have indicated that currently smoking women cease menstruating from I to 2
years earlier than otherwise similar nonsmokers. Expressed as relativje risk. women
ased 43 to 54 years who \mohe become menopausal at about twice the rate of never
smokers (Willett et al. IYX3: Bailey. Robinson. Ves\ey 1977: Hartz et al. 1987:
Andersen. Transbol. Christiansen 1983: Baron 1990).

Several features of the data suggest that [hi\ is a causal relationship. By using both
cohort and cross-sectional methodology with a variety of subject populations. the
results have been replicated repeatedly in studies in several areas of the United States
and Europe. Dose-response effects have generally been found. with heavy smokers
experiencing an even earlier menopause on average than light smokers. HowevJer. these
trends have not a1uay.s been assessed with formal test\ of statistical significance in the
reports describing the data. Several studie\ demonstrating thi4 association have con-
trolled for potential covlariates. That premenopausal smokers may be more lihely than
nonsmokers to have a hysterectomy does not appear to explain the relationship (Krailo
and Pike 1983).

Pathophgsiologic Framework

There are at lea\t three way\ in which cigarette smoking could lead to an early natural
menopuu\e. Experiments M ith laboratory rodents indicate that the polycyclic aromatic
hydrocarbons found in cigarette smoke may be directly, toxic to ovarian follicles
(Mattison IYXO). Mattison and colleagues found that intraperitoncal injection of
benzo(a)pyrene. .i-meth~lcholanthrene. or 7.12.dimethylbenz(a)anthtracene led to
ovarian folliculur atresia (Matti~nl and Thorgcirsson lY7X. lY7Y: Gulyas and Mattison
lY7Y). Earlier uncontrolled studic`s of prolonged exposure of mice to cigarette smohe
led to similar findings (Es\enbcrg. Fa~an. 5lalerstein IYC I ). which were also seen in a
later controlled study of rut\ rsubbarao IYKX). However. other investigator\ failed to
find ovarian atrophy in rodent\ chronically rxposcd to cigarette mohe (Haag. Larson.
Weatherby 1960: Dontenuill ct al. IY73a). and in most studies. parentersl nicotine or
tobacco extract has had minimal effect on the ovaries of experimental animals (Es\en-
berg. Fagan. Maler\tein I Y5 I : Thienes I Y60: Lar\on. Haag. Silv ette I Y6 I ; Larson and
Silvette 196X).

The other two postulated mechanims for premature menopause do not involve direct
ovarian toxicity. Cigarette smoking may interfere with IuteiniLing hormone release at


TABLE 1X-Summary of studies reporting relationship of cigarette smoking
      and age at natural menopause

Dantell t IY7X1

Lindqui\t and
Rell!It\\on t IY7Y)

Kaufmm et 21
t I YXO,

Adenu and
Gtlla~hrr t I `)X2)

Willert et al. t IYXi)

I O.YYS health
\creenrrs

McKinlay. Bifano.
McKinla) (19X.5)

E\er\on et 31. (19X6)

Hiatt and Fireman
(10X6)

Stanford et al. t 19x7 )

Brambilla and
McKinlny (19X9)

s.350 generul
populatton subject\

261 population wbjcct\

5.316 HMO health
scrrenees

I.7




I .3"





?.(I'

I P



1i.X'


1.2"



1.7'





I .o



I .4







I .7



1.I

NYS'



0.3



I .s


least in rodents exposed to parenteral nicotine or cigarette smoke (Andersson et al. 1980:
Andersson et al. 1984; Andersson et al. 1988: Eneroth et al. 1977a.b; Kanematsu and
Sawyer 1973: Blake. Norman. Sawyer 1974: Blake 1974: Blake et al. 1971a.b: McLean.
Rubel. Nikitovitch-Winer 1977). This effect appears to be due to a nicotinic effect on
neurotransmitter release. A return to a more normal function after the end of exposure
to smoke or nicotine has not been documented. but it seems likely that such a nicotinic
effect on the brain would not be permanent. Therefore. it is possible that in humans.
smoking could cause a reversible interference in the pituitary-ovarian axis, which could
lead to a cessation of menses. Several investigators found that smoking has been
associated with menstrual it-regularity earlier in reproductive life (Wood 1978; Pet-
tersson. Fries, Nillius 1973: Brown. Vessey. Stratton 19X8: Hammond 1961).

Smoking has also been associated with disturbances of estradiol metabolism. Mich-
novicz and colleagues (1986) found that premenopausal smokers tend to metabolize
estradiol through pathways producing more catechol-estrogen metabolites than non-
smokers. This change would be expected to result in a relative antiestrogenic influence
because of the lack ofestrogenic potency of the catechol-estrogens compared with the
estrogenic metabolites. such as estriol. which are produced in larger amounts in
nonsmokers. There is also evidence that nicotine may inhibit aromatase. an enzyme
important in the synthesis of estrogen\ (Barbieri. McShane. Ryan 1986: Barbieri.
Gochberg. Ryan 1986). Again. the recovery of normal enzymatic function after
cessation of smoking has not been studied. However. it is postulated that these or
similar disturbances could result in enough antagonism of estrogen effect to cause an
early cessation of menstrual cycling in women already in the perimenopausal years
(Baron. LaVecchia. Levi 1990)

Studies of Former Smokers

Former smoker\ experience menopause only slightly earlier than never smokers
(Table 13). In a study of hospitalized women. Jick. Porter. and Morrison ( 1977) found
that former smokers had a median age at menopause between that of never smokers
and that of women currently smoking half a pack of cigarettes per day. Kaufman and
coworkers ( 19X0) reported on hospitalized women aged 60 to 69 years. Data from 10
women who stopped smoking before age 35 indicated that the mean age at menopause
was 0.2 year\ earlier than in never makers. after adjustment for parity and body habitu\
(Kaufman et al. 1980). In a cross-sectional study ofwomen attending a screening clinic.
Adena and Gallagher (19X2) found ex-smokers to have a median age of natural
menopause 0.3 year\ earlier than never makers. Finally. Hiatt and Fireman (1986)
found among a group of enrollees in a prepaid health plan attending a screening clinic
that es-\moken reached menopause about 0.5 years earlier than never smokers. Thu\.
natural menopause appears to occur. at most. 6 month\ earlier in ex-smokers than in
never smokers.

Limited findings on relative risk of early menopause in former smokers are av,ailable
(Willett et al. 1983; Baron. LuVecchia. Levi 1990). From data presented hy, Lindquist
and Bengtsson ( 1979) regarding %-year-old women. it can be calculated that compared
with never smokers. former smokers had a relative ri& of early menopause of I.8


TABLE Id.-Summary of studies of age at natural menopause among former
        smokers

              Vumtxr of         COViiKiI~~
Reference          ei-rmoker\         considered            Fmdllw\

Lmdqw\t and Bengtwn
f 1071))

Kaufman et al.
I IYXO)

Adena and Gallagher
IIYX?~

Willett et al.
(10X3)

30







I 0







NR

I6.034

Age. welght.
nulliparity

Sib           None

Pan). reyon.
Quetelst'\ Index

None

Htatt and Fireman
(19861

Mean age at menopauw
wit\ 0.2 \`rarllrr among
e\-moher\ than amcmy
never vnohw

Odd\ ratto of herng
menopausal for current
\moher\ v,. never
smoker\ ua\ I. IO

Mean aFe at menopause
u3k 0.5 y earlier among
ex-\mohrr\ than among
never vnoher\

(95percent confidence interval, (CI), 1.14.7). In a prospective study of American
nurses, Willett and coworkers ( 1983) found ex-smokers to have a relative risk of early
menopause of I.1 (95percent CI, 0.98-1.23) compared with never smokers after
adjustment for age, weight, and nulliparity. In this study, those who stopped smoking
in the 2 years previously retained a modest increase in risk of early menopause
(RR=1.4); after a longer period of abstinence, there was no effect associated with
previous smoking (Willett et al. 1983).

All the investigations of smoking and menopause have relied on self-report of
menstrual status and smoking history. It is unlikely that misclassification with regard
to these features would seriously distort the findings regarding current smoking, but the
results for former smoking may be more susceptible to artifact. In particular. some of
the study participants who claimed to be former smokers might actually have continued
to smoke, or they might have quit for health reasons related to an early natural

399


menopause. Like current smokers. former smoher\ may be more likely to be passively
exposed to passive 3mohing than ne\er smokers. thus po\cihly affecting menopaui;al
age. These factors would tend to lead to an exaggeration of the apparent impact of
former smoking on menopausal age (Chapter 2). Therefore. the results summarized
above may overstate the degree to which former smoking is associated with any
disturbance in menopausal age.

It appears that age at menopause in former smokers is closer to that of never smokers
than to current smokers. and the data are consistent with a decline in the risk of early
menopause with the cessation of smoking. The effect of smoking on menopausal age
may be partly or wholly rever\ible with cessation of smoking during the premenopausal
years. However. some pertinent data are lacking. Most of the studies did not consider
how long it takes after cessation of smoking for the ri\k of early natural menopause to
decrease. Ko studies have verified that the women bho stopped smoking had a lifetime
smoking exposure similar to that of women who continued smoking.

PART IL MALE

Introduction

Cigarette smoking has been considered to be associated with impairment of male
sexual functioning. and tobacco abstinence has been recommended for men attempting
to maximize sexual performance (Larson. Haag. Silvette 1961: Sterling and Kobayajhi
1975: Ochsner I97 1a.b). An association between mokin2 and impaired sexual per-
formance among men ha\ been publicized in the lay pre\\ (Reuben 19X8). Althouph
some data pro\ ide e\ idcnce for this association. the> are inconclusix e.

Pathophysiologic Framework

Three general type\ of mechanisms habe been proposed to explain the harmful effect
of cigarette 9ilohing on wxtial performance. impotence. and sperm qualit),. First.
smoking ma! expose the te\te\ to compounds that are directI! toxic to the 5pern-
producing germinal epithelium. to earl! sperm form\. or to the hormone-producing
Leydig cell\. The effect\ on sperm rna\~ be ;I manifestation of ;I genotoxic effect of
cigarette <make constituents (Ohe and Herha 197X: DcSlarini 19x3 ).

Second. smoking C;IIIW\ atherosclerotic peripheral \ a\cular disease (Chapter 6): thi\
may translate into adiminished va\cuIar \uppl> to the genital\. a\ reflected b! the penile
brachial index (PBI) and other 1 atcular mt`a\urement\. A diminished va\cuIar \uppl>
to the genitals would compromi\c w~ual performance and spermatogenesis and hor-
mone production. Although athero\clero\i5 i\ often considered a fixed lesion. several
studie\ ha\,c \ugcsted that :ithcro\clerotic plaque\ ma! regre\\ v, ith approprlatr
lifestyle change\ (Barndt et aI. 1077: Nihhil;l 19X0: Kram~h et al. 19X I: Chapter 6).
However. no studies ha\e been conducted on the effect of \mohing cessation on
regression of atherosclerotic lesion\.