American Journal of Epidemiology

Copyright C 1999 by The Johns Hopkins University School of Hygiene and Public Hearth
All rights reserved

Vol. 149, No. 3

Pnnted in U.SA.

Exposure to Environmental and Mainstream Tobacco Smoke and Risk of

Spontaneous Abortion

Gayle C. Windham,1 Julie Von Behren,1 Kirsten Waller,1'2 and Laura Fenster1
The authors examined the risk of spontaneous abortion from environmental tobacco smoke (ETS) exposure
in a prospective study of over 5,000 women conducted in California during 1990-1991. Among nonsmokers,
there was little association by hours of ETS exposure at home or work (adjusted odds ratio (OR) for any
exposure = 1.01, 95% confidence interval (Cl) 0.80-1.27), or by paternal smoking. However, the risks
associated with ETS exposure were increased among nonsmokers who had moderate alcohol or heavy caffeine
consumption. A moderate association with maternal smoking was observed (adjusted OR for >5 cigarettes per
day = 1.3, 95% Cl 0.91-1.9). Am J Epidemiol 1999; 149:243-7.

prepaid health plan in California when they called to

make their first prenatal appointment. To be eligible
for the study, pregnant women had to be at least 18
years old, 12 weeks gestation or less (mean = 8
weeks), and Spanish- or English-speaking. Telephone
interviews were completed within a few weeks of initial contact for 5,342 women; reasons for noncompletion included 18 percent refusals, about 10 percent
ineligible, and 3 percent lost to follow-up.
Pregnancy outcomes were ascertained primarily by
computerized hospital admission records and abstraction of medical records. Less than one percent of outcomes could not be determined. These pregnancies
were excluded as were therapeutic abortions and
ectopic or molar pregnancies (n = 198). Pregnancies
that ended by 20 completed weeks of gestation or earlier were defined as spontaneous abortions (n = 499).
The remaining pregnancies resulted in 4,613 livebirths
and 32 stillbirths.
Most of the consumption questions on the interview
(e.g., alcohol, caffeinated beverages, and cigarettes
smoked) were asked for two time periods: the week
before interview (considered "during" pregnancy) and
the week at last menstrual period. The smoking status
of the spouse was also ascertained during pregnancy,
but the amount he smoked was ascertained only for the
3 months before pregnancy. Exposure to ETS was
ascertained as the average number of hours per day the
respondent was near other people smoking at home
and at work since the last menstrual period. Hours at
the two locations were summed to create a total daily
ETS exposure. ETS exposure was examined only
among women who reported not smoking at both time
periods (n = 4,209).

Maternal smoking during pregnancy is considered a

hazard to the fetus (1,2). Studies of spontaneous abortion have generally found an association: reviews of
the literature suggest increased risks on the order of
25-50 percent, with some studies showing greater
effects with heavier smoking (3, 4). Because of these
risks, exposure to environmental tobacco smoke (ETS)
is also of concern. To our knowledge, only two studies
have examined spontaneous abortion in relation to
ETS exposureone a case-control study that we conducted in California (5) and the other a prospective
study carried out in Sweden (6). Both studies had
similar findings of about a 50 percent increased risk,
but had various limitations including only limited
exposure ascertainment in the case-control study.
Therefore, we have examined this issue further in a
prospective study of spontaneous abortion which
ascertained usual daily hours of ETS exposure at home
and work. In addition, we present data on maternal and
paternal smoking.
MATERIALS AND METHODS

Details of data collection have been published elsewhere (7, 8) and are summarized briefly here. Pregnant
women were recruited during 1990-1991 from a large
Received for publication November 18, 1997, and accepted for
publication June 15, 1998.
Abbreviations: Cl, confidence interval; ETS, environmental tobacco smoke; OR, odds ratio; SAB, spontaneous abortion.
1
Environmental Health Investigations Branch, Department of
Health Services, Oakland, CA.
2
Currentty with Public Health Institute, Berkeley, CA.
Reprint requests to Dr. Gayle Windham, Department of Health
Services, 1515 Clay Street, Suite 1700, Oakland, CA 94612.

243

Downloaded from http://aje.oxfordjournals.org/ by guest on May 2, 2015

abortion, spontaneous; pregnancy; tobacco smoke pollution

244

Windham et al.

The hours of ETS exposure were categorized and

crude risk ratios and 95 percent confidence intervals
for spontaneous abortion were calculated relative to
nonexposed. Any exposure versus none was entered in
logistic regression models because there was little
increase in risk (univariately) by greater hours
exposed. Odds ratios were adjusted for maternal age,
prior history of spontaneous abortion, alcohol and caffeine consumption during pregnancy, and gestational
age at interview. Potential effect modifiers of a priori
interest from previous smoking studies were alcohol
and caffeine consumption, which were examined for
departure from additivity of effects (9, 10). Effects of
active smoking were examined using similar logistic
regression models.
RESULTS

variable
Maternal age (years)
<35
Pregnancy history
No prior pregnancy
No SAB, but pregnancy
1 prior SAB
2 SABs
Maternal race
White
Hispanic
Black
Asian or Filipino
Education
<High school
High school graduate
Some college or graduate
Employed at interview
Yes
No
Gestational age (weeks) at
interview
^8
>8
Smoke before or during
pregnancy
Yes
No
ETS exposure (hours/day)
in nonsmokers
0
0.5
1.0-1.5
2.0-5.5
56

No. of
pregnancies

SABs(%)

4,583
561

8.8
17.5

1,320
2,754
815
247

8.4
9.5
9.8
18.6

3,390
945
330
475

9.4
8.3
13.0
12.0

338
1,789
3,012

7.7
9.2
10.2

3,861
1,281

10.3
7.8

3,287
1,847

11.1
7.3

933
4,209

10.4
9.6

3,016
313
311
344
210

9.6
11.2
8.4
9.0
9.5

amounts smoked. Adjustment decreased the risk

slightly. Women who reported that they quit smoking
during early pregnancy were not at increased risk of
having a spontaneous abortion. The risk from smoking
(>5/day) appeared greater for losses which occurred
after 10 weeks (adjusted odds ratio (OR) = 1.6, 95 percent confidence interval (CI) 1.0-2.4) than for earlier
abortions (adjusted OR = 0.9, 95 percent CI 0.6-1.7).
There was slight effect modification of the active
smoking association by caffeine consumption of >150
mg/day, but not when categorized as >300 mg/day.
There was no effect modification by alcohol consumption during pregnancy, nor by spouse smoking habits.
DISCUSSION

This prospective study found little evidence for an

association of spontaneous abortion with ETS expoAm J Epidemiol

Vol. 149, No. 3, 1999

Downloaded from http://aje.oxfordjournals.org/ by guest on May 2, 2015

The proportion of spontaneous abortions was

9.7/100 (9.6/100 among nonsmokers). The median
gestational age at loss was 11 weeks, with 72 percent
of spontaneous abortions occurring in the first
trimester. Risk factors for spontaneous abortion in
these data (table 1) were similar to those generally
found in the literature (11). Among nonsmoking
women, 13.5 percent reported ETS exposure at home
and 23 percent of working women reported ETS exposure at work (figure 1), with about 28 percent exposed
overall. Although more women reported exposure at
work than home, much of the exposure tended to be of
short duration.
The proportion of spontaneous abortions did not
increase consistently with increasing ETS exposure
(table 1). When examined by source of exposure, miscarriage risk was somewhat elevated at >6 hours per
week of home exposure compared with none (11.4/100
vs. 9.4/100), but greater hours of work exposure were
not associated with greater risk of spontaneous abortion. The adjusted odds ratios for any ETS exposure by
source were all close to null (table 2). Some effect
modification by alcohol and caffeine consumption was
found (table 3). There was no difference in the ETS
association for earlier versus later losses.
Among nonsmokers, 15.5 percent of women reported that the father of the pregnancy smoked. The risk of
spontaneous abortion was not increased with his smoking status during pregnancy, nor with amount smoked
by the father before pregnancy (table 4). Combining
women who reported either ETS exposure at home or
spouse smoking or both did not identify any increased
risks for spontaneous abortion.
In comparison, women who smoked >5 cigarettes
per day during pregnancy were 40 percent more likely
to have a spontaneous abortion than nonsmokers (table
5), but there was little dose-response trend for greater

TABLE 1. Proportion of spontaneous abortions (SABs) by

selected demographic factors and environmental tobacco
smoke (ETS) exposure during pregnancy, California,
1990-1991

Tobacco Smoke Exposure and Risk of Spontaneous Abortion

245

Exposed

Total

FIGURE 1. Proportion of nonsmokers who reported exposure to environmental tobacco smoke (ETS) at home, work, and combined, by hours
exposed, in a prospective study of pregnancy outcome, California, 1990-1991.

TABLE 2. Adjusted* odds ratios (OR) for spontaneous

abortion (SAB) and environmental tobacco smoke (ETS)
exposure during pregnancy, among nonsmokers, California,
1990-1991
Variable

Any ETS at home

Any ETS at work
Any ETS, either place

No.

SAB
(%)

Adjusted
OR

95%Clt

571
743
1,178

10.9
9.0
9.6

1.15
0.88
1.01

0.86-1.55
0.66-1.17
0.80-1.27

* Adjusted for maternal age, prior spontaneous abortion, alcohol

and caffeine consumption, and gestational age at interview, and
compared with nonexposed for each source.
t Cl, confidence interval.

sure, except among women who also consumed alcohol

or caffeine in moderate to high amounts. The increase
in risk seen with active maternal smoking is consistent
with previous reports (3,4). Women who consume alcohol or caffeine may be more susceptible to further risks
from ETS than other women. However, effect modification of alcohol and caffeine on the relation with
smoking was not seen clearly in this study nor in previous studies (5, 11-13), so these findings are tenuous.
We are aware of only two previous studies of spontaneous abortion with data on ETS exposure other than
spousal smoking, which were limited to nonsmoking
mothers and adjusted for confounders (5, 6). Both
reported a moderate association of spontaneous abortion with ETS exposure, which we would have been
able to detect with greater than 80 percent power. In
Am J Epidemiol

Vol. 149, No. 3, 1999

the Swedish study (6), the association was observed

only with workplace exposure (adjusted relative risk =
1.5, 95 percent CI 1.0-2.4), which was defined as
spending most of the time at work around smokers.
The California study (5) did not ask about sources of
exposure separately, but rather whether subjects were
exposed to ETS for >1 hours per day at home or work
(adjusted OR = 1.6, 95 percent CI 1.2-2.1). These
findings are of similar or greater magnitude to the
associations found between spontaneous abortion and
active smoking in the same data sets.
ETS contains many of the same chemical constituents as those in mainstream smoke, but the physiochemical nature of sidestream smoke compared with
mainstream smoke leads to differences in the ratio of
constituents, which may be higher in sidestream
smoke (14). Sidestream smoke undergoes dilution and
further chemical transformation to form ETS. Because
cotinine levels measured in nonsmokers who are
exposed to ETS are generally one to two orders of
magnitude lower than in smokers (15), the effect of
ETS exposure might be expected to be much lower
than that of active smoking. However, this presupposes a linear relation and that cotinine (a metabolite of
nicotine) is the primary toxicant of interest (14).
The current study is based on more detailed assessment of ETS exposure than the previous California
study (5), but may be subject to some misclassification (and nullification of risks) because exposures outside the home or workplace were not ascertained, nor

Downloaded from http://aje.oxfordjournals.org/ by guest on May 2, 2015

Work
ETS Exposure

Home

246

Windham et al.
TABLE 3. Additive models for effect modification of caffeine and alcohol on association of
environmental tobacco smoke (ETS) and spontaneous abortion (SAB), in nonsmokere, California,
1990-1991
Caffeine consumption
ETS

Caffeine
(mg/day)

None
Any
None
Any

300

300
>300
>300

No.

2,955
1,138
61

40

SABs

9.6
8.9
9.8
27.5

Alcohol consumption

Adjusted'
' 95% Clf
ORt

0.93
0.84
3.4

0.73-1.2
0.35-2.0
1.7-7.0

RERIf = 3.4-0 .93-0.84 + 1 = 2.6 (0.2-5.0)

Alcohol
(drinks/wtc)

NO.

3,002
1,168
14
10

S3
S3

>3
>3

SABs
(%)

9.6
9.3
14.3
30.0

Adjusted
OR

0.98
1.0
2.9

95% Cl

0.78-1.2
0.22^1.8
0.72-11.6

RERI = 2.9-0.98-1.0 + 1 = 1.9 (0.2-3.6)

* Adjusted for maternal age and alcohol use.

t OR, odds ratio; Cl, confidence Interval; RERI, relative excess risk due to interaction and 95% Cl of this risk
difference (see references 9 and 10).
X Adjusted for maternal age and caffeine use.
Reference group

Paternal smoking
(cigarettes/day)

No.

SABs(%)

0
1-10
11-20
>20

3,550
405
186
55

9.6

Adjusted*
0 R +

95% CI*

TABLE 5. Adjusted* odds ratios (OR) for spontaneous

abortion (SAB) and amount of cigarettes smoked during
pregnancy, California, 1990-1991
Maternal smoking
(cigarettes/day)
0

9.4

0.98

0.73-1.3

9.1
10.9

1-4
5-9

0.97

0.41-2.3

10

* Amount father smoked during 3 months before last menstrual

period.
t Adjusted by logistic regression for maternal age, prior fetal loss,
alcohol and caffeine consumption, and gestational age at interview.
X OR, odds ratio; Cl, confidence interval.
Reference group.

were details about the intensity of exposure. The proportion of nonsmokers who reported ETS exposure is
lower than in some studies, but is similar to our study
conducted 4 years earlier and to other studies in
California of nonpregnant women (5, 17). California
has lower smoking rates than other states and more
severe smoking restrictions (17, 18), so that exposure
to ETS may not only be less frequent but also of a
lower magnitude or intensity. To date, the literature on
the possible association of ETS exposure and fetal
loss is somewhat limited. Additional studies should
include highly exposed women to provide power for
detecting a possible association and more detailed
exposure assessment.

REFERENCES
1. US Department of Health and Human Services. The health
consequences of smoking for women: a report of the Surgeon
General. Atlanta, GA: USDHHS, Public Health Service,
Office of the Assistant Secretary for Health, Office of
Smoking and Health, 1980.

No

SABs
(%)

Adjusted
OR

4,607
209
119
208

9.4
9.6
13.5
13.5

+
0.91

} 13

95% Clt

0.56-1.5
0.91-1.9

* Adjusted for maternal age, prior fetal loss, alcohol and caffeine
consumption during pregnancy, and gestational age at interview,
t Cl, confidence interval.
X Reference group.

2. Stillman RJ, Rosenberg MJ, Sachs BP. Smoking and reproduction. Fertil Steril 1986;46:545-66.
3. Windham GC. Effects of tobacco smoke on risk of spontaneous abortion. In: Current issues in smoking and reproductive
health. ARHP Clinical Proceedings 1996:13-15.
4. DiFranza JR, Lew RA. Effect of maternal cigarette smoking on
pregnancy complications and sudden infant death syndrome. J
Fam Pract 1995;40:385-94.
5. Windham GC, Swan SH, Fenster L. Parental cigarette smoking
and the risk of spontaneous abortion. Am J Epidemiol 1992;
135:1394-1403.
6. Ahlborg G Jr, Bodin L. Tobacco smoke exposure and pregnancy outcome among working women. A prospective study at
prenatal care centers in Orebro County, Sweden. Am J
Epidemiol 1991;133:338-47.
7. Fenster L, Schaefer C, Mathur A, et al. Psychologic stress in
the workplace and spontaneous abortion. Am J Epidemiol
1995;142:1176-83.
8. Windham GC, Von Behren J, Fenster L, et al. Moderate maternal alcohol consumption and risk of spontaneous abortion.
Epidemiology 1997;8:509-14.
9. Rothman KJ. Modern epidemiology. Boston, MA: Little,
Brown, 1986.
10. Hosmer DW, Lemeshow S. Confidence interval estimation of
interaction. Epidemiology 1992;3:452-6.
11. Kline J, Stein Z. Spontaneous abortion. In: Bracken MB, ed.
Perinatal epidemiology. New York: Oxford University Press,
1984:23-51.
12. Anokute CC. Epidemiology of spontaneous abortions: the
effects of alcohol consumption and cigarette smoking. J Nat!
MedAssoc 1986;78:771-5.
13. Harlap S, Shiono PH. Alcohol, smoking, and incidence of
Am J Epidemiol Vol. 149, No. 3, 1999

Downloaded from http://aje.oxfordjournals.org/ by guest on May 2, 2015

TABLE 4. Proportion of spontaneous abortions (SABs) by

paternal smoking* among maternal nonsmokers, California,
1990-1991

Tobacco Smoke Exposure and Risk of Spontaneous Abortion

spontaneous abortions in the first and second trimester. Lancet
1980;2:173-6.
14. Guerin MR, Jenkins RA, Tomkins BA. The chemistry of environmental tobacco smoke: composition and measurement.
Boca Raton: Lewis Publishers, 1992.
15. Jarvis MJ, Russell MAH. Measurement and estimation of
smoke dosage to non-smokers from environmental tobacco
smoke. E a r i Respirf>isr (Suppr) 1984; 133:68-75.
16. Haddow JE, Knight GJ, Palomaki GE, et al. Second-trimester

247

serum cotinine levels in nonsmokers in relation to birth weight.

Am J Obstet Gynecol 1988;159:481-4.
17. Bums D, Pierce JP. Tobacco use in California 1990-1991.
Sacramento, CA: California Department of Health Services.
1992.
18. Office of Environmental Health Hazard Assessment Health
effects of exposure to environmental tobacco smoke.
Sacramento, CA: California Environmental Protection
Agency, 1997.

Downloaded from http://aje.oxfordjournals.org/ by guest on May 2, 2015

Am J Epidemiol

Vol. 149, No. 3, 1999