Oil companies have released billions of gallons of untreated
wastes and oil directly into the environment of the Ecuado-
rian Amazon. This cross-sectional study investigated the
environmental conditions and reproductive health of
women living in rural communities surrounded by oil fields
in the Amazon basin and in unexposed communities. Water
from local streams was analyzed for total petroleum hydro-
carbons (TPH). The women, aged 17 to 45 years, had
resided for at least three years in the study communities.
Socioeconomic and reproductive histories of the last three
pregnancies were obtained from interviews. Information
from the questionnaire was available for 365 exposed and
283 non-exposed women. The study was conducted from
November 1998 to April 1999. Streams of exposed commu-
nities had TPH concentrations above the allowable limit.
After adjustment for potential confounders, the pregnancies
of women in exposed communities were more likely to end
in spontaneous abortion (OR: 2.47; 95% CI: 1.61– 3.79; p <
0.01). No association was found between stillbirth and expo-
sure. An environmental system to control and eliminate the
sources of pollution in the area is needed. Key words: repro-
ductive; spontaneous abortion; oil; Amazon; Ecuador.
INT J OCCUP ENVIRON HEALTH 2002;8:312–319
he Amazonian tropical forests of Ecuador are
among the most biologically diverse natural
ecosystems on earth and home of many indige-
nous and peasant peoples. The Amazon is also home to
hundreds of oil fields, the most important source of
income for the country. Since 1972, foreign companies
together with Ecuador´s national oil company have
extracted more than two billion barrels of crude oil
from the Ecuadorian Amazon. The 1970s oil price
boom lifted Ecuador from being one of the poorest
countries in Latin America—per capita income rose
from $290 in 1972 to $1,490 in 1982, decreasing to
$1,390 in 1995. Today, oil continues to account for
nearly 50% of the nation’s export earnings and gov-
ernment budget. However, during this process, mil-
lions of gallons of untreated toxic wastes, gas, and oil
have been released into the environment.
Both peasants and indigenous people have reported
that many local streams and rivers, once rich in fish,
now support little or no aquatic life; cattle are reported
to be dying from drinking from contaminated streams
These are typically the same waters people
use for drinking, cooking, and bathing. Peasants have
reported that bathing in these waters causes skin
rashes, especially after heavy rains, which accelerate the
flow of wastes from nearby pits into the streams.
In 1993, a local community health workers’ associa-
tion conducted a descriptive study in their communi-
ties, including communities exposed to oil contamina-
tion and controls. The study suggested that exposed
communities had elevated morbidity and mortality
rates and increases in rates of spontaneous abortions.
In 1994, the New York City–based Center for Eco-
nomic and Social Rights released a report document-
ing high levels of toxic contamination and related
health problems in Ecuador’s Amazon. Concentrations
of polynuclear aromatic hydrocarbons were found in
drinking, bathing and fishing waters that were 10 to
10,000 times greater than those considered acceptable
according to the U.S. Environmental Protection
Several studies of animals have provided evidence
for an association between adverse reproductive out-
comes and exposures to oil pollutants. Crude oil
administered orally to pregnant rats decreased fetal
weight and length, and multiple exposures also caused
a significant reduction in maternal body weight.
Other studies have demonstrated pronounced effects
of crude oil on the reproductive capacities of birds
(deformed bills, incomplete ossification and feather
formation, dead embryos) after oral administration or
application on the shells of eggs surface.
Few epidemiologic studies have examined the associ-
ation between exposures to oil pollutants and outcomes
of pregnancy, particularly among women living close to
petrochemical industries. In one study conducted in
Outcomes of Pregnancy among Women
Living in the Proximity of Oil Fields in the
Amazon Basin of Ecuador
MIGUEL SAN SEBASTIÁN, PHD, BEN ARMSTRONG, PHD, CAROLYN STEPHENS, PHD
Received from the Instituto de Epidemiología y Salud Comuni-
taria “Manuel Amunárriz,” Quito Ecuador (MSS), and the Environ-
mental Epidemiology Unit, London School of Hygiene and Tropical
Medicine, London, U.K. (BA, CS).
Supported by the Spanish organization Medicus Mundi Gipuzkoa.
Address correspondence and reprint requests to Miguel San
Sebastián, PhD, Instituto de Epidemiología y Salud Comunitaria
“Manuel Amunárriz,” Apdo. 17-10-7410, Nicolás López 349 y Avenida
La Prensa, Quito, Ecuador; telephone: (02) 257689; fax: (02)
449763; e-mail: <firstname.lastname@example.org>.
Sweden, the miscarriage rate was slightly elevated in the
exposed area, although the study concluded that ambi-
ent community exposures were not associated with an
increased risk of unfavorable pregnancy outcome.
However, studies from Bulgaria have shown significantly
higher prevalences of toxemia, spontaneous abortion,
and prematurity among populations living in areas pol-
luted by petrochemical industries.
To our knowl-
edge, no study of reproductive outcomes in populations
living in close proximity of oil fields exists.
The study presented in this paper aimed to investi-
gate the environmental conditions of the area and
determine whether residence near an oil field was asso-
ciated with an increased risk of adverse reproductive
outcomes in peasants’ communities of the Amazon
basin of Ecuador. This research is part of a larger study
assessing the health impacts of oil pollution in rural
communities of Ecuador.
Study Area and Population
The study was carried out in communities of peasants sit-
uated in the Orellana and Sachas districts of the Orellana
province, and in the Shushufindi district of the Sucum-
bíos province, both in the northeastern part of Ecuador.
This area was chosen because of local concern and the
long term and high density of oil-drilling activities.
Peasants are organized in small communities, where
each peasant owns 50 hectares of land. The total popu-
lation of the area is approximately 50,000.
people live in close proximity to oil-production facili-
ties. Most communities in the area lack electricity and-
piped water supplies and have difficulty accessing
Selection of Communities
Two groups of communities were selected for the study:
communities based in areas with potential exposures to
toxic contaminants from oil fields and unexposed com-
munities selected as controls. The people living in the
control communities had sociodemographic and geo-
graphic characteristics similar to those of the people in
the exposed communities.
An exposed community was defined as a community
within 5 km of an oil field, following a downstream
direction. A non-exposed community was defined as a
community at least 30 km upstream from any oil field.
All studied communities were considerable distances
away from other chemical industries.
The target population was women aged 17 to 45
years resident for periods of at least three years in the
study communities. This population was selected
because: 1) they were at reproductive age; younger and
older women were excluded because spontaneous
abortions tend to be less common at those ages
they were easier to contact because of the local work
activities; and 3) the criterion of living at least three
years in the same community was chosen as a proxy
measure for long-term exposure.
The sampling method used was a two-stage sampling
procedure. Initially, a list of communities was prepared,
stratified into exposed and non-exposed. Nine exposed
and 14 unexposed communities were selected ran-
domly and all women aged 17 to 45 years old who had
lived for at least three years in the selected communi-
ties were included.
Community leaders were used to identify women who
met the eligibility criteria for age and length of residence
and asked these women to participate in the study.
The sample size calculation was based on the
number of reproductive outcomes required to detect a
double difference in spontaneous abortions between
the exposed and unexposed groups at the 5% level with
80% power. It was calculated using the baseline preva-
lence of spontaneous abortion of 10% as reported for
females living in the Orient.
The sample size was doubled to adjust for the cluster
nature of the sampling. The calculations were based on
the Epi-Info 6 program. This gave a requirement of 438
pregnancy outcomes per group. To allow for 70% par-
ticipation, the sample size was increased to 625 for each
group to obtain a more realistic estimate. To reach that
number, the history of the three most recent pregnan-
cies was collected.
To assess whether the communities surrounding oil
fields were exposed to pollution, samples of water from
the places used by the communities to obtain water for
drinking, bathing, and washing clothes were collected.
The water analysis included determination of total
petroleum hydrocarbons (TPH) and was carried out by
the Water and Soil Laboratory of the P. Miguel Gamboa
Technical School, Coca. The method to measure TPH
comprised extraction with 1,1,2-trichlor-trifluorethane
and determination by infrared spectrophotometry
(limit of detection 0.001). Special bottles for water sam-
ples taken were provided by the laboratory. Laboratory
technicians were kept blind to the water origins.
Because of economic and technical limitations, it
was not possible to measure the levels of land and air
During the months of February to April 1999, samples
of different rivers used by the exposed and non-exposed
communities were collected. The samples were taken in
the winter season without visible crude oil presence in
the rivers. Twenty streams from the nine exposed com-
munities and two streams from two non-exposed com-
munities were investigated. Existing data from water
analysis reports from the areas were also reviewed.
VOL 8/NO 4, OCT/DEC 2002 Pregnancy Outcomes Near Oil Fields • 313
The field work was conducted between November 1998
and April 1999. A structured questionnaire was devel-
oped for administration to the female head of the
The questionnaire comprised two parts. The first
part elicited sociodemographic details (age, ethnic
group, length of residence, marital status) and socioe-
conomic characteristics (educational level, female
occupation, husband’s occupation, and living condi-
tions). The second part elicited information about the
reproductive histories of the women during residence
in the community. Information was obtained about the
number of pregnancies to the time of interview, the
outcomes (live births, spontaneous abortions, and still-
births) of the three most recent pregnancies, the date
and gestational age at the end of each pregnancy, and
whether liveborn children were still alive. The three
most recent pregnancies were selected to maximize
recall of pregnancy and exposure information. Smok-
ing and alcohol use habits and medications taken
during pregnancy were also investigated. This part of
the questionnaire was adapted from Doyle et al.
The questionnaire, in Spanish, took approximately
30 minutes to administer. The questionnaire was piloted
in one community of the area. The study was presented
as a health status survey. The women selected for the
study were asked to participate in a personal interview.
A date for the interview was arranged, and the women
agreed to be interviewed at a central location in the
community. Respondents were interviewed in a private
room of the school or the community center. Informed
consent was obtained from all study participants.
Confidentiality of all information collected was
maintained, and sick persons received free medical
attention and treatment. At the completion of the
investigation, the communities were informed of the
preliminary results of the study.
Definition of Reproductive Outcomes
Pregnancy outcomes were recorded according to the
following definitions: 1) pregnancy was defined as the
delayed-“period” perception by the subject more than
three months from the last menstrual period; 2) spon-
taneous abortion was defined as fetal loss at 28 weeks’
gestation or earlier; 3) stillbirth was defined as a fetal
loss after 28 weeks of gestation, but without any evi-
dence of life at birth, and 4) a full-term baby was one
born alive after the 36th week of gestation.
Only pregnancies that ended before December 31,
1998, and occurred during the residence of the women
in the communities were included.
Neonatal deaths were differentiated from stillbirths
by reports of respiratory efforts or crying by the infant
Only self-reported miscarriages were considered in
the study due to the lack of hospital records or medical
Elective abortions, multiple pregnancies, and preg-
nancies of women having used an intrauterine device at
the time of conception were excluded from analyses
because of the high rate of spontaneous abortions
among women who become pregnant while using the
latter method of contraception.
The statistical analysis treated each pregnancy as one
unit. Multiple pregnancies of the same woman were
treated as independent observations, even though this
is not strictly true. Statistical analyses were also carried
out separately for each of the three most recent differ-
ent pregnancies to take into account the variability in
the risk of spontaneous abortion between women.
Prevalence rates for the three most recent preg-
nancy outcomes occurring in the study communities
were compared between women living in contaminated
and noncontaminated areas. Odds ratios (ORs) for the
pregnancy outcomes in the exposed group were calcu-
lated with 95% confidence intervals (CIs) and p-values.
Potential confounders included those that had been
identified in international studies as confound-
Thus, age at interview, age at pregnancy,
pregnancy order, year of pregnancy, socioeconomic
status (level of education, mother’s and father’s occu-
pations, living conditions) were used as potential con-
founders. Multiple logistic regression was used to esti-
mate ORs adjusted for several potential confounders
simultaneously. In the analytic process, standard errors
were adjusted for the clustered nature of the sampling
using the Huber–White method.
History of previous spontaneous abortion was exam-
ined but was not included in the statistical model, as
these losses might have been caused in part by the
exposure to oil pollutants and might be correlated with
the index outcome under study and thus have resulted
in biased risk estimates.
The results of the water analysis in the exposed com-
munities are presented in Table 1. No TPH contami-
nation was found in rivers close to non-exposed com-
munities, in two samples taken. In the “exposed”
area, 18 streams close to eight communities were
contaminated with TPH, ranging from a concentra-
tion of 0.02 parts per million (ppm) in the Manduro
1 stream to 2.883 ppm in the Basura river. No con-
tamination was found in two streams in the other
314 • San Sebastián et al. INT J OCCUP ENVIRON HEALTH
Review of existing data collected independently by
Zehner et al.
in 1998 from 46 streams showed con-
tamination in those located in areas of oil activities,
while in areas without such activities no water contami-
nation by TPH was found.
Characteristics of the Population
Nine communities in the exposed area (out of 87 com-
munities) and 14 in the control area (out of 125) were
included in the study. Of 610 exposed and 439 non-
exposed women identified as within the age range, 428
(70.2%) and 347 (79.0%), respectively, were inter-
viewed. Of these, 60 women (14.0%) from the exposed
area were subsequently excluded because of living less
than three years in the communities; 56 (16.1%) from
the non-exposed area were excluded for the same
reason. Finally, three women in the exposed group and
eight in the non-exposed group were excluded because
their forms were incomplete or unreadable. Informa-
tion from the questionnaire was therefore available for
365 (59.8%) and 283 (64.4%) of the potential partici-
Those exposed showed little difference in lengths of
residence, ags, ethnicity, marital status, or educational
levels from the controls (Table 2). However, the
exposed women worked less in agriculture (72.6%)
than did the non-exposed women (86.3%). The
women’s husbands in the exposed communities tended
to work for oil companies more (7.8%) than did the
husbands in the control group (1.3%).
Living conditions were assessed through three
indicators: type of house, possession of refrigerator,
and availability of latrine. Living conditions, as meas-
ured by these variables, were better in the exposed
communities than in the control communities. None
of the women was classified as a cigarette smoker or
The exposed and non-exposed communities showed
differences in sources of water for drinking, bathing,
and washing. Women from the exposed communities
were less likely to use water from the rivers (Table 2).
Over all, 555 women (85.6%) reported having at least
one pregnancy, with little difference between groups.
Of the women reporting at least one pregnancy, 508
(78.3%) had had at least one liveborn child and 111
(17.1%) a fetal loss (spontaneous abortion or stillbirth).
Table 3 shows details of the individual pregnancies
(including the three most recent reported) according
to exposures of the mothers. The total number of preg-
nancies reported was 1,377. Of these pregnancies,
7.5% ended as spontaneous abortions and 1.8% ended
Pregnancies of women living in exposed communi-
ties were more likely to end in spontaneous abortion
than were those of women living in comparison com-
munities (OR: 2.34; 95% CI: 1.48–3.71; p < 0.01). No
association was found between stillbirth and exposure
(OR: 0.85; 95% CI: 0.35–2.05; p = 0.83).
Logistic regression analysis was used to examine the
combined effects of the potential confounding factors
and exposure on spontaneous abortion. After adjust-
ment, the estimated OR was slightly higher than the
crude value and the association between spontaneous
abortion and living in the proximity of oil fields
remained highly significant (OR: 2.47; 95% CI: 1.61–
3.79; p < 0.01).
These results were also observed when the analysis
was stratified by pregnancy number (numbering from
the last) (Table 4). Higher risks of spontaneous abor-
tion were found in the three different groups; the risks
in the first two pregnancies were statistically significant.
No evidence of interaction between exposures and
the investigated potential confounders was found with
respect to their effects on spontaneous abortion.
VOL 8/NO 4, OCT/DEC 2002 Pregnancy Outcomes Near Oil Fields • 315
TABLE 1. Concentrations of Total Petroleum
Hydrocarbon (TPH)* in the Streams of Communities
Surrounding Oil Fields, Ecuador 1999
Identification (Stream) TPH (ppm)
Escuela 28-M 0
Pozo 66 0.04
Río Negro 1.438
Victoria 1 0.051
Victoria 2 1.426
Itaya 1 0.043
Itaya 2 0.028
Escuela 18-N 0.036
Lumu pueblo 0.066
Lumu 3 0.055
Manduro 1 0.02
Pisc Manduro 0.434
Manduro 2 0.108
*The permitted limit for hydrocarbons in drinking water
according to the European Community laws is 0.01 parts per
316 • San Sebastián et al. INT J OCCUP ENVIRON HEALTH
Analysis of the river water showed heavy exposure to oil
chemicals among the residents of the exposed commu-
nities. In some streams, hydrocarbon concentrations
reached 144 and 288 times the limit permitted by the
European Community regulation.
These data suggest
that residents of communities close to oil fields are
exposed to pollutant levels originating from oil-related
activities that significantly exceed the internationally
recognized safety limits. Though the initial time of
such exposures is not known, numerous reports have
indicated they may date from the beginning of the oil
exploration in the area in the 1970s.
The lesser use of river water in the exposed commu-
nities suggests that women who live in these communi-
ties are aware of its contamination and try to use other
sources of water; however, this is not always possible.
The study revealed a risk for spontaneous abortion
2.34 times higher among women living in communities
exposed to oil pollutants. After adjusting for the differ-
ent confounders, the significant difference remained
(OR: 2.47; 95% CI: 1.61–3.79). No association was
observed for stillbirth.
TABLE 2. Sociodemographic Characteristics of the Study Population
Exposed Group (%) Comparison Group (%)
(n= 365) (n= 283)
17-20 44 (12.0%) 43 (15.1%)
21-30 146 (40.0%) 105 (37.1%)
31-40 119 (32.6%) 89 (31.4%)
41-45 56 (15.3%) 46 (16.2%)
Mean age (SD) 30.6 years (7.9) 30.7 years (8.3)
Time of residence (years)
4–10 116 (31.7%) 108 (38.1%)
11–20 186 (50.9%) 142 (50.1%)
> 20 63 (17.2%) 33 (11.6%)
Mean residence (SD) 14.4 years (6.2) 13.5 years (6.1)
Mestizo 357 (97.8%) 276 (97.5%)
Black 6 (1.6%) 2 (0.7%)
Indigenous 2 (0.5%) 5 (1.8%)
Single 38 (10.4%) 30 (10.6%)
Married 305 (83.6%) 237 (83.7%)
Widowed 22 (6.0%) 16 (5.7%)
None 17 (4.7%) 8 (2.8%)
Primary non-finished 73 (20.0%) 67 (23.7%)
Primary 215 (58.9%) 173 (60.1%)
Secondary non-finished 41 (11.2%) 28 (9.9%)
Secondary 19 (5.2%) 7 (2.5%)
Persons at home, mean (SD) 6.3 (4.5) 6.2 (2.4)
Agriculture 264 (72.3%) 244 (86.2%)
Other 101 (27.6%) 39 (13.8%)
Agriculture 241 (75.1%) 216 (88.5%)
Oil company 29 (9.0%) 3 (1.2%)
Palm company 8 (2.5%) 0 (0%)
Other 43 (13.4%) 25 (10.3%)
Cement house 49 (13.4%) 20 (7.0%)
Refrigerator 126 (34.5%) 39 (13.7%)
Latrine 177 (48.4%) 111 (39.2%)
Drink from river 27 (7.3%) 70 (24.0%)
Bath in river 103 (28.0%) 162 (55.6%)
Wash in river 132 (35.9%) 191 (65.6%)
VOL 8/NO 4, OCT/DEC 2002 Pregnancy Outcomes Near Oil Fields • 317
These findings are consistent with earlier reports
from the area suggesting an increased risk of sponta-
neous abortions in women living in communities sur-
rounded by oil fields.
Corresponding studies of residents near oil fields
are rare, and have concentrated on industrialized
countries. More problematic, existing studies tend to
be based on lower levels of exposures than those in
Ecuador. In Sweden, a study concluded that the expo-
sure levels near a petrochemical industry were not asso-
ciated with an increased risk of unfavorable pregnancy
However, in Bulgaria an investigation of
the association between exposures to emissions from
petrochemical industries and outcomes of pregnancy
showed a higher prevalence of spontaneous abortions
among residents near the industries.
risk of spontaneous abortion for women workers with
frequent exposure to petrochemicals compared with
those working in non-chemical-related plants was also
found in China.
In addition, studies of animals sup-
port the evidence for a high risk of adverse reproduc-
tive outcomes when exposed to oil pollutants.
No association was found between stillbirth and living
in the proximity of oil fields. Due to its low frequency, still-
birth has been considered a weak indicator of develop-
mental toxicity in relation to environmental chemicals.
Limitations of the study design and the methods of
data collection, due to logistic and economic short-
comings, create some potential for biases in this study.
The similarity of the sociodemographic variables
among the study groups indicates that non-exposed
communities were an adequate reference population.
Response rates were quite high were and similar in the
exposed and control areas (70.2% vs 79.0%), limiting
the potential for non-response bias. Reasons for non-
participation were unknown.
The study was community-based, and cases were
selected from communities randomly chosen. Only
current residents of the target or comparison areas
were eligible, and migration might have been consid-
erable. However, there seems no reason for migration
to be related to both pregnancy outcome and living in
an exposed area, so it should not cause bias in the asso-
ciation between exposure and outcome.
One important threat to the validity of the findings
in this type of study comes from reporting (recall) bias,
since recall is likely to be incomplete, and people who
believe they are exposed might be more likely to recall
We tried to limit this problem
by presenting the study to the communities as a general
health study within a primary health care program. In
addition, the women in the exposed communities were
not aware of spontaneous abortion as an outcome
related to oil pollution.
However, the overall propor-
tions of pregnancies reported as ending in sponta-
neous abortions in the unexposed communities were
lower than those in other similar surveys in developing
countries. Percentages of self-reported pregnancies ter-
minating in miscarriage have ranged from 6.3 in Peru
to 9.1 in Colombia and Venezuela.
It has also been
reported that when the event is ascertained retrospec-
tively by means of a questionnaire, the rates of sponta-
neous abortion are between 5% and 10%.
study, the rate in the unexposed population was 4.4%,
suggesting a true low risk or underreporting.
Recall might also be expected to increase with prox-
imity of the pregnancy to the date the questionnaire
was administered. The presence of the association with
exposure after stratifying by pregnancy order, number-
ing from the last, suggests that it is not due to this bias.
The study design did not allow us to address certain
other questions. First, the validity of the reported spon-
TABLE 3. Outcomes of Pregnancies by Exposure Status
Exposed Unexposed Total
_________________ _________________ _________________
No. (%) No. (%) No. (%)
Total reported pregnancies 791 (100) 586 (100) 1,377 (100)
Live births 700 (88.4) 548 (93.5) 1,248 (90.6)
Spontaneous abortions (< 28 weeks) 78 (9.8) 26 (4.4) 104 (7.5)
Stillbirths (≥ 28 weeks) 13 (1.6) 12 (2.0) 25 (1.8)
TABLE 4. Risk of Spontaneous Abortion in the Last Three Pregnancies by Exposure Status*
_______________ _______________ Crude Adjusted
No. (%) No. (%) OR (95% CI) OR† (95% CI)
Last pregnancy (first) 24 (7.9) 12 (4.9) 1.64 (0.80–3.36) 1.62 (0.70–3.75)
Previous one (second) 24 (8.9) 7 (3.5) 2.65 (1.11–6.32) 2.76 (1.03–7.39)
Previous one (third) 30 (14.4) 7 (5.0) 3.15 (1.33–7.48) 3.66 (0.97–13.73)
All three 78 (9.8) 26 (4.4) 2.34 (1.48–3.71) 2.47 (1.61–3.79)
*Standard errors adjusted for clustering.
†Adjusted for age at interview, age at pregnancy, pregnancy order, year of pregnancy, educational level, woman’s and her hus-
band’s occupations, and living conditions.
taneous abortion was not possible to address due to the
lack of medical records.
Several studies from industrialised countries have
reported that such a problem does not necessarily
cause a serious distortion.
In a study among labora-
tory workers in Sweden, the accuracy of reporting of
miscarriages was high.
Ninety-four percent of self-
reported spontaneous abortions could be confirmed by
reviewing medical records in a study conducted among
workers at two semiconductor manufacturing plants in
the United States.
Self-report of fetal loss was also reli-
able in a study conducted on dry cleaning workers in
Second, we could not assess whether early
(subclinical) fetal loss might be affected by oil pollu-
tants. There is a high probability for women not to rec-
ognize the event as such but to perceive it as a delayed
menstrual period if it occurs very early in pregnancy.
Accurate exposure assessment is always a major con-
cern in epidemiologic studies, especially when the rel-
evant exposure occurred in the past. No data on how
people have been affected by past exposure to the
chemicals in the area exist. In addition, there is little or
no information about chronic toxicity from the diverse
chemical substances spilled from the oil fields. Even
the exact nature of the chemical substances spilled by
oil companies in the Amazon basin of Ecuador is
In our study, the same exposure status was
assigned to every individual within the same study area,
even though the individuals certainly did not all have
the same level of exposure to oil pollutants.
None of the potential confounders examined—age
at interview, age at pregnancy, pregnancy order, year of
pregnancy, socioeconomic status—could explain the
association between spontaneous abortion and living in
the proximity of oil fields. However, some residual con-
founding may remain due to misclassification of some
factor in the analysis.
Repeated spontaneous abortions in the same women
are treated as independent events in our analyses, which
can lead to spuriously narrow confidence intervals.
However, the allowance for clustering by community
allows implicitly for this source of additional variation,
so confidence intervals should not be misleading. The
persistence of the association of spontaneous abortions
with exposure after stratifying by pregnancy number
(from the last), statistically significant in two of three
strata, gives further reassurance on this point.
This study has demonstrated the presence of con-
tamination by oil pollutants in communities close to oil
fields, at levels high enough to cause alarm. It also pro-
vides some evidence of an increased risk of sponta-
neous abortions in women living in the proximity of
the oil fields, after adjustment for other better-known
risks common in developing-country settings. Further
research is necessary to confirm these results in other
communities experiencing similar exposures. There is
also a need for studies to contribute to clearer under-
standing of the overall implications of this form of
development for local health, particularly for women.
The oil industry argues that it has a role to play in
development, but it should not be at the expense of
unnecessary contamination, exposures, or health
impacts. We concur with the community’s wish to
address urgently environmental control and remedia-
tion of contamination in the exposed areas.
The authors thank the participants for their collaboration and Sandi
Yura and FUSA for logistic support. Special thanks to Dr. Pat Doyle
and Prof. Lowell Sever for their valuable comments on an early
1. Kimerling J. Amazon Crude. New York: Natural Resources
Defense Council, 1991.
2. Jochnick C, Normand R, Zaidi S. Rights violations in the
Ecuadorian Amazon: the human consequences of oil develop-
ment. Health Hum Rights. 1994;1:82-100.
3. Kimerling J. Rights, responsibilities, and realities: environmen-
tal protection law in Ecuador’s Amazon oil fields. Southwestern
J Law Trade Americas. 1995;2:293-384.
4. Unión de Promotores Populares de Salud de la Amazonía Ecu-
atoriana. Culturas bañadas en petróleo: diagnóstico de salud
realizado por promotores. Quito, Ecuador: Abya-Yala, 1993.
5. Centro de Derechos Económicos y Sociales. Violaciones de
derechos en la Amazonía Ecuatoriana. Quito, Ecuador: Abya-
6. Scheiner CA. Petroleum and petroleum products: a brief review
of studies to evaluate reproductive effects. In: Christian MS, Gal-
braith WM, Voytek P, Mehlman MA (eds). Advances in Modern
Environmental Toxicology. Vol. III, Assessment of Reproductive
and Teratogenic Hazards. Princeton, NJ: Princeton Scientific
7. Khan S, Irfan M, Rahimtula AD. The hepatotoxic potential of a
Prudhoe Bay crude oil: effect on mouse liver weight and com-
position. Toxicology. 1987;46:95-105.
8. Hoffman DJ. Embryotoxic and teratogenic effects of petroleum
hydrocarbons in mallards (Anas platyrhynchos). J Toxicol Envi-
ron Health. 1979;5:835-44.
9. Lee Y, O’Brien PJ, Payne JF, Rahimtula AD. Toxicity of petro-
leum crude oils and their effect on xenobiotic metabolizing
enzyme activities in the chicken embryo in ovo. Environ Res.
10. Walters P, Khan S, O’Brien PJ, Payne JF, Rahimtula AD. Effec-
tiveness of a Prudhoe Bay crude oil and its aliphatic, aromatic
and heterocyclic fractions in inducing mortality and aryl hydro-
carbon hydroxylase in chick embryo in ovo. Arch Toxicol.
11. Axelsson G, Molin I. Outcome of pregnancy among women
living near petrochemical industries in Sweden. Int J Epidemiol
12. Tabacova S, Vukow M. Developmental effects of environmental
pollutants. In: East–West European Initiative for Research in
Reproductive Health. Special Program for Research, Develop-
ment and Research Training in Human Reproduction. Geneva,
Switzerland: World Health Organization, 1991:37-8.
13. Tabacova S, Balabaeva L. Environmental pollutants in relation
to complications of pregnancy. Environ Health Perspect.
14. San Sebastián M. Informe Yana Curi: impacto de la actividad
petrolera en la salud de poblaciones rurales de la Amazonía
ecuatoriana. Quito, Ecuador: Abya-Yala, 2000.
15. Instituto Nacional de Estadísticas y Censos. Cifrando y
descifrando Napo. Quito, Ecuador: Instituto Nacional de
Estadísticas y Censos, 1995.
16. Cabodevilla MA. Coca: la región y sus historias. Pompeya,
Ecuador: Cicame, 1997.
17. Kline J, Stein Z, Susser M. Conception to Birth: Epidemiology of
318 • San Sebastián et al. INT J OCCUP ENVIRON HEALTH
VOL 8/NO 4, OCT/DEC 2002 Pregnancy Outcomes Near Oil Fields • 319
Prenatal Development. New York: Oxford University Press,
18. Doyle P, Roman E, Beral V, Brookes M. Spontaneous abortion in
dry cleaning workers potentially exposed to perchloroethylene.
Occup Environ Med. 1997;54:848-53.
19. Schnorr TM, Grajewski BA, Hornung RW, et al. Video display
terminals and the risk of spontaneous abortion. N Engl J Med.
20. Rothman KJ, Greenland S. Modern epidemiology. Philadelphia,
PA: Lippincott–Raven, 1998.
21. Restrepo M, Muñoz N, Day NE, Parra JE, de Romero L, Nguyen-
Dinh X. Prevalence of adverse reproductive outcomes in a pop-
ulation occupationally exposed to pesticides in Colombia.
Scand J Work Environ Health. 1990;16:232-8.
22. Correa A, Gray RH, Cohen R, et al. Ethylene glycol ethers and
risks of spontaneous abortion and subfertility. Am J Epidemiol.
23. Stata Corporation. Stata Reference Manual. Texas: Stata Press,
24. Weinberg CR. Toward a clearer definition of confounding. Am
J Epidemiol. 1993; 137:1-8.
25. Arbuckle TE, Savitz DA, Mery LS, Curtis KM. Exposure to phe-
noxy herbicides and the risk of spontaneous abortion. Epi-
26. Zehner R, Villacreces LA. Estudio de la calidad de aguas de río
en la zona de amortiguamiento del Parque Nacional Yasuní.
Primera fase: monitoreo de aguas. Coca, Ecuador: Laboratorio
de Aguas y Suelos P. Miguel Gamboa-Fepp, 1998.
27. Dirección General de Medioambiente. Estudio de impacto
ambiental 42. Quito, Ecuador: Dirección General de Medioam-
28. Xu X, Cho SI, Sammel M, et al. Association of petrochemical
exposure with spontaneous abortion. Occup Environ Med.
29. World Health Organization. Principles for Evaluating Health
Risks to Progeny Associated with Exposure to Chemicals during
Pregnancy. Geneva, Switzerland: World Health Organization,
30. Wilcox AJ, Horney LF. Accuracy of spontaneous abortion recall.
Am J Epidemiol. 1984;120:727-33.
31. San Sebastián M. Health perceptions of women living in the
proximity of oil fields. London, U.K.: London School of
Hygiene and Tropical Medicine, 1999 [unpublished].
32. Hobcroft J. The proximate determinants of fertility. In: Cleland
J, Scott C (eds). The World Fertility Survey: An Assessment. New
York: Oxford University Press, 1987.
33. Hemminki K, Niemi ML, Saloniemi I, Vainio H, Hemminki E.
Spontaneous abortions by occupation and social class in Fin-
land. Int J Epidemiol. 1980;9:149-53.
34. Hewson D, Bennet A. Childbirth research data: medical records
or women’s reports? Am J Epidemiol. 1987;125:484-91.
35. Axelsson G, Rylander R. Outcome of pregnancy in women
engaged in laboratory work at a petrochemical plant. Am J Ind
36. Wilcox AJ, Weinberg CR, O’Connor JF, et al. Incidence of early
loss of pregnancy. N Engl J Med. 1988;319:189-94.
European Research Conferences sponsored by the
Programme of European Research Conferences
(EURESCO) are run by the European Science Foun-
dation. Each conference consists of a series of meet-
ings, held typically every other year. Conferences on
the topics of Biomedicine and Health, Geosciences and
Environment, and Social Sciences may be of particular
interest to occupational and environmental scientists.
Deadline for applications is generally three months
before a conference. For further information, contact:
J. Hendekovic, EURESCO, 1 quai Lezay-Marnesia,
67080 Strasbourg Cedex, France; tel: 33 388 76 71 35;
fax: 33 388 36 69 87; e-mail: <euresco@ esf.org> or Web
The Nordic Institute for Advanced Training in Occu-
pational Health has announced the 2000 Advanced
Courses and Symposia in Occupational Health and
Safety. To order a booklet containing details of a broad
array of courses and topics, contact: Peter Lundqvist,
Course Director, NIVA, Topeliuksenkatu 41 aA, FIN-
00250 Helsinki, Finland; tel: 358 9 47471; fax: 358 9
4747 497; e-mail: <email@example.com>; Website:
National School of Occupational Health, University of
Utrecht, offers International Short Courses covering
topics such as Biological Agents in the Workplace, Ven-
tilation Design, Occupational and Environmental Epi-
demiology, and Principles of Exposure Assessment. For
further information, contact: Division Environmental
and Occupational Health, Utrecht University, P.O. Box
80176, 3508 TD Utrecht, The Netherlands; tel: +31
317482080; fax: +31 317485278; e-mail: <M.Lumens@
University of California, San Francisco, Advances in
Occupational and Environmental Medicine is an inten-
sive review of the specialty of occupational and envi-
ronmental medicine. The next course will be held
April 28–May 2, 2003. For course information, contact:
UCSF CME Office, Box 0656, San Francisco, CA 94143-
0656; tel: (415) 476-5208; fax: (415) 476-6426; e-mail:
<firstname.lastname@example.org>. Scholarships are available
to occupational physicians from developing countries.
The next program will be held April 28 to May 2, 2003.
For further information on scholarships, contact: Joe
LaDou, MD, 563 Mountain Laurel Drive, Aspen, CO
81611; tel: (970) 920-3585; fax: (970) 920-4645; e-mail:
What in the World, continued from page 394