American Journal of Epidemiology
Copyright © 2003 by the Johns Hopkins Bloomberg School of Public Health
All rights reserved
Vol. 158, No. 2
Printed in U.S.A.
DOI: 10.1093/aje/kwg122
Recreational Physical Activity and Endometrioma Risk
Preet K. Dhillon1,2 and Victoria L. Holt1,2
1
2
Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA.
Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA.
Received for publication August 16, 2002; accepted for publication January 22, 2003.
Endometriosis, defined by implantation of endometrial tissue outside the uterus, is commonly found on the
ovaries and is characterized by a range of symptoms and severity, including chronic pelvic pain and infertility.
The role of modifiable exogenous factors such as exercise in the development of this disease is not well
understood. The authors conducted a 1990–1994 case-control study (77 cases, 735 controls) to evaluate the risk
of cystic ovarian endometriosis (endometrioma) associated with recreational physical activity among females
aged 18–39 years enrolled in the Group Health Cooperative of Puget Sound, Washington. Physical activity was
evaluated according to its duration, frequency, and intensity during two periods of exposure: 2 years prior to the
reference date and at ages 12–21 years. Women who reported frequent, high-intensity activity during the 2 years
prior to the reference date had a 76% reduced endometrioma risk (odds ratio = 0.24, 95% confidence interval:
0.08, 0.71) compared with women who engaged in no high-intensity activity. A nonsignificant 53% reduction in
risk was observed for women who reported such activity at ages 12–21 years (odds ratio = 0.47, 95% confidence
interval: 0.18, 2.23). Activities of lower intensity, frequency, and duration were not associated with a woman’s risk
of endometrioma.
case-control studies; endometriosis; exercise
Abbreviations: CI, confidence interval; MET, metabolic equivalent; OR, odds ratio.
Endometriosis is a gynecologic disease primarily affecting
women of reproductive age that is marked by the presence of
functioning endometrial glands outside the uterus, usually
within the peritoneal cavity (1, 2). While the disease may
exist asymptomatically in many women, it is quite debilitating in others; symptoms include heavy menstrual flow,
severe menstrual cramps, chronic pelvic pain, dyspareunia,
and infertility (2). An analysis of the US National Health
Interview Survey in 1984–1992 revealed that over half of the
women who reported an endometriosis diagnosis in the past
year experienced at least 1 day of bed rest (mean, 17.8
affected days) (3). The spectrum of this disease ranges from
superficial implants to deep infiltrating lesions in areas
including the peritoneal surface, the ovaries, and the
rectovaginal septum. Different lesion locations may denote
differing pathogenesis (4), and cystic ovarian endometriosis
(endometrioma) is generally associated with more severe
disease (5, 6). The specific associations of various risk
factors for endometriosis therefore may vary by the location
of lesions.
The etiology and pathogenesis of endometriosis are not
well understood, and prevalence estimates vary according to
the study population, underlying the difficulty in determining risk factors for the disease. Nonetheless, the role of
endogenous characteristics in the growth and maintenance of
endometriosis has been widely reported (7). These characteristics include factors related to menstruation, such as early
age at menarche (8–14) and shorter cycle length (8, 9, 11–
16), and hormonal factors such as circulating estrogen (17).
Environmental and behavioral factors that can potentially
influence endogenous factors are of particular interest in
preventing and controlling this disease. Regular exercise,
which is associated with reduced cumulative exposure to
menstrual flow, decreased ovarian stimulation, and estradiol
production (18), is one such factor.
Prior studies of the association between endometriosis and
physical activity did not distinguish between different forms
Correspondence to Dr. Victoria L. Holt, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North (MP-474), P.O. Box 19024,
Seattle, WA 98109-1024 (e-mail: [email protected]).
156
Am J Epidemiol 2003;158:156–164
Physical Activity and Endometrioma Risk 157
of the disease (superficial vs. deep infiltrating lesions, site of
disease, etc.) and yielded mixed results (8, 13, 14). A population-based, cross-sectional survey in a Norwegian county
found no association (13), whereas two case-control studies
in the United States reported a decreased endometriosis risk
associated with strenuous exercise for at least 3 hours a week
(8, 14). The previous studies were based on self-reported or
infertile cases of all types of endometriosis combined and
considered only current, regular, and strenuous exercise. Our
study focused on one specific type of significant endometriotic disease (endometriomas), captured all newly diagnosed
medically confirmed cases within a defined population, and
examined a broad range of physical activity over two time
periods of possible etiologic relevance as well as varying
levels of intensity, duration, and frequency. In this study, we
evaluated the risk of endometriomas associated with physical activity, both in the recent past and during adolescence
and early adulthood.
MATERIALS AND METHODS
Ascertainment of cases and controls
We conducted a case-control study of endometrioma risk
associated with physical activity as part of a larger investigation of ovarian cysts and benign neoplasms in female
enrollees of the Group Health Cooperative of Puget Sound, a
mixed-model health maintenance organization in Washington State. In the parent study, potential cases were
defined as women aged 18–39 years who had been enrolled
in the health cooperative for at least 6 months and were
newly diagnosed with a functional cyst or benign neoplasm
between January 1, 1990, and June 30, 1994, by physical,
ultrasound, or surgical examination. Controls were
frequency matched to functional cyst cases on age and
approximate date of enrollment in the Group Health Cooperative. Reference dates for cases were based on their diagnosis dates, and reference dates for controls were assigned to
correspond to these dates. Women without at least one ovary
or who had not received any care at the health cooperative in
the past 5 years were excluded from the control group.
Women who were non-English-speaking, pregnant, postmenopausal, or diagnosed with polycystic ovaries were also
excluded.
We obtained the participation of 83.9 percent of eligible
benign neoplasm cases (n = 162) and 82.2 percent of eligible
controls (n = 755) through written informed consent. The
current analysis was based on the subset of benign neoplasm
cases with cystic ovarian endometriosis, which consisted of
77 women with a surgically confirmed endometrioma diagnosis (International Classification of Diseases, Ninth Revision, code 617.1) and 735 controls without a history of
endometriosis. The study was approved by the Group Health
Cooperative and the Fred Hutchinson Cancer Research
Center human subjects review committees.
Physical activity
In-person interviews were conducted to collect information on demographic characteristics as well as medical,
Am J Epidemiol 2003;158:156–164
menstrual, reproductive, contraceptive, and physical activity
histories prior to the reference date. Women were asked to
report their physical activity during the 2 years prior to the
reference date and between the ages of 12 and 21 years
according to the following question: “During this time, did
you do any strenuous physical activities, exercise or sports
on a regular basis, that is, at least 24 times a year? Include
walking for pleasure to and from work/school, if at least 1
mile [1.61 km]. Include indoor and outdoor sports.” The
questions were based in part on the Minnesota Leisure Time
Physical Activity Questionnaire, which has been widely
used and evaluated in the exercise science and public health
fields (19, 20). Women reported up to four activities; for
each, the ages at which they started and stopped, the time
spent per episode, the frequency, and the number of months
per year that they engaged in the activity were recorded.
To account for varying levels of energy expenditure, a
metabolic equivalent (MET) was assigned to each activity. A
MET is the ratio of the metabolic rate for a specific activity
to a standard resting metabolic rate of 1.0 kcal per kilogram
of body weight per hour; one MET is defined as the rate
during quiet sitting. Therefore, a 3.0-MET activity would
require three times the metabolic energy expenditure of
sitting quietly. Intensity codes (in METs) are available in the
updated version of the Compendium of Physical Activities
(21), which includes 605 specific activities with intensity
codes ranging from 1.0 (sitting) to 15.0 (speed skating). In
the present study, all activities were categorized as low,
medium, or high intensity. On the basis of previous studies
(20, 22, 23) and on data indicating that conditioning or heavy
activities require more than 50 percent of the average
maximal oxygen intake (19), we defined high-intensity
activities as those with intensity codes of ≥6.0 METs, such as
running, bicycling, and tennis. Activities of moderate intensity (4.0–5.9 METs) consisted of those such as gardening,
dancing, and softball; low/light activities (<4.0 METs)
included golf, bowling, and walking for pleasure. For a
complete list of activities and intensity codes, refer to the
Appendix.
Statistical analysis
Unconditional logistic regression analysis was used to
calculate odds ratios as an estimate of the relative risk and 95
percent confidence intervals for the association between
physical activity and the risk of endometrioma (24). Separate
analyses were conducted for the 2-year period prior to the
reference date and the period between the ages of 12 and 21
years. The analyses were based on five different measurements of physical activity during each period: 1) any physical activity versus none; 2) the highest intensity level of
activity defined as low intensity, moderate intensity, or high
intensity versus none (four mutually exclusive categories);
3) the total number of hours spent engaging in high-intensity
activity (in tertiles) versus no high-intensity activity; 4) the
total number of episodes of high-intensity activity (in
tertiles) versus no high-intensity activity; and 5) highintensity physical activity ≥three times/week, ≥30 minutes/
episode, and ≥10 months/year for 2 years (for the 2-year
period) or for ≥5 years (for the period between the ages of 12
158 Dhillon and Holt
TABLE 1. Distribution* of selected demographic and other
characteristics among endometrioma cases and controls,
Group Health Cooperative of Puget Sound, Washington State,
1990–1994
TABLE 1. Continued
Cases
(n = 77)
Controls
(n = 735)
≤56.9
26.0
30.2
>56.9–62.8
15.6
21.0
>62.8–71.0
23.4
23.4
>71.0
35.1
25.5
≤1.61
19.5
29.9
>1.61–1.66
23.4
27.2
>1.66–1.71
27.3
23.7
>1.71
29.9
19.3
Nonsmoker
68.8
60.8
≤0.5
20.8
18.6
Risk factor
Weight (kg)†
Risk factor
Cases
(n = 77)
Controls
(n = 735)
Age at reference date (years)
≤24 (7 cases, 129 controls)
9.1
17.6
25–29 (15 cases, 145 controls)
19.5
19.7
30–34 (22 cases, 186 controls)
28.6
25.3
35–39 (33 cases, 275 controls)
42.9
37.4
Race†
White
85.7
83.0
Black
5.2
8.9
Asian
5.2
4.6
Native American/Eskimo
0.0
0.5
Hispanic
2.6
1.2
Other
1.3
1.8
Marital status†
Married
52.0
62.2
Living as married
10.4
0.80
Single
37.7
29.9
Highest educational level†
≤High school
13.0
18.1
Technical/some college
28.6
35.9
College graduate
58.4
45.9
Family income†
<$15,000
6.5
6.1
$15,000–$29,999
26.0
24.7
$30,000–$45,000
26.0
29.7
>$45,000
41.6
38.7
Current
5.2
23.2
Former
81.9
64.3
Never
13.0
12.6
Height (m)†
Cigarette smoking (packs/day)†
>0.5–<1
2.6
5.5
≥1
7.8
15.1
0
66.2
45.4
1
15.6
16.2
2
15.6
26.4
3
2.6
9.2
≥4
0.0
2.9
No
93.6
98.4
Yes
6.5
1.7
No. of livebirths†
Anorexia†
* All values are expressed as percentages; some percentages do
not total 100 because of rounding.
† Age standardized to the distribution among cases using the
direct method.
Oral contraceptive use†
Table continues
and 21 years) versus no high-intensity activity during the
relevant period. Tertiles were used to compare the distribution of number of hours and episodes, and the cutpoints were
based on the distribution of these exposures among the
controls.
To account for the difference in age between cases and
controls, we adjusted the proportion of controls to the age
distribution for endometrioma cases based on the following
groups: ≤24 years, 25–29 years, 30–34 years, and 35–39
years. Age standardization was conducted for the univariate
presentation of results shown in table 1.
Two sets of regression models were constructed for the
final analyses, one based on physical activity during the 2year period prior to the reference date and the second on
activity between the ages of 12 and 21 years. For physical
activity during the 2-year period, all women were included in
the analyses (n = 812). For physical activity between the
ages of 12 and 21 years, only those women aged 24 years or
older were included in the analyses (n = 707) to ensure no
overlap between recent activity and activity at 12–21 years.
Potential confounding variables were included in the final
regression model if their inclusion resulted in a ≥10 percent
change in the risk estimate (25). All models for the 2-year
period included age, number of livebirths (zero, one, two,
three, and four or more), oral contraceptive use (never,
former, current), and cigarette smoking (nonsmoker, >0–0.5,
>0.5–<1.0, and ≥1.0 packs per day based on the average
number of cigarettes usually smoked per day for women who
ever reported smoking). Estimates of the relative risk for the
12–21-year period were adjusted for age only. Other variables considered but not included in the final models were
race, income, education, menstrual period and cycle length,
anorexia, height, infertility, and number of pelvic examinations during the 5 years prior to the reference date. Weight
and menstrual pain were not considered for inclusion as
confounders because they could be either in the causal
pathway between physical activity and endometrioma or a
Am J Epidemiol 2003;158:156–164
Physical Activity and Endometrioma Risk 159
TABLE 2. Multivariate-adjusted odds ratios* of endometriomas according to aspects of study
participants’ physical activity during the 2 years prior to the reference date, Group Health
Cooperative of Puget Sound, Washington State, 1990–1994
Cases (n = 77)
Controls (n = 735)
Physical activity level
OR†
95% CI†
No.
%‡
No.
%‡
No
27
35.1
289
39.3
1.00
Yes
50
64.9
446
60.7
0.90
None
27
35.1
289
39.3
1.00
Exclusively low
12
15.6
69
9.4
1.72
0.80, 3.71
7
9.1
60
8.2
0.93
0.37, 2.33
31
40.3
317
43.1
0.74
0.42, 1.31
None
46
59.7
418
56.9
1.00
≤120
14
18.2
109
14.8
0.95
0.48, 1.85
>120–264
11
14.3
103
14.0
0.70
0.34, 1.46
6
7.8
105
14.3
0.37
0.15, 0.92
None
46
59.7
418
56.9
1.00
≤160
12
15.6
109
14.8
0.89
>160–288
15
19.5
104
14.2
0.90
0.47, 1.76
4
5.2
104
14.2
0.24
0.08, 0.71
Any physical activity
0.54, 1.52
Activities of exclusively low, medium,
and any high intensity
Medium or medium/low
Any high
Total no. of high-intensity hours over
the 2-year period§
>264
Total no. of high-intensity times over
the 2-year period§
>288
0.44, 1.81
* Unconditional logistic regression models for physical activity in the 2 years prior to the reference date
were adjusted for age (continuous), number of livebirths (0, 1, 2, 3, ≥4), oral contraceptive use (never,
former, current), and number of packs of cigarettes smoked per day (none, ≤1/2, >1/2–<1, ≥1).
† OR, odds ratio; CI, confidence interval.
‡ Some percentages do not total 100 because of rounding.
§ Tertiles based on distribution among controls.
marker of disease. We did not find significant (p < 0.05)
interactions between physical activity during the 2-year
period and body mass index (weight (kg)/height (m)2)
(dichotomized as <25 vs. ≥25) or between physical activity
from ages 12–21 years and age at the reference date.
RESULTS
Participants
Table 1 shows the distribution of age and age-adjusted
demographic characteristics among cases and controls.
Cases were older and less likely to be Black. In addition,
they were more often single and reported a higher level of
education. Cases also tended to be heavier and taller based
on the distribution of weight and height in quartiles. On the
other hand, controls were heavier smokers and had had a
higher number of livebirths. They were more likely to be
current oral contraceptive users and less likely to have a
history of anorexia. Among cases, 3.9 percent had a prior
diagnosis of ovarian endometriosis (data not shown).
Am J Epidemiol 2003;158:156–164
Physical activity during the 2 years prior to the reference
date
We found no association between “any physical activity”
during the 2 years prior to the reference date and risk of
endometrioma after controlling for age, number of livebirths, oral contraceptive use, and cigarette smoking (odds
ratio (OR) = 0.90, 95 percent confidence interval (CI): 0.54,
1.52; table 2). A nonsignificant decreased risk was found for
women who reported any high-intensity activity (OR = 0.74,
95 percent CI: 0.42, 1.31; table 2), whereas those who
engaged in exclusively low and/or moderate activities had no
difference in risk relative to those who reported no physical
activity.
We evaluated the effects of high-intensity activity based
on varying levels of duration and frequency in women who
reported any physical activity associated with intensity
codes of ≥6.0 METs. Women who reported spending the
highest number of hours engaged in high-intensity activity
(an average of 2.5 hours per week) had a significantly
reduced endometrioma risk (OR = 0.37, 95 percent CI: 0.15,
160 Dhillon and Holt
TABLE 3. Multivariate-adjusted odds ratios* of endometriomas according to aspects of study
participants’ physical activity between 12 and 21 years of age, Group Health Cooperative of Puget
Sound, Washington State, 1990–1994
Cases† (n = 72)
Controls† (n = 635)
Level of physical activity
OR‡
No.
%§
No.
%§
No
25
34.7
239
37.6
1.00
Yes
47
65.3
396
62.4
1.27
95% CI‡
Any physical activity
0.75, 2.15
Activities of exclusively low, medium, and
any high intensity
None
25
34.7
239
37.6
1.00
Exclusively low
6
8.3
26
4.1
2.35
0.88, 6.28
Medium or medium/low
7
9.7
64
10.1
1.15
0.47, 2.80
34
47.2
306
48.2
1.19
0.68, 2.08
None
38
52.8
329
51.8
1.00
≤240
13
18.1
101
15.9
1.19
0.61, 2.34
>240–816
12
16.7
101
15.9
1.13
0.56, 2.28
9
12.5
104
16.4
0.82
0.38, 1.77
None
38
52.8
329
51.8
1.00
≤240
11
15.3
111
17.5
0.94
>240–732
18
25.0
95
15.0
1.80
0.97, 3.35
5
6.9
100
15.8
0.47
0.18, 2.23
Any high
Total no. of high-intensity hours between
ages 12 and 21 years¶
>816
Total no. of high-intensity times between
ages 12 and 21 years¶
>732
0.46, 1.91
* Unconditional logistic regression models for physical activity between the ages of 12 and 21 years were
adjusted for age (continuous).
† Analysis restricted to women aged 24 years or older.
‡ OR, odds ratio; CI, confidence interval.
§ Some percentages do not total 100 because of rounding.
¶ Tertiles based on distribution among controls.
0.92). A larger decrease in risk was observed for women who
engaged in high-intensity activity more than 288 times over
the 2-year period (an average of three or more times per
week); these women were 76 percent less likely to be diagnosed with endometrioma than women who engaged in no
high-intensity activity at all (OR = 0.24, 95 percent CI: 0.08,
0.71).
Physical activity between 12 and 21 years of age
After we controlled for age, physical activity between the
ages of 12 and 21 years was not associated with risk of
endometrioma (“any physical activity”; OR = 1.27, 95
percent CI: 0.75, 2.15; table 3). The risk did not differ significantly by the intensity level of activity. When we focused on
high-intensity activity, no significant reduction in risk was
associated with increased duration or frequency of activity,
although a lower risk was observed for women in the highest
tertile (105 minutes per week) of high-intensity hours (OR =
0.82, 95 percent CI: 0.38, 1.77). A similar trend was
observed for women reporting the highest frequency of
activity (>1.5 times per week) of physical activity during
adolescence (OR = 0.47, 95 percent CI: 0.18, 2.23). We
found no significant interaction between age at the reference
date and physical activity during this period (data not
shown).
Consistent, high-intensity exercise
To evaluate the effects of physical activity in accordance
with public health recommendations (26), we focused on
women who participated in high-intensity activity for at least
30 minutes per episode, three or more times per week, and 10
or more months per year. The risk of endometrioma was
significantly lowered by 70 percent for consistent, highintensity exercisers in comparison to women reporting no
high-intensity activity during the 2 years prior to the reference date (OR = 0.30, 95 percent CI: 0.12, 0.74; table 4). A
similar analysis focusing on the period between 12 and 21
years of age yielded a 35 percent reduced risk, although the
association was nonsignificant (OR = 0.65, 95 percent CI:
0.19, 2.23).
Am J Epidemiol 2003;158:156–164
Physical Activity and Endometrioma Risk 161
TABLE 4. Multivariate-adjusted odds ratios of endometriomas associated with consistent, high-intensity physical
activity by study participants, Group Health Cooperative of Puget Sound, Washington State, 1990–1994
Cases (n = 77)
Controls (n = 735)
OR*
No.
%
No.
95% CI*
%
Physical activity during the 2-year period prior to the reference date
High-intensity activity ≥3 times/week, ≥30 minutes/episode, ≥10
months/year for 2 years†
No high-intensity activity‡
Yes
46
59.7
418
56.9
1.00
6
7.8
121
16.5
0.30
0.12, 0.74
Physical activity between the ages of 12 and 21 years§
High-intensity activity ≥3 times/week, ≥30 minutes/episode, ≥10
months/year for ≥5 years¶
No high-intensity activity‡
Yes
38
52.8
329
51.8
1.00
3
4.2
43
6.8
0.65
0.19, 2.23
* OR, odds ratio; CI, confidence interval.
† Unconditional logistic regression models for physical activity in the 2 years prior to the reference date were adjusted for age
(continuous), number of livebirths (0, 1, 2, 3, ≥4), oral contraceptive use (never, former, current), and number of packs of cigarettes
smoked per day (none, <1/2, >1/2–<1, ≥1).
‡ Reference group includes women who participated in no activity or exclusively low- or medium-intensity activity.
§ Analysis restricted to women aged 24 years or older.
¶ Unconditional logistic regression models for physical activity between 12 and 21 years were adjusted for age (continuous).
DISCUSSION
We observed a significant 70 percent decreased risk of
endometrioma associated with recent, frequent, and regular
high-intensity physical activity (≥three times/week, ≥30
minutes/episode, ≥10 months/year for 2 years). About half of
these women exercised 3 or fewer hours a week (median,
3.08 hours per week) and still experienced a reduction in
endometrioma risk. A decreased risk was also observed for
engaging in such activity between the ages of 12 and 21
years, although the estimate did not reach statistical significance. Reduced risks were not observed for activities of
lower intensity. Our findings concerning current or recent
exercise are in accordance with those of two previous epidemiologic studies (8, 14). Cramer et al. (8) found a 40 percent
lower risk for women who reported “regular exercise” for 3–
7 hours per week (OR = 0.6, 95 percent CI: 0.4, 0.8) and an
80 percent risk reduction for those who exercised more than
7 hours per week (OR = 0.2, 95 percent CI: 0.1, 0.6) when
compared with nonexercisers. Signorello et al. (14) reported
a nonsignificant 60 percent reduction in risk associated with
“vigorous exercise” for 4 or more hours a week (95 percent
CI: 0.2, 1.2).
One advantage of the current study is that the intensity
level of physical activity was based on the determination of
METs for specific activities by using the recently updated
Compendium of Physical Activities (21) and not on a study
subject’s interpretation of “vigorous” or “regular” exercise.
We were also able to evaluate the combined effects of duration, frequency, and intensity level of physical activity and
consequently activity levels suggested in public health
recommendations (26). The inverse associations between
physical activity and endometriotic disease in the current
study are similar to those found in the two previous studies
Am J Epidemiol 2003;158:156–164
(8, 14) even though cases in our study were women with
more severe disease and cases in previous studies had infertility-associated endometriosis, often a mild or moderate
form of disease. In a cross-sectional study in Norway (13),
investigators compared women with and without a selfreported diagnosis of endometriosis and did not find a
reduced risk associated with “regular” exercise (OR = 0.7,
95 percent CI: 0.3, 1.5). In that study, the investigators did
not define “regular” exercise or specify exercise prior to
diagnosis as the relevant exposure period. In none of the
three studies (8, 13, 14) did the authors address whether the
presence of preexisting disease may have influenced the
levels of recent physical activity reported by endometriosis
cases. Our study was able to address this issue by obtaining
exercise patterns during adolescence, a time when activity
would unlikely be affected by undiagnosed disease.
This population-based study was designed to include a
representative sample of defined cases with one type of
serious endometriotic disease (endometriomas) as opposed
to previous studies, which were based on self-reported or
infertility-associated endometriosis. The majority of cases in
our study (96.1 percent) had no prior history of ovarian
endometriosis, and all cases of disease were confirmed
surgically, increasing diagnostic validity. Although cases
were limited to a subset of women with more severe disease,
they represented a homogeneous group etiologically, which
enabled a more precise evaluation of potential risk factors.
Controls were randomly selected and thus considered representative of the defined source population (female enrollees
in the Group Health Cooperative). Although controls were
considered disease free, without a diagnostic evaluation
confirming the absence of disease, one cannot exclude the
possibility that some of them may have harbored undiagnosed endometriosis. One would expect routine and stren-
162 Dhillon and Holt
uous activity to take place less often among these women,
which would result in bias toward the null.
A potential explanation for the decreased endometrioma
risk we observed in relation to strenuous physical activity is
a reverse temporal sequence of events whereby cases did not
engage in physical activity because of symptoms indicative
of preexisting, undiagnosed disease. To address this possibility, we conducted two additional analyses to assess the
impact of symptoms on consistent, high-intensity physical
activity during the 2 years prior to diagnosis. In one such
analysis, we excluded women who reported severe pain or
heavy bleeding associated with menstruation (n = 65). We
found a similar reduction in risk associated with regular,
high-intensity activity for the remaining women (OR = 0.32,
95 percent CI: 0.13, 0.80). Although we did not have information on chronic pelvic pain, we asked cases whether a
symptom—including pelvic pain, pressure, fullness; vaginal
bleeding heavier than during a period; delayed or irregular
menstruation; and other symptoms—prompted the visit that
ultimately led to diagnosis. In an additional analysis, we
excluded cases who reported having a symptom for more
than 3 months (n = 24) and found a slightly attenuated effect
consistent with a 61 percent reduction in endometrioma risk
(OR = 0.39, 95 percent CI: 0.14, 1.07). We were unable to
assess the presence of chronic pelvic pain, which might have
led to a reduced pattern of physical activity among some
women with endometriomas prior to their diagnosis, contributing to the observed decrease in risk associated with current
activity. However, past research has found that only 7
percent of women with ovarian endometriosis report severe
pelvic pain (27), reassuring us that this factor is unlikely to
fully explain our results. Moreover, we found a trend toward
reduced risks for women reporting consistent and strenuous
activity between the ages of 12 and 21 years, a time when
temporal ambiguity is unlikely, supporting the plausibility of
a true inverse association.
We did not find an association between endometrioma risk
and exclusively low- to medium-intensity activity in either
of the two periods. This may reflect varying effects of
different levels of exercise intensity on the female reproductive system. In some women, routine and athletic activity
results in suppressing the hypothalamic release of gonadotropin-releasing hormone, which limits pituitary secretion of
luteinizing hormone and follicle-stimulating hormone, in
turn reducing ovarian stimulation and estradiol production
(18). By disrupting the hypothalamic-pituitary-ovarian axis,
regular and strenuous exercise can prolong the follicular
phase and lead to interruption or suppression of menstrual
cycles (18). If physical activity at higher intensity levels
results in an altered hormonal milieu, a woman’s risk of
endometriosis may be reduced because of lower levels of
hormonal stimuli (e.g., endogenous estrogen levels), which
influence both growth and development. This situation may
be more apparent in the current study since cases were
defined as women with endometriomas, which occur in the
ovaries, a microenvironment with very high steroid hormone
concentrations (28). In addition, to the extent that retrograde
menstruation plays a role in development of endometriosis,
exercise-related menstrual irregularities could lower a
woman’s risk of ovarian endometriosis by reducing the
amount of menstrual reflux over time. Activities of moderate
intensity may not correlate to similar changes in risk because
of transient or clinically insignificant hormonal changes or
because of no measurable disturbances in the menstrual
cycle.
Definitive diagnosis of endometriosis requires direct visualization of lesions through an operative intervention, so
women with a surgical diagnosis of the disease represent a
subset of all women harboring the disease. The possibility of
asymptomatic disease and the absence of sensitive screening
tests that are noninvasive limited our ability to draw conclusions about the etiologic relevance of factors such as strenuous physical activity, which is associated with the clinical
incidence of disease. Although its role in the pathogenesis of
endometriosis needs to be explored further, the effects of
strenuous exercise on circulating hormone levels and on the
menstrual cycle are supported by studies that have found
delayed menarche as well as primary and secondary amenorrhea in female athletes (29–32).
We explored the effects of strenuous activity on characteristics of the menstrual cycle in controls. Women who
reported routine, high-intensity activity during the 2 years
prior to the reference date had a shorter “period” length (34.7
vs. 30.6 percent with 4 or fewer days) and a lighter volume
of menstrual flow (11.6 vs. 5.6 percent experienced “light”
bleeding) than those who did not participate in high-intensity
activity. For activity between 12 and 21 years of age, regular
exercisers were more likely to have a later (age ≥14 years)
menarche (28.0 vs. 19.6 percent) and onset of regular periods
(46.0 vs. 33.1 percent).
In our study, physical activity data were derived from selfreports, which are subject to more misclassification than
measurements based on physical activity logs or exercise
tests. In one validity study, subject responses on the Minnesota Leisure Time Physical Activity Questionnaire were
compared with those from six 48-hour records and a 4-week
version of the questionnaire (20). The authors noted a strong
correlation in heavy activities reported by the three methods
but weak or no associations for moderate and light activities.
The lack of an association for lower-intensity activities in
our study may stem from a greater degree of nondifferential
misclassification of these activities.
At least two explanations are possible for the absence of a
significant association between physical activity at ages 12–
21 years and endometrioma risk in our study. First, exercise
patterns during adolescence and early adulthood often fluctuate widely from year to year, so any summary estimate is a
crude measure of activity during this highly variable period.
To address this issue, we separated the period into two
segments, 12–16 years and 17–21 years, and conducted the
same set of analyses; the overall conclusions remained the
same (data not shown). Second, self-reports of physical
activity between the ages of 12 and 21 years are subject to
more recall error than activities performed in the recent past.
If cases and controls inaccurately recall past activity to a
similar extent, then measurement of physical activity during
that period can result in bias toward the null.
In summary, our study findings suggest a reduced
endometrioma risk associated with regular, high-intensity
physical activity during adulthood and, to a lesser extent,
Am J Epidemiol 2003;158:156–164
Physical Activity and Endometrioma Risk 163
adolescence. Endometriosis constitutes a wide spectrum of
pathologic and clinical disease, and this study was based on
a relatively homogeneous subgroup of women diagnosed
with cystic ovarian endometriosis. Although the ovaries are
the most common site for endometriotic implants, the
etiology of ovarian endometriosis may differ from disease
that develops at other sites, especially when its unique endocrine environment is considered. Different microenvironments may imply differing causal pathways for the
development and maintenance of endometriosis. There is a
need to investigate whether the potential benefits of vigorous
and consistent physical activity extend to other affected sites
and types of this disease as well.
ACKNOWLEDGMENTS
This work was funded by grant R01 HD 25959 from the
National Institute of Child Health and Human Development
and was conducted at the Fred Hutchinson Cancer Research
Center.
The authors thank Dr. Noel Weiss for his valuable
comments.
REFERENCES
1. Jenkins S, Olive DL, Haney AF. Endometriosis: pathogenetic
implications of the anatomic distribution. Obstet Gynecol 1986;
67:335–8.
2. Adamson GD. Diagnosis and clinical presentation of
endometriosis. Am J Obstet Gynecol 1990;162:568–9.
3. Kjerulff KH, Erickson BA, Langenberg PW. Chronic gynecological conditions reported by U.S. women: findings from the
National Health Interview Survey, 1984–1992. Am J Public
Health 1996;86:195–9.
4. Nisolle M, Donnez J. Peritoneal endometriosis, ovarian
endometriosis, and adenomyotic nodules of the rectovaginal
septum are three different entities. Fertil Steril 1997;68:585–96.
5. Morales AJ, Murphy AA. Endoscopic treatment for endometriosis. Obstet Gynecol Clin North Am 1999;26:121–33.
6. Koninckx PR, Meuleman C, Demeyere S, et al. Suggestive evidence that pelvic endometriosis is a progressive disease,
whereas deeply infiltrating endometriosis is associated with
pelvic pain. Fertil Steril 1991;55:759–65.
7. Holt VL, Jenkins J. Endometriosis. In: Goldman MB, Hatch
MC, eds. Women and health. San Diego, CA: Academic Press,
1999:226–39.
8. Cramer DW, Wilson E, Stillman RJ, et al. The relation of
endometriosis to menstrual characteristics, smoking, and exercise. JAMA 1986;255:1904–8.
9. Parazzini F, La Vecchia C, Franceschi S, et al. Risk factors for
endometrioid, mucinous and serous benign ovarian cysts. Int J
Epidemiol 1989;18:108–12.
10. Parazzini F, Ferraroni M, Fedele L, et al. Pelvic endometriosis:
reproductive and menstrual risk factors at different stages in
Lombardy, northern Italy. J Epidemiol Community Health
1995;49:61–4.
11. Matorras R, Rodiquez F, Piuoan JI, et al. Epidemiology of
Am J Epidemiol 2003;158:156–164
endometriosis in infertile women. Fertil Steril 1995;63:34–8.
12. Arumugam K, Lim JM. Menstrual characteristics associated
with endometriosis. Br J Obstet Gynecol 1997;104:948–50.
13. Moen MH, Schei B. Epidemiology of endometriosis in a Norwegian county. Acta Obstet Gynecol Scand 1997;76:559–62.
14. Signorello LB, Harlow BL, Cramer DW, et al. Epidemiologic
determinants of endometriosis: a hospital-based case-control
study. Ann Epidemiol 1997;7:267–74.
15. Darrow SL, Vena JE, Batt RE, et al. Menstrual cycle characteristics and the risk of endometriosis. Epidemiology 1993;4:135–
42.
16. Sangi-Haghpeykar H, Poindexter AN. Epidemiology of
endometriosis among parous women. Obstet Gynecol 1996;85:
983–92.
17. DiZeraga GS, Barber DL, Hodgen GD. Endometriosis: role of
ovarian steroids in initiation, maintenance and suppression.
Fertil Steril 1980;33:649–53.
18. Warren MP, Perlroth NE. The effects of intense exercise on the
female reproductive system. J Endocrinol 2001;170:3–11.
19. Taylor HL, Jacobs DR Jr, Schucker B, et al. A questionnaire for
the assessment of leisure time physical activities. J Chronic Dis
1978;31:741–55.
20. Richardson MT, Leon AS, Jacobs DR Jr, et al. Comprehensive
evaluation of the Minnesota Leisure Time Physical Activity
Questionnaire. J Clin Epidemiol 1994;47:271–81.
21. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of
physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc 2000;32:S498–516.
22. Leon AS, Connett J, Jacobs DR Jr, et al. Leisure-time physical
activity levels and risk of coronary heart disease and death. The
Multiple Risk Factor Intervention Trial. JAMA 1987;258:
2388–95.
23. Littman AJ, Voigt LF, Beresford SA, et al. Recreational physical activity and endometrial cancer risk. Am J Epidemiol 2001;
154:924–33.
24. Breslow NE, Day NE, eds. Statistical methods in cancer
research. Vol 1. The analysis of case-control studies. Lyon,
France: International Agency for Research on Cancer, 1980.
(IARC scientific publication no. 32).
25. Maldonado G, Greenland S. Simulation study of confounderselection strategies. Am J Epidemiol 1993;138:923–36.
26. Pate RR, Pratt M, Blair SN, et al. Physical activity and public
health. JAMA 1995;273:402–7.
27. Fedele L, Parazzini F, Bianchi S, et al. Stage and localization of
pelvic endometriosis and pain. Fertil Steril 1990;53:155–8.
28. Koninckx PR, Kennedy SH, Barlow DH. Endometriotic disease: the role of peritoneal fluid. Hum Reprod Update 1998;4:
741–51.
29. Warren MP. The effects of exercise on pubertal progression and
reproductive function in girls. J Clin Endocrinol Metab 1980;
51:1150–7.
30. Baker ER, Mathur RS, Kirk RF, et al. Female runners and secondary amenorrhea: correlation with age, parity, mileage, and
plasma hormonal and sex hormone-binding globulin concentrations. Fertil Steril 1981;36:183–7.
31. Boyden TW, Pamenter RW, Stanforth P, et al. Sex steroids and
endurance running in women. Fertil Steril 1983;39:629–32.
32. Loucks AB, Mortola JF, Girton L, et al. Alterations in the hypothalamic-pituitary-ovarian and the hypothalamic-pituitaryadrenal axes in athletic women. J Clin Endocrinol Metab 1989;
68:402–11.
164 Dhillon and Holt
APPENDIX
APPENDIX TABLE 1. Intensity codes for different types of
physical activities reported by study participants, Group Health
Cooperative of Puget Sound, Washington, 1990–1994
APPENDIX TABLE 1. Continued
Metabolic
equivalent
Physical activity
Physical activity
Metabolic
equivalent
Aerobics
6.5
Jump rope
10.0
Backpacking (pack weighing ≥20 pounds*)
7.0
Karate
10.0
Badminton
4.5
Lacrosse
8.0
Baseball, softball
5.0
Mountain climbing, rock climbing
8.0
Basketball, game play or officiating
7.5
Paddleball
6.0
Basketball, nongame
6.0
Racquetball
7.0
Bicycling for pleasure or to and from work
6.0
Rowing, canoeing, kayaking for competition
Bowling (one game = 10 minutes)
3.0
Rowing, canoeing, kayaking for pleasure
3.5
Carpentry, home building
3.0
Running
8.0
Cheerleading
6.0
Sailing (racing or practicing for racing only)
5.0
Dancing, any type
4.5
Scuba diving
7.0
Farm chores, including chopping wood, clearing land
5.5
Skating, roller or ice
7.0
Fencing, javelin
6.0
Snorkeling
5.0
Field hockey
8.0
Snow skiing, cross country
8.0
Fishing (in stream only)
6.0
Snow skiing, downhill
6.0
Football
8.0
Snowmobiling
3.5
Gardening, mowing lawn, raking (except riding mower)
4.5
Soccer, kickball
Golf—hitting balls
3.0
Squash
Golf—riding power cart (nine holes = 1.5 hours)
3.5
Swimming and water aerobics
6.0
Golf—walking (nine holes = 1.5 hours)
4.5
Table tennis
4.0
Tennis, doubles (set = 30 minutes)
6.0
Tennis, singles (set = 20 minutes)
8.0
Gymnastics, march and drill team (competitive color
guard)
Handball
4.0
12.0
Health club exercise, except aerobics and weightlifting
5.5
Heavy lifting
6.5
Hiking
6.0
Home exercise, health club exercise
3.5
Horseback riding, rodeo
12.0
7.0
12.0
Track
10.0
Volleyball
3.0
Walking for pleasure
3.5
Walking to and from work or school
4.0
4.0
Walking up stairs (at least five flights/day, one flight =
0.5 minute)
8.0
Hunting
5.0
Water skiing (ride = 5 minutes)
6.0
Jogging (running and walking)
6.5
Weight lifting
3.0
Yoga
4.0
Judo
10.0
Table continues
* One pound = 0.45 kg.
Am J Epidemiol 2003;158:156–164