Professional Documents
Culture Documents
Objective: To evaluate the association between ectopic pregnancy (EP) and clinical and historical factors among
women presenting with pain and/or bleeding in early pregnancy.
Design: Nested case– control study.
Setting: University medical center.
Patient(s): Women with symptomatic early pregnancies of unknown location presenting for care between January
1, 1990 and July 31, 1999.
Intervention(s): None.
Main Outcome Measure(s): Clinical and historical risk factors were compared between women with EP and
women with ongoing intrauterine pregnancies or spontaneous abortions.
Result(s): The following factors were associated with increased risk of EP: prior EP (odds ratio, 2.98 [95%
confidence interval, 1.88 – 4.73] for one prior EP and 16.04 [5.39 – 47.72] for 2 or more), pelvic inflammatory
disease history (1.5 [1.11–2.05]), pain at presentation (1.42 [1.06 –1.92]), vaginal bleeding at presentation (1.42
[1.04 –1.93]), and hCG of 501–2,000 mIU/mL (1.73 [1.24 –2.42]). Age younger than 25 years (0.59 [0.41– 0.85])
and a history of abortion were protective from EP (0.58 [0.38 – 0.90]). Prior nontubal pelvic surgery, past
intrauterine device use, prior cesarean section, and current cervical infection demonstrated no association
with EP.
Conclusion(s): Evaluation of women with a symptomatic early pregnancy confirms and refutes some of the
classical risk factors for EP. Prior EP is a strong risk factor, whereas pelvic inflammatory disease has an
unexpected weak association. Previous abortion was found to have a negative association, whereas nontubal
surgery, cesarean section, and a history of or concomitant cervical infection have no association. Knowledge of
historical and clinical factors associated with EP may aid in early diagnosis. (Fertil Steril威 2006;86:36 – 43. ©
2006 by American Society for Reproductive Medicine.)
Key Words: Ectopic pregnancy, risk factors, diagnosis
Ectopic pregnancy (EP) is defined as a pregnancy that from 17,800 in 1970 to 88,400 in 1992 (1, 2). Conserva-
occurs outside of the uterus. Ruptured EP is a leading tive medical and surgical treatments now are widely avail-
cause of morbidity and mortality for women during the able for ectopic pregnancies that are diagnosed early. It
first trimester of pregnancy (1). Approximately 2% of all has been shown clearly that early diagnosis and outpatient
pregnancies in the United States are diagnosed as ectopic treatment of an early unruptured EP can lead to reduced
(1). Recent studies indicate that the incidence of EP has morbidity and mortality, preservation of future fertility,
increased sixfold during the last 25 years, and hospital and substantial cost savings compared with inpatient care
admissions for this condition have increased dramatically, (1, 3– 6). Counseling of high-risk patients before concep-
tion and early screening of patients with heightened risk
Received September 19, 2005; revised and accepted December 5, 2005. of EP can significantly affect patient prognosis and med-
Supported by National Institutes of Health (Bethesda, MD) grant R01: ical costs.
HD-36455-05.
Reprint requests: Kurt T. Barnhart, M.D., M.S.C.E., Penn Fertility Care, The apparent increase in the incidence of EP may be a
Department of Obstetrics and Gynecology, University of Pennsylva-
result of improved diagnostic aids, heightened awareness
nia Medical Center, 3701 Market Street, Suite 810, Philadelphia,
Pennsylvania 19104 (FAX: 215-615-4200; E-mail: kbarnhart@mail. by clinicians, or increased prevalence of risk factors of the
obgyn.upenn.edu). disease (7–9). There is a large body of literature reporting
MATERIALS AND METHODS First, descriptive statistics were used to summarize the
entire population and to compare cases and controls. Uni-
Approval to conduct this study was obtained from the insti- variate associations were evaluated by using Student’s t test
tutional review board of the University of Pennsylvania. A for continuous variables. Categorical variables were evalu-
database is maintained at the University of Pennsylvania of ated by using Pearson’s 2 test of association. Wald statistics
all women in their first trimester of pregnancy (positive were constructed to test for an overall association between
pregnancy test or history of a missed period) who present the categorical risk factor and diagnosis. To check for dif-
with pain and/or bleeding. Data are entered directly into the ferences in univariate comparisons resulting from classifica-
computerized database by clinical staff who are caring for tion of information, some variables (age, parity, history of
the patient. surgery) were analyzed as both continuous and categorical
variables.
Potential risk factors for EP were identified from the
history, clinical presentation, and diagnostic tests. Poten- Stratified analyses were performed to test for confound-
tial predictors derived from the historical factors were ing and effect modification. Historical and clinical-pre-
age, gravity, parity, number of live births, number of CSs, sentation variables first were tested individually to check
number of spontaneous abortions (SAB), number of elec- for interaction. For the purpose of analysis of categorical
tive abortions, number of previous ectopic pregnancies, values, one category was chosen as the reference cate-
PID (defined as inpatient treatment), outpatient treatment gory. Reference categories included age 25–29 years and
of gonorrhea and/or chlamydia, history of intrauterine hCG of ⬍500 mIU/mL. When no interaction was noted,
device use, and history of pelvic surgery (excluding CS). both historical and clinical variables were combined. A
Age (y)
⬍20 43 11.8 443 27.8 0.3 0.18–0.39 ⬍.0001
20–25 87 23.8 502 31.5 0.5 0.34–0.64 .0003
25–30 112 30.7 304 19.1 1.0 Ref
30–34 84 23.0 210 13.9 1.1 0.80–1.70 .4
ⱖ35 39 10.7 133 8.4 0.8 0.78–1.52 .8
Race
African-American 229 62.6 853 51.4 1.41 0.60–2.18 .69
Others 12 3.3 51 3.1 Ref
Prior live births 192 52.3 718 43.5 1.43 1.14–1.79 .002
Parity
0 174 47.4 946 57.0 Ref
1 94 25.6 319 19.2 1.6 1.21–2.12 .001
2 45 12.3 192 11.6 1.27 0.89–1.83 .19
3 29 7.9 128 7.7 1.23 0.8–1.9 .35
4 or more 25 6.8 74 4.5 1.84 1.14–2.97 .01
Prior voluntary interruption of
pregnancy
0 294 82.4 1,268 79.6 Ref
1 33 9.2 216 13.5 0.66 0.45–0.97 .04
2 or more 30 8.4 110 6.9 1.18 0.77–1.8 .45
History of spontaneous abortion
0 269 83.8 1,232 85.2 Ref
1 36 11.2 162 11.2 1.02 0.69–1.5 .93
2 or more 16 5 52 3.6 1.41 0.79–2.51 .24
History of ectopic pregnancy
0 306 83.4 1,570 94.7 Ref
1 48 13.1 82 4.9 3 2.1–4.4 ⬍.0001
2 or more 13 3.5 6 0.4 11.17 402–29.5 ⬍.0001
History of pelvic surgery 88 24 286 17.2 1.51 1.15–1.99 .003
History of prior cesarean section
0 342 93.2 1,545 93.2 Ref
1 16 4.4 89 5.4 0.81 0.47–104 .45
2 or more 9 2.4 24 1.4 1.69 0.78–3.68 .18
Past use of intrauterine device 20 5.5 67 4 1.37 0.82–2.29 .23
History of pelvic inflammatory disease 90 24.5 327 19.7 1.32 1.01–1.73 .04
History of outpatient treatment for
gonorrhea and/or chlamydia
0 289 79 1,343 81 Ref
1 60 16.4 262 15.8 1.06 0.78–1.45 .69
2 18 4.9 46 2.8 1.82 1.04–3.19 .04
3 or more 0 0 7 0.4 0 0 .99
Note: Ref ⫽ reference group.
Barnhart. Risk factors predictive of ectopic pregnancy. Fertil Steril 2006.
logistic regression model then was generated using man- process was repeated until all the variables had a P value
ual selection of confounding variables and backward step- of ⱕ.05. A variable was retained in the model as a
wise selection of variables. At each step, the largest confounder if it significantly affected the coefficient esti-
P-value variable was removed from the table, and this mates of other variables by ⱖ15% (10).
38 Barnhart et al. Risk factors predictive of ectopic pregnancy Vol. 86, No. 1, July 2006
TABLE 2
Unadjusted clinical presentation predictors in women at risk of ectopic pregnancy.
Pain as the presenting symptom 244 66.5 1,046 63.1 1.16 0.92–1.48 .22
Bleeding at presentation 256 75.7 1,119 69.7 1.34 1.04–1.78 .03
(moderate to severe)
Current chlamydia cervical 13 4.1 109 7.7 0.51 0.28–0.91 .02
infection
Current gonorrhea and/or 22 6.9 151 10.8 0.61 0.38–0.97 .04
chlamydia cervical infection
HCG at presentation (mIU/mL)
0–500 155 42.2 828 49.9 Ref
501–2,000 103 28.1 314 18.9 1.75 1.32–2.32 ⬍.0001
2,001–4,000 40 10.9 167 10.1 1.3 0.87–1.88 .21
ⱖ4,000 69 18.8 350 21.1 1.05 0.77–1.44 .74
Note: Ref ⫽ reference group.
Barnhart. Risk factors predictive of ectopic pregnancy. Fertil Steril 2006.
An odds ratio of ⱖ1 denotes that a particular risk factor is history of EP, an initial quantitative hCG value of ⬎500
more prevalent in cases (EP) than in controls (SAB or mIU/mL, parity of one, prior hospitalization for PID, or the
ongoing IUP). Statistical analysis was performed by using presence of pain as the presenting symptom or moderate to
SAS software (Cary, NC), and a two-tailed P value of ⬍.05 severe bleeding. Factors negatively associated with EP were
was considered statistically significant. age younger than 25 years and prior history of a voluntary
interruption of pregnancy.
RESULTS Notable putative risk factors for EP that did not demon-
A total of 2,026 patients who presented with pain and/or strate a statistically significant association are listed in
bleeding in the first trimester of pregnancy from January 1, Table 4. These are gravidity, prior SAB, prior live births,
1990 to July 31, 1999 were evaluated. Of these, 367 patients past use of intrauterine device, prior history of pelvic sur-
were diagnosed with EP, and 1,659 women were diagnosed gery, prior CS, history of outpatient treatment for cervical
with an IUP (467 with an ongoing IUP and 1,192 with an infection with chlamydia and/or gonorrhea, or current infec-
SAB). tion with chlamydia and/or gonorrhea.
Age (y)
⬍20 0.34 0.22–0.52 ⬍.0001
20–24 0.59 0.41–0.85 .01
25–29 Ref
30–34 1.18 0.79–1.76 .42
ⱖ35 1.00 0.61–1.64 .99
Voluntary interruption of pregnancy
0 Ref
1 0.58 0.38–0.90 .02
2 or more 0.99 0.61–1.6 .96
History of ectopic pregnancy
0 Ref
1 2.98 1.88–4.73 ⬍.0001
2 or more 16.04 5.39–47.72 ⬍.0001
History of pelvic inflammatory disease: yes 1.50 1.11–2.05 .01
Parity
0 Ref
1 1.71 1.21–2.42 .003
2 1.13 0.72–1.78 .60
3 0.95 0.56–1.59 .83
4 or more 1.26 0.68–2.36 .46
Bleeding at presentation (moderate to severe): yes 1.42 1.04–1.93 .03
Pain as the presenting symptom: yes 1.42 1.06–1.92 .02
hCG at presentation (mIU/mL)
0–500 Ref
501–2,000 1.73 1.24–2.42 .001
2,001–4,000 1.38 0.88–2.16 .16
ⱖ4,000 0.97 0.67–1.39 .86
Note: Ref ⫽ reference group.
Barnhart. Risk factors predictive of ectopic pregnancy. Fertil Steril 2006.
There is extensive literature regarding potential risk fac- tent tubal pathology, which may lead to an increased incidence
tors for EP, including a meta-analysis summarizing the re- of subsequent EP in the same patient.
sults from 27 case– control studies and nine cohort studies Prior PID long has been thought to have a strong associ-
(12). Our results confirm and refute some of these putative ation with EP. The strength of association was noted to be
risk factors. 2.5 (2.1–3.0) in the meta-analysis (12), with an OR ranging
Previous EP was confirmed to have one of the strongest in the literature from 2.0 to 10.1 (13–15, 22–25). Although
associations of all the possible risk factors of EP. The risk of we also noted a statistically significant positive association
experiencing a repeat EP increased dramatically with the between PID and EP, the strength of that association was
number of prior ectopic gestations. We found that women surprisingly weak (OR 1.5; 95% CI, 1.11–2.05). It is possi-
with an EP were almost three times as likely to have had one ble that this association has been overestimated in other
prior EP compared with controls (odds ratio [OR], 2.98; 95% studies as a result of improper selection of controls. It also is
confidence interval [CI], 1.88 – 4.73) and were 16 times more possible that it is not a strong risk factor in an inner-city
likely to have had two prior EPs compared with women with population with a relatively high rate of pelvic infection,
an IUP (OR, 16.04; 95% CI, 5.39 – 47.72). Our findings confirm such as the population studied at our institution.
other research studies that have found a history of prior EP to be Previous tubal surgery also has been reported to be a
strongly associated with subsequent EP (OR ranging from significant risk factor for EP, with an estimated OR of 4.7
2.4 –25.0) (13–21). The meta-analysis estimated the risk to be (2.4 –9.5) according to one study (12). However, the associ-
6.6 (95% CI, 5.2– 8.4) (12). Recurrent EP likely reflects persis- ation between other types of pelvic surgery and CS has been
40 Barnhart et al. Risk factors predictive of ectopic pregnancy Vol. 86, No. 1, July 2006
TABLE 4
Adjusted odds ratios of factors not associated with risk of ectopic pregnancy.
inconsistent. The strength of association of nontubal abdom- (OR 1.6; 95% CI, 1.0 –1.6) (12). Perhaps the use of non-
inal or pelvic surgery has ranged from no association (13) pregnant controls in previous studies resulted in bias. In
to an OR as high as 2.4 –5.0 (14, 15, 22, 26 –28). Simi- addition, we found no association of previous SAB with risk
larly, the association between CS and EP has a wide range for EP. This also is contrary to findings of some published
that includes negative associations (24, 29 –31) to positive reports (14, 24, 41). Interestingly, we did find a positive
associations (22, 27, 32). Our data do not support the asso- association with parity.
ciation between nontubal surgery and EP (OR, 0.95; 95% CI,
Another important part of our study was the evaluation of
0.67–1.35). In addition, prior CS and history of multiple CS
both historical risk factors as well as signs and symptoms of
do not appear to be associated with EP.
presentation as risk factors for EP. We initially evaluated
Various studies have shown that the incidence of EP these factors separately but found that these variables were
increases with maternal age (24, 33, 34, 35). Our study too independent; therefore, we combined them in our final
shows that women with EP tend to be older than women with model. Our study population was restricted to women who
IUPs. In particular, compared with women aged 25–29 presented with pain and/or bleeding in the first trimester of
years, women aged younger than 20 years had an odds ratio pregnancy. However, some women presented with only
of EP 0.34 (95% CI, 0.22– 0.52), and women aged 20 –24 bleeding or only pain. We noted that moderate and/or severe
years had an odds ratio of EP of 0.59 (95% CI, 0.41– 0.85). bleeding (compared with mild or absent bleeding) was sig-
No significant associations were noted for women older than nificantly and positively associated with the presence of an
age 30 years. The cause of this association remains largely EP. In addition, the presence of pain as the chief complaint
unknown. also was positively associated with EP. The presence of pain
may be a result of ectopic gestation and was expected.
There has been controversy regarding the association be-
However, heavy bleeding often is thought to be a symptom
tween EP and a history of voluntary interruption of preg-
of a spontaneous miscarriage, not of an EP. Thus, it is
nancy. In this study, a prior induced abortion was signifi-
imperative that all patients with presenting complaints of
cantly negatively associated with risk of EP. Moreover, more
pain or bleeding with a period of amenorrhea should be
than one prior induced abortion did not increase the risk of
treated as having a potential EP.
EP in this population. Past studies have reported mixed
results (14, 15, 23, 24, 26, 28, 36 – 40), with the meta- An interesting novel finding of this study is the evaluation
analysis estimating a borderline, though statistically signifi- of the association of past and concomitant pelvic infections
cantly increased risk of EP with a history of induced abortion with the risk of EP. Although we found that a past history of
42 Barnhart et al. Risk factors predictive of ectopic pregnancy Vol. 86, No. 1, July 2006
abortion. Tokyo, Japan: Family Planning Association of Japan, 40. Savolainen E, Saksela E. Ectopic pregnancy: relationship to the pre-
1966:49 – 63. ceding contraception. Ann Chir Gynaecol 1978;67:198 –202.
38. Chow WH, Daling JR, Cates W Jr, Greenberg RS. Epidemiology of 41. Mol BWJ, Hajenius PJ, Engelsbel S, Ankum WM, Van der Veen F,
ectopic pregnancy. Epidemiol Rev 1987;9:70 –94. Hemrika DJ, et al. Serum human chorionic gonadotrophin measurement
39. Eskes TK, van Oppen AC. Ectopic pregnancy: not only a tubal disease. in the diagnosis of ectopic pregnancy when trans-vaginal sonography is
Eur J Obstet Gynecol Reprod Biol 1984;18:391– 4. inconclusive. Fertil Steril 1998;70:972– 81.