Body Mass Index and Risk for Oral Contraceptive Failure:
A Case–Cohort Study in South Carolina
LARISSA R. BRUNNER HUBER, PHD, CAROL J. HOGUE, PHD, ARYEH D. STEIN, PHD,
CAROLYN DREWS, PHD, AND MIRIAM ZIEMAN, MD
PURPOSE: Studies have suggested that obesity is associated with an increased risk for oral contraceptive
(OC) failure. We conducted a case–cohort study in South Carolina to examine the association between
body mass index (BMI) and OC failure by using population-based data sources.
METHODS: Our cohort sample from the source population consists of 205 women who reported using
OCs to prevent pregnancy on the 1999 Behavioral Risk Factor Surveillance System survey. The 153 women
who reported using OCs at the time of conception on the 2000 Pregnancy Risk Assessment Monitoring
System survey represent the case sample that arose from the source population. Logistic regression was
used to obtain odds ratios (ORs) and 95% confidence intervals (CIs).
RESULTS: In unadjusted models with normal BMI (20 to 24.9 kg/m2) as the comparison, greater BMI was
associated significantly with OC failure (overweight [25 to 29.9 kg/m2], OR Z 2.54; 95% CI, 1.18–5.50; and
obese [>30 kg/m2], OR Z 2.82; 95% CI, 1.05–7.58). After adjustment for education, income, and race/ethnicity, associations were attenuated and no longer statistically significant.
CONCLUSIONS: In this heterogeneous population, we found a suggestion that overweight and obese
women may be at increased risk for OC failure. However, long-term prospective studies are needed to study
this association in diverse populations.
Ann Epidemiol 2006;16:637–643. Ó 2006 Elsevier Inc. All rights reserved.
KEY WORDS:
Oral Contraceptives, Obesity, Women’s Health.
INTRODUCTION
Each year since 1988, nearly three million of the six million
pregnancies occurring in the United States annually have
been classified as unintended (1, 2), with nearly half these
unintended pregnancies occurring in the 90% of sexually active women of reproductive age who use some type of contraceptive (3). Although researchers attributed these
contraceptive failures to noncompliance, ineffective use,
and discontinuation (2, 4–6), few studies investigated
whether biologic factors, including body weight, may be
involved.
Secondary analyses of efficacy trials of Norplant (WyethAyerst Labs., St. David’s, PA) and the transdermal patch
suggest that greater body weight is associated with increased
risk for contraceptive failure (7–10). More recently, Holt
et al. (11, 12) showed an association between greater body
weight and increased risk for oral contraceptive (OC) failure
in two separate studies of residents of Washington State who
were enrolled in a large health maintenance organization.
Whereas Norplant is a progestin-only method of contraception in which progestin is released daily, OCs and the transdermal patch contain both estrogen and progestin. The
transdermal patch is similar to OCs in that a woman receives
a dose of estrogen and progestin daily for 21 days, followed
by a dose-free week.
The studies by Holt et al. (11, 12) may not be applicable
to the general population because participants were mainly
white and of higher socioeconomic status. The present study
investigates the association between body mass index (BMI)
and OC failure in a more diverse population by using two
population-based data sources.
METHODS
From the Department of Epidemiology, Emory University, Charlotte,
NC.
Address correspondence to: Larissa R. Bruner Huber, PhD, The University of North Carolina at Charlotte, 9201 University City Boulevard, Charlotte, NC 28223-0001. Tel.: (704) 687-6190; fax: (704) 687-6122. E-mail:
[email protected].
Research for this paper was partially funded through HRSA grant no. 2
T02 MC 00003-04 0, Dissertation Support for Applied Maternal and Child
Health Epidemiology; and the Sigma Delta Epsilon Fellowship, awarded by
Sigma Delta Epsilon/Graduate Women in Science.
Received August 8, 2005; accepted December 15, 2005.
Ó 2006 Elsevier Inc. All rights reserved.
360 Park Avenue South, New York, NY 10010
We conducted a case–cohort study of the association between BMI and risk for OC failure by linking data from
the 1999 Behavioral Risk Factor Surveillance System
(BRFSS) and 2000 Pregnancy Risk Assessment Monitoring
System (PRAMS) for South Carolina. A case–cohort study
incorporates features of both the case–control and cohort
designs. As in a traditional case–control design, cases in
a case–cohort study represent incident cases. However, in
1047-2797/06/$–see front matter
doi:10.1016/j.annepidem.2006.01.001
638
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
Selected Abbreviations and Acronyms
BRFSS Z Behavioral Risk Factor Surveillance System
BMI Z body mass index
CI Z confidence interval
OR Z odds ratio
OC Z oral contraceptive
PRAMS Z Pregnancy Risk Assessment Monitoring System
a case–cohort design, ‘‘controls’’ (i.e., the cohort) are sampled from the population at risk at the start of the risk period,
not from the population at risk at the end of the risk period,
as with a traditional case–control study.
BRFSS and PRAMS provide population-based estimates
of the prevalence of OC use in the general population and in
women reporting a live birth, respectively. Because these
two ongoing surveillance systems use random sampling to
identify the respondent sample, results may be extrapolated
to the general population. Our cohort sample consists of
women who reported using OCs to prevent pregnancy in
1999, and our case sample includes women who indicated
that they had an OC failure in 2000. We assume that both
the cohort and case samples are representative of the source
population from which the cases arose. Thus, the cohort provides information on the distribution of BMI in the source
population, and the lag in calendar time accounts for the duration of pregnancy.
Identification of OC Users in the General Population
The cohort includes nonpregnant women between the ages
of 18 and 45 years from South Carolina who responded to
the BRFSS survey and indicated they currently were using
OCs to prevent pregnancy. BRFSS is an ongoing state-based
telephone surveillance system that collects data on behaviors and conditions that place adults at high risk for the
chronic diseases, injuries, and preventable infectious diseases that are the leading causes of morbidity and mortality
in the United States (13). In 1999, a total of 3468 individuals from South Carolina participated in BRFSS, with
a state-specific survey response rate of 59.4% (response
rate across all states, 44.5% to 95.1%). A total of 223 OC
users represent our population at risk (14–18).
Identification of OC Users Among Women
Delivering Live Births
Cases are South Carolina women delivering live-born infants who indicated they were using OCs at the time of conception. PRAMS is a population-based survey of women
who delivered live-born infants (19). Each state that participates in PRAMS draws a sample of 100 to 250 new mothers
every month from a frame of eligible birth certificates. From
2 to 6 months after delivery, women are mailed self-
AEP Vol. 16, No. 8
August 2006: 637–643
administered questionnaires that collect information on
their experiences and behaviors before, during, and shortly
after pregnancy. Additional information is collected from
birth certificates. In 2000, a total of 2195 women from South
Carolina participated in PRAMS. This participation reflects
a response rate of 74.7% (response rate across participating
states, 63.1% to 82.0%). Cases are the 179 PRAMS respondents who reported using OCs at the time of conception.
Rationale for Use of Specific Data Sets
South Carolina is the only state that includes a question on
the specific type of contraceptive used at the time of conception in its PRAMS surveillance. Although this question has
been included in the South Carolina PRAMS for multiple
years, we were unable to use multiple years for this study because BRFSS questions on contraception, available as optional modules since the early 1990s, were not included in
the South Carolina BRFSS until 1999. As a result, we
were limited to using data from the 1999 BRFSS and the
2000 PRAMS questionnaires.
Measurement of Exposure and Covariates
Self-reported prepregnancy body weight and height were
used to calculate BMI (in kilograms per square meter). We
considered age, marital status, education, income, race/ethnicity, and smoking as potential confounding factors (4, 5).
Information for these variables was obtained for members of
the cohort and cases from the BRFSS and PRAMS data sets,
respectively.
Analysis
Women were excluded if the record had missing measurements for height (n Z 1), weight (n Z 17), or smoking
(n Z 3). Additionally, women not aged between 18 and
45 years when they gave birth (n Z 17) or with race/ethnicity other than black or white (n Z 6) were excluded. Thus,
358 women (153 cases, 205 cohort) remained for analysis.
Unadjusted odds ratios (ORs) and 95% confidence intervals (CIs) were obtained by using logistic regression to provide a crude association of BMI with OC failure and identify
other risk factors for OC failure. Potential confounding factors that altered the BMI–OC failure OR estimates by 10%
or more were included in the final logistic models (20). Ultimately, education, income, and race/ethnicity were confirmed as confounders in this data set. When BMI was
considered as a dichotomous variable, age also met criteria
for being a confounder. These confounding factors were
treated as categorical variables. We used multivariable logistic regression to obtain adjusted ORs and 95% CIs to model
the association between BMI and OC failure while accounting for confounding.
AEP Vol. 16, No. 8
August 2006: 637–643
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
Because of the use of the case–cohort study design, ORs
may be interpreted as risk ratios, without the need for
a rare disease assumption (14–18). We made the assumption
that none of our cases appeared in our cohort. We believe we
were justified in this assumption because it is highly unlikely
that a woman would have been selected randomly to participate in both the 1999 BRFSS and 2000 PRAMS. Thus, we
did not need to use alternative formulae for obtaining variance estimates to adjust for this overlap. SUDAAN Software for the Statistical Analysis of Correlated Data,
release 8.0.2 (Research Triangle Institute; Research Triangle Park, NC) was used in all analyses to account for weights
used in PRAMS and BRFSS to adjust for differences in probability of selection, nonresponse, and noncoverage.
RESULTS
Women reporting an OC failure were younger than the cohort of OC users (Table 1). In addition, women reporting an
OC failure were more likely than the cohort of OC users to
have yearly incomes less than $35,000, have less education,
and be black. Women were more likely to be overweight or
TABLE 1. Risk for oral contraceptive failure in relation
to selected characteristics, South Carolina Pregnancy Risk
Assessment Monitoring System and Behavioral Risk
Factor Surveillance System
Characteristic
Age (years)
18–30
O30
Marital status
Married
Other
Education
<High school
OHigh school
Income ($)
!35,000
>35,000
Missing
Race/ethnicity
White
Black
Smoking status
Nonsmoker
Smoker
Body mass index (kg/m2)
!20
20–24.9
25–29.9
>30
Cases
(n Z 153)a
Odds ratio
(95% confidence
interval)
Cohort
(n Z 205)
129 (85.6)
24 (14.5)
117 (62.5)
88 (37.5)
3.55 (1.53–8.27)
1.00 (referent)
61 (46.5)
92 (53.5)
108 (53.4)
97 (46.6)
1.00 (referent)
1.32 (0.70–2.47)
97 (61.9)
56 (38.1)
68 (36.7)
137 (63.3)
2.80 (1.45–5.41)
1.00 (referent)
119 (80.2)
19 (9.0)
15 (10.8)
83 (39.1)
95 (47.4)
27 (13.5)
10.82 (3.94–29.70)
1.00 (referent)
4.23 (1.16–15.46)
69 (51.2)
84 (48.8)
156 (74.1)
49 (26.0)
1.00 (referent)
2.72 (1.39–5.35)
120 (73.0)
33 (27.0)
169 (82.5)
36 (17.6)
1.00 (referent)
1.73 (0.85–3.55)
31 (18.4)
50 (30.1)
44 (35.5)
28 (16.0)
39 (20.3)
102 (48.3)
45 (22.4)
19 (9.1)
1.46 (0.58–3.64)
1.00 (referent)
2.54 (1.18–5.50)
2.82 (1.05–7.58)
Values expressed as number (percent) unless noted otherwise.
a
Weighted percents may not total 100 because of rounding.
639
obese if they had yearly incomes less than $35,000 or were
black (Table 2).
In unadjusted models (Table 1), women with BMIs in the
overweight or obese range had statistically significant increased risks for OC failure compared with women in
the normal-BMI range. Overweight women had an
increased risk of 2.54 (95% CI, 1.18–5.50), and obese
women had nearly a three-fold risk (OR Z 2.82; 95% CI,
1.05–7.58).
When adjusted for education, income, and race/ethnicity, the risk for overweight and obese women was attenuated
and no longer statistically significant (Table 3). Overweight
women had nearly twice the risk for OC failure (OR Z 1.87;
95% CI, 0.73–4.78), whereas obese women had 1.58 times
the risk for OC failure compared with women in the normal
BMI range (95% CI, 0.49–5.10). When BMI was dichotomized, results were similar. Women who were overweight
or obese had 1.90 times the risk for OC failure after adjustment for education, income, race/ethnicity, and age (95%
CI, 0.82–4.41).
When we stratified by education and income, we found
no indication of effect modification by these factors (data
not shown). Analyses stratified by race/ethnicity found
that in white women, overweight and obese individuals
had 2.64 times the risk for OC failure (95% CI, 1.16–
6.05) compared with normal or underweight individuals.
However, in black women, overweight and obese individuals had only a slight increased risk for OC failure (OR Z
1.29; 95% CI, 0.43–3.88).
TABLE 2. Association between selected demographic
characteristics and overweight/obesity, South Carolina
Pregnancy Risk Assessment Monitoring System and
Behavioral Risk Factor Surveillance System
Characteristic
Age (years)
18–30
O30
Marital status
Married
Other
Education
<High school
OHigh school
Income ($)
!35,000
>35,000
Missing
Race/ethnicity
White
Black
Smoking status
Nonsmoker
Smoker
Odds ratio (95% confidence interval)
0.88 (0.47–1.65)
1.00 (referent)
1.00 (referent)
1.41 (0.74–2.69)
1.33 (0.68–2.62)
1.00 (referent)
3.38 (1.62–7.06)
1.00 (referent)
2.59 (0.44–15.21)
1.00 (referent)
2.96 (1.32–6.61)
1.00 (referent)
1.91 (0.87–4.20)
Overweight/obesity is considered to be a body mass index of 25 kg/m2 or greater.
640
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
TABLE 3. Risk for oral contraceptive failure in
relation to body mass index, South Carolina Pregnancy Risk
Assessment Monitoring System and Behavioral Risk
Factor Surveillance System
Characteristic
Body mass index (kg/m2)
!20
20–24.9
25–29.9
>30
!25
O25
Multivariate-adjusted odds ratioa
(95% confidence interval)
1.07 (0.31–3.73)
1.00 (referent)
1.87 (0.73–4.78)
1.58 (0.49–5.10)
1.00 (referent)
1.90 (0.82–4.41)
a
Odds ratio adjusted for education, race/ethnicity, and income. When body mass
index was dichotomized, the odds ratio also was adjusted for age.
DISCUSSION
We found an association between greater BMI and OC failure in this population-based study. However, the associations were attenuated and lost statistical significance after
adjustment for education, income, and race/ethnicity.
Research showed that low socioeconomic status and
black race/ethnicity were both associated independently
with greater BMI in women (21–23). Additional research
into the obesity–socioeconomic status relationship also suggested that obesity may influence socioeconomic status (24,
25). In the case of a possible obesity–OC failure association,
education, income, and race/ethnicity are related closely to
the exposure of interest. Thus, the correlation between these
variables may be so strong as to result in overadjustment,
thereby obscuring the true obesity–OC failure association.
Alternatively, it is possible that obesity influences socioeconomic status. If this is the case, such factors as education and
income may be on the causal pathway between obesity and
OC failure. Consequently, control for these factors would be
improper.
To evaluate whether adjustment for socioeconomic factors represented overcontrol, we attempted to build a twostage regression model in which the first stage represents a
regression of obesity on education, income, and race/ethnicity. We then regressed OC failure on the residual from
that model. However, because all three predictors are categorical, the resulting predicted obesity has limited number
of discrete values; hence, there was little reduction in variation between the observed obesity and residuals of the
first-stage regression. Therefore, this approach was not helpful in differentiating between adjusted and unadjusted
effects (data not shown).
This study has several limitations. Nondifferential misclassification of the exposure is possible because height
and weight were self-reported by study participants, and
PRAMS respondents were recalling their prepregnancy
AEP Vol. 16, No. 8
August 2006: 637–643
measurements. A number of studies indicated that self-reported height and weight give an accurate representation
of true height and weight because women tend to slightly
overreport their height and underreport their weight by
a few pounds (26). In addition, studies found that women
can accurately recall past body weight (27, 28). Thus, BMI
derived from self-report will underestimate true BMI, particularly for women who are overweight or obese because these
women tend to underreport their weight more than women
of normal weight. Any resulting nondifferential misclassification is likely to bias the results toward the null, although
bias away from the null is possible because the exposure category is polytomous. However, because there are a sufficient
number of women of normal, overweight, and obese BMI
who did not have an OC failure, this possibility is minimized
(29).
As mentioned, PRAMS samples are from women who delivered live-born infants. Thus, the present study lacks information on women who may have had an OC failure, but had
their pregnancy end in an outcome other than a live birth.
To our knowledge, no literature investigates whether BMI
is related to pregnancy outcome in women who experienced
a contraceptive failure. To explore this possibility further,
we used data from the 1995 National Survey of Family
Growth, a survey of US women between the ages of 15
and 44 years. Mean prepregnancy BMI in 390 women who
reported having a contraceptive failure was 24.0 kg/m2 in
women experiencing a live birth, 23.2 kg/m2 in women
who underwent an induced abortion, and 24.9 kg/m2 in
those who experienced spontaneous abortion (p Z 0.28).
Based on this analysis, there was no indication that prepregnancy BMI was related to pregnancy outcome in women
who experienced a contraceptive failure. Thus, it seems unlikely that PRAMS data would contain a disproportionate
number of OC failures ending in live births among women
of a certain BMI category.
Use of standard questionnaires and trained interviewers
limits the potential for information bias. However, PRAMS
and BRFSS collect data by using two different methods, and
the two surveillance systems have different response rates.
These differences in data-collection and response rates
may have unpredictable effects on estimates if a participant’s
willingness to respond is related to BMI or OC failure.
Although we controlled for a number of variables associated with both BMI and OC failure, we were not able to control for all potential confounding factors because we were
limited by the questions asked in PRAMS and BRFSS.
We lacked information on underlying fecundity, parity,
dual-method use, frequency of intercourse, and adherence
to an optimal OC regimen. To the extent that these variables may be related to weight status, failure to control for
these variables could result in an underestimate or overestimate of the true association between BMI and OC failure.
AEP Vol. 16, No. 8
August 2006: 637–643
With respect to possible confounding by frequency of intercourse or adherence to an OC regimen, sensitivity analyses
that assumed various estimates for the OC failure–frequency
of intercourse and OC failure–adherence associations did
not indicate that either variable was a confounder of the
BMI–OC failure association. In addition, a recent longitudinal study of OC users found no association between BMI and
frequency of intercourse or adherence (30).
Additionally, PRAMS and BRFSS do not collect information on the estrogen content of pills, a potential effect
modifier of the BMI–OC failure association. However, since
the late 1980s, most prescriptions for OCs have been for
preparations containing 35 mg or less per dose (31, 32). In
1998, approximately 90% of OC prescriptions were written
for preparations containing 30 or 35 mg of estrogen (31).
Thus, estrogen content is not likely to be an effect modifier
in our study. However, the growing popularity of very-lowdose OC preparations (31) could make estrogen dose an effect modifier of the BMI–OC failure association in future
studies.
Previous literature on body weight and risk for OC failure
is sparse. A study of women who participated in the Oxford
Family Planning Association contraceptive study found no
association between body weight and OC failure (33). To
be eligible, women had to be between 25 and 39 years of
age, married, white, British, and a current user of OCs, a diaphragm, or an intrauterine device. It is unclear from the
published report whether self-reported weight at the time
of recruitment in 1968 to 1974 was used to reflect a woman’s
weight when she had an OC failure. Additionally, the investigators did not discuss whether ‘‘accidental pregnancy’’ referred only to live births or whether this also encompassed
spontaneous and induced abortions. Furthermore, 75% of
the exposure to OCs in this study was to preparations containing 50 mg or greater of estrogen (34). OCs containing
that much estrogen rarely have been used in the last two
decades.
In 2002, Holt et al. (11) investigated the association between body weight and risk for OC failure in a retrospective
cohort of 618 women in Washington State. After adjustment for parity, they found that OC users who weighed
70.5 kg or greater (>155 pounds) had 1.6 times the risk
for OC failure compared with OC users who weighed less
than 70.5 kg (95% CI, 1.1–2.4). However, the investigators
were not able to control for several major potential confounding factors, including fecundity, adherence, poverty,
and frequency of intercourse.
Holt et al. (12) recently published results of a case–control study (248 cases, 533 controls) conducted among OC
users enrolled at a health maintenance organization in
Washington State. During in-person interviews, the investigators were able to collect information on frequency of intercourse and adherence to an OC regimen. These two
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
641
variables did not ultimately meet the investigators’ criteria
for being a confounder and were not included in their final
models. The investigators found that women with a BMI
greater than 27.3 kg/m2 had 1.58 times the risk for having
an OC failure compared with women with a BMI less than
27.3 kg/m2 (95% CI, 1.11–2.24) after adjustment for age,
parity, and reference year.
Brunner and Hogue (35) used 1995 National Survey of
Family Growth data to examine the role of obesity in OC
failure in 1916 women (35). In this study, obese women
(BMI > 30 kg/m2) had nearly twice the risk for OC failure
compared with women in the 20- to 24.9-kg/m2 BMI category (hazard ratio Z 1.80; 95% CI, 1.01–3.20). However, after adjustment for age, marital status, education, poverty,
race/ethnicity, parity, and dual-method use (use of a contraceptive method in addition to OCs), this increased risk was
attenuated and no longer statistically significant (hazard
ratio Z 1.51; 95% CI, 0.81–2.82). Although the investigators had limited information on adherence to an OC regimen, a secondary analysis conducted among 1995
National Survey of Family Growth respondents who
answered a series of adherence questions found no indication that BMI was associated with adherence.
Although it is possible that there is no association between obesity and OC failure, there are biologic factors
that could contribute to greater failure rates in heavier
women. One factor is that adipose tissue is metabolically active and is a site of steroid hormone storage. In addition, possible increased volume of distribution in larger women and
the sequestration of contraceptive hormones into adipose
tissue may decrease serum hormone levels to less than the
therapeutic threshold (7–12). Alternatively, adipose tissue
is a significant site of estrogen production. Obese women
in general have elevated levels of estrogen and relatively
low levels of sex-hormone–binding globulin, the protein
that binds estrogen (36–38). Low levels of this protein result
in a high concentration of free estrogen in the circulation. It
is possible that these changes in estrogen metabolism may
interfere with OC effectiveness through undetermined
mechanisms. It has been postulated that the half-life of
OCs could be shortened by either the increased metabolic
rate of heavier women (39–41) or the increased clearance
of all hepatically metabolized drugs, including OCs, in
heavier women (42, 43). One or both of these mechanisms
may cause insufficient hormone levels to maintain adequate
contraceptive efficacy.
These proposed mechanisms may contribute to increased
contraceptive failure rates, such as those observed in secondary analyses of efficacy studies of Norplant and the transdermal contraceptive patch, which found some indication that
increasing body weight is associated with increased risk for
contraceptive failure (7–10). However, the observed associations were based on small numbers and were unadjusted for
642
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
potential confounding factors. Moreover, it is interesting
to note that both these contraceptive methods deliver
a steady-state concentration of hormones to the bloodstream. Perhaps this type of drug delivery and metabolism
is most susceptible to the previously mentioned biologic
mechanisms.
In conclusion, we found an association between greater
BMI and OC failure. Although the associations were attenuated after adjustment for education, income, and race/ethnicity, they were similar in magnitude to other studies of the
obesity–OC failure association. In addition, when we stratified on race/ethnicity, we confirmed associations seen in predominantly white populations (11, 12). Although risks for
black women also were slightly increased, conclusions about
risks for black women were hindered by a small sample size.
Unlike other studies of the obesity–OC failure association, the present study used population-based data sets and
examined the association in a more heterogeneous population in terms of race/ethnicity and education. Although it
is possible that there is no association between BMI and
OC failure, it appears that body weight may have a biologically relevant role. Large prospective studies are needed in
diverse populations to adequately address the obesity–OC
failure association. In addition, this study highlights the
complexities related to examining this association among
women of differing race/ethnicity and socioeconomic status.
Future studies need to be large enough to allow for stratified
analysis by these variables. If an obesity–OC failure association is confirmed in prospective studies, pharmacologic studies that detail how OCs are metabolized in women of varying
BMIs will be necessary to determine the exact biologic
mechanism involved.
REFERENCES
AEP Vol. 16, No. 8
August 2006: 637–643
8. Du MK, Chow LP, Zheng HM, Chen CHC. A 10-year follow-up study
of contraceptive Norplant implants. Int J Gynaecol Obstet. 2000;68:
249–256.
9. Audet MC, Moureau M, Koltun WD, Waldbaum AS, Shangold G, Fisher
AC, et al. Evaluation of contraceptive efficacy and cycle control of a transdermal contraceptive patch vs. an oral contraceptive: A randomized controlled trial. JAMA. 2001;285:2347–2354.
10. Zieman M, Guillebard J, Weisberg E, Shangold GA, Fisher AC, Creasy
GW. Contraceptive efficacy and cycle control with the Ortho Evra/Evra
transdermal system: The analysis of pooled data. Fertil Steril.
2002;77(Suppl):S13–18.
11. Holt VL, Cushing-Haugen KL, Daling JR. Body weight and risk of oral
contraceptive failure. Obstet Gynecol. 2002;99:820–827.
12. Holt VL, Scholes D, Wicklund KG, Cushing-Haugen KL, Daling JR. Body
mass index, weight, and oral contraceptive failure risk. Obstet Gynecol.
2005;105:46–52.
13. Centers for Disease Control and Prevention. Behavioral Risk Factor
Surveillance System User’s Guide. Atlanta, GA: US Department of
Health and Human Services, Centers for Disease Control and Prevention;
1998.
14. Kupper LL, McMichael AJ, Spirtas R. A hybrid epidemiologic study design
useful in estimating relative risk. J Am Stat Assoc.. 1975;70:524–528.
15. Miettinen OS. Design options in epidemiologic research: An update.
Scand J Work Environ Health. 1982;8(Suppl 1):S7–14.
16. Hogue CJR, Gaylor DW, Schulz KF. Estimators of relative risk for casecontrol studies. Am J Epidemiol. 1983;118:396–407.
17. Hogue CJR, Gaylor DW, Schulz KF. The case-exposure study: A further
explication and response to a critique. Am J Epidemol. 1986;124:877–
883.
18. Prentice RL. A case-cohort design for epidemiologic cohort studies and
disease prevention trials. Biometrika. 1986;73:1–11.
19. Lipscomb LE, Johnson CH, Morrow B, Colley Gilbert B, Ahluwalia IB,
Beck LF, et al. PRAMS 1998 Surveillance Report. Atlanta, GA: Division
of Reproductive Health, National Center for Chronic Disease Prevention
and Health Promotion, Centers for Disease Control and Prevention; 2000.
20. Maldonado G, Greenland S. Simulation study of confounder selection
strategies. Am J Epidemiol. 1993;138:923–936.
21. Harrell JS, Gore SV. Cardiovascular risk factors and socioeconomic status
in African American and Caucasian women. Res Nurs Health.
1998;21:285–295.
22. Sundquist J, Johannson SE. The influence of socioeconomic status, ethnicity and lifestyle on body mass index in a longitudinal study. Int J Epidemiol. 1998;27:57–63.
1. Henshaw SK. Unintended pregnancy in the United States. Fam Plann
Perspect. 1998;30:24–29 46.
23. Robert SA, Reither EN. A multilevel analysis of race, community disadvantage, and body mass index among adults in the US. Soc Sci Med.
2004;59:2421–2434.
2. Forrest JD. The role of hormonal contraceptives: Epidemiology of unintended pregnancy and contraceptive use. Am J Obstet Gynecol.
1994;170:1485–1489.
24. Gortmaker SL, Must A, Perrin JM, Sobol AM, Dietz WH. Social and economic consequences of overweight in adolescence and young adulthood. N
Engl J Med. 1993;329:1008–1012.
3. Harlap S, Kost K, Forrest JD. Preventing Pregnancy, Protecting Health:
A New Look at Birth Control in the United States. New York: Alan
Guttmacher Institute; 1991.
25. Stunkard AJ, Sorensen TIA. Obesity and socioeconomic status: A complex relation. N Engl J Med. 1993;329:1036–1037.
4. Ranjit N, Bankole A, Darroch JE, Singh S. Contraceptive failure in the
first two years of use: Differences across socioeconomic subgroups. Fam
Plann Perspect. 2001;33:19–27.
5. Fu H, Darroch JE, Haas T, Ranjit N. Contraceptive failure rates: New estimates from the 1995 National Survey of Family Growth. Fam Plann Perspect. 1999;31:56–63.
26. Engstrom JL, Paterson SA, Doherty A, Trabulsi M, Speer KL. Accuracy of
self-reported height and weight in women: An integrative review of the
literature. J Midwifery Womens Health. 2003;43:338–345.
27. Perry GS, Byers TE, Mokdad AH, Serdula MK, Williamson DF. The validity of self-reports of past body weights by US adults. Epidemiology.
1995;6:61–66.
6. Hillard PJA. Oral contraception noncompliance: The extent of the problem. Adv Contracept. 1992;8(Suppl 1):S13–20.
28. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC.
The validity of recalled weight among younger women. Int J Obes Relat
Metab Disord. 1995;19:570–572.
7. Gu S, Sivin I, Du M, Zhang L, Ying L, Meng F, et al. Effectiveness of Norplant implants through seven years: A large-scale study in China. Contraception. 1995;52:99–103.
29. Correa-Villasenor A, Stewart WF, Franco-Marina F, Seacat H. Bias from
nondifferential misclassification in case-control studies with three exposure levels. Epidemiology. 1995;6:276–281.
AEP Vol. 16, No. 8
August 2006: 637–643
Brunner Huber et al.
BODY MASS INDEX AND ORAL CONTRACEPTIVE FAILURE
643
30. Huber LRB, Hogue CJ, Stein AD, Drews C, Zieman M, King J, et al., Contraceptive use and discontinuation: Findings from the Contraceptive History, Initiation, and Choice Study. Am J Obstet Gynecol. In press.
36. Moran LJ, Norman RJ. The obese patient with infertility: A practical approach to diagnosis and treatment. Nutr Clin Care. 2002;6:290–297.
31. Grimes DA (ed). Trends in oral contraceptive prescribing: An update.
Contracept Rep. 1999;10:12–14.
38. Frisch RE. Body fat, menarche, fitness, and fertility. Hum Reprod.
1987;2:521–533.
32. Gerstman BB, Gross TP, Kennedy DL, Bennett RC, Tomita DK, Stadel
BV. Trends in the content and use of oral contraceptives in the United
States, 1964-88. Am J Public Health. 1991;81:90–98.
39. Ravussin E, Burnand B, Schutz Y, Jequier E. Twenty-four hour energy expenditure and resting metabolic rate in obese, moderately obese, and control subjects. Am J Clin Nutr. 1982;35:566–573.
33. Vessey M, Painter R. Oral contraceptive failures and body weight: Findings
in a large cohort study. J Fam Plann Reprod Health Care. 2001;27:90–91.
40. Owen OE, Kavle E, Owen RS, et al. A reappraisal of caloric requirements
in healthy women. Am J Clin Nutr. 1986;44:1–19.
34. Vessey M, Painter R, Mant J. Oral contraception and other factors in relation to back disorders in women: Findings in a large cohort study. Contraception. 2000;60:331–335.
41. Dietz WH. Obesity. J Am Coll Nutr. 1989;8(Suppl):S13–21.
35. Brunner LR, Hogue CJ. The role of body weight in oral contraceptive
failure: Results from the 1995 National Survey of Family Growth. Ann
Epidemiol. 2005;15:492–499.
37. Frisch RE. Fatness and fertility. Sci Am.. 1988;258:88–95.
42. Abernathy DR, Greenblatt DJ. Drug disposition in obese humans: An update. Clin Pharmacokinet. 1986;11:199–213.
43. Speerhas R. Drug metabolism in malnutrition and obesity: Clinical concerns. Cleve Clin J Med. 1995;62:73–75.