Prospective Study of Oral Contraceptives and Hypertension Among Women in the United States
Background Oral contraceptives induce hypertension in approximately 5% of users of high-dose pills that contain at least 50 μg estrogen and 1 to 4 mg progestin, and small increases in blood pressure have been reported even among users of modern low-dose formulations. However, neither the responsible hormone in the oral contraceptive nor particular subgroups of women who might be susceptible to the hypertensive effect of oral contraceptives have been identified.
Methods and Results In a prospective cohort study in the United States, 68 297 female nurses aged 25 to 42 years and free of diagnosed hypertension, diabetes, coronary heart disease, stroke, and cancer at baseline were followed up for 4 years. During 231 006 person-years of follow-up, 1567 incident cases of hypertension were diagnosed. Compared with women who had never used oral contraceptives, the age-adjusted relative risk was 1.5 (95% CI=1.2 to 1.8) for current use and 1.1 (95% CI=0.9 to 1.2) for past use. After adjustment for age, body mass index, cigarette smoking, family history of hypertension, parity, physical activity, alcohol intake, and ethnicity, current users of oral contraceptives had an increased risk of development of hypertension (RR=1.8; 95% CI=1.5 to 2.3) compared with women who had never used them. The multivariate relative risk for past users was 1.2 (95% CI=1.0 to 1.4). There were no important modifying effects of age, family history of hypertension, ethnicity, or body mass index.
Conclusions Current users of oral contraceptives had a significant, moderately increased risk of hypertension. However, among this group, only 41.5 cases per 10 000 person-years could be attributed to oral contraceptive use. Risk decreased quickly with cessation of oral contraceptives, and past users appeared to have only a slightly increased risk.
Oral contraceptives induce hypertension in approximately 5% of users of high-dose pills that contain at least 50 μg estrogen and 1 to 4 mg progestin.1 2 3 4 The formulations of OCs have changed markedly since their introduction more than 30 years ago; current products contain as little as 20% of the dose of estrogen and less than 16% of the progestin contained in earlier preparations. However, small increases in blood pressure have been reported even with monophasic pills that contain 30 μg estrogen.4 5 6 Neither the responsible hormone in the OC nor particular subgroups of women who might be susceptible to the hypertensive effect of OCs have been identified. However, women with a history of high blood pressure during pregnancy, those with a family history of hypertension, and black women may respond to the hormonal stimulus of OCs with a greater increase in blood pressure than other groups of women.2 7 8
Previous prospective cohort studies did not include long-term users of low-dose estrogen-progestin combinations and progestin-only formulations.1 2 3 4 7 8 9 Although OC use declined in American women from 1973 to 1982, use increased among the youngest age group (aged 12 to 19 years) from the 1970s to the early 1980s.10 Because this group of early users will also tend to have a longer duration of use, it is important to determine whether such use has an impact on risk of increased blood pressure. We studied the relationship between OC use and risk of hypertension prospectively among the 116 678 participants in the Nurses' Health Study II, who were aged 25 to 42 years when enrolled in 1989.
NHS II is an ongoing prospective cohort study designed to examine associations between lifestyle and diet and the occurrence of breast cancer and other major illnesses. The participants are 116 678 female registered nurses aged 25 to 42 years who were living in 1 of 14 states when they responded to a baseline questionnaire in 1989. Follow-up questionnaires were sent in 1991 and 1993, with response rates of 93% and 92%, respectively.
We excluded from the analysis women who reported a diagnosis of hypertension (n=6535), cancer (except nonmelanoma skin cancer) (n=1046), myocardial infarction (n=574), stroke (n=321), or diabetes (n=1003) on or before 1989, the start of the follow-up period. Women who had not had a physical examination within the 2 years before the baseline questionnaire was completed (n=12 584) were excluded to avoid a potential bias due to the greater opportunity for hypertension diagnosis among OC users, because OC users may undergo routine blood pressure measurements at least every year. We also excluded women who at baseline reported current use of any antihypertensive medication, including thiazide diuretics, furosemide-like diuretics, and β-blockers (n=5900). Women who reported on the baseline questionnaire that they were currently pregnant or had been pregnant for at least 6 months within the previous 2 years were also excluded to avoid pregnancy-induced hypertension (n=28 149). One or more of the exclusion criteria were met by 48 381 of the participants, which left 68 297 eligible nurses who were followed up for hypertension incidence over the subsequent 4-year period.
At the beginning of the next 2-year time period (1991 to 1993), we updated information on OC use and excluded women who reported on the 1991 follow-up questionnaire any diagnosis of the above diseases since 1989 (n=490), because these women may have altered their OC use due to their illness. At that time, we also excluded women who reported that they were currently pregnant or had been pregnant for at least 6 months since 1989 (n=6123).
Measurement of OC Use and OtherExposure Variables
A complete history of OC use was obtained at baseline. For each year from age 13 to age at baseline, duration of OC use in two categories (2 months to <10 months and 10 months to a full year) was recorded. Less than 2 months of OC use was considered nonuse. A booklet with photos, names, and the pharmacological contents of the 227 OCs that had ever been marketed in the United States up to the time of the study was provided to the participants. This detailed list included separate codes for 21- versus 28-day pills with the same pharmacological preparation and dose and separate codes for changes in packaging of the same preparation and dose as well as for different pharmacological preparations or doses sold under the same brand name. For each age at which an OC was used for at least 2 months, we asked the nurses to indicate from the booklet which brand was used, and if multiple brands were used during that year of age, to indicate the brand that was used for the longest period of time. Information about subsequent use and brand of OC used was obtained from the next biennial follow-up questionnaire. We calculated the duration of each reported interval and then summed all of the intervals from both the baseline and follow-up questionnaire to ascertain duration of current and past use. We used the brand information to categorize OC use in terms of type, dose, biological potency of estrogen and progestin, and formulation (monophasic, biphasic, and triphasic combination or progestin only). We examined the relation between these variables and risk of hypertension in both current and past OC users.
In a review of the relative potency of progestins used in OCs, Dorflinger11 summarized the human data on delay of menses, glycogen deposition, and lipids. The potency approximations from all three techniques were remarkably similar, which indicated that norethindrone, norethindrone acetate, and ethynodiol diacetate are approximately equivalent in potency and that norgestrel is roughly 5 to 10 times and levonorgestrel 10 to 20 times the potency of norethindrone. We therefore used the following equivalency to order the OCs into three categories of progestational potency (low, medium, or high): 1 mg norethindrone=1 mg norethindrone acetate=1 mg ethynodiol diacetate=0.2 mg norgestrel=0.1 mg levonorgestrel. For norgestrel and levonorgestrel, these were the lower limits of potency.
Ethynyl estradiol has approximately 0.7 times the potency of mestranol; however, estrogenic potency depends not only on the dose and type of estrogen but also on the estrogenic activity of the type of progestin in the OC. We therefore used Dickey's method (as expanded by the American Medical Association Department of Drugs), which is based on the entire formulation of the OC, to classify the OCs into three categories of estrogenic potency (low [≤30 μg], medium [>30 to ≤50 μg], and high [>50 μg, in ethynyl estradiol equivalents]).12 13 14 Estrogenic potency was measured by mouse uterine assay.
Validation of OC History
We assessed the validity of self-reporting of OC history by conducting a detailed telephone interview of 215 women at least 8 months after completion of the baseline questionnaire (D.J.H., MBBS, et al, unpublished data, 1996). The interview used reproductive and other life events as cues for recall of contraceptive history and asked for the name of the brand used the most in any given year. Of 177 women who reported by telephone interview that they had ever used OCs, 175 (99%) had reported this on the baseline questionnaire. Of 38 never users by telephone interview, 37 (97%) reported never use on the baseline questionnaire. Mean duration of use among women who had ever used OCs was similar by questionnaire (mean±SD, 54.8±41.0 months) and telephone interview (mean±SD, 51.8±40.6 months), with a correlation of 0.94. For a subset of women for whom we were able to obtain OC prescription records (n=110), the medical records confirmed the use of an identical or equivalent brand in 74% of the intervals of reported use. However, this percentage is most likely an underestimation, because women may receive prescriptions for different brands from more than one source or may change brands or prescribers during an interval of use.
Blood Pressure and Diagnosis of Hypertension
In 1989, the nurses in the present study were asked to report their current usual blood pressure within the following categories: for systolic blood pressure, unknown or not checked within 2 years, <105, 105 to 114, 115 to 124, 125 to 134, 135 to 144, 145 to 154, 155 to 164, 165 to 174, or ≥175 mm Hg; and for diastolic blood pressure, unknown or not checked within 2 years, <65, 65 to 74, 75 to 84, 85 to 89, 90 to 94, 95 to 104, or ≥105 mm Hg. We also inquired whether subjects had been diagnosed as having high blood pressure before 1980, from 1980 to 1984, or from 1985 to 1989. On subsequent biennial questionnaires (1991 and 1993), we asked whether subjects had been diagnosed as having high blood pressure within the past 2 years.
Validation of Diagnosis of Hypertension
Blood pressure status was defined by self-reported responses to questionnaires. The self-reported diagnosis of hypertension was validated in the NHS I,15 a complementary cohort of 121 700 female nurses aged 30 to 55 years in 1976. The medical records of a sample of the NHS I nurses who reported a diagnosis of high blood pressure on the 1982 questionnaire were reviewed. Recorded blood pressure was >160/95 mm Hg in 39 (76%) of the 51 records reviewed and >140/90 in all of them. To investigate the likelihood of unreported high blood pressure, we measured blood pressure in an age-stratified sample of 194 participants who were living in the greater Boston area. Among 161 women without a previous self-report of high blood pressure, 11 (7%) had a blood pressure level >140/90 mm Hg, but none of these women had a blood pressure >160/95 mm Hg. Thus, the self-reported diagnosis of elevated blood pressure appeared to be a valid measure in this population of registered nurses.
We conducted two separate analyses: a cross-sectional analysis of self-reported blood pressure in 1989, and a prospective analysis of newly diagnosed hypertension during the 4 years of follow-up. For the cross-sectional analysis, we examined the associations between systolic and diastolic blood pressures reported in 1989 and OC use among those nurses not diagnosed with hypertension. We excluded an additional 89 women who did not know their blood pressure at baseline, which left 68 208 nurses for the cross-sectional analysis. We used multiple linear regression models to control for confounding by the following variables: age, BMI (weight in kilograms divided by the square of height in meters), cigarette smoking, alcohol intake, family history of hypertension, ethnic group, parity, and physical activity (metabolic equivalent score16 ). Quadratic terms for age and BMI were included to account for the nonlinearity of the age-BMI–blood pressure relation (see footnote to Table 1⇓ for categories of covariates).
For the prospective analysis, we used diagnosis of hypertension during the 4 years of follow-up as the outcome. We calculated the RR of hypertension by comparing the incidence rate of hypertension among women in each category of OC use with the corresponding rate in never users. For covariates that remained constant throughout the duration of the study, cases and person-years of follow-up were assigned to the exposure level observed at baseline in 1989. For covariates that varied with time (age, BMI, cigarette smoking, parity, physical activity, and alcohol intake) cases and person-years were reassigned after 2 years according to the updated exposure values reported on the next follow-up questionnaire. Proportional hazards models17 were used to control simultaneously for age, BMI, cigarette smoking, alcohol intake, family history of hypertension, ethnic group, parity, and physical activity (see footnote to Table 2⇓ for categories of covariates). Ethnicity, age, family history of hypertension, and BMI were strong predictors of hypertension in this population; therefore, we analyzed differences in risk associated with category of OC use within categories of these variables.
In addition, some analyses involved the examination of several aspects of OC use simultaneously: estrogenic potency, progestational potency, and duration of OC use. Because the referent category was the same for all of these variables (never users), we created cross-categories by crossing each level of each variable with the other. A summary RR estimate was then calculated by use of the inverse variance of the estimate within each cross-category to weight the odds ratio for each cross-category.
Women with higher baseline blood pressure have a greater risk of future hypertension. Because women with higher blood pressure at baseline may be less likely to have OCs prescribed for them, baseline blood pressure can be considered a potential confounder of the relation between OC use and hypertension. However, baseline blood pressure can also be considered an intermediate variable in that OC use could raise baseline blood pressure and thereby increase the risk of hypertension. We therefore did not include baseline systolic and diastolic blood pressures as independent variables in the primary prospective analyses, because their inclusion may cause an underestimation of the association between OC use and risk of future hypertension. However, to assess whether baseline blood pressure confounded the association with future hypertension, we conducted an additional analysis that included these variables in the multivariate model.
We calculated the 95% CI for each RR ratio.18 A single trend variable, coded as the category of exposure (1, 2, etc), was included to calculate tests of trend across categories of OC use. We calculated rate differences by subtracting the age-adjusted incidence rate for never users from the corresponding rate in each category of OC use. The population attributable risk was calculated by use of the formula given by Miettinen.19 All probability values are two-tailed.
During 231 006 person-years of follow-up, 1567 incident cases of hypertension were diagnosed. The 4-year cumulative incidence ranged from 0.5% for women aged 24 to 29 years to 1.9% for women aged 40 to 44 years. Among women without diagnosed hypertension, baseline mean systolic blood pressure was 111.9 mm Hg at age 25 to 29 and 113.8 mm Hg at age 45 to 49 years, whereas diastolic blood pressure was 69.8 mm Hg at age 25 to 29 and 72.0 mm Hg at age 45 to 49 years. The Pearson correlation coefficient between baseline systolic and diastolic pressures was .70 (P<.0001).
The cross-sectional analysis examined the relation between OC use and self-reported blood pressure at baseline. After adjustment for age, parity, alcohol intake, cigarette smoking, BMI, physical activity, ethnicity, and family history of hypertension, systolic blood pressure among current users of OCs was 0.7 mm Hg higher (95% CI=0.4 to 1.0) than among never users of OCs. The comparable difference for diastolic blood pressure was 0.4 mm Hg (95% CI=0.2 to 0.6). Among past users of OCs, systolic blood pressure was 0.2 mm Hg lower (95% CI= 0.03 to 0.4) and diastolic pressure was 0.3 mm Hg lower (95% CI=0.2 to 0.5) than among those who had never used OCs. Both systolic and diastolic pressures showed similar patterns with increase in the duration of current use. After an initial nonsignificant decrease in risk for those with <2 years of OC use, the statistically significant increase in blood pressure was fairly constant regardless of the increase in the duration of use (Table 1⇑).
The prospective analysis examined the relation between OC use and subsequent diagnosis of hypertension. After adjustment for age, BMI, cigarette smoking, family history of hypertension, parity, physical activity, alcohol intake, and ethnicity, current users of OCs had an increased risk of developing hypertension (RR=1.8; 95% CI=1.5 to 2.3) compared with women who had never used them (Table 2⇑). This RR was higher than the age-adjusted RR of 1.5 (95% CI=1.2 to 1.8), largely because of confounding by BMI; that is, OC users tended to be leaner than nonusers. However, among the current OC users, only 41.5 cases per 10 000 person-years could be attributed to OC use. The multivariate RR for past users was 1.2 (95% CI=1.0 to 1.4) compared with never users. When we repeated this analysis and controlled for baseline systolic and diastolic blood pressures, the RRs were virtually unchanged.
We then distinguished four categories of duration of current OC use (Table 3⇓), and as in the cross-sectional analysis, we did not find a significant trend for an increase in the risk of hypertension with an increase in the duration of OC use (Ptrend=.11). However, women with 6 or more years of OC use (RR=2.1; 95% CI=1.6 to 2.7) appeared to be at the highest increased risk. To examine whether the increased risk among women with longer duration of OC use was due to a higher prevalence of high-potency OC use among this group, we repeated the above analysis while controlling for duration and potency simultaneously. We found no attenuation in the RR, which suggested that a difference in the use of high-potency OCs could not explain the higher risk in long-term current users.
We examined the association between potency (see “Methods”) and risk of hypertension among current users of OCs (Table 4⇓). There was a suggestion of increase in risk with an increase in the potency of progestin, but this was not statistically significant (Ptrend=.13 among OC users). Compared with never users of OCs, women who took OCs with <1 mg progestational potency had a multivariate RR of 1.6 (95% CI=1.0 to 2.2); for 1 mg of progestin, the RR was 2.5 (95% CI=1.8 to 3.4); and for >1 mg progestin, the RR was 2.0 (95% CI=1.4 to 3.0). When we added duration of OC use to the model, the summary RRs for each level of progestational potency were almost identical to the model that did not include duration of use. Finally, adjustment for estrogenic potency did not alter the RRs.
For estrogen, the majority of women were taking combination OCs with low to medium estrogenic potency (≤50 μg in ethynyl estradiol equivalents per day). Within this group, the multivariate RR of hypertension did not vary appreciably (Table 4⇑). The RRs for low and medium estrogenic potency were virtually unchanged when adjusted for duration of use or progestational potency.
We repeated the potency analysis in the cross-sectional data using change in systolic or diastolic blood pressures as the outcome variable. We found that the relative effects of various doses of estrogen and progestin were virtually the same as in the prospective data. Again, controlling for duration of use had virtually no effect.
We found no decrease in risk with an increase in the amount of time elapsed since the last use of OCs, which is consistent with an acute effect. Compared with women who never used OCs, the multivariate RR among past users who had used OCs within the past 2 years was 1.1 (95% CI=0.8 to 1.5); for 2 to <4 years, the RR was 1.1 (95% CI=0.8 to 1.4); for 4 to <6 years, the RR was 1.1 (95% CI=0.8 to 1.4); and for 6 or more years, the RR was 1.2 (95% CI=1.0 to 1.4).
In this population, positive family history of hypertension and black ethnicity were strong predictors of subsequent hypertension. For those with a positive family history, the age-adjusted RR was 2.2 (95% CI=2.0 to 2.5). For those with black ethnicity, the age- and BMI-adjusted RR was 2.1 (95% CI=1.7 to 2.6). Women aged 40 to 44 years had an RR of 3.9 (95% CI=3.2 to 4.9) compared with women aged 25 to 29 years, and women in the highest quintile of BMI had an RR of 8.0 (95% CI=6.6 to 9.5) compared with women in the lowest quintile. Because it has been suggested that the influence of OCs differs among women according to ethnicity, family history of hypertension, obesity, and age,2 7 8 we analyzed differences in OC use and risk of hypertension within categories of these variables. Among black women, the multivariate RR for current OC use was 0.9 (95% CI=0.3 to 2.6), whereas it was 1.9 (95% CI=1.5 to 2.4) among white women. However, the differences in risk by ethnicity were not statistically significant (P=.81 for test of heterogeneity). Risk of hypertension due to OC use also was not altered by family history of hypertension, category of BMI, or age.
We further examined current use of OCs by type of formulation: monophasic, biphasic, or triphasic combination. There were too few users of progestin-only contraceptives in the present study to enable us to estimate an RR. Women who used monophasic combination pills had the largest increase in risk of hypertension compared with never users of OCs (RR=2.3; 95% CI=1.7 to 3.0). Women who used biphasic or triphasic combination OCs had RRs of 1.7 (95% CI=1.2 to 2.4) and 1.9 (95% CI=0.9 to 4.1), respectively.
We found an increased risk of hypertension among current users of OCs that was highest among long-term users and decreased shortly after OC cessation. Risk of hypertension increased with increases in the potency of progestin.
Our cross-sectional analysis excluded women with hypertension, because hypertension is recognized as a relative contraindication to OC use. However, women with high blood pressure (but not full-blown hypertension) are also more likely to be advised not to take OCs. This would tend to bias the observed relations in the cross-sectional analysis toward the null. However, women who take OCs may be monitored more closely and thus are more likely to be diagnosed with hypertension than a nonuser who develops hypertension. This would tend to increase the observed RR in the prospective analysis. We addressed these concerns by restricting both analyses to nurses who reported having had a physical exam within the 2 years before the baseline questionnaire was completed. Furthermore, this potential bias is unlikely to account for the patterns seen when the relative effects of different formulations of OCs were compared.
In the prospective analysis, the exposure and covariate information was not subject to bias, because it was collected before the diagnosis of disease. In contrast, in the cross-sectional analysis in which exposure and outcome were assessed simultaneously, report of disease may have influenced report of history of OC use or vice versa. However, results from these two independent analyses yielded highly consistent results. Just as in the prospective analysis, differences in both mean systolic and diastolic blood pressures between current and never users of OCs increased with increase in the potency of progestin.
We could not standardize the diagnostic criteria for hypertension; these vary among the subjects' physicians. If the misclassification of outcome is not related to OC use, it will result in dampening any observed association between OCs and hypertension. However, if the tendency of a physician to diagnose hypertension for the same blood pressure reading is more likely among OC users, the magnitude of any observed association between OCs and hypertension will be overestimated.
We relied on self-reported blood pressure. Although a direct measurement is more accurate, a single blood pressure reading may be unstable because of the large within-person variability in measurements over time.20 Because blood pressure is the outcome, random error in self-reporting of blood pressure does not bias the estimates of effect of OC use in linear regression analysis, although it does increase their standard errors. The validation studies provided substantial reassurance as to our reliance on self-reported hypertension and blood pressure in this population of nurses.
The mechanisms involved in the production and maintenance of OC-induced hypertension are not well understood. The renin-angiotensin system has been implicated; users of high-dose OCs may have greatly elevated levels of angiotensinogen.21 Evidence suggests that changes in blood pressure related to OCs are reversible in a short time. Weir et al1 found that blood pressure returned to pretreatment levels within 3 months in a controlled prospective study of 32 women who discontinued combination OCs after 1 to 3 years of use; the mean systolic pressure fell by 9.7 mm Hg (P<.001) and the diastolic pressure by 2.9 mm Hg (P<.05) compared with the measurements made 1 month before OC use was stopped. This may account for the lack of an association between time since last use, past use, and incidence of hypertension in the present study. We found a slightly lower blood pressure level among past users in the cross-sectional analysis, which may be due to screening before OC use and selection of other forms of contraception for women with high blood pressure.
Groups of women who generally have higher distributions of blood pressure have been hypothesized to be more susceptible to the hypertensive effects of OCs. However, Khaw and Peart7 performed a cross-sectional study of 461 women who routinely attended two family-planning clinics and found no significant differences in mean systolic and diastolic pressures according to contraceptive use between black and white women. In 2676 black women who attended a family-planning clinic in Atlanta, Ga, and who were followed up for 6 to 24 months, Blumenstein et al22 found no significant difference in change in mean arterial blood pressure between those who used OCs and those who used nonhormonal contraceptive methods.22 We also did not find an increased risk among blacks in the present study.
Both cross-sectional studies and clinical trials have examined the effects of combined preparations that contain various levels of both estrogen and progestin.1 2 3 5 7 9 The Royal College of General Practitioners Oral Contraceptive Study,8 a 5-year follow-up study of 23 000 women, raised the possibility that the dose of progestin in the combined pill might influence the incidence of hypertension. Among women who took 50 μg estrogen, the ratios of the rate of hypertension in relation to progestin dose were 1.00 (referent, <3.0 mg progestin), 1.07 (3.0 mg progestin), and 1.32 (>3.0 mg progestin). There was no evidence that the dose of estrogen was related to incidence of hypertension. In cross-sectional studies, a dose-response effect between blood pressure and level of progestin is also suggested. Among 63 women who used a combined preparation that contained 50 μg of estrogen, Meade et al3 observed higher systolic pressures among those who took formulations with 3 to 4 mg progestin compared with those who took preparations with 1 μg progestin. That study, however, was based on a small number of subjects and did not adequately rule out selection bias. Similarly, Khaw and Peart,4 in a cross-sectional study of women who took 30 μg of estrogen in combination with 150 or 250 μg of progestin, found that women aged less than 35 years who took the higher progestin dose had significantly higher mean blood pressures compared with women who did not use OCs (113.7/73.6 versus 109.3/70.1 mm Hg; P<.05), and although not significant, they had higher means than the 150 μg group (111.4/71.1 mm Hg).
In contrast, prospective controlled trials among healthy, normotensive women of reproductive age have not shown an association between dose of progestin and risk of hypertension.1 4 6 Wilson et al4 reported a significant increase in blood pressure over a 1-year period in women who took combined OCs that contained 30 μg of estrogen, irrespective of progestin dose. Bloch6 observed a slight fall (<2 mm Hg systolic and <4 mm Hg diastolic) in blood pressure among women who took a similar dose, also regardless of progestin dose. Weir et al1 found no significant difference in change in blood pressure in women who took 50 μg estrogen with different types and doses of progestin. All of these trials were limited, however, by the small numbers of subjects involved.
Most prior studies concerning OCs and hypertension have been unable to take into account steroid type and dose simultaneously or the estrogenic activity of the progestin component. The method we selected to determine potency is based on the entire formulation of the OC. We found that all levels of progestational potency and even low levels of estrogenic potency were associated with significantly increased risk of hypertension.
Triphasic OCs have been observed to be less likely than the monophasic formulations to affect blood pressure because of the lower overall monthly intake of progestin.4 We found that users of monophasic combination OCs had an increased risk of hypertension, whereas use of biphasic and triphasic combinations conveyed a smaller but still statistically significant increase in risk. Longitudinal studies23 24 of progestin-only OCs with 6 months to 1 year of follow-up have shown no effect of these OCs on the development of elevated blood pressure. There were too few users of progestin-only OCs in the present study to enable us to estimate this risk.
Current use of OCs is significantly associated with a moderately increased risk of hypertension in the present cohort, which underscores the importance of careful monitoring of young women who are taking OCs. However, among current OC users, only 41.5 cases of hypertension per 10 000 person-years could be attributed to OC use. Moreover, after cessation of OCs, risk appears to diminish quickly.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|NHS||=||Nurses' Health Study|
This investigation was supported by research grant CA50385 from the National Cancer Institute. Dr Chasan-Taber was supported by the National Institute of Environmental Health Services, National Research Service Award ES-07069 from the National Institutes of Health. Dr Colditz and Dr Hunter are supported by faculty research awards from the American Cancer Society (No. FRA398 and No. FRA455, respectively).
Reprint requests to Dr Lisa Chasan-Taber, Department of Epidemiology, Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115. E-mail email@example.com.
- Received December 5, 1995.
- Revision received January 26, 1996.
- Accepted February 3, 1996.
- Copyright © 1996 by American Heart Association
Weir RJ, Briggs E, Mack A, Naismith L, Taylor L, Wilson E. Blood pressure in women taking oral contraceptives. Br Med J. 1974;1:533-535.
Meade TW, Haines AP, North WRS, Chakrabarti R, Howarth DJ, Stirling Y. Haemostatic, lipid and blood-pressure profiles of women on oral contraceptives containing 50 μg or 30 μg oestrogen. Lancet. 1977;ii:948-951.
Khaw K-T, Peart WS. Blood pressure and contraceptive use. Br Med J. 1982;285:403-407.
Royal College of General Practitioners. Oral Contraceptives and Health. London, England: Pitman Medical Publishing Co Ltd; 1974.
Briggs M, Briggs M. Oestrogen content of oral contraceptives. Lancet. 1977;ii:1233. Letter.
American Medical Association. AMA Drug Evaluations. 5th ed. Chicago, Ill: American Medical Assn; 1983:965-967.
American Medical Association. Drug Evaluations Annual 1991. Milwaukee, Wis: American Medical Assn; 1991:982-986.
Dickey RP. Managing Contraceptive Pill Patients. Minneapolis, Minn: Creative Informatics; 1987:206-211.
Colditz GA, Martin P, Stampfer MJ, Willett WC, Sampson L, Rosner B, Hennekens CH, Speizer FE. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. 1986;123:894-900.
Wolf AM, Hunter D, Colditz GA, Manson JE, Stampfer MJ, Corsano KH, Rosner B, Kriska A, Willett WC. Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol. 1994;23:991-999.
Miettinen O. Estimability and estimation in case referent studies. Am J Epidemiol. 1976;103:226-235.
Miettinen OS. Proportion of disease caused or prevented by a given exposure, trait or intervention. Am J Epidemiol. 1974;99:325-332.
Rosner B, Polk BF. Predictive value of routine blood pressure measurements in screening for hypertension. Am J Epidemiol. 1983;117:429-442.
Woods JW. Oral contraceptives and hypertension. Hypertension. 1988:11(suppl II):II-11-II-15.
Blumenstein BA, Douglas MB, Hall WD. Blood pressure changes and oral contraceptive use: a study of 2676 black women in the southeastern United States. Am J Epidemiol. 1980;112:539-552.
Gaspard UJ, Deville JL, DuBois MH. Clinical experience with triphasic oral contraceptives (Trigynon) in six hundred cycles. In: Haspels AA, Rolland R, eds. Benefits and Risks of Hormonal Contraception: The Proceedings of an International Symposium. Amsterdam, Netherlands: MTP Press; 1982:61-77.
Zador G. Clinical performance of a triphasic administration of ethinyl oestradiol and levonorgestrel in comparison with the 30+150 μg fixed dose regimen. In: Haspels AA, Rolland R, eds. Benefits and Risks of Hormonal Contraception: The Proceedings of an International Symposium. Amsterdam, Netherlands: MTP Press; 1982:43-55.