Administration of EE at pharmacologic levels causes an increase in the production of procoagulant factors such as factor V, factor VIII, factor X, and fibrinogen.111
In addition, anticoagulant factors and proteins significantly decrease. The progestin component modulates the EE-induced clotting factor responses.112
The net effects depend upon the estrogen used and route of administration (see Chapter 2
). EE-containing contraceptives rapidly cause prothrombotic changes in the coagulation system manifested by immediate downstream effects such as increased prothrombin fragment 1 + 2, increased D-dimer, and changes in endogenous thrombin potential, a global marker of thrombophilia.112,113
Studies that have tried to differentiate the effects of different HC formulations on thrombophilia have had mixed results.112,114,115,116,117,118
The coagulation system changes in a procoagulant direction with CHC use, but the average changes are small and values tend to remain with normal ranges. Which changes are clinically important is not entirely clear.112
Changes in individual factors have weak or unknown relationships to VTE risk, and thus one must be cautious about overinterpreting the clinical importance of the any laboratory changes, as none are validated surrogates or proven predictors for VTE risk.
Pharmacodynamic studies of EE-containing contraceptives are consistent in showing procoagulant effects. E2-containing oral contraceptives are less well studied but have similar or slightly less effect on the coagulation system compared to EE-containing contraceptives.119,120
Whether the various progestins paired with EE modulate the procoagulant effect of estrogen is controversial with studies reporting small differences of uncertain clinical relevance.121,122
Studies that evaluated a specific measure of thrombin generation, the activated protein C sensitivity ratio, have reported that CHCs containing EE and low-androgenic gonanes or DRSP cause more thrombin generation than older CHCs.123
This research test is a global measure of coagulation system activation, and studies show that greater activation is directly related to VTE risk in individuals with hereditary thrombophilias.124,125
However, this test has not been validated as a surrogate for VTE risk associated with different CHC formulations.
Estetrol (E4) is another estrogen being studied as an oral contraceptive. A detailed hematologic study demonstrated that E4 combined with DRSP caused little or no procoagulant changes from baseline, while the comparison oral contraceptive containing EE/DRSP led to substantial changes from baseline.126
Current experience with this agent is limited to clinical trials; we will not be able to learn if the favorable laboratory results will translate into lower VTE risk until this combination comes into widespread clinical use.
Transdermal and transvaginal CHC delivery systems avoid hepatic firstpass effects. As such, initial hopes were that these nonoral contraceptives would have less effect on coagulation and on VTE risk than EE-containing oral contraceptives. However, in actuality, the second-pass metabolism of EE when delivered via a nonoral route is quite extensive. A randomized crossover study found that patch and ring use were associated with greater increases in the activated protein C sensitivity ratio than with an EE/LNG oral contraceptive.127
Another crossover study showed similar adverse changes of multiple coagulation factors when comparing the EE/NLGM patch and an EE/NGM oral contraceptive.128
These results suggested that using the patch or ring would not avoid the VTE risks of EE-containing COCs.
The monthly contraceptive injections are less studied; a WHO randomized trial published in 2003 compared the hemostasis effects among 259 women using two monthly injectables to the effects of an EE/NET oral contraceptive among 119 women.129
In the 9-month study, the injection users had slightly smaller unfavorable changes in their coagulation system than did the oral contraceptive users. While that study measured many individual factors, it did not assess the more global measures of coagulation system activation such as thrombin generation (measured by APCsr) or D-dimer or fragment 1+2.
Epidemiologic Studies of CHC and Thrombosis
Older epidemiologic evaluations of oral contraceptives and vascular disease indicated that venous thrombosis was an effect of estrogen, limited to current users, with a disappearance of the risk by 3 months after
Thromboembolic disease was believed to be a consequence of EE administration with the level of risk related only to the EE dose.132,133,134
In the first years of oral contraception, the available products, containing high doses of EE or mestranol, were associated with a 4- to 6-fold increased risk of venous thrombosis.135
Smoking was documented to produce an additive increase in the risk of arterial thrombosis136,137,138
but had no effect on the risk of VTE.139,140
Because of the increased risks for venous thrombosis, myocardial infarction (MI), and stroke, lower dose formulations (<50 µg estrogen) came to dominate the market, and clinicians became more careful in their screening of patients and prescribing of oral contraception. Two forces, therefore, were at work simultaneously to bring greater safety to women utilizing oral contraception: (1) the use of lower dose formulations and (2) the avoidance of COCs by high-risk patients. Because of these two forces, the Puget Sound study in the United States documented a reduction in venous thrombosis risk to 2-fold greater than nonusers.141
Is there still a risk of VTE with the current low-dose oral contraceptive formulations and nonoral CHC? Epidemiologic studies have included case-control studies, very large multiyear prospective trials, and even larger database evaluations (Table 7.1)
Each study type has its strengths and weaknesses. VTE is fortunately rare, with about 1 event per 1,000 CHC users per year, but this relative rarity makes comparator studies difficult. The very large studies that use data directly from existing databases are often the only quick and cost-efficient way to evaluate the associations between various CHC types and these outcomes. Despite these advantages, database studies usually cannot evaluate important confounding factors (for VTE, these may include family history, obesity, smoking, and other risk factors not captured by insurance claims or registries). Database studies also have only indirect measures of the CHC exposure—that is, they typically know about prescriptions, but do not have direct information about whether and when a woman actually used the CHC. The incidence of VTE in database studies is usually lower than that found in other study types.142,143,144
In contrast, case-control studies may be disadvantaged by recall bias145,146,147
while the smaller number of VTEs available for analysis typically hamper prospective cohort studies, as seen with Dinger et al.148
Dramatic changes in CHC availability and in clinical practice frequently can occur during the time frame of any study, which can create analytical challenges. Thus, the differences in results across studies remain a topic for vigorous methodologic discussion. Keeping these limitations in mind, we present some of the key papers below.
The World Health Organization (WHO) Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception was a hospital-based, case-control study with subjects collected from 21 centers in 17 countries in Africa, Asia, Europe, and Latin America.145
A subevaluation of data from 10 centers in 9 countries assessed the risk of idiopathic
VTE associated with a formulation containing EE 30 µg and LNG (doses ranging from 125 to 250 µg) compared to the risk associated with preparations containing EE 20 or 30 µg and either DSG or GSD.149
The OR for VTE, adjusted for body weight and height, for EE-LNG formulation users compared with nonusers was 3.5. Compared to nonusers, the OR for EE-DSG and EE-GSD products were 9.1. Thus, the risk of VTE with EE-LNG use was 2.6 times lower than with EE-DSG and EE-GSD use.
Table 7.1 Epidemiologic Studies of CHC and VTE Risk
Range of Effect Estimates
Database Studies (Relative Risks)
Prospective Studies (Hazard Ratios)
Case-Control Studies (Odds Ratios)
Oral EE with levonorgestrel (reference)
Oral EE with desogestrel/gestodene
Oral EE with drospirenone
Oral E2 valerate with dienogest
Vaginal EE with etonogestrel
Transdermal EE with norelgestromin
See text for details of individual studies.
The Transnational Study of Oral Contraceptives and the Health of Young Women analyzed 471 cases of deep vein thrombosis and/or VTE from the United Kingdom and Germany.146
These investigators found an OR of 3.2 for VTE among women using CHC containing EE 35 µg or less and a progestin other than DSG or GSD. They reported that the OR was 1.5-fold greater for women using EE-DSG and EE-GSD products than other products.
In Denmark, Lidegaard et al.150
performed a hospital-based, case-control study of women with confirmed diagnoses of VTE in 1994 and 1995 (in Denmark, all women with this diagnosis are hospitalized, and therefore, very few, if any, cases were missed). They identified a 2-fold increased VTE risk in current oral contraceptive users, across EE doses ranging from 20 to 50 µg. The increased risk was concentrated in the first year of use. Because there were more short-term users of the new progestins and more long-term users of the older progestins, adjustment for duration of use resulted in no significant differences between the progestin types. Factors associated with an increased VTE risk included coagulation disorders, treated hypertension during pregnancy, family history of VTE, and a greater BMI. Notably, conditions not associated with an increased risk of VTE included smoking, migraine, diabetes, hyperlipidemia, parity, or age at first birth. This study could not establish the absence or presence of a dose-response relationship comparing the EE 20 µg dose to higher doses. However, a 5-year update reported that EE 20 µg had a lower risk than did products with EE 30 to 40 µg.151
Additionally, this update found that the risk of venous thrombosis associated with current COC use declined with increasing duration of use, VTE risk was slightly lower with EE/LNG than EE/DSG or EE/GSD oral contraceptives, smoking more than 10 cigarettes per day increased VTE risk, and progestin-only contraceptive products did not increase VTE risk.
Case-control studies using VTE cases derived from the computer records of general practices in the United Kingdom concluded that the increased risk associated with oral contraceptives was the same for all types, and that the pattern of risk with specific oral contraceptives suggested confounding because of “preferential prescribing” (defined later).152,153
In these studies, matching cases and controls by year of birth eliminated differences between different types of oral contraceptives. A similar analysis based on 42 cases from a German database also found no difference between new progestin and older progestin oral contraceptives.154
Thus, in these two studies, more precise adjustments for age eliminated a confounding bias. An assessment of the incidence of VTE in the United Kingdom before and after the decline in EE/DSG and EE/GSD product use could detect no impact on the statistics (neither an increase nor a decrease).155
A reanalysis of the Transnational Case-Control Study considered the duration and patterns of oral contraceptive use, focusing on first-time users of oral contraceptives with EE doses less than 50 µg.156,157
Statistical analysis with adjustment for duration of use in the 105 cases who were firsttime users could find no differences between different progestin products. A similar reanalysis of the United Kingdom General Practice Database could demonstrate no difference between different oral contraceptive formulations.157
A case-control study in Germany assessed the outcome when the cases were restricted to hospitalized patients compared to results when all cases, both in-hospital and out-of-hospital, were considered.158
indicated that hospital-based studies overestimated the risk of VTE, and that there was no difference comparing progestins when all cases were included.
Former users discontinue oral contraceptives for a variety of reasons and often are switched to what clinicians perceive to be “safer” products, a practice called “preferential prescribing.”159,160,161
Individuals who do well with a product tend to remain with that product. Thus, at any one point in time, individuals on an older product will be relatively healthy and free of side effects—the “healthy user effect.” This is also called attrition of susceptibles because higher risk individuals with problems are gradually eliminated from the group.162
Comparing users of older and newer CHC products requires careful analysis to adjust for important differences between these individuals.
Because EE/DSG and EE/GSD products were marketed as less androgenic and therefore “better” (a marketing claim not substantiated by epidemiologic studies), clinicians chose to provide these products to higher risk patients and older women.159,160
In addition, clinicians switched patients perceived to be at greater VTE risk from older oral contraceptives to the newer formulations with EE/DSG and EE/GSD. Furthermore, these products were prescribed more often to young women who were starting oral contraception for the first time (these young women will not have experienced the test of pregnancy or previous oral contraceptive use to help identify those who have an inherited predisposition to venous thrombosis). These changing practice patterns exert different effects over the lifetime of a product, and meaningful analytical adjustments are extremely difficult to achieve.
The initial VTE risk epidemiologic studies were impressive in their agreement. All indicated increased relative risks (RRs) or ORs associated with EE/DSG and EE/GSD compared with EE/LNG. Nevertheless, results of the early studies may have been influenced by the same unrecognized biases. Consistent conclusions may be the result of consistent errors.
Shortly after the first oral contraceptive with DRSP (EE 30 µg/DRSP 3 mg) was marketed, 40 VTE cases in DRSP users (2 of which were fatal) were reported in Europe in 2002.163
The Dutch College of General Practitioners issued a statement encouraging clinicians not to prescribe the EE-DRSP product. However, this story is similar to that reported with EE/DSG and EE/GSD (“third-generation” progestins), only to learn that preferential prescribing and the healthy user effect probably biased the early studies. In postmarketing surveillance of EE 30 µg/DRSP 3 mg, only one VTE case occurred in a million cycles compared with five among users of other oral contraceptives.163
In a subsequent monitoring study, the VTE incidence in new DRSP users was comparable to that seen in users of other low-dose oral contraceptives.164
The prospective European Active Surveillance (EURAS) cohort study (see below) enrolled only new OC users containing a variety of progestins, including DRSP and LNG.165
The cardiovascular event incidence
was similar for all progestins. An American cohort study also focused on new OC users and found that VTE occurred at a similar low rate when comparing DRSP users to other oral contraceptives.166
The Danish investigators have continued their interest in hormonal contraception and venous thrombosis and performed a national database study using the reliable Danish national registries of events from 1995 to 2005.143
As in the earlier Danish case-control study,151
the VTE risk in current COC users decreased with duration of use and with EE dose and was slightly higher with EE/DSG, EE/GSD, EE/DRSP, and EE-cyproterone acetate (CPA) products. Did this study escape the problems of preferential prescribing and the healthy user effect (attrition of susceptibles)? The incidence of thrombotic events in the comparator group (EE-LNG users) was lower than that reported in other studies, possibly because this group did demonstrate a healthy user effect, but also because database studies often ascertain fewer cases than studies that use active surveillance. The study was not limited to new users. This study was unable to control for BMI or family history of thrombosis, two important markers for women at high risk of venous thrombosis. Preferential prescribing may be a confounder in the Danish database studies; however, the problem of a healthy user effect is also possible.
A case-control study from the Netherlands also reported higher risks of venous thrombosis in EE/DSG, EE/GSD, EE/DRSP, and EE/CPA users compared with EE-LNG users.147
The authors supported their results by citing findings from their own institution that users of oral contraceptives containing EE/DRSP and EE/CPA had lower levels of free protein S and free tissue factor pathway inhibitor associated with greater resistance to activated protein C compared with EE/LNG users.118
The RRs in that study were surprisingly high, more so than all other reports involving low-dose oral contraceptives. Once again, the healthy user effect is a possible confounder because the study was not limited to new users. The authors claimed to mitigate any attrition of susceptibles by analyzing only short-term users. Even though the validity of this approach can be debated, the results indicated nonsignificant increased risks with EE/DRSP and EE/CPA compared with EE/LNG, and any conclusion was limited by a small number of short-term users. In that study, the risk associated with products containing EE 20 µg was not increased compared to nonusers.
In an effort to better ascertain risk, several true prospective studies have compared the thrombosis effects of various CHC formulations. Strengths of these studies include the ability to adjust for baseline confounding and rigorous adjudication of outcome with published methodology. The European Active Surveillance (EURAS) study enrolled 58,674 European women initiating a new prescription for combined oral contraception and contacted subjects every 6 months to assess safety outcomes and found no significant difference in VTE risk between DRSP and LNG COCs (hazard ratio [HR], 1.0; 95% confidence interval [CI], 0.6 to 1.8) and DRSP with
other oral contraceptives (HR, 0.8; 95% CI, 0.5 to 1.3).165
The subsequent International Surveillance Study of Women Taking Oral Contraceptives (INAS-OC) study followed more than 85,000 women in the United States and Europe and found no increased risk of VTE in DRSP users compared with LNG users (HR, 0.8; 95% CI, 0.4 to 1.3).167
The Transatlantic Active Surveillance on Cardiovascular Safety of NuvaRing study used a similar design and found no significant difference in the risk of VTE between women using the EE/ENG contraceptive vaginal ring compared with combined OCPs users (adjusted HR, 0.8; 95% CI, 0.5 to 1.5).168
Most recently, the International Active Surveillance study “Safety of Contraceptives: Role of Estrogens” (INAS-SCORE) investigated the cardiovascular risks associated with the use of estradiol-containing pills and found an adjusted HR of 0.5 (95% CI, 0.2 to 1.5) for a COC-containing E2V and DNG compared to EE/LNG pills.148
Several studies have compared VTE risk between oral and nonoral contraceptive users.169
Six studies assessing the contraceptive patch had mixed results with two studies suggesting a VTE risk among patch users about double the risk in EE/LNG oral contraceptive users, but the other such studies found similar VTE risks among patch and oral contraceptive users. Only three studies assessed VTE risk among vaginal contraceptive users; one of these found a doubling of risk and the others found no increase in risk compared to COC users. While excess VTE risk among patch and ring users (compared to COC users) is uncertain, it is clear that the patch and ring are not safer than oral contraceptives for VTE. A 2017 systematic review did not identify any studies of VTE risk among women using monthly injectable CHC.169
Conclusion: The venous thrombosis risk associated with modern EE-containing hormonal contraceptives is increased about 2-fold and is greatest in the first year of use.165,170,171 VTE risk increases with increasing body weight and age and increases with EE dose. The contribution of different progestins to the VTE risk seems to be small; studies of different design find different progestin effects, which may reflect bias and unmeasured confounding. The important takeaway is that EE increases venous thrombosis risk; the impact of progestins remains controversial. Although progestins may modulate the hepatic effects of estrogens, prospective studies fail to show any significant differential effect on thrombosis risk. Thus far, a single prospective study suggests that estradiol-containing oral contraceptives may have slightly lower VTE risk than EE-containing contraceptives.
The general population incidence of VTE is higher than previously estimated; this may be due to obesity and other lifestyle changes as well as the prevalence of sensitive diagnostic methods. VTE risk associated with low-dose oral and nonoral CHCs is lower than previously reported and concentrated among high-risk individuals (e.g., obesity and inherited or acquired thrombophilias). Because women over 40 and women with a BMI greater than 30 kg/m2 have higher baseline VTE risks, CHC is not a first-choice contraceptive for them.