Susan Jick1
(1)
Boston University School of Public Health, 11 Muzzey Street, Lexington, 02421, MA, USA
Susan Jick
Email: sjick@bu.edu
Introduction
Oral Contraceptives (OC) were introduced in the 1960s and were a major advance in family planning technology for women. The first pill released in the United States, Enovid®, contained high doses of estrogen and progestin; specifically, mestranol and norethynodrel; four times more estrogen and ten times more progestin than today’s OCs (Buttar and Seward 2009; http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm?fuseaction=Search.DrugDetails). Because OCs were a new class of drug there was no existing knowledge or experience with their adverse risk profile, in particular in relation to the risk of cardiovascular disease (CVD). The first cases of cardiovascular events in OC users, including venous thromboembolism (VTE), myocardial infarction and stroke were observed soon after the pills were marketed (Records unit of the Research Advisory Service of the Royal College of General Practitioners 1967; Inman and Vessey 1968; Vessey and Doll 1968, 1969; Sartwell et al. 1969; Inman et al. 1970). It was thought at that time that the high doses of hormone were responsible (Records unit of the Research Advisory Service of the Royal College of General Practitioners 1967). In response to this concern, pills with lower doses of estrogen and different progestins were introduced to the market with the goal to reduce the CVD risk.
VTE is the most common of the three adverse CVD outcomes and thus is the focus of this chapter. Indeed the risk of VTE decreased with use of the OCs introduced in the 1980s; those with lower doses of estrogen (around 35–50 μg) and with the newer progestins levonorgestrel and norethindrone (BCDSP 1973; Stolley et al. 1975; Wharton and Blackburn 1988; Vessey et al. 1986). The risk of myocardial infarction and stroke have also decreased and will not be discussed further in this chapter as these outcomes are very rare in young healthy women and are thus difficult to study and not a major public health concern.
In the mid-1990s the discussion around OCs and VTE risk refocused on progestin type when concerns were raised that newer, so called third generation OCs, might increase the risk of VTE compared to OCs containing the older progestin levonorgestrel (Meade et al. 1980; Sitruk-Ware 2004; WHO 1995; Jick et al. 1995; Bloemenkamp et al. 1995; Lidegaard et al. 1998; Farmer et al. 1997; Spitzer et al. 1996; Jick et al. 2000; Kemmeren et al. 2001). Studies have since found that risks vary according to progestin formulation (Lidegaard et al. 2011; Vasilakis-Scaramozza and Jick 2001; Seaman et al. 2004; Vasilakis et al. 1999; Rubig 2003; Heineman and Dinger 2004). Below, the history of OC development and its relationship to the risk of VTE is described from the 1960s when the pill was first marketed until the present day, many years and many OC formulations later.
Oral Contraceptive Safety in the 1960s and 1970s
The first studies of oral contraceptives in relation to VTE were published in the late 1960s and early 1970s (Records unit of the Research Advisory Service of the Royal College of General Practitioners 1967; Inman and Vessey 1968; Vessey and Doll 1968, 1969; Sartwell et al. 1969; Inman et al. 1970). In 1967 the Royal College of General Practitioners (RCGP) published results of a case–control study of VTE in relation to OC use (Records unit of the Research Advisory Service of the Royal College of General Practitioners 1967). While this study was based on a small numbers of cases and controls, the results found that women who were taking OCs were at increased risk of VTE. Another study by Inman and Vessey found a strong association between OC use and fatal pulmonary embolism in previously healthy women (Inman and Vessey 1968). Vessey and Doll then published a more comprehensive study of OCs in relation to VTE in 1968 (Vessey and Doll 1968) which was updated in 1969 (Vessey and Doll 1969). These studies found that OC use was associated with an increased risk VTE of around 6–7 fold. The studies evaluated current use of OCs in idiopathic hospitalized and confirmed cases of VTE and compared OC use to women hospitalized for reasons other than VTE. These formal case-control studies quantified the magnitude of the risk of the then available OCs.
Sartwell et al. published another case-control study in 1969 (Sartwell et al. 1969) using similar case definition and methods that yielded similar results and further explored the effects of duration of use. These studies established both that OCs increase the risk of VTE and identified important risk factors and covariates for the outcome. Publications in the 1970s by Inman et al., the Boston Collaborative Drug Surveillance Program, and Stolley et al. added additional data and expanded on the known risk factors and biases present in studying this exposure-outcome relationship (Inman et al. 1970; BCDSP 1973; Stolley et al. 1975).
These early publications provided the framework for all future studies of OCs in relation to cardiovascular events. See Table 6.1 for a summary of these studies. For example, increasing age is strongly associated with an increased risk of VTE while OC use declines with increasing age. Thus the risk of VTE is highest in the age group of women with the lowest exposure. Therefore, age must be carefully controlled to prevent confounding. Calendar time is another important variable to control in all OC studies because OC use changes over time. OC use in cases of VTE must be compared to use in non-cases at the same point in time to avoid non-comparability. Similarly OC use differs in different regions of a country and the world, and women in different countries may have different risk factors for VTE and different OC use patterns, thus case women should be compared to women in the same geographic area to achieve valid results.
Table 6.1
Early studies of oral contraceptives in relation to venous thromboembolism
|
Authors |
Citation |
Title |
Results |
Exposure and case details |
|
Records unit of the Research Advisory Service of the Royal College of General Practitioners |
Collective Investigation 1967;13:267–79 |
Oral contraception and thromboembolic disease |
17 exposed out of 147 cases vs 17 exposed out of 294 controls |
OC exposure was stratified by current, recent and past use and compared to non-use. Cases were women 15–49 reported by their GP to have had a VTE |
|
Inman WH, Vessey MP |
BMJ 1968;2:193–9 |
Investigations of death from pulmonary, coronary, and cerebral thrombosis and embolism in women of child bearing age |
95 confirmed PEs and 209 coronary thrombosis or MI; 2 controls per case. There was a strong association between use of OCs and death due to pulmonary embolism |
OC exposure was use of an OC. Cases were women aged 20–44 in 1966 with idiopathic fatal thrombosis or pulmonary embolism identified from death certificates |
|
Vessey MP, Doll R |
Br Med J 1968;2(5599):199–205 |
Investigation of relation between use of oral contraceptives and thromboembolic disease |
42 cases of idiopathic VTE and 23 controls Risk of hospital VTE ~ 6–7 times greater in OC users than non-users |
Current OC exposure was evaluatedCases were hospitalized with confirmed idiopathic VTE UK 1964–1966, subjects aged 16–40 years |
|
Vessey MP, Doll R |
Br Med J 1969;2(5658):651–7 |
Investigation of relation between use of oral contraceptives and thromboembolic disease. A further report |
84 idiopathic VTE cases and 168 controls: RR 6.3 for OC use compared to nonuse |
Current OC exposure was evaluated in idiopathic VTE cases from London hospitals 1964–1967 aged 16–40 years |
|
Sartwell PE et al. |
Am J Epidemiol 1969:90:365–380 |
Thromboembolism and oral contraceptives: an epidemiologic case–control study |
175 cases and matched controls. RR for VTE was 4.4 for current use compared to nonuse |
Current OC use was evaluatedCases were idiopathic VTE in US hospitals aged 15–44. Matched to controls on age, race and hospital |
|
Boston Collaborative Drug Surveillance Program |
Lancet 1973;1(7817):1399–404 |
Oral contraceptives and venous thromboembolic disease… Report from the BCDSP |
43 Cases and 842 controls. Age standardized RR = 11 (95 % CI 5.2–25) for users compared to non-users |
Current OC use was evaluatedCases were hospitalized idiopathic VTE in the US aged 20–44. Controls were women hospitalized for diagnoses not related to CVD or OC use |
|
Stolley PD, Tonascia JA, Tockman MS, Sartwell PE et al |
Am J Epidemiol 1975;102:197–208 |
Thrombosis with low-estrogen oral contraceptives |
All cases (N = 457) and 1290 controls: RR = 1.8 for OC <100 μg and 2.1 for 100 μg OCs. Idiopathic cases only (N = 103) and 288 controls: RR = 4.7 and 10.1 for < 100 and 100 μg OCs |
Current OC use was evaluatedCases were US hospitalized VTE. Matched to controls on age, hospital and race, aged 15–40. Idiopathic only, and separately, all cases were analyzed |
PE pulmonary embolism, RR relative risk, MI myocardial infarction
Other important risk factors that were identified in the early studies include smoking, obesity, and pre-existing cardiovascular disease. Pill use in women at high risk for thromboembolism thus became contraindicated and women with hypertension, a history of VTE or other cardiovascular disease, for example, became less likely to receive OCs. From this point forward it became important to exclude women with cardiovascular disease from studies of OCs and VTE, since they would be less likely to receive an OC than a healthy woman and their inclusion in a study would lead to biased results.
Early studies established that only current OC use was associated with increased risk of VTE and that the risk returned to baseline shortly after OC discontinuation (Sartwell et al. 1969). This has important implications for both study design and for women who are concerned about persistent effects of OCs. For woman taking the pill they should know that any increase in VTE risk conferred by the pill will return to baseline almost immediately after discontinuing the pill. For the researcher, studies of OCs should focus on current pill use.
Another important insight was that women who have other proximate causes for their VTE, such as recent surgery, pregnancy, injury, or trauma, should be excluded from studies of the relative effects of OCs on the risk of VTE (BCDSP 1973; Stolley et al. 1975). In the study by Stolley et al. (1975), it was demonstrated for example, that the inclusion of non-idiopathic cases greatly diluted the results such that the risk estimates were close to the null. Further investigation revealed that the risks in the non-idiopathic cases were around 1.0 while the risks in the idiopathic cases were elevated, consistent with previous studies. If restriction is not possible, then stratification by idiopathic (no other proximate cause or strong risk factors) versus non-idiopathic (recent other cause or strong risk factor present) case status, is essential to obtain valid results. This is because the risk conferred by the other causes will overwhelm any effect of an OC and therefore the distribution of OC use in non-idiopathic cases will tend to be similar to the OC use in the non-cases, leading to a bias of the true effect toward the null. Further, the risk in people with strong risk factors is biased away from receiving any OC or toward receiving OCs perceived to confer less risk. Differences in results of many published studies can be explained by differences in case definition (case validation and inclusion criteria), exposure definition (current only users or past users), and selection of reference group. See Table 6.1 for a summary of the early studies, their results and basic methodological characteristics.
Changes in OC Formulations Over Time: Second and Third Generation Pills
The 1980s and 1990s
The early findings that linked OC use to increased risk of thromboembolic events led to the development of newer OCs with lower doses of oestrogen in the 1980s. The second round of OCs to come to market, so called second generation OCs, contained 50 μg or less estrogen, in combination with the progestins levonorgestrel or norethindrone. The goal of the new formulations was to reduce adverse cardiovascular events. At the same time, women with known risk factors were less frequently prescribed OCs and the rates of VTE consequently declined. That is, the risk was lower in the second generation compared to the high dose first generation OCs, and women at high risk were less frequently prescribed the pill. While new OCs were developed in the 1980s, few studies of OCs and VTE were published in this decade (Wharton and Blackburn 1988; Vessey et al. 1986; Meade et al. 1980). The results of these studies were similar to earlier studies and did not expand materially on the knowledge of the OC VTE risk.
Despite the lowered VTE risk, second generation OCs still conferred an increased risk of VTE compared to non-use of the pill, and thus newer “generations” of OCs with newer progestins were developed and marketed in attempt to further reduce the VTE risk. While the theory was that the new OC formulations, which were less androgenic (Sitruk-Ware 2004), would consequently confer lower cardiovascular risk, the reality did not meet the expectation. In fact the next, so called third generation OCs were found to increase rather than decrease the risk of VTE compared to the second generation pills (WHO 1995; Jick et al. 1995; Bloemenkamp et al. 1995; Lidegaard et al. 1998).
In the mid 1990s, research conducted by scientists at the World Health Organization found that women who received third generation OCs that contained either gestodene or desogestrel (new progestins), had higher risks of VTE compared to women who received second generation OCs which contained either levonorgestrel or norethindone (WHO 1995). The WHO study was published along with two independent research papers, each of which found that the new third generation OCs increased the risk of VTE by around two fold compared to the second generation pills (Jick et al. 1995; Bloemenkamp et al. 1995). There ensued several years of publications with varied results and debate over the true risk of third versus second generation OCs (Lidegaard et al. 1998, 2011; Farmer et al. 1997; Spitzer et al. 1996; Jick et al. 2000; Kemmeren et al. 2001). A summary of these studies is provided in Table 6.2.
Table 6.2
Summary of third generation OC studies
|
Authors |
Citation |
Title |
Results |
Exposure and case details |
|
WHO |
Lancet 1995; 346:1582–1588 |
Effect of different progestagens in low oestrogen oral contraceptives on venous thromboembolic disease |
Multiple results and comparisons presented but overall risk were around 2 fold for 3rd gen compared to levo OCs |
Case-control study of current use, and non-use. Excluded non-idiopathic cases multi country 1989–93 aged 20–44 |
|
Jick H, Jick S, Gurewich V, Myers MW, Vasilakis C |
Lancet 1995;346:1589–1593 |
Risk of idiopathic cardiovascular death and nonfatal venous thromboembolism in women using oral contraceptives with differing progestagen components |
75 cases and 300 controls. OR = 2.2 for desogestrel compared to users of levo. OR = 2.1 for gestodene compared to users of levo. Cohort RR for 3rd gen vs levo = 1.9 |
Cohort study with nested case-control analysis. Current OC use evaluated, excluded non-idiopathic cases UK 1991–94 aged 15–44 |
|
Bloemenkamp KWM, Rsoendaal FR, Helmerhorst FM, Buller HR, Vandenbroucke JP |
Lancet 1995; 346;1593–1596 |
Enhancement by factor V leiden mutation of risk of deep-vein thrombosis associated with oral contraceptives containing a third-generation progestagen |
37 Deso cases, 20 levo cases OR 2.2 |
Case–control study of current OC use, excluded cases with prior VTE and recent pregnancy. Netherlands 1998–92, aged 15–49 |
|
Lidegaard O, Edstrom B, Kreiner S |
Contraception 1998;57:291–301 |
Oral contraceptive and venous thromboembolism. A case-control study |
117 cases 3rd gen, 30 2nd gen cases. OR 1.44 |
Case-control study of current, past and non OC use. Excluded only cases with prior VTE and MI. Denmark 1994–95 aged 15–44 |
|
Farmer RDT, Lawrenson RA, Thompson CR, Kennedy JG, Hambleton IR |
Lancet 1997;349:83–88 |
Population-based study of risk of venous thromboembolism associated with various oral contraceptives |
85 cases and 313 controls. 54 3rd gen cases, 29 2nd gen cases OR = 1.34. Cohort RR = 1.68 |
Cohort study with nested case-control analysis. Current OC use evaluated and, excluded cases with other proximate causes but included cases with risk factors. UK 1991–95, aged 10–54 |
|
Spitzer WO, Lewis MA, Heinemann AJ, Thoroughgood M, MacRae KD |
BMJ 1996;312:83–88 |
Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case-control study |
127 cases and 249 controls: 98 3rd gen vs 64 2nd gen cases OR 1.5 |
Case-control study of current use. Included all cases, none excluded. UK and Germany, 1992–1995 aged 16–44 |
|
Jick H, Kaye JK, Vasilakis-Scaramozza C, Jick S |
BMJ 2000;321:1190–1195 |
Risk of venous thromboembolism among users of third generation oral contraceptives compared with users of oral contraceptives with levonorgestrel before and after 1995: cohort and case-control analysis |
54 3rd gen and 17 levo cases pre Oct. 199510 3rd gen and 25 Levo cases post 1995. ORs 2.2 and 2.8 respectively |
Cohort study with nested case-control analysis. Current OC use evaluated, excluded non-idiopathic cases. UK 1993–1999 aged 15–39 |
|
Kemmeren JM, Algra A, Grobbee DE |
BMJ 2001;323:131–4 |
Third generation oral contraceptives and risk of venous thrombosis: meta analysis |
Concludes that risk of VTE is higher in users of third vs second generation OCs |
Magnitude of risk depends on duration of use |
|
Lidegaard O, Nielsin LH, Wessel CW et al. |
BMJ 2011;343:d6423 |
Risk of venous thrombo-embolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study, 2001–9 |
939 3rd gen and 167 levo cases. Risk for 3rd gen OCs compared to levo was around 2 |
Cohort study of current OC use. Excluded women with prior thrombotic events, recent pregnancy, cancer and surgery, Denmark 2001–2009 aged 15–49 |
Close examination of these studies reveals that variation in choice of comparison groups and case definitions between studies can explain the variations in the findings. The studies by the WHO, Jick, and Bloemenkamp all evaluated current OC use and included only or primarily idiopathic cases (the Bloemenkamp study had fewer exclusions than the other studies), and used a common referent (levonorgestrel OCs). Later studies, in contrast, included many or all VTE cases regardless of other proximate causes and strong risk factors. The strength of the associations for the third versus second generation OCs can be correlated with the number of exclusions applied. Those with the least exclusions for risk factors and other proximate causes yielded risks closest to 1.0. The risk estimates increased as the proportion of idiopathic cases increased. Studies by, Lidegaard, and Spitzer (Lidegaard et al. 1998; Spitzer et al. 1996) included many cases with other proximate causes of the VTE as well as co-morbidities that might bias the selection of the OC prescribed. The risk estimates in these studies were close to 1.0.
One theory that was proposed to explain the higher risk in third versus second generation OC users was that biased prescribing was responsible. This theory, however, was not supported by later results where third generation OC users were still at higher risk compared to second generation pill users after the publication of the first studies in 1995 (Jick et al. 2000; Lidegaard et al. 2011). If the early studies had been subject to the proposed biased prescribing, where less healthy women received the third generation pills because they were thought to be less harmful, then prescribing should have changed after the initial studies were published. If prescribing had been biased, that is influenced by perceived VTE risk and health of the OC user then, after the 1995 publications, the women with risk factors should have been preferentially prescribed second generation OCs. Thus if prescribing bias were the explanation for the results, this would have resulted in lower VTE risks in users of third generation OCs. In fact, later studies continued to find increased risks of VTE among users of third generation OCs (Jick et al. 2000; Lidegaard et al. 2011) suggesting that prescribing bias did not explain the results.
There continues to be some disagreement about the magnitude of the risk of third compared to second generation OCs, but over time general agreement has emerged that there is an increased risk for third compared to second generation OCs of around twofold more or less (Kemmeren et al. 2001). While use of third generation OCs declined ORs for VTE in third versus second generation OCs did not change, despite the reduced use and possibility of selective prescribing.
Cyproterone-Containing and Progestin-Only Oral Contraceptives
In the 1990s cyproterone-containing OCs were marketed. These OCs were also found to increase the risk of VTE compared to second generation OCs and in some countries they are no longer marketed for contraception (Lidegaard et al. 2011; Vasilakis-Scaramozza and Jick 2001; Seaman et al. 2004). These OCs are still marketed and licensed for contraception in the UK, NZ and other markets, but it is not a first line indication (acne is the indication). In practice these OCs may be prescribed to women with acne who also need contraception, but this is also decreasing as better treatments for acne become available.
There has been little controversy about the finding that cyproterone-containing OCs increase the risk of VTE. Conversely, progestin-only OCs (known as the progesterone-only pill (POP) in many countries – see Chap. 5) have been shown to confer no increase in risk of VTE when compared to non-users of OCs (Lidegaard et al. 2011; Vasilakis et al. 1999). Table 6.3 summarise the studies performed to investigate the risk of VTE with cyproterone-containing and progestin-only OCs.
Table 6.3
Summary of cyproterone and progesterone-only OC studies of VTE
|
Authors |
Citation |
Title |
Results |
|
Cyproterone OCs |
|||
|
Vasilakis-Scaramozza, Jick H |
Lancet 2001;358:1427–9 |
Risk of VTE with cyproterone or levonorgestrel contraceptives |
OR for cyproterone vs levo OCs was 3.9 (95 % CI 1.1–13.4) adjusted for BMI, smoking, polycystic ovaries, hirsutism, acne |
|
Lidegaard O, Nielsin LH, Wessel CW et al. (Danish cohort study, 2001–9) |
BMJ 2011;343:d6423 |
Risk of venous thrombo-embolism from use of oral contraceptives containing different progestogens and oestrogen doses |
Risk was around twofold compared to levonorgestrel OCs and 4.1 (95 % CI 3.37–4.98) compared to non-use adjusted for age, year and level of education |
|
Seaman HE, de Vries CS, Farmer RDT |
Pharmocoepi and Drug Safety 2004;13:427–36 |
Venous thromboembolism associated with cyproterone acetate in combination with ethinylestradiol (Dianette): observational studies using the UK General Practice Research Data Base |
RR was around 2.5 for Cyp OCs vs other OCs in women with and without acne, hirsutism, PCOS. No exclusions to case definition. OR for c-c analysis 1.71 |
|
Progesterone only OCs |
|||
|
Vasilakis C et al. |
Lancet 1999;354:1610–1611 |
Risk of idiopathic venous thromboembolism in users of progestagens alone |
OR for contraceptive progestins vs no use was 1.3 (95 % CI 0.3–6.8) |
|
Lidegaard O, Nielsin LH, Wessel CW et al. Danish cohort study, 2001–9 |
BMJ 2011;343:d6423 |
Risk of venous thrombo-embolism from use of oral contraceptives containing different progestogens and oestrogen doses |
ORs <1.0 for different progestogens (norethisterone, desogestrel and levonorgestrel) adjusted for age, year and level of education |
Women who have contra- indications for combined oral contraceptive may be prescribed these progestin only pills. It should be noted however that the effectiveness of the progestin-only pills is lower than for combined OCs (see Chap. 5).
Arrival of the Next Generation of OCs: the 2000s and Drospirenone OCs
The 2000s saw a new generation of OCs come to market. There was hope, because of the antiandrogenic properties of drospirenone, that drospirenone-containing OC (Yaz® and Yasmin®) would be safer in relation to VTE and confer other benefits to women such as better control of pre-menstrual syndrome and acne (Sitruk-Ware 2004; Rubig 2003). The new OC formulation quickly became popular among young women in United States and Europe and use of the drug increased rapidly. Unfortunately so did the reports of VTE. The earliest studies of VTE in users of drospirenone-containing OCs provided reassurance that the risk of the newer OC was similar to that of the older second generation OCs (Heineman and Dinger 2004; Dinger et al. 2007; Seeger et al. 2007; Dinger et al. 2010). However, later studies yielded risks ranging from around 1.5 to over 3.0, depending on the study methods (Lidegaard et al. 2009; van Hylckama Vliegl et al. 2009; Jick and Hernandez 2011; Parkin et al. 2011; Gronich et al. 2011).
A careful review of these studies reveals important differences in the case and exposure definitions. The first studies included all cases of VTE including those with prior VTE, cardiovascular disease and other proximate causes such as surgery, lower limb injury, and trauma (non-idiopathic cases) (Heineman and Dinger 2004; Dinger et al. 2007). As has been demonstrated, inclusion of these cases can bias and dilute any differences in effects between different OCs. These studies may have also included non-current OC users, but this information cannot be ascertained from the published literature. Later studies applied different degrees of exclusion to the case definitions and the risk of drospirenone compared to levonorgestrel pills increased with the increasing number of exclusions. The Seeger study (Seeger et al. 2007) included all VTE cases except for those with a prior VTE, and included all users of all OCs other than drospirenone OCs in the reference group including women who had taken third generation OCs. All these factors would explain the null result for drospirenone OCs.
The studies of Lidegaard and van Hylckama Vliegl (Lidegaard et al. 2009; van Hylckama Vliegl et al. 2009) excluded women with prior cancer and cardiovascular disease and recent pregnancy, and results yielded small increased risks in drospirenone OC users compared to levonorgestrel OCs. The studies of Parkin and Jick (Jick and Hernandez 2011; Parkin et al. 2011) excluded women with these factors in addition to other exclusions such as recent trauma, surgery, limb injury, and women with other chronic diseases. The effects were strongest in the latter two studies (Jick and Hernandez 2011; Parkin et al. 2011). It is important to recognize the subtle differences in case definition when assessing these studies and to understand how they impact the results. When attempting to assess the presence of adverse drug effects it is important to identify the healthiest population possible so that the presence and magnitude of risk can be established without concern for bias, confounding and effect modification. Studies restricted to idiopathic cases thus yield the most interpretable and informative results. To estimate the risk of VTE in women with risk factors and other proximate causes, a clinical trial may be the only way to obtain unbiased results and it is unlikely that such a study will ever be conducted. At this time there seems to be some acceptance that there is a higher risk of VTE conferred by drospirenone-containing OCs but the magnitude of the risk is not yet established (Table 6.4).
Table 6.4
Summary of drospirenone-containing OC studies of VTE
|
Authors |
Citation |
Title |
Results |
Exposure and case details |
|
Heineman LA, Dinger J |
Drug Saf 2004;27:1001–18 |
Safety of a new oral contraceptive containing drospirenone |
12 VTEs on DRSP 11 on levo RR for DRSP vs levo = 1.0 |
Drospirenone vs levonorgestrel users. Timing of use not defined. All cases included regardless of prior history |
|
Dinger JC, Heinemann LA, Kühl-Habich D |
Contraception. 2007; 75(5):344–54 |
The safety of a drospirenone containing oral contraceptive: final results from the European Active Surveillance Study on oral contraceptives based on 142,475 women-years of observation |
26 VTEs on DRSP 25 on Levo. 52 other. HR = 1.0 |
Drospirenone vs levonorgestrel timing of use not defined. All cases included regardless of prior history of VTE or other comorbidities |
|
Seeger JD, Loughlin J, Eng PM, Clifford CR, Cutone J, Walker AM |
Obstet Gynecol. 2007;110(3):587–93 |
Risk of thromboembolism in women taking ethinylestradiol/ and other oral contraceptives |
14 VTEs on DRSP, 30 on Levo current users, HR = 1.0 |
Drospirenone vs all other OCs. current OC users. Included all first time VTEs: no exclusions |
|
Lidegaard Ø, Løkkegaard E, Svendsen AL, Agger C |
BMJ. 2009; 339:b2890 |
Hormonal contraception and risk of venous thromboembolism: national follow-up study |
2045 VTEs; 103 on DRSP, 201 on Levo, RR 1.64 |
Multiple comparisons including drospirenone vs levonorgestrel current OC users. Excluded some non-idiopathic cases |
|
van Hylckama VA, Helmerhorst FM, Vandenbroucke JP, Doggen CJM, Rosendaal FR |
BMJ. 2009 13;339:b2921 |
The venous thrombotic risk of oral contraceptives, effects of oestrogen dose and progestogen type: results of the MEGA case-control study |
1103 VTEs; current users 19 on DRSP 485 on Levo OR = 1.7 |
Drospirenone vs levonorgestrel current OC users. Excluded some non-idiopathic cases |
|
Dinger J, Assman A, Mohner S, Minh TD |
J FamPlannReprod Health Care, 2010;36:123–9 |
Risk of venous thromboembolism and the use of dienogest- and drospirenone-containing oral contraceptives: results from a German case-control study |
85 VTEs 25 on DRSP and 60 on Levo; DRSP vs Levo OR = 1.0 |
Drospirenone vs levonorgestrel timing of use not defined. All cases included regardless of prior history of VTE or other comorbidities |
|
Jick SS, Hernandez R |
BMJ 2011;340:d2151 |
Risk of non-fatal venous thromboembolism in women using oral contraceptive containing drospirenone compared with women using oral contraceptives containing levonorgestrel: case-control study using United States claims data |
121 VTEs on DRSP, 65 on Levo OR = 2.4 |
Drospireonone vs levonorgestrel current OC users. Excluded non-idiopathic cases |
|
Parkin L, Sharples K, Hernandez R, Jick SS |
BMJ 2011;340:d2139 |
Risk of venous thromboembolism in users of oral contraceptives containing drospirenone or levonorgestrel: nested case-control study based on the UK General Practice Research Database |
17 VTEs on DRSP, 44 on Levo OR = 3.3 |
Drospirenone vs levonorgestrel current OC users. Excluded non-idiopathic cases |
|
Lidegaard O, Nielson LH, Skovlund CW, Skjeldestad FE, Lokkegaard E |
BMJ 2011;343:d6423 |
Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001–9 |
196 VTEs on DRSP, and 123 on Levo OR = 2.2 |
Multiple comparisons including drospirenone vs levonorgestrel current OC users. Some VTE non-idiopathic cases excluded |
|
Gronich N, Lavi I, Rennert G |
Can Med Assoc J 2011;183:E1319–25 |
Higher risk of venous thrombosis associated with drospirenone-containing oral contraceptives: a population-bases cohort study |
99 VTEs on DRSP, N not reported for second generation OC OR =1.65 |
Women aged 12–50. Current OC use drospirenone vs second generation OCs. Included all first time VTEs: no exclusions |
Recent Study of Currently Available OCs
In a large Danish cohort study using data from Danish national registries of women aged 15–49 in 1995–2009 (Lidegaard et al. 2011), authors identified 4,307 cases of first time VTE after excluding women at high risk for VTE (those with prior CVD, cancers and coagulation disorders) and cases that occurred post sterilization, or during pregnancy or postpartum. Current OC exposure was evaluated according to progestin type, timing of use, duration, dose of estrogen, and prior OC use, compared to non-exposure or to levonorgestrel-OC exposure. OCs containing desogestrel, gestodene, cyproterone and drospirenone all had increased risks of around twofold for VTE compared to levonorgestrel OCs. These results are consistent with studies that carefully selected idiopathic cases, defined exposure as current use, and compared the OC risk in relation to levonorgestrel OCs (WHO 1995; Jick et al. 1995, 2000; Bloemenkamp et al. 1995; Kemmeren et al. 2001; Lidegaard et al. 2009, 2011; Vasilakis-Scaramozza and Jick 2001; van Hylckama Vliegl et al. 2009; Jick and Hernandez 2011; Parkin et al. 2011; Gronich et al. 2011).
The use of the Danish data enabled the researchers to identify a large number of women in a well-defined population, to identify virtually all cases of VTE, and to ascertain complete information on OC exposure with a common reference group. These factors distinguish it from earlier studies that did not find differences in the effects of the various OCs (Lidegaard et al. 1998; Spitzer et al. 1996; Dinger et al. 2007, 2010; Seeger et al. 2007). However, it should be noted that the authors did include some cases with other proximate causes of VTE, such as women with recent surgery or lower limb injury. Inclusion of these non-idiopathic cases would tend to dilute the relative effects of the various contraceptives since other causes would likely overwhelm any incremental differences in risk. Thus it is possible that the true relative risks are higher than those presented in the Danish study. While only some non-idiopathic cases were excluded the authors did exclude many cases who were at high risk of VTE and whose medical history could have confounded the OC-VTE relation.
How to Evaluate Risks Between Various OC Formulations
In order to critically read and evaluate the epidemiological literature there are some basic methods with which one should be familiar. There are two main study designs in the area of observational research (non-intervention studies); case-control and cohort studies (Rothman et al. 2008). In case-control studies cases of an outcome (VTE in these studies) are identified and compared to controls (non-cases) with respect to an exposure (OCs in these studies). If there is more exposure in the cases than the controls then it is said that the exposure increases the risk of the outcome. In a cohort study, a group of people exposed to, for example OCs, is followed forward to identify occurrences of a new outcome such as VTE and a rate is calculated. This rate is compared to the rate of VTE in people with no OC exposure. If the rate is higher in the exposed group compared to the unexposed group then it is said that the exposure increases the risk of the outcome under study.
In both study designs it is of utmost importance to ensure that the cases and controls or exposed and unexposed, are as similar as possible, so that other factors do not explain the difference in the exposure. So for example, the cases and controls or exposed and unexposed patients must all have the same age distribution so that one does not compare the rate of an outcome in non-comparable age groups since the rate of most outcomes varies greatly across different ages. The comparison of different studies of OCs and VTE must be made with careful consideration to the differences in study methods. The most important components to consider are the definition of exposure, the definition of outcome, and the selection of the comparison or referent group.
Exposure
How is OC exposure defined? It has been shown that current OC use confers an increased risk of VTE, but that the risk diminishes quickly upon discontinuation of the pill. Thus the relevant exposure definition for studies of OCs in relation to VTE is current use. Also, there are many OC formulations. It is important to be sure that similar formulations are compared and studied.
Case definition
The definition of the outcome must be clearly described and differences between studies should be identified. There are several important proximate causes of and risk factors for VTE. The most common causes of VTE in healthy young women (which describes most OC users), are recent pregnancy, surgery, lower limb injury, and major trauma. Important risk factors include a history of previous VTE, cardiovascular disease, renal failure, or auto immune disease. Thus it is important to exclude women with any of these proximate causes or strong risk factors from studies of drugs in relation to VTE. Inclusion of people with these important risk factors could bias the study result in either direction depending on the risk factor, the drug and the outcome. In the case of OCs and VTE, women with a history of VTE or cardiovascular disease would be unlikely to be prescribed an OC thus one could find a spurious protective effect of OC use in relation to a non-exposed referent if these women were included in a study. The inclusion of non-idiopathic cases (those with another proximate cause of the VTE) could bias the true result toward the null, since the incremental effect of either OC in women with other proximate causes (which could comprise a large proportion of all cases), is likely to be minimal.
Selection of the referent group
In some published studies of drospirenone in relation to VTE, third generation OCs were included in the comparison group, some used levonorgestrel and others had a non-exposed referent. Each of these comparators confers a different risk for VTE, thus the relative risk of another OC, such as the drospirenone OC, will vary according to which OC it is compared. When compared to the lowest risk OC (a second generation pill) the relative risk is greater than when compared to a third generation OC, but not as great as when compared to non-users of OCs. Third-generation OCs have been shown to increase the risk of VTE compared to levonorgestrel OCs, consequently, a comparator comprised of these OCs would lead to a lower relative risk of drospirenone OCs. Conversely, if the comparison group was made up of women who did not use any OC the effect measure, or relative risk, would be greater for drospirenone OC users. In other words, one must be aware of the referent group in each study to properly interpret the results. In epidemiology one must always ask: compared to what?
Regulatory Responses to the Risk of VTE with Oral Contraceptives
In 1995, before the three studies were published that found risks of around twofold for third generation compared to second generation OCs, the UK’s Medicines Control Agency (MCA) sent a “Dear Doctor letter” to all GPs in the UK, warning of the increased risk of third compared to second generation OCs. The UK Committee on Safety of Medicines (CSM) recommended switching women from third generation pills to other OCs and warned that women should consult their Doctors “to see if a change of pill is necessary.” (http://www.mhra.gov.uk/Safetyinformation/Safetywarningsalertsandrecalls/Safetywarningsandmessagesformedicines/CON019572; The pharma letter 1995) In the absence of published data to provide context, many women in the UK panicked and stopped taking the third generation pills. In the UK use of third generation OCs decreased by more than 80 % and because these pills were so widely used at that time, the supply of alternate pills was not sufficient to make up the deficit created by the pill’s discontinuation and some women were left with no OC replacement. Data suggest that many unplanned pregnancies resulted from this unfortunate series of events (Furedi 1999). It is interesting to note that the UK CSM recommendation was later reversed. The UK “Pill Scare” is discussed further in Chap. 19 which addresses communicating the risks and benefits of women’s medicines.
Regulatory action in other countries was not so dramatic and did not lead to mass discontinuation of the third generation pills or sudden material changes in OC prescribing. In Germany where the regulatory authorities had already recommended against taking third generation pills, the Federal Institute for Pharmaceutical and Medical Products issued a warning in October 1995 and prohibited prescribing of third generation pills to first time pill users under the age of 30. This action was reversed upon further review (BBC News Health U turn over pill scare). In the United States the authorities did not take immediate action. Norway followed the UK’s actions and in France the Agence du Medicament sent out a ‘communique de press’ in October 1995 which summarized the results of the three studies, and said that they and the European Agency would review the evidence. In the meantime they warned against stopping current OCs. The EMEA did not take further action as they thought the studies were flawed (http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2009/12/WC500017336.pdf).
In the late 2000s after several studies found increased risks of VTE in drospirenone-containing pills, the Food and Drug Administration (FDA) in the United States added a warning to the label of drospirenone pills (http://www.fda.gov/drugs/drugsafety/ucm299305.htm). Similar warnings were added in Canada (http://www.bayer.ca/files/YASMIN-PT3-ENG-03FEB2014-169558.pdf?#), and other European countries (http://www.yazontrial.com/2010/04/yasmin-label-change-in-the-european-union-eu/).
The only pills to have been removed from the market in some countries due to regulatory action, rather than due to improvements in formulations, have been the cyproterone-containing pills. It is important to note that while some OCs have higher risks of VTE compared to others, the risk is low for all OCs and lower than the risk of VTE in pregnancy.
Clinical Considerations
Clinical issues for prescribing oral contraceptives are discussed in detail in Chap. 5. Some key clinical issues relating to the risk of VTE with OCs are summarised here as follows:
· Whilst the absolute risk of VTE with all OCs is very low, this is a well-recognised, serious and potentially fatal adverse effect of these medicines.
· Prescribers should take the relevant steps to identify women at increased risk of VTE (take a careful history from each woman) and undertake appropriate risk management.
· Prescribers should identify women who should not be prescribed a combined oral contraceptive (COC) due to the risk of VTE and prescribing should be in accordance with the relevant Medical Eligibility Criteria and international and local guidelines (see Chap. 5).
· For women who should not take an OC due to a higher risk of VTE, prescribers should find another suitable form of contraception.
· Women requesting oral contraceptives should be advised about the risk of VTE: explain it is rare, but may be serious or fatal if it occurs.
· Put the risks into context with the benefits of OCs and the higher risk of VTE in pregnancy.
· Explain that the increased risk of VTE with OCs is highest when the pill is first started and returns to baseline after the pill is stopped.
· OCs should not be stopped without first consulting a doctor (or other health professional) as the woman will then be at risk of becoming pregnant if no other form of contraception is used.
· Explain the risks factually and in terms women can understand.
· Prescribe the OC with the lowest risk of VTE i.e. a second generation pill (remember efficacy is similar between different brands of combined OCs).
· Inform women of the signs and symptoms of VTE (e.g. unilateral swollen leg or acute breathlessness) and encourage early presentation to medical services.
Conclusions
That OCs increase the risk of venous (and arterial) thrombosis has been well recognised and well researched since OCs were first marketed. Since the earliest reporting of the higher risk of VTE in women taking COCs, the doses and formulations of both the oestrogen and progestin components of COCs have changed, initially resulting in a decrease in the risk of thrombotic events and later leading to differences in the risk between COCs. There are also well recognized risks factors for VTE that should be considered before choosing to take COCs. A history of cardiovascular disease and obesity are two important risk factors, as are a family history of VTE and immobility. Presence of certain chronic diseases such as cancer, kidney disease and autoimmune diseases may also increase a woman’s risk of VTE and thus OC use should be carefully considered in affected women.
Despite the increased risk of VTE in COCs users and the higher risk of drospirenone and third generation COCs, VTE is rare in healthy young women and the risks of VTE in pregnancy and postpartum are substantially higher than in OC users. Thus, preventing unintended pregnancies must remain a high priority and women should have multiple contraceptive options and be informed of each of their risks. Regardless of whether the thrombotic risk of third generation and drospirenone COCs compared to levonorgestrel COCs is increased by 1.5 or 3-fold, the absolute risk is still low.
Take Home Messages
· The risk of VTE in healthy young women (age 35 years or younger) is very low (see Table 6.5 above)
Table 6.5
Summary of approximate VTE rates with combined oral contraceptive pills
|
Exposure category |
Rate of all VTE a per 10,000 women aged < 35 |
|
Women not using a combined hormonal pill/patch/ring and are not pregnant |
About 2 out of 10,000 women |
|
Women using a COC containing levonorgestrel, norethisterone or norgestimate |
About 5–7 out of 10,000 women |
|
Women using a COC containing etonogestrel or norelgestromin |
About 6–12 out of 10,000 women |
|
Women using a COC containing drospirenone, gestodene or desogestrel |
About 9–12 out of 10,000 women |
|
Women using a COC containing chlormadinone, dienogest or nomegestrol |
Not yet known |
Source: European Medicines Agency report 2013
aAll VTE include VTEs in women with risk factors and other proximate causes such as recent surgery, pregnancy and trauma
· Combined oral contraceptives (COCs) increase the risk of VTE and this risk varies according to the oestrogen dose and type of progestin (see Table 6.5 above).
· Second generation COCs have the lowest risk of VTE of all currently available COCs, increasing the risk (compared to no use) by about threefold (see Table 6.5 above).
· Third generation and drosperinone COCs have a 1.5–3 fold higher risk of VTE than second generation pills (see Table 6.5 above).
· Despite the increased risk of VTE in COC users, the risk is still lower than in pregnancy and the post-partum period. In pregnancy the risk of VTE is 7–27 per 10,000 women.
· Other risk factors also increase the risk of VTE, including family history of VTE, cardiovascular disease, obesity and immobility.
· If there are no contraindications to COC use, doctors should prescribe a second generation pill initially
· Explain the risks (and benefits) of COCs factually and in terms each woman can understand
· Advise women about the signs and symptoms of VTE and that if these occur they should seek medical help immediately.
· Also explain that stopping OCs will result in loss of contraceptive protection and pregnancy may occur unless alternative contraceptive methods are used.
· The increased risk of VTE returns to baseline after cessation of therapy.
References
BBC News Health U-turn over pill scare: http://news.bbc.co.uk/2/hi/health/313848.stm
BCDSP (1973) Oral contraceptives and venous thrombo-embolic disease… Report from the BCDSP. Lancet 1(7817):1399–1404
Bloemenkamp KWM, Rsoendaal FR, Helmerhorst FM, Buller HR, Vandenbroucke JP (1995) Enhancement by factor V leiden mutation of risk of deep-vein thrombosis associated with oral contraceptives containing a third-generation progestagen. Lancet 346:1593–1596PubMedCrossRef
Buttar A, Seward S (2009) Enovid: the first hormonal birth control pill. Embryo project encyclopedia (2009-01-20). ISSN: 1940–5030. http://embryo.asu.edu/handle/10776/1956
Dinger JC, Heinemann LA, Kühl-Habich D (2007) The safety of a drospirenone containing oral contraceptive: final results from the European Active Surveillance Study on oral contraceptives based on 142,475 women-years of observation. Contraception 75(5):344–354PubMedCrossRef
Dinger J, Assman A, Mohner S, Minh TD (2010) Risk of venous thromboembolism and the use of dienogest- and drospirenone-containing oral contraceptives: results from a German case-control study. J Fam Plann Reprod Health Care 36:123–129PubMedCrossRef
Farmer RDT, Lawrenson RA, Thompson CR, Kennedy JG, Hambleton IR (1997) Population-based study of risk of venous thromboembolism associated with various oral contraceptives. Lancet 349:83–88PubMedCrossRef
Furedi A (1999) Social consequences. The public health implications of the 1995 ‘pill scare’. Hum Reprod Update 5:621–626PubMedCrossRef
Gronich N, Lavi I, Rennert G (2011) Higher risk of venous thrombosis associated with drospirenone-containing oral contraceptives: a population-bases cohort study. Can Med Assoc J 183(18):E1319–E1325CrossRef
Heineman LA, Dinger J (2004) Safety of a new oral contraceptive containing drospirenone. Drug Saf 27:1001–1018CrossRef
http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm?fuseaction=Search.DrugDetails
http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2009/12/WC500017336.pdf
http://www.fda.gov/drugs/drugsafety/ucm299305.htm
http://www.mhra.gov.uk/Safetyinformation/Safetywarningsalertsandrecalls/Safetywarningsandmessagesformedicines/CON019572
http://www.yazontrial.com/2010/04/yasmin-label-change-in-the-european-union-eu/
Inman WH, Vessey MP (1968) Investigation of deaths from pulmonary, coronary, and cerebral thrombosis and embolism in women of child-bearing age. BMJ 2:193–199PubMedCentralPubMedCrossRef
Inman WH, Vessey MP, Westerholm B, Engelund A (1970) Thromboembolic disease and the steroidal content of oral contraceptives a report to the committee on safety of drugs. BMJ 2:203–209PubMedCentralPubMedCrossRef
Jick SS, Hernandez R (2011) Risk of non-fatal venous thromboembolism in women using oral contraceptive containing drospirenone compared with women using oral contraceptives containing levonorgestrel: case-control study using United States claims data. BMJ 342:d2151PubMedCentralPubMedCrossRef
Jick H, Jick S, Gurewich V, Myers MW, Vasilakis C (1995) Risk of idiopathic cardiovascular death and nonfatal venous thromboembolism in women using oral contraceptives with differing progestagen components. Lancet 346:1589–1593PubMedCrossRef
Jick H, Kaye JK, Vasilakis-Scaramozza C, Jick S (2000) Risk of venous thromboembolism among users of third generation oral contraceptives compared with users of oral contraceptives with levonorgestrel before and after 1995: cohort and case-control analysis. BMJ 321:1190–1195PubMedCentralPubMedCrossRef
Kemmeren JM, Algra A, Grobbee DE (2001) Third generation oral contraceptives and risk of venous thrombosis: meta analysis. BMJ 323:131–134PubMedCentralPubMedCrossRef
Lidegaard Ø, Edstrøm B, Kreiner S (1998) Oral contraceptive and venous thromboembolism. A case-control study. Contraception 57:291–301PubMedCrossRef
Lidegaard Ø, Løkkegaard E, Svendsen AL, Agger C (2009) Hormonal contraception and risk of venous thromboembolism: national follow-up study. BMJ 339:b2890PubMedCentralPubMedCrossRef
Lidegaard Ø, Nielson LH, Skovlund CW, Skjeldestad FE, Løkkegaard E (2011) Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001–9. BMJ 343:d6423PubMedCentralPubMedCrossRef
Meade TW, Greenberg G, Thompson SG (1980) Progestogens and cardiovascular reactions associated with oral contraceptives and a comparison of the safety of 50- and 30- microgram oestrogen preparations. BMJ 280:1157–1161PubMedCentralPubMedCrossRef
National Institute for Health and Care Excellence (NICE) (2014) Addendum to clinical guideline 30, long-acting reversible contraception. NICE, London
Parkin L, Sharples K, Hernandez R, Jick SS (2011) Risk of venous thromboembolism in users of oral contraceptives containing drospirenone or levonorgestrel: nested case-control study based on the UK General Practice Research Database. BMJ 340:d2139CrossRef
Records unit of the Research Advisory Service of the Royal College of General Practitioners (1967) Oral contraception and thrombo-embolic disease. J R Coll Gen Pract 13(3):267–279
Rothman KJ, Greenland S, Lash TL (2008) Modern epidemiology. Lippincott Williams & Wilkins, Philadelphia
Rubig A (2003) Drospirenone: a new cardiovascular-active progestin with antialdosterone and antiandrogenic properties. Climacteric 6(Suppl 3):49–54PubMed
Sartwell PE, Masi AT, Arthes FG, Greene GR, Smith HE (1969) Thromboembolism and oral contraceptives. An epidemiologic case-control study. Am J Epidemol 90(5):365–380
Seaman HE, de Vries CS, Farmer RDT (2004) Venous thromboembolism associated with cyproterone acetate in combination with ethinylestradiol (Dianette): observational studies using the UK General Practice Research Data Base. Pharmocoepidemiol Drug Saf 13:427–436CrossRef
Seeger JD, Loughlin J, Eng PM, Clifford CR, Cutone J, Walker AM (2007) Risk of thromboembolism in women taking ethinylestradiol/ and other oral contraceptives. Obstet Gynecol 110(3):587–593PubMedCrossRef
Sitruk-Ware R (2004) Pharmacological profile of progestins. Maturitas 47:277–283PubMedCrossRef
Spitzer WO, Lewis MA, Heinemann AJ, Thoroughgood M, MacRae KD (1996) Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case–control study. BMJ 312:83–88PubMedCentralPubMedCrossRef
Stolley PD, Tonascia JA, Tockman MS, Sartwell PE, Rutledge AH, Jacobs MP (1975) Thrombosis with low-estrogen oral contraceptive. Am J Epidemiol 102:197–208
The pharma letter. Panic follow UK’s oral contraceptive warning. 30 Oct 1995
van Hylckama Vliegl A, Helmerhorst FM, Vandenbroucke JP, Doggen CJM, Rosendaal FR (2009) The venous thrombotic risk of oral contraceptives, effects of oestrogen dose and progestogen type: results of the MEGA case-control stud. BMJ 339:b2921
Vasilakis C, Jick H, Melero-Monte MM (1999) Risk of idiopathic venous thromboembolism in users of progestagens alone. Lancet 354:1610–1611
Vasilakis-Scaramozza C, Jick H (2001) Risk of venous thromboembolism with cyproterone or levonorgestrel contraceptives. Lancet 358:1427–1429PubMedCrossRef
Vessey MP, Doll R (1968) Investigation of relation between use of oral contraceptives and thromboembolic disease. Br Med J 2(5599):199–205PubMedCentralPubMedCrossRef
Vessey MP, Doll R (1969) Investigation of relation between use of oral contraceptives and thromboembolic disease. A further report. Br Med J 2(5658):651–657PubMedCentralPubMedCrossRef
Vessey M, Mant D, Smith A, Yeates D (1986) Oral contraceptives and venous thromboembolism: findings in a large prospective study. Br Med J (Clin Res Ed) 292:526CrossRef
Wharton C, Blackburn R (1988) Lower dose pills. Popul Rep A 16(7):1–31
WHO (1995) Effect of different progestagens in low oestrogen oral contraceptives on venous thromboembolic disease. Lancet 346:1582–1588CrossRef