Arch Gynecol Obstet (2014) 289:413419 DOI 10.

1007/s00404-013-2983-9

REPRODUCTIVE MEDICINE

Risk of venous thrombosis in users of hormonal contraceptives in German gynaecological practices: a patient database analysis
M. Ziller V. Ziller G. Haas J. Rex K. Kostev

Received: 21 May 2013 / Accepted: 24 July 2013 / Published online: 4 August 2013 Springer-Verlag Berlin Heidelberg 2013

Abstract Purpose Recent studies showed differences in the risk of venous thrombosis between different combined hormonal contraceptives. Database studies comprising large cohorts can add relevant aspects from daily clinical practice. The purpose of this study was to evaluate different progestogen in combination with ethinylestradiol on the risk of venous thrombosis in Germany. Methods Computerized data from 68,168 contraceptive users in gynecological practices throughout Germany (Disease Analyzer Database) were analyzed. The adjusted odds ratios for risk of thrombosis were estimated in users of different oral contraceptive (OC) formulations relative to users of levonorgestrel-containing preparations. Results In total, 38 (0.06 %) of the 68,168 contraceptive users had a recorded diagnosis of thrombosis within 365 days after the initial prescription. The adjusted risk was 1.95 for desogestrel (95 % CI 0.527.29), 2.97 for dienogest (95 % CI 0.969.24), 1.57 for drospirenone (95 % CI 0.465.38), 2.54 for chlormadinone (95 % CI 0.729.04), and 3.24 for norgestimate (95 % CI 0.5917.75) compared to levonorgestrel. None of those ndings reached statistical signicance. The maximum absolute increase versus levonorgestrel was 6 cases per 10,000 women (n.s.).

Conclusion The study shows the low incidence rates of thrombosis in OC users. Since there is no signicant difference, this study does not conrm an increased risk but shows only a tendency for this risk of third- and fourthgeneration OC versus levonorgestrel-containing products. Keywords Contraception Thrombosis Disease Analyzer

Introduction For decades, the risk of thrombosis in combined contraceptives has attracted medial attention. Unfortunately, the often over-exaggerated risk has lead to increased uncertainty among both patients and physicians. As early as the 1990s, preparations with desogestrel and gestodene were discredited due to an acute thrombotic risk, and a warning was including in the prescribing information. Public discussion and a loss of condence in the contraceptive pill led to the use of less reliable methods of contraception and, ultimately, to an increase in pregnancy and induced abortions [1]. Today, the majority of oral contraceptives (OCs) contain ethinylestradiol, whereas the progesterone component varies and carries different side effects. Even though modern OCs are much better tolerated, regular intake is still associated with side effects. Studies indicate a clear association between combined OCs (COCs) and the increased risk of thrombosis [26]. At an incidence of 110 cases/10,000 woman years (in comparison, 35 cases/10,000 woman years without COCs), thrombosis in pill users can be interpreted as a rare event. A pregnancy would lead to 840 cases/10,000 woman years and, immediately after birth, one could expect [40 cases/10,000 women years [2]. The

M. Ziller V. Ziller K. Kostev Department of Gynecological Endocrinology, Reproductive Medicine and Osteoporosis, Philipps-University of Marburg, Marburg, Germany G. Haas J. Rex K. Kostev (&) dter Landstr. 108, Epidemiology, IMS Health, Darmsta 60598 Frankfurt, Germany e-mail: kkostev@de.imshealth.com

123

414

Arch Gynecol Obstet (2014) 289:413419

highest incidence with contraceptives has been described in the rst months of treatment and diminishes continuously over time [6, 7]. In relation to the dependence of thrombosis risk, a number of studies have proposed an increased risk when using different progestogens. Hence, the discussion is still open given that a number of epidemiological studies have shown no difference in thrombosis risk between different progestogens [811], whereas others found an increased risk with the so-called third- and fourth-generation COCs (e.g., gestodene, desogestrel, and drospirenone) compared with second-generation levonorgestrel-containing preparations [3, 7, 12]. Recently, discussions about ethinylestradiol dose and progesterone type potentially increasing the risk of thromboembolism have been spurred by the publication of a Danish cohort study [7] and the results of a MEGA case control study from The Netherlands [13]. The overall conclusion indicated that currently available oral contraceptives still have a major impact on thrombosis occurrence, and many women may not use the safest contraception given the risk of venous thrombosis. The substance drospirenone was the particular focus of the American Food and Drug Administration (FDA). As a result of recent reports, all brands containing drospirenone needed to include detailed information in their labels about the possibility of an increased risk of blood clots [14]. The evidence for the risk of thrombosis associated with OCs is still conicting, and further research is needed. Epidemiological and database studies including large cohorts could be most useful to further contribute and additionally investigate clinically important factors associated with the risk of thrombosis. The purpose of this study is to evaluate whether different progestogen in combination with ethinylestradiol raise the risk of venous thrombosis events (VETs) compared with second-generation progesterone levonorgestrel in Germany.

interfaces and provide daily routine information about patients diseases and therapies. A practice transmits patient data stored in the physicians computer to IMS on a monthly basis. Before transmission, the data are encrypted for privacy purposes and information from patients les from a doctors practice is similar in scope and detail. In total, the database includes data from approximately 3,000 practices and 20 million patients in Germany collected over the last 10 years. The validity of the Disease Analyzer Database has been analyzed and the outcomes have been published elsewhere [16]. In addition, it has been the basis of a number of studies and peer-reviewed scientic publications in the eld of epidemiology in, for example, venous thromboembolism research [1720]. Selection of study patients All women who visited their gynecologist between 1 January 2005 and 31 December 2010 were included in the analysis. Women between the ages of 16 and 45 with a prescription for a combined oral hormonal contraceptive and receiving followup at least 1 year after the initial prescription were included and followed until the end of the study period or upon receiving a conrmed diagnosis of thrombosis. They were allowed to have prescriptions for other contraceptives, but should not have taken a study substance prior to the index date. Furthermore, women with a conrmed diagnosis of thrombosis including arterial embolism and thrombosis (ICD 10: I74), phlebitis and thrombophlebitis (ICD 10: E80), and portal vein thrombosis (ICD 10: E81) or with a history of antithrombotic agent use (ATC: B01) prior to the index date were excluded. Women were eligible for inclusion as potential cases if they were aged 1645 years as of the index date. To be included in the study, patients had to have at least 1 year of recorded medical information before the index date and at least 1 year after the index date. In total, 68,168 women were included in the analysis. Study outcome The main analysis estimated the risk of thrombosis in users of combined oral contraceptives including ethinylestradiol and desogestrel, drospirenone, dienogest, chlormadinone, cyproterone, and norgestimate in relation to current users of levonorgestrel preparations within 1 year after their initial prescriptions. Norethisterone, lynestrenol, and gestodene were not analyzed due to the small patient numbers in the database. Statistical analysis Data were analyzed using SAS software version 9.2 (SAS Institute, Cary, NC, USA). Descriptive analysis presented

Patients and methods Disease Analyzer Database The Disease Analyzer Database (IMS HEALTH) collects drug prescriptions, diagnoses, and basic medical and demographic data directly obtained from the computer systems of general practitioners and specialists throughout Germany [15]. Diagnoses (ICD-10), prescriptions (Anatomical Therapeutic Chemical (ATC) Classication System), and the quality of reported data are continuously monitored by IMS based on a number of quality criteria (e.g., completeness of documentation, linkage of diagnoses and prescriptions). The data are obtained directly from the computers in the physicians practices via standardized

123

Arch Gynecol Obstet (2014) 289:413419

415

absolute and relative frequencies for the diagnosis of thrombosis in each treatment group; 95 % binomial lower and upper condence limits for prevalence were used. Logistic regression was used to compute odds ratios and 95 % condence intervals in the analyses. The key analysis estimated the risk of non-fatal thrombosis in current users of the dened contraceptive relative to current users of levonorgestrel preparations. The variables included in the regression model were thrombosis (yes or no), type of oral contraceptive (levonorgestrel as the reference group; desogestrel, drospirenone, dienogest, etonogestrel, chlormadinone, cyproterone, norgestimate, or norelgestromin), and the following covariates: age at index date; body mass index (BMI); insurance status (private versus statutory health); region (West versus East Germany); history of hormonal contraceptive use (ATC: G03A or cyproterone); history of use of other sexual hormones or modulators of the genital system (ATC: G03 excluding G03A); follow-up examination after surgery (ICD 10: Z080, Z090) within 365 days prior to the index date; and history of pregnancy, childbirth, and the puerperium period (ICD 10: O). In addition, heart diseases as of the index date were included as covariates: ischemic heart diseases (ICD 10: I20I25); pulmonary heart disease and diseases of pulmonary circulation (ICD 10: I26I28); other forms of heart disease (ICD 10: I30I52); cerebrovascular diseases (ICD 10: I60I69); diseases of arteries, arterioles, and capillaries (ICD 10: I70I79, excluding I74); diseases of the veins, lymphatic vessels, and lymph nodes not elsewhere classied (ICD 10: I80I89, excluding I80I81); and other unspecied disorders of the circulatory system (I95I99). Good practice methods for retrospective database studies were considered [21].
Fig. 1 Patient selection criteria (Disease Analyzer Database, IMS HEALTH)

Results The database included 347 gynecological practices with 426 doctors continuously reporting to IMS HEALTH during the study period, comprising 1,740,615 patients (Fig. 1). Patients Incidence of thrombosis Table 1 summarizes the characteristics of the 68,168 study participants. The mean age was 29.6 years (SD 7.4). Chlormadinone users were the youngest with a mean age of 28.2 years (SD), while norgestimate users were the oldest with a mean age of 29.5 years (SD). Among those with a recorded BMI, the mean body mass index was 23.5 (SD 4.7) and there were no signicant differences between groups. In addition, 8.3 % of the total study population was Out of the 68,168 contraceptive users included in this study, 38 (0.06 %) had a recorded diagnosis of thrombosis within 365 days after their rst prescriptions. The number of patients with diagnosed thrombosis varied between the groups (Fig. 2). In levonorgestrel users, thrombosis was documented in 0.03 % (95 % CI 0.000.06); in norgestimate users, 0.09 (95 % CI 0.000.22); and in dienogest, privately insured; 7.2 % among levonorgestrel users and 10.1 % among drospirenone users. In total, 45.7 % of the study patients used a different contraceptive compound prior to the index date, while 54.3 % were naive contraceptive users.

123

416

Arch Gynecol Obstet (2014) 289:413419

Table 1 Baseline characteristics of contraceptive users in German gynecological practices: IMS HEALTH Disease Analyzer
Variables N Age (years) Body mass index (kg/m2) Private health insurance (%) Region (West Germany) (%) Prescriptions within 365 days prior to index date Hormonal contraceptives Other sex hormones/ modulators of the genital system Diagnoses prior to ID Follow-up examination after surgery Pregnancy, childbirth and the puerperium Heart diseases 2.7 19.8 4.7 2.7 16.5 3.6 2.7 21.3 4.8 2.5 19.0 4.8 2.9 19.1 4.6 2.4 20.5 4.8 2.7 20.0 5.0 2.9 28.7 6.6 45.7 7.9 40.3 7.2 32.9 7.0 48.3 7.4 48.2 8.0 49.7 7.9 56.5 10.5 52.8 10.4 Total 68,168 29.6 (7.4) 23.5 (4.7) 8.3 83.4 Desogestrel 8,375 30.7 (7.1) 23.7 (4.8) 7.8 84.2 Levonorgestrel 13,222 30.2 (7.7) 23.8 (4.9) 7.2 83.8 Drospirenone 15,572 29.2 (7.5) 23.7 (4.7) 10.1 87.0 Dienogest 13,785 29.2 (7.6) 23.3 (4.7) 8.8 81.4 Chlormadinone 9,194 28.2 (7.3) 23.3 (4.3) 8.0 83.7 Cyproterone 5,837 28.7 (7.3) 23.3 (4.9) 7.7 84.1 Norgestimate 2,183 29.5 (7.2) 23.9 (4.7) 5.4 65.4

Fig. 2 Share of women with a recorded diagnosis of thrombosis within 12 months after initial prescription of dened contraceptive substance

0.09 % (95 % CI 0.040.14). There were no signicant differences in the incidence of thrombosis between the groups. Risk of thrombosis The multivariate odds ratios of the logistic regression models are shown in Fig. 3. The adjusted risk for thrombosis diagnosis was elevated but not signicantly higher for patients using desogestrel, drospirenone, dienogest, chlormadinone, cyproterone, or norgestimate compared with levonorgestrel. The risk of thrombosis increased with age

when age was included as a continuous variable (OR 1.10; 95 % CI 1.051.14; p \ 0.001). Regression analyses showed that the thrombosis risk in contraceptive use was not affected by other covariates. Ethinylestradiol dosages were distributed as follows: 21,180 (33.6 %) 20 lg, 34,584 (54.8 %) 30 lg, and 7,306 (11.6 %) 35 lg and higher. There were no signicant differences in dosing between the progestogen groups. When analyzed for thrombosis risk, there was no signicant effect of ethinylestradiol dose on thrombosis risk (dose 35 and higher versus 20: 1.19 (0.373.80), p = 0.885; dose 30 versus 20: 1.42 (0.672.99), p = 0.358).

123

Arch Gynecol Obstet (2014) 289:413419

417

Discussion Over the past four decades, numerous studies have examined the risk of venous thromboembolism in users of combined oral contraceptives. There is clear evidence for a higher risk among current users compared with non-users. Early studies focused on the effects of estrogen components although recent reports demonstrated that the progestogen content may also be important. Focus sharpened on the progestogen levels in the mid-1990s, when several publications reported an increased risk of venous thromboembolism in users of combined oral contraceptives including desogestrel or gestodene compared with the older progestogen levonorgestrel. These results were replicated in subsequent studies and led to intensive scientic and legal scrutiny [22, 23]. Our community-based study evaluated the risk of thrombosis in patients receiving combined oral contraceptives. In 68,168 users over a 5-year study period, 38 thromboembolic events were recorded. When grouped for different progestogens, the number of patients with diagnosed thrombosis varied between the groups. Our results showed differences between the substances with an elevated risk for drospirenone, cyproterone, chlormadinone, dienogest and norgestimate versus levonorgestrel. These differences did not reach statistical signicance. Levonorgestrel is part of the second generation of OCs and generally considered safer with regard to thrombosis risk compared with third-generation OCs including desogestrel and norgestimate [24]. Second-generation OCs still tend to have a residual androgenic effect which, in some users, can lead to bothersome side effects such as acne, hirsutism, and weight gain [25]. Third-generation oral contraceptives were introduced to reduce the androgenic side effects;

17-a-hydroxyprogesterone derivatives such as chlormadinone and cyproterone show an anti-androgenic effect. The improved side effect prole, especially regarding skin appearance and weight change due to reduced uid retention (edema), have made the use of fourth-generation contraceptives very popular. Several studies have shown that the risk of stroke and myocardial infarction seems to have decreased. However, this new generation of contraceptives seems to increase the risk of thrombosis [26]. According to our ndings, it is difcult to determine the risk for thrombosis with regard to the progestogen used in various oral contraceptives. The differences in absolute numbers were small with a maximum of six cases per 10,000 women and did not reach statistical signicance. A doubling in risk can be a threatening number; but, on the other hand, a vast majority of users had a chance to benet from a specic treatment prole. This study has several limitations. Assessment of comorbidity relied on ICD codes by primary care gynecologists only. Patients could be observed in only one physicians practice; when they received prescriptions from or diagnoses by other doctors or in hospitals, those prescriptions and diagnoses were not necessarily documented by the gynecologists included in this analysis. Data pertaining to socioeconomic status and lifestyle-related risk factors (i.e., smoking, alcohol use, physical activity) were also lacking. However, in Germany, gynecologists are routinely responsible for prescriptions of contraceptives and do record side effects, such as thrombosis diagnoses. According to the regression analysis in our data set, the only signicant risk factor for thrombosis was age. The expected factor of ethinylestradiol dosing did not show any signicant differences. This is probably also due to the number of cases. Thrombosis

Fig. 3 Association between contraceptive use with incidence of thrombosis in patients treated in German gynecological practices: logistic regression analyses (levonorgestrel is the reference group)

123

418

Arch Gynecol Obstet (2014) 289:413419 Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Lancet 346(8990):15751582 Rabe T, Luxembourg B, Ludwig M et al (2011) Contraception and thrombophiliaa statement from the German Society of Gynecological Endocrinology and Reproductive Medicine (DGGEF e. V.) and the Professional Association of the German Gynaecologists (BVF e. V.). J Reprod Med Endocrinol 8(Sonderheft 1):178218 Wu C, Grandi S, Filion K, Abenhaim H, Joseph L, Eisenberg M (2013) Drospirenone-containing oral contraceptive pills and the risk of venous and arterial thrombosis: a systematic review. BJOG 120(7):801811 Plu-Bureau G, Maitrot-Mantelet L, Hugon-Rodin J, Canonico M (2013) Hormonal contraceptives and venous thromboembolism: an epidemiological update. Best Pract Res Clin Endocrinol Metab 27(1):2534 Lidegaard , Nielsen LH, Skovlund CW, Skjeldestad FE, Lkkegaard E (2011) Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study, 2001-9. BMJ 343:d6423 hl-Habich D (2007) The safety of Dinger JC, Heinemann LA, Ku a drospirenone-containing oral contraceptive: nal results from the European Active Surveillance Study on Oral Contraceptives based on 142,475 women-years of observation. Contraception 75:344354 hner S et al (2010) Risk of venous Dinger J, Assmann A, Mo thromboembolism and the use of dienogest- and drospirenonecontaining oral contraceptives: results from a German case control study. J Fam Plann Reprod Health Care 36:123129 Farmer RD, Todd JC, Lewis MA et al (1998) The risks of venous thromboembolic disease among German women using oral contraceptives: a database study. Contraception 57:6770 Seeger JD, Loughlin J, Eng PM et al (2007) Risk of thromboembolism in women taking ethinylestradiol/drospirenone and other oral contraceptives. Obstet Gynecol 110:587593 Jick H, Jick SS, Gurewich V et al (1995) Risk of idiopathic cardiovascular death and nonfatal venous thromboembolism in women using oral contraceptives with differing progestogen components. Lancet 346:15891593 van Hylckama Vlieg 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 casecontrol study. BMJ 339:b2921 Lenzer J (2011) US panel rules that warnings on two birth control pills be strengthened. BMJ 343:d8104 Kostev K, Haas G (2011) Medical care in Germany. Optimus, ttingen Go der-Bernhardi D (2009) Validity and Becher H, Kostev K, Schro representativeness of the Disease Analyzer patient database for use in pharmaco-epidemiological and pharmacoeconomic studies. Int J Clin Pharmacol Ther 47(10):617626 Clayton TC, Gaskin M, Meade TW (2011) Recent respiratory infection and risk of venous thromboembolism: casecontrol study through a general practice database. Int J Epidemiol 40(3):819827 Nightingale AL, Lawrenson RA, Simpson EL (2000) The effects of age, body mass index, smoking and general health on the risk of venous thromboembolism in users of combined oral contraceptives. Eur J Contracept Reprod Health Care 5(4):265274 Lawrenson R, Farmer R. Venous thromboembolism and combined oral contraceptives: does the type of progestogen make a difference? Contraception. 2000;62(2 Suppl):21S28S (discussion 37S38S) Farmer RD, Lawrenson RA, Todd JC, Williams TJ, MacRae K (1999) Oral contraceptives and venous thromboembolic disease. Analyses of the UK General Practice Research Database and the UK Mediplus database. Hum Reprod Update 5(6):688706

in young adults is rare. Even with 1.7 million patients in the database and nearly 70,000 patients analyzed, only 38 thrombosis events were detected. This small number provides reduced statistical power, but also shows that numerous gynecologists will rarely diagnose thrombosis associated with contraception use. In total, 347 practices recorded 38 cases of thrombosis associated with combined oral hormonal contraceptives over 5 years. Thus, the incidence of venous thrombosis found in this community-based study is similar to other reports (7/10,000 women years), suggesting that the study captured the majority of events [27]. Further important limitation is the missing information of ethinylestradiol dose in the used data. Oral contraceptives are also in use for the symptomatic treatment of dysmenorrhea or menstrual disorders. Androgen-like symptoms such as acne, seborrhea, and hirsutism can be effectively reduced with third- and fourth-generation contraceptives [28]. It is, therefore, clinically important to differentiate the specic side effect proles for different progestogens and dene an individual treatment strategy. If thrombosis risk factors are lacking, preparations should be chosen according to their side effect prole rather than thrombosis risk. Our results are in line with several recent reports and support the thesis of differences between different progestogens and thrombosis risk. Still, the data also showed again that the risk is small and the differences in absolute numbers are of questionable clinical relevance in daily clinical practice. Given that venous thromboembolism is a very rare event and associated with individual risk factors, it is unreasonable to expect a denitive answer to the progestogens and thrombosis question from a randomized controlled trial. We, therefore, may never be able to exclude residual confounding factors as a possible explanation for the higher VTE rates found with newer progestogens. Luckily, in clinical practice, this does not matter much. For COCs, as for any treatment, health professionals should rst recommend the safest and most effective treatment [29], and chose a patient-specic strategy after adequate anamnesis and patient information.
Conict of interest We declare that we have no conict of interest.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14. 15. 16.

17.

References
1. Szarewski A, Mansour D (1999) The pill scare: the responses of authorities, doctors and patients using oral contraception. Hum Reprod Update 5(6):627632 2. Khader YS, Rice J, John L, Abueita O (2003) Oral contraceptives use and the risk of myocardial infarction: a meta-analysis. Contraception 68(1):1117 3. World Health Organization (1995) Venous thromboembolic disease and combined oral contraceptives: results of international multicentre casecontrol study. World Health Organization

18.

19.

20.

123

Arch Gynecol Obstet (2014) 289:413419 21. Motheral B, Brooks J, Clark MA (2003) A checklist for retrospective database studiesreport of the ISPOR Task Force on Retrospective Databases. Value Health 6(2):9097 22. Inman WHW, 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 2:203209 23. Hannaford PC (2011) The progestogen content of combined oral contraceptives and venous thromboembolic risk. BMJ 25(343):d6592 24. Fazio G, Ferrara F, Barbaro G (2010) Prothrombotic effects of contraceptives. Curr Pharm Des 16(31):34903496 25. Thorneycroft IH (1999) Update on androgenicity. Am J Obstet Gynecol 180(2 Pt 2):288294

419 26. Leblanc ES, Laws A (1999) Benets and risks of third-generation oral contraceptives. J Gen Intern Med 14(10):625632 27. Shapiro S, Dinger J (2010) Risk of VTE among users of oral contraceptives. J Fam Plann Reprod Health Care 36(2):103 user M, Braendle W, Kuhl H, Mu ck AO, Neulen J, Thaler 28. Birkha C, Wildt L (2011) Aktuelle Empfehlungen zur hormonalen chskre rcher Gespra Kontrazeption44. Arbeitstreffen des Zu kologische Endokrinologie r Gyna ises vom Mai 2010. Journal fu 5(1):1725 29. Dinger J (2009) Oral contraceptives and venous thromboembolism: old questions revisited. J Fam Plann Reprod Health Care 35(4):211213

123