Venous thromboembolic disease in users of low-estrogen combined estrogen-progestin oral contraceptives

Contraception 70 (2004) 3–10

Original research article

Venous thromboembolic disease in users of low-estrogen combined estrogen-progestin oral contraceptives Stephen Sidneya,*, Diana B. Petittib, Gerald A. Soffc, Deborah L. Cundiffc, Kimberly K. Tolana, Charles P. Quesenberry, Jr.a a

Division of Research, Kaiser Permanente Medical Care Program (Northern California), 2000 Broadway, Oakland, CA 94612, USA b Research and Evaluation Department, Kaiser Permanente Medical Care Program (Southern California), Pasadena, CA, USA c Northwestern University, Chicago, IL, USA Received 24 September 2003; received in revised form 6 January 2004; accepted 2 February 2004

Abstract Objective: To assess the relationship between venous thromboembolic disease (VTE) and use of low-estrogen dose (⬍50 ␮g) combined estrogen-progestin oral contraceptives (OC) and three thrombosis-related gene mutations in a United States population. Design: This case-control study was conducted in 1998 –2000 among women ages 15– 44 years who were members of the Kaiser Permanente Medical Care Program [KPMCP] (Northern and Southern California). Cases were women with incident VTE; about three times as many women frequency matched for age were randomly selected as controls from the KPMCP membership in the same years. Data were collected in a 1 h face-to-face interview; blood was drawn to extract DNA to test for gene polymorphisms. The analysis data set comprised 196 cases (mean age 35.3 years) and 746 controls (mean age 36.2 years). Results: The adjusted odds ratio (OR) for VTE associated with current OC use was 4.07 (95% confidence interval [CI]: 2.77– 6.00). The OR associated with OC use was higher for women who were obese than in the nonobese (p ⫽ 0.01 for likelihood test for interaction) and in women without predisposing medical conditions (p ⫽ 0.02 for interaction). The adjusted OR for VTE was 7.10 (95% CI: 2.33–21.61) in women with factor V Leiden (G1691A) mutation, 2.83 (95% CI: 0.70 –11.63) in women with prothrombin G20210A mutation and 0.26 (95% CI: 0.10 – 0.65) in women with the MTHFR C677T mutation. The OR for VTE in OC users with factor V Leiden mutation (11.32) was elevated more than in OC users without the mutation (3.20) and women with the mutation who were non-OC users (8.42), but confidence intervals overlapped. Conclusions: The risk of VTE is increased in users of low-estrogen OC formulations. Obese women appear to be at greater risk of VTE when using OCs. © 2004 Elsevier Inc. All rights reserved. Keywords: Oral contraceptives; Venous thromboembolism; Factor V; Prothrombin; MTHFR; Case-control studies

1. Introduction Venous thromboembolism (VTE), consisting of deep venous thrombosis (DVT) and pulmonary embolus (PE), is the most common vascular disease in women of childbearing age. It occurs at a rate of approximately 1 per 10,000 women-years among nonpregnant women who do not use oral contraceptives [1]. Oral contraceptives (OCs) are a popular form of contraception in the United States since their introduction in the 1960s. Early epidemiological studies showed that OC use * Corresponding author. Tel.: ⫹1-510-891-3753; fax: ⫹1-510-8913761. E-mail address: [email protected] (S. Sidney). 0010-7824/04/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2004.02.010

increased the risk of VTE by a factor of 4 – 8 range, with risk estimates ranging from 1.6 to 11 [2–22]. The estrogen component of combined estrogen-progestin OCs was considered to be the mediator of VTE risk, and doses were decreased. Formulations with 50 ␮g or more of estrogen are now rarely used. There are relatively few studies that have assessed VTE risk specifically in users of low-estrogen formulations [2,16,18 –22], and only one study [21] has involved women in the United States. These studies have generally found relative risks in the 3– 4 range. The possibility that the risk of VTE in OC users might be concentrated in a relatively few women who might be identified prior to initiation of OC use is underexplored. We conducted a population-based case-control study to determine the relationship of low-estrogen dose OCs with

4

S. Sidney et al. / Contraception 70 (2004) 3–10

incident VTE in women in a large prepaid medical care program. We also assessed the association with VTE of three gene mutations and other known and potential VTE risk factors and the modifying effect these factors have in OC users.

2. Materials and methods The institutional review boards of Kaiser Permanente (Northern and Southern California) approved the project. All subjects provided written consent for the interview and separate written consent for having blood drawn to extract DNA and assess gene mutations. We attempted to identify all episodes of VTE that occurred between March 1998 through June 2000 in female members, 18 – 44 years of age, of the Northern and Southern California regions of the Kaiser Permanente Medical Care Program (KPMCP). In both regions, sources for case identification included hospital admission and discharge records and claims for non-Kaiser Permanente hospitalizations. For Northern California, outpatient visit records were also used to identify cases. For Southern California, diagnosis codes for outpatient visits were not available in a computerized database, but patients who were given enoxaparin for outpatient anticoagulation were identified as possible VTE cases. Screening for eligibility as cases was performed by review of medical records and imaging studies for VTE. Only women with definite and probable VTE were to be included. Diagnostic criteria for definite VTE were diagnostic venogram (for DVT) and diagnostic pulmonary arteriogram (for PE). Diagnostic criteria for probable VTE were clinic suspicion of DVT with an imaging study consistent with DVT (compression ultrasonography, impedance plethysmography or radioisotope scan) and clinical suspicion of PE and high-probability pulmonary ventilation-perfusion scan. After the study onset, diagnostic spiral CT scan was added as a criterion for probable DVT. We excluded nonincident cases and patients who were pregnant on the date of the VTE episode. We frequency matched four potential controls, randomly selected from Kaiser Permanente female members in the same 5-year age stratum, with the intent of interviewing approximately three controls for each case. Eligible subjects and controls were interviewed in person by trained interviewers using a standardized instrument. Subjects were interviewed as soon as possible after their VTE event; the mean interval between the index date (date of medical treatment for cases and date of assignment to an interviewer for controls) and the interview date was 82.0 days for cases and 77.6 days for controls. If a woman who was a case died or was unable to communicate verbally, an attempt was made to interview a proxy, preferably the woman’s husband or her live-in companion if she was married or living as married. If the woman was not married

or living as married, or if the husband or live-in companion did not feel they could provide accurate information, the women’s daughter, mother, sister or a close friend was approached to carry out a proxy interview, with preference in the order stated. Surrogate interviews were not conducted for controls. All interview questions were asked in relation to the index date. In interviews of nonsurrogates, the calendar method, which gathers contraceptive-use information by structuring questions in relation to significant life events, was used to obtain information on all contraceptive methods ever used. A picture book of all OC formulations ever marketed in the US was used to facilitate recall of the formulation of the OCs used currently and in the past. Surrogate interviews gathered information only about ever OC use and OC use in the month prior to the index date. The questionnaire included questions about medical history, prolonged immobilization, demographics, health behaviors and habits and use of aspirin and vitamin supplements [23,24]. Current and occasional smokers were combined in the analysis because of the small number of occasional smokers. Body mass index (BMI) (kg/m2) was calculated using self-reported height and weight. Medical record review and self-reported history of the following medical conditions were used to define a composite variable indicating the presence of a preexisting condition that might increase the risk of VTE as follows: within 6 weeks postpartum; history of cancer, myocardial infarction, stroke, congenital heart disease, congestive heart failure, blood clotting disorder, systemic lupus erythematosis, diabetes, gallbladder disease, varicose veins or peripheral vascular disease; recent fracture of the spine, pelvis, femur or tibia; spinal cord injury; recent surgery (orthopedic, thoracic, abdominal, genitourinary); venulitis. The prevalence of myocardial infarction, peripheral vascular disease, spinal cord injury and venulitis were each 1% or less in both cases and controls, so these data were not individually shown in the tables, though they were included in the composite variable. Self-reported history of any of the following conditions during the 6 weeks prior to the index date was used to define a composite variable indicating recent prolonged immobilization: leg fracture or other leg injury; overnight hospitalization; prolonged bed rest; airplane flight of 4 h or longer. We attempted to obtain a blood sample for DNA extraction at the time of the interview from all women who were cases and approximately one of three women in each 5-year age group who were controls. Blood samples were obtained in an ethylenediaminetetraacetic acid tube, refrigerated and shipped to Northwestern University. DNA was extracted from whole blood using a Puregene DNA Isolation (Gentra Systems, Inc, Minneapolis, MN, USA). Purified DNA was then amplified using polymerase chain reaction primers specific for the fragment of the gene that contained the site of the mutation of interest. Initially, analysis for two gene mutations was performed: the factor V Leiden (G1691A) mutation and the methylenetetrahydrofolate reductase

S. Sidney et al. / Contraception 70 (2004) 3–10

(MTHFR) C677T mutation [25,26]. Identification of the prothrombin gene G20210A mutation was added later [27]. Women were considered to be MTHFR positive if they were homozygous for the C677T polymorphism, factor V Leiden-positive if either homozygous or heterozygous for the G1691A polymorphism, and prothrombin positive if either homozygous or heterozygous for the G20210A polymorphism. The exposure odds ratio (OR) was used to estimate relative risk [28]. Logistic regression was used to estimate ORs and calculate 95% confidence intervals in age-adjusted and multivariable analyses. In the final models, we adjusted for race or ethnic group because of known racial and ethnic differences in the incidence of VTE and genetic polymorphisms. We adjusted for income as an indicator of socioeconomic status (SES) because of the role of SES as a predictor of many health outcomes, and its association in our data with both OC use and VTE. BMI was included in multivariable models because of its association with OC use and its strong association with VTE in our database, as well as its known association with VTE in other studies. We considered other questionnaire variables (Table 1) and planned a priori to include them in the final adjustment models if they affected the ␤ coefficient for current OC use by 10% or more, referring to the relative difference between the adjusted and the unadjusted OR estimate. None, including family history of VTE, met this criterion. The interaction between current OC use and age, BMI and gene mutations was assessed in the logistic regression models with the addition of the corresponding cross-product terms.

3. Results A total of 379 possible VTE cases were identified, of whom 299 (80.8%) were interviewed. Interviews were not completed for the remaining 80 cases for the following reasons: patient refusal (n ⫽ 56); physician refusal (n ⫽ 5); surrogate refusal (n ⫽ 6); inability to locate (n ⫽ 4); ineligibility for the interview (n ⫽ 9, including terminal illness [n ⫽ 1], language other than English or Spanish [n ⫽ 1], death [n ⫽ 1], patient out of area [n ⫽ 2], death [n ⫽ 1], and failure to meet criteria as definite or probable VTE upon initial medical record review [n ⫽ 3]). In the analysis reported here, 103 of the interviewed cases were excluded for the following reasons: prior VTE (n ⫽ 48); pregnancy (n ⫽ 4); failure to meet criteria as definite or probable VTE after investigator review of medical records (n ⫽ 5); hysterectomy and/or bilateral oophorectomy and/or receiving hormone replacement therapy (n ⫽ 29); missing OC use data (n ⫽ 1), use of an OC dose with 50 ␮g or more of estrogen (n ⫽ 2) or inability to determine the amount of estrogen (n ⫽ 14). This left 196 interviewed cases for the analysis. Of these, 11 were surrogate interviews. Of 1519 potential controls identified, 115 were not eli-

5

gible for the following reasons: pregnancy (n ⫽ 38); not a member of KPMCP in the study period (n ⫽ 21); out of the area (n ⫽ 24); language other than English or Spanish (n ⫽ 20); mental incapacitation (n ⫽ 4); terminal illness (n ⫽ 4); death (n ⫽ 1); age/gender error (n ⫽ 3). Interviews were completed for 819 of the remaining 1404 potentially eligible women (58.3%). Of the 819 interviewed potential controls, 73 were excluded from this analysis for the following reasons: pregnancy (n ⫽ 4); age ⬎45 years (n ⫽ 1); prior history of VTE (n ⫽ 6); history of hysterectomy and/or bilateral oophorectomy and/or receiving hormone replacement therapy (n ⫽ 49); use of an OC with 50 ␮g or more of estrogen (n ⫽ 1) or inability to determine the amount of estrogen (n ⫽ 12). This left 746 interviewed cases for the analysis. Table 1 shows the characteristics of the VTE cases and their controls, along with distribution of OC use in controls. The prevalence of current OC use was 44% in cases and 18% in controls. Factors associated with current OC use included younger age, lower BMI, lower parity, more than occasional alcohol use, vigorous physical activity at least once a week, ethnicity, marital status, education, income and family history of VTE. The age-adjusted ORs for VTE were increased in relation to high BMI, low education level, low income, prolonged immobilization, possible contributing medical conditions and family history of VTE (Table 1). ORs were decreased in relation to current smoking, more frequent alcohol use, regular and vigorous physical activity, Hispanic or Asian ethnicity (relative to non-Hispanic whites), marital status (married or living as married, relative to never married), aspirin use and parity. The age-adjusted OR for VTE in current OC users compared with noncurrent users was 3.62 (95% confidence interval [CI]: 2.56 –5.13). After additional adjustment for race/ethnicity, income and BMI, the OR in current OC users compared with noncurrent users was 4.07 (95% CI: 2.77– 6.00) (Table 2). When current users were compared with never users of OCs, the adjusted OR was 3.22 (95% CI: 1.89 –5.49). The adjusted OR for VTE in past OC users compared with nonusers was not elevated. Among current OC users, the adjusted OR for VTE was 5.43 (95% CI: 2.12–13.94) in women with less than12 months use, decreasing to 3.12 (95% CI: 1.99 – 4.88) with 60 months or greater use. Genetic testing was performed on 153 of the 196 cases (78.1%) and 312 of the 746 (41.8%) controls in the analysis data set. Among controls, there was no association between current or past OC use being positive for Factor V, MTHFR or prothrombin gene mutations. The adjusted ORs for VTE were 7.10 (95% CI: 2.33–21.61) in women who were factor V Leiden positive, 2.83 (95% CI: 0.70 –11.63) in women who were prothrombin positive and 0.26 (95% CI: 0.10 – 0.65) for women who were MTHFR positive. The age-adjusted OR for VTE was 11.32 in women who were current OC users and factor V Leiden positive relative to nonusers of OCs who were factor V Leiden negative

6

S. Sidney et al. / Contraception 70 (2004) 3–10

Table 1 Characteristics of women with VTE and of controls according to OC use, age-adjusted odds ratios and 95% confidence interval for each characteristic Characteristic

Age (y), mean ⫾ SD (per year) Body mass index (kg/m2), mean ⫾ SD (per kg/m2) Years of OC use, mean ⫾ SD (per year) Parity (years), mean ⫾ SD (per year) Smoking (%) Current/occasional Former Never Alcohol use (%) Never or occasional 1–3 drinks per month 1–3 drinks per week ⬎3 drinks per week Physical activity (%) Regular physical activity (ref: no regular physical activity) Moderate ⱖ 1 x/wk (ref: ⬍1 x/wk) Vigorous ⱖ 1 x/wk (ref: ⬍1 x/wk) OC use (%) Never use of OCs Ever use of OCs Current use of OCs Former use of OCs Ethnicity (%) Hispanic Black Asian White, non-Hispanic Other/unknown Marital status (%) Never married Separated, divorced or widowed Married or living as married Educational level (%) Some high school High school graduate Some college or trade school ⱖHigh school graduate Unknown Income (%) ⬍$20,000 $20,000–34,999 ⱖ$35,000 Unknown Prolonged immobilization (composite) (%)a Possible contributing medical conditions (composite) (%)b Aspirin use (%) Index date Prior week Supplements (%) Folic acid Vitamins Family history of VTE (%) BMI (kg/m2) (%) ⬍25 25–30 ⬎30

Cases (n ⫽ 196)

Controls

Odds ratio

95% CI

0.98 1.08 0.94 0.89

0.96–1.00 1.06–1.10 0.89–0.99 0.78–1.01

OC use Never (n ⫽ 143)

Former (n ⫽ 469)

Current (n ⫽ 134)

All (n ⫽ 746)

35.3 ⫾ 7.0 31.5 ⫾ 9.2 5.0 ⫾ 5.7 1.3 ⫾ 1.5

35.6 ⫾ 7.4 26.6 ⫾ 6.7 0.0 1.3 ⫾ 1.4

37.3 ⫾ 6.1 27.8 ⫾ 6.6 4.7 ⫾ 4.5 1.8 ⫾ 1.3

33.2 ⫾ 7.2 24.9 ⫾ 5.0 9.2 ⫾ 6.7 1.1 ⫾ 1.2

36.2 ⫾ 6.7 27.0 ⫾ 6.5 4.7 ⫾ 5.4 1.6 ⫾ 1.3

6.6 23.0 70.4

11.9 11.9 76.2

16.6 22.2 61.2

11.9 23.9 64.2

14.9 20.5 64.6

0.41 1.08 1.00

0.22–0.75 0.73–1.58 Ref

67.9 15.3 13.8 3.1

76.2 11.2 9.1 3.5

59.3 17.3 16.9 6.4

50.0 30.6 15.7 3.7

60.9** 18.6 15.2 5.4

1.00 0.72 0.83 0.55

Ref 0.46–1.12 0.52–1.31 0.23–1.32

58.2

66.4

65.3

64.2

65.3

0.72

0.52–1.00

56.1 23.5

49.7 35.7

59.5 36.9

59.7 41.0

57.6 37.4

0.94 0.50

0.68–1.29 0.35–0.72

15.8 84.2 44.4 39.8

– – – –

– – – –

– – – –

19.2 80.8 18.0 62.9

1.00 1.29 2.99 0.77

Ref 0.84–1.98 1.86–4.81 0.49–1.22

17.9 16.8 1.5 57.1 6.6

34.3 11.9 15.4 31.5 7.0

29.2 13.0 5.5 45.6 6.6

24.6 3.7 5.2 57.5 9.0

29.4** 11.1 7.4 45.0 7.1

0.47 1.19 0.16 1.00 0.71

0.31–0.71 0.75–1.89 0.05–0.54 Ref 0.37–1.36

35.9 11.8 52.3

32.2 9.8 58.0

16.7 15.0 68.3

26.1 14.2 59.7

21.4** 13.8 64.8

1.00 0.52 0.49

Ref 0.29–0.91 0.33–0.71

7.1 24.0 39.3 28.6 1.0

9.8 16.1 34.3 39.9 0.0

6.4 29.6 41.8 21.8 0.4

9.0 21.6 44.0 25.4 0.0

7.5** 25.6 40.8 25.9 0.3

1.00 0.98 0.99 1.18 3.52

Ref 0.51–1.97 0.54–1.94 0.63–2.35 0.39–31.64

14.3 21.4 58.2 6.1 49.5 52.6

9.1 16.8 62.9 11.2 14.7 17.5

4.9 18.8 72.1 4.3 14.3 19.1

6.7 20.9 65.7 6.7 14.2 19.4

6.0* 18.8 69.2 6.0 14.4 18.2

1.00 0.48 0.37 0.43 5.93 5.27

Ref 0.27–0.87 0.22–0.62 0.19–0.94 4.18–8.41 3.74–7.42

4.3 8.4

3.5 12.6

6.2 17.3

3.0 12.7

5.1 15.6

0.87 0.51

0.40–1.91 0.29–0.88

7.3 40.7 15.0

4.9 42.7 4.2

5.8 48.0 11.4

7.5 47.0 10.5

5.9 46.8 9.8*

1.30 0.80 1.69

0.70–2.44 0.58–1.11 1.06–2.70

25.5 24.0 50.5

49.7 29.4 21.0

41.7 25.6 32.7

60.5 23.1 16.4

46.6** 25.9 27.5

1.00 1.78 3.47

Ref 1.14–2.77 2.35–5.10

S. Sidney et al. / Contraception 70 (2004) 3–10

7

Table 1 continued Characteristic

Parity (%) 0 1–2 3⫹

Cases (n ⫽ 196)

42.4 38.3 19.4

Controls

Odds ratio

95% CI

OC use Never (n ⫽ 143)

Former (n ⫽ 469)

Current (n ⫽ 134)

All (n ⫽ 746)

40.6 39.9 19.6

19.7 53.4 26.9

41.8 45.5 12.7

27.7** 49.4 23.0

1.00 0.52 0.58

Ref 0.36–0.76 0.37–0.90

Data are presented as mean ⫾ SD or percentage. a Leg fracture, overnight hospitalization, prolonged bedrest, airplane ride during 6 weeks prior to index date. b Six weeks postpartum, cancer, stroke, congenital heart disease, congestive heart failure, clotting disorder, systemic lupus erythematosis, diabetes gallbladder disease, varicose veins, fracture, surgery, myocardial infarction, peripheral vascular disease, spinal cord injury, venulitis. * p ⬍ 0.05 for comparisions of never users, former users and current users among controls. **p ⬍ 0.001 for comparisions of never users, former users and current users among controls.

(Table 3). This OR was higher than the OR for VTE for current OC use in women who were factor V Leiden negative (3.20) and also higher than the OR for VTE in women who were factor V Leiden positive but nonusers of OCs (8.42). However, the 95% confidence intervals overlapped. The prothrombin gene mutation did not exhibit the same pattern: the highest OR for VTE was in nonusers of OCs who were prothrombin positive. In models, the two-way interaction between OC use and each of the genetic mutations was nonsignificant (p ⫽ 0.37 for factor V Leiden, 0.15 for prothrombin G20210A and 0.94 for MHTFR C677T). We also examined ORs for VTE for current OC use in women younger than 35 years or 35⫹ years, in women with and without a predisposing condition, with and without prolonged immobilization, women at or below and above a BMI of 30 kg/m2, and according to progestin type, comparing OCs containing norethindrone and levonorgestrel (Table 4). An interaction between OC use and predisposing condition was present (likelihood test for interaction, p ⫽ 0.02).

Current or noncurrent use Current use Noncurrent use Current or past use and no use Current use Past use No use Any or no use Any use No use Duration of current use (months) ⬍12 12–59 ⱖ60 a

4. Discussion The results of this analysis are consistent with other recent studies that show three- to fourfold increase in the

Table 3 Age-adjusted odds ratios of VTE by gene strata

Table 2 Results of multivariate analysis according to OC use OC use

OC use also had a significant interaction with BMI evaluated as a continuous variable (p ⫽ 0.01). When treated as a dichotomous variable, the odds ratio for VTE associated with OC use was nearly double in those who were ⬎30 kg/m2 compared to the OR in those who were ⱕ30 kg/m2. We found no evidence of interaction between OC use and age or recent prolonged immobilization. In current OC users, ORs for VTE for formulations containing norethindrone and for formulations containing levonorgestrel were 3.39 and 5.06, respectively, but the confidence intervals overlapped.

Stratum

No. of cases

No. of controls

Odds ratio

95% CI

OC OC OC OC OC OC OC OC OC OC OC OC

76 59 8 10 69 62 5 3 80 65 4 4

244 61 3 3 223 58 2 2 211 53 37 11

1.00 3.20 8.42 11.32 1.00 3.64 8.43 5.10 1.00 3.29 0.29 1.00

Ref 2.04–5.03 2.18–32.56 3.00–42.81 Ref 2.30–5.77 1.60–44.41 0.83–31.18 Ref 2.09–5.18 0.10–0.83 0.30–3.27

a

Model OR

95% CI

4.07 1.00

2.77–6.00 Ref

3.22 0.73 1.00

1.89–5.49 0.44–1.21 Ref

1.25 1.00

0.78–2.01 Ref

5.43 5.73 3.12

2.12–13.94 2.98–10.99 1.99–4.88

Adjusted for age, race/ethnicity, income and BMI.

noncurrent/FVL⫺ current/FVL⫺ noncurrent/FVL⫹ current/FVL⫹ noncurrent/PT⫺ current/PT⫺ noncurrent/PT⫹ current/PT⫹ noncurrent/MTHFR⫺ current/MTHFR⫺ noncurrent/MTHFR⫹ current/MTHFR⫹

FVL ⫽ factor V Leiden; positive if carrier— homozygous or heterozygous— of G1691A polymorphism;PT ⫽ prothrombin; positive if carrier— homozygous or heterozygous— of G20210A polymorphism; MTHFR ⫽ methylenetetrahydrofolate reductase; positive if homozygous for C677T polymorphism.

8

S. Sidney et al. / Contraception 70 (2004) 3–10

Table 4 Adjusted odds ratios of VTE for current OC use relative to noncurrent use Subgroup Age (y) ⬍35 ⱖ35 Presence of predisposing condition Yes No Prolonged immobilization Yes No Body mass index (kg/m2) ⱕ30 ⬎30 Progestin type Norethindrone Levonorgestrel a

Adjusted odds ratioa

95% CI

3.56 4.56

1.93–6.59 2.73–7.63

1.85 7.84

0.93–3.67 4.49–13.69

4.37 4.35

2.06–9.27 2.60–7.26

3.34 6.04

2.04–5.46 3.11–11.72

3.39 5.06

2.11–5.46 3.07–8.35

Adjusted for age, race/ethnicity, income and BMI.

risk of VTE in current user of low-estrogen dose OCs. In this study, also consistent with other recent studies [19,29,30], the risk of VTE was highest in the first year of OC use and diminished with increasing duration of use. In the last decade, much attention has focused on the effect on VTE risk of the progestin component of combination estrogen-progestin OCs. Low-estrogen dose OC formulations containing gestodene or desogestrel increase the risk of VTE more than low-estrogen dose formulations containing other progestins [31,32]. We were unable to estimate the OR for VTE in current users of low-estrogen OC formulations containing gestodene or desogestrel because none of the OC users in this study population used these formulations. OC formulations containing gestodene were never marketed in the United States. The health plan formulary during the years of this study did not include formulations containing desogestrel. During the period of this study, about 93% of health plan members had prescription drug coverage and would pay only a small copayment for OCs if they filled their prescriptions at a Kaiser Permanente pharmacy. In our study population, there was no difference in the risk of VTE between low-estrogen dose OCs containing norethindrone and levonorgestrel. Recent studies done outside the United States reflect limited use of formulations with norethindrone. The major strengths of this study are the high response rate of the cases, the nearly universal use of low-dose estrogen pills (under 50 ␮g per day), equal access to medical care for the base population, and the high level of diagnostic testing in suspected VTE cases. In the World Health Organization Collaborative Study, which is the largest study of low-estrogen dose OCs and VTE, fewer than half of cases had definitive diagnostic procedures, and the

proportion of cases having these procedures was higher among OC users than nonusers [2]. Limitations of this study include possible recall bias and diagnostic bias. To explore the possibility of recall bias, we used information from a review of medical records, which found that 47% of cases and 26% of controls had a prescription for OCs in the 2 years prior to the index date, compared with self-reports of current OC use for 44% of cases and 18% of controls. These data suggest that there might be some upward bias in our estimate of the risk of VTE in OC users. Diagnostic bias would arise if clinicians were more likely to consider a diagnosis of VTE in OC users than in nonusers. We cannot rule out this possibility but believe that it is less than in studies that had lower rates of diagnostic testing. Diagnostic bias would not explain the findings with regard to gene mutations and is unlikely to affect overall conclusions about risk estimates comparing different progestins. The factor V Leiden and prothrombin G20210A mutations both result in prothrombotic conditions, and have been established as risk factors for VTE [33]. The point estimates for the increased risk of VTE associated with factor V Leiden mutation and the prothrombin G20210A mutation are comparable to those reported in a pooled analysis of 8 case-control studies with more than 2300 cases [34]. The 11-fold increase in the risk of VTE among OC users who were factor V Leiden carriers compared to nonusers of OCs who were negative for factor V Leiden is lower than the 34-fold increase reported by Vandenbroucke et al. [18], but comparable to the 10-fold increase found in the pooled analysis [34]. In our study, 10 of 69 VTE cases in current OC users who had genetic testing occurred in women with factor V Leiden mutation. This raises the question of whether women should be screened for factor V Leiden mutation before initiating OC use. Tests for mutations in the factor V Leiden gene are not, however, widely available and are costly. Moreover, the usefulness of factor V Leiden screening of OC users would depend on the prevalence of the mutation, which is known to vary widely according to ethnicity. In the Physicians’ Health Study and Nurse’s Health Study, for example, the prevalence of factor V Leiden was 5.3% in whites, 2.2% in Hispanics, 1.2% in blacks and 0.5% in Asian Americans [33]. Among the more than 300 controls tested in our study, the prevalence of factor V Leiden was 2.1% in whites, 2.0% in Hispanics, and 0% in blacks and Asians. The finding of a lower risk of VTE associated with the MTHFR C677T mutation was not expected. Some studies have shown an association of the MTHFR mutation with increased risk of VTE [35,36] but most have found no association [37– 40]. We do not have a mechanistic explanation for our results, which should be considered tentative until confirmed. Several studies show a graded association between BMI and VTE risk in women [2,30,41]. We are unaware of other

S. Sidney et al. / Contraception 70 (2004) 3–10

reports of a significant interaction of OC use with BMI for VTE. The mechanism by which obesity increases the risk of VTE is not clear and the mechanism by which OC use would interact with obesity to increase risk is equally uncertain. In our study, almost half of all VTE cases in current OC users (41/96) occurred in women with a BMI ⬎30. Somewhat surprisingly, we found that the risk of VTE associated with OC use was higher in women without a possible predisposing medical condition than in those with a predisposing medical condition. The relative risk for VTE among patients with acute medical conditions was 17 (6.5, 46) in a nested case-control analysis of the General Practice Research database [42]. In many early studies of VTE in OC users, women with predisposing conditions were excluded. Other studies that addressed this issue in a comparable way could not be identified.

[3]

[4]

[5]

[6]

[7] [8]

[9]

5. Summary and conclusions Our study is consistent with other recent studies and confirms that low-estrogen dose OCs increase the risk of VTE about fourfold in women of childbearing age. It shows that there is no difference in the risk of VTE between low-estrogen dose OCs containing norethindrone or levonorgestrel. It confirms the role of the prothrombotic genetic factors, factor V Leiden mutation and the prothrombin G20210A mutation, in increasing the risk of VTE in women in this age group and suggests that OC users with factor V Leiden mutation may be at particularly high risk of VTE when using OCs. Notwithstanding the interaction between OC use and these mutations, population screening is not recommended. The interaction between obesity and OC use and the fact that obese women accounted for almost half of all cases of VTE in OC users are noteworthy in the context of the current epidemic of obesity in the United States.

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17]

[18]

Acknowledgments [19]

The authors acknowledge Luisa Hamilton and Kathy Heller for project management; Socorro Ramirez, Karla Villarta and Faye Jack for interviewing participants; and Carolyn Salazar for abstracting medical records. Supported by grant number R01 HL057286 from the National Heart Lung and Blood Institute.

[20]

[21]

References [1] WHO Scientific Group on Cardiovascular Disease and Steroid Contraception. Cardiovascular disease and steroid hormone contraception. World Health Organ Tech Rep Ser 1998;87:1– 89. [2] World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Venous thromboem-

[22]

[23]

9

bolic disease and combined oral contraceptives: results of international multi-centre case-control study. Lancet 1995;346:1575– 82. Records Unit and Research Advisory Service of the Royal College of General Practitioners. Oral contraception and thromboembolic disease. J Coll Gen Pract 967;13:267–79. Vessey MP, Doll R. Investigation of relation between use of oral contraceptives and thromboembolic disease: a further report. Br Med J 1969;2:651–7. Sartwell PE, Masi AT, Arthes FG, Greene GR, Smith HE. Thromboembolism and oral contraceptives: an epidemiologic case-control study. Am J Epidemiol 1969;90:365– 80. Vessey MP, Doll R, Fairbairn AS, Glober G. Postoperative thromboembolism and the use of oral contraceptives. Br Med J 1970;3: 123– 6. Bottiger LE, Westerholm B. Oral contraceptives and thromboembolic disease. Acta Med Scand 1971;190:455– 63. Greene GR, Sartwell PE. Oral contraceptive use in patients with thromboembolism following surgery, trauma, or infection. Am J Public Health 1972;62:680 –5. Boston Collaborative Drug Surveillance Programme. Oral contraceptives and venous thromboembolic disease, surgically confirmed gallbladder disease, and breast tumours. Lancet 1973;1:1399 – 407. Stolley PD, Tonascia JA, Tockman MS, Sartwell PE, Rutledge AH, Jacobs MP. Thrombosis with low-estrogen oral contraceptives. Am J Epidemiol 1975;102:197–208. Royal College of General Practitioners’ Oral Contraception Study. Oral contraceptives, venous thrombosis, and varicose veins. J R Coll Gen Pract 1978;28:393–9. Petitti DB, Wingerd J, Pellegrin F, Ramcharan S. Oral contraceptives, smoking, and other factors in relation to risk of venous thromboembolic disease. Am J Epidemiol 1978;108:480 –5. Porter JB, Hunter JR, Danielson DA, Jick H, Stergachis A. Oral contraceptives and nonfatal vascular disease: recent experience. Obstet Gynecol 1982;59:299 –302. Porter JB, Hunter JR, Jick H, Stergachis A. Oral contraceptives and nonfatal vascular disease. Obstet Gynecol 1985;66:1– 4. Vessey M, Mant D, Smith A, Yeates D. Oral contraceptives and venous thromboembolism: findings in a large prospective study. Br Med J 1986;292:526. Helmrich SP, Rosenberg L, Kaufman DW, Strom B, Shapiro S. Venous thromboembolism in relation to oral contraceptive use. Obstet Gynecol 1987;69:91–5. Thorogood M, Mann J, Murphy M, Vessey M. Risk factors for fatal venous thromboembolism in young women: a case-control study. Int J Epidemiol 1992;21:48 –52. Vandenbroucke JP, Koster T, Briet E, Reitsma PH, Bertina RM, Rosendaal FR. Increased risk of venous thrombosis in oral-contraceptive users who are carriers of factor V Leiden mutation. Lancet 1994;344:1453–7. Spitzer WO, Lewis MA, Heinemann LA, Thorogood M, MacRae KD. Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case-control study. Br Med J 1996; 312:83– 8. Meade TW, Greenberg G, Thompson SG. Progestogens and cardiovascular reactions associated with oral contraceptives and a comparison of the safety of 50- and 30-␮g oestrogen preparations. Br Med J 1980;281:1157– 61. Gerstman BB, Piper JM, Tomita DK, Ferguson WJ, Stadel BV, Lundin FE. Oral contraceptive estrogen dose and the risk of deep venous thromboembolic disease. Am J Epidemiol 1991;133:32–7. Lewis MA, Heinemann LA, MacRea KD, Bruppacher R, Spitzer WO. The increased risk of venous thromboembolism and the use of third generation progestagens: role of bias in observational research. Contraception 1996;54:5–13. Petitti DB, Sidney S, Bernstein A, Wolf S, Quesenberry C, Ziel HK. Stroke in users of low dose oral contraceptives. N Engl J Med 1996;335:8 –15.

10

S. Sidney et al. / Contraception 70 (2004) 3–10

[24] Sidney S, Petitti DB, Quesenberry CP Jr, Klatsky AL, Ziel HK, Wolf S. Myocardial infarction in users of low-dose oral contraceptives. Obstet Gynecol 1996;88:939 – 44. [25] Bertina RM, Koeleman CP, Koster T, et al. Mutation in blood coagulation factor V associated with resistance to activated protein C. Nature 1994;369:64 –7. [26] Arruda VR, von Zuben PM, Chiaparini LC, Annichino-Bizzacchi JM, Costa FF. The mutation Ala677–⬎Val in the methylene tetrahydrofolate reductase gene: a risk factor for arterial disease and venous thrombosis. Thromb Haemost 1997;77:818 –21. [27] Poort SR, Rosendaal FR, Reitsma PH, Bertina RM. A common genetic variation in the 3⬘-untranslated region of the prothrombin gene is associated with elevated plasma prothrombin levels and an increase in venous thrombosis. Blood 1996;88:3698 –703. [28] Breslow NE, Day NE. Statistical methods in cancer research. Volume 1. The analysis of case-controls studies. Oxford (UK): Oxford University Press, 1993. [29] Jick H, Jick SS, Gurewich V, Myers MW, Vasilakis C. Risk of idiopathic cardiovascular death and nonfatal venous thromboembolism in women using oral contraceptives with differing progestagen components. Lancet 1995;346:1589 –93. [30] Lidegaard O, Edstrom B, Kreiner S. Oral contraceptives and venous thromboembolism: a five-year national case-control study. Contraception 2002;65:187–96. [31] Hennessy S, Berlin JA, Kinman JL, Margolis DJ, Marcus SM, Strom BL. Risk of venous thromboembolism from oral contraceptives containing gestodene and desogestrel versus levonorgestrel: a metaanalysis and formal sensitivity analysis. Contraception 2001;64:125– 33. [32] Kemmeren JM, Algra A, Grobbee DE. Third generation oral contraceptives and risk of venous thrombosis: meta-analysis. BMJ 2001; 323:131– 4. [33] Ridker PM, Miletich JP, Hennekens CH, Buring JE. Ethnic distribution of factor V Leiden in 4047 men and women. Implications for venous thromboembolism screening. JAMA 1997;277:130 –7.

[34] Emmerich J, Rosendaal FR, Cattaneo M, et al. Combined effect of factor V Leiden and prothrombin 20210A on the risk of venous thromboembolism—pooled analysis of 8 case-control studies including 2310 cases and 3204 controls. Study group for pooled-analysis in venous thromboembolism. Thromb Haemost 2001;86:809 –16. [35] Gemmati D, Serino ML, Trivellato C, Fiorini S, Scapoli GL. C677T substitution in the methylenetetrahydrofolate reductase gene as a risk factor for venous thrombosis and arterial disease in selected patients. Haematologica 1999;84:824 – 8. [36] Couturaud F, Oger E, Abalain JH, et al. Methylenetetrahydrofolate reductase C677T genotype and venous thromboembolic disease. Respiration 2000;67:657– 61. [37] McColl MD, Ellison J, Reid F, Tait RC, Walker ID, Greer IA. Prothrombin 20210 G3 A, MTHFR C677T mutations in women with venous thromboembolism associated with pregnancy. BJOG 2000; 107:565–9. [38] Linfert DR, Tsongalis GJ. Coexistence of the methylenetetrahydrofolate reductase single-nucleotide polymorphism (C677T) in patients with the factor V Leiden or prothrombin G20210A polymorphisms. Diagn Mol Pathol 2001;10:111–5. [39] Varela ML, Adamczuk YP, Forastiero RR, et al. Major and potential prothrombotic genotypes in a cohort of patients with venous thromboembolism. Thromb Res 2001;104:317–24. [40] Ray JG, Langman LJ, Vermeulen MJ, Evrovski J, Yeo EL, Cole DE. University of Toronto Thrombophilia Study in Women (GUTTSI): genetic and other risk factors for venous thromboembolism in women. Curr Control Trials Cardiovasc Med 2001;2:141–9. [41] Nightingale AL, Lawrenson RA, Simpson EL, Williams TJ, MacRae KD, Farmer RD. 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 2000;5:265–74. [42] Black C, Kaye JA, Jick H. Clinical risk factors for venous thromboembolus in users of the combined oral contraceptive pill. Br J Clin Pharmacol 2002;53:637– 40.