UPDATE: contraception

Does the contraceptive patch raise the risk of thromboembolism?

Jick SS, Kaye JA, Russman S, Jick H. Risk of nonfatal venous thromboembolism in women using a contraceptive patch and oral contraceptives containing norgestimate and 35 microg of ethinyl estradiol. Contraception. 2006;73:223–228.

Jick S, Kaye JA, Li L, Jick H. Further results on the risk of nonfatal venous thromboembolism in users of the contraceptive transdermal patch compared to users of oral contraceptives containing norgestimate and 35 microg of ethinyl estradiol. Contraception. 2007;76:4–7.

Cole JA, Norman H, Doherty M, Walker AM. Venous thromboembolism, myocardial infarction, and stroke among transdermal contraceptive users. Obstet Gynecol. 2007;109(2 Pt 1):339–346.

Both the media and regulatory agencies have raised concerns about whether the contraceptive patch heightens the risk of thromboembolism and is less effective in women above a certain body weight.

The controversy surrounding thromboembolic disease stems from a pharmacokinetics study by van den Heuvel and colleagues that compared serum ethinyl estradiol levels in users of the patch, vaginal ring, and a combination OC containing 30 μg of ethinyl estradiol and 150 μg of levonorgestrel.⁶ Women randomized to the patch had serum ethinyl estradiol levels 1.6 times higher than women randomized to an OC, and 3.4 times higher than women randomized to the ring.

These findings led the FDA to update package labeling of the patch to warn health-care providers and patients that the patch exposes women to 60% more estrogen and may increase the risk of thromboembolic events. Oddly, the FDA did not require any labeling change to combination OCs to indicate that they contain up to twice as much estrogen as the contraceptive ring.

A set of studies finds no elevated risk

Although the study by van den Heuvel and associates raised the possibility of increased blood clots in patch users, no association between the two had been corroborated at the time it was published.⁶ Since then, three epidemiological studies have explored the potential link between thromboembolic events and use of the patch.

In the first of these studies, Jick and colleagues used the PharMetrics database to extract data on users of the patch and norg-estimate-containing OCs. This database contains drug prescription information, patient demographic data, and ICD-9 billing codes submitted by managed care health plans. A nested case-control study design was used to compare patch and pill users and control for confounding variables.

The base population was women 15 to 44 years old who were new users of the patch or a norgestimate-containing OC. A thromboembolic event was diagnosed if the patient’s record included a diagnosis code for pulmonary embolus, deep vein thrombosis, or an emergency room visit or diagnostic testing indicating venous thromboembolism (VTE). These diagnosis codes, combined with the prescription of long-term anticoagulation therapy, strengthened the identification of cases. As many as four controls were selected for each case.

The 215,769 women included in this study contributed 147,323 woman-years of exposure to the drugs. There were 31 and 37 cases of VTE identified in the patch and pill groups, respectively, with an incidence of 52.8 for every 100,000 woman-years in the patch group and 41.8 for every 100,000 woman-years in the pill group and an unadjusted, matched odds ratio of VTE in patch versus pill users of 0.9. When the data were adjusted for duration of drug exposure, the odds ratio did not change.

A follow-up study by Jick and associates, published in 2007, had the same study design and included 17 additional months of data. Another 56 cases of VTE were diagnosed. The odds ratio for patch users, compared with pill users, was 1.1. When data from the two studies were combined, 73 and 51 total cases of VTE had occurred in the pill and patch groups, respectively. The overall odds ratio was 1.0.

A third study finds significantly heightened risk

Cole and associates studied insurance claims data from UnitedHealthcare, a database containing medical and prescription claim information as well as patient demographics. Because researchers used pharmacy dispensing records, they were able to include women 15 to 44 years old who had received at least one prescription for the contraceptive patch or a norgestimate-containing OC with 35 μg of ethinyl estradiol.

Unlike the studies by Jick and colleagues, the study by Cole and associates considered all women eligible, even if they had used OCs in the past. Cases of VTE, stroke, and acute myocardial infarction (AMI) were abstracted from this group, identified from insurance claim information, and confirmed by chart review. Review of medical records is an important strength of this study; no such review was done in the studies by Jick and colleagues. Four controls were matched to each case, by age and duration of contraceptive use.