drospirenona

8/3/2019 drospirenona

1/8

RE SE ARCH

Risk of non-fatal venous thromboembolism in women usingoral contraceptives containing drospirenone compared withwomen using oral contraceptives containing levonorgestrel:case-control study using United States claims data

Susan S Jick, director and senior epidemiologist Rohini K Hernandez, epidemiologist

ABSTRACT

Objective To compare the risk of non-fatal venous

thromboembolism in women receiving oralcontraceptives containing drospirenone with that in

women receiving oral contraceptives containing

levonorgestrel.

Design Nested case-control and cohort study.

SettingThe study was based on information from

PharMetrics, a United States based company that collects

information on claims paid by managed care plans.

Participants The study encompassed all women aged 15

to 44 years who received an oral contraceptive containing

either drospirenone or levonorgestrel after 1 January

2002. Cases were women with current use of a study oral

contraceptive and a diagnosis of venous

thromboembolism in the absence of identifiable clinicalrisk factors (idiopathic venous thromboembolism). Up to

four controls were matched to each case by age and

calendar time.

Main outcome measures Odds ratios comparing the risk

of non-fatal venous thromboembolism in users of the two

contraceptives; incidence rates and rate ratios of non-

fatal venous thromboembolism for users of each of the

study contraceptives.

Results 186 newly diagnosed, idiopathic cases of venous

thromboembolismwere identifiedin the studypopulation

and matched with 681 controls. In the case-control

analysis, the conditional odds ratio for venous

thromboembolism comparing use of oral contraceptivescontaining drospirenone with use of those containing

levonorgestrel was 2.3 (95% confidence interval 1.6 to

3.2). The incidence rates for venous thromboembolism in

the study population were 30.8 (95% confidence interval

25.6 to 36.8) per 100 000 woman years among users of

oral contraceptives containing drospirenone and 12.5

(9.61 to 15.9) per 100 000 woman years among users of

oral contraceptives containing levonorgestrel. The age

adjusted incidence rate ratio for venous

thromboembolism for current use of oral contraceptives

containing drospirenone compared with those containing

levonorgestrel was 2.8 (2.1 to 3.8).

Conclusions The risk of non-fatal venousthromboembolism among users of oral contraceptives

containing drospirenoneseems to be aroundtwicethat of

users of oral contraceptives containing levonorgestrel,

after the effects of potential confounders and prescribingbiases have been taken into account.

INTRODUCTION

The first oral contraceptives were introduced in theearly 1960s and contained high doses of both oestro-gen and progestogen. The doses of oestrogen werefound to be associated with an increased risk of venousthromboembolism.1 Over the subsequent years, oralcontraceptives containing smaller doses of oestrogenand progestogen were introduced to the market in anattempt to reduce cardiovascular risk. The type of pro-gestogen became the focus of discussion in the mid-

1990s, when concern was raised that women takingthird generation oral contraceptives (which containeddesogestrel or gestodene) were at an increased risk ofvenous thromboembolism compared with those tak-ing second generation oral contraceptives (which con-tained levonorgestrel). Several studies found anincreased risk, whereas others argued that confound-ing by indication or other biases could account for thefindings.2-6 In the end, the consensus was that anincreased risk of venous thromboembolism existed inusers of third generation oral contraceptives. The riskof venous thromboembolism waslater also found to beincreased for use of oral contraceptives containing

cyproterone.7

Thus, post-marketing surveillance tomonitor newer contraceptives as they are introducedto the market is important, particularly for an estab-lished risk such as venous thromboembolism.

Four published studies have examined the risk ofvenous thromboembolism among women taking thenewer oral contraceptives containing drospirenonecompared with those taking other oral contraceptives(including third generation oral contraceptives), withinconclusive results.8-11 Dinger et al and Seeger et alfound no association between venous thromboembo-lism and drospirenone compared with other oralcontraceptives. Two more recent studies published in

the BMJeach found a small increased risk. A study byLidegaard et al reported a relative risk of 1.6 (95%

Boston Collaborative DrugSurveillance Program, Boston

University School of Medicine, 11Muzzey Street, Lexington, MA02421, USA

Correspondence to: S S [email protected]

Cite this as: BMJ2011;340:d2151doi:10.1136/bmj.d2151

BMJ | ONLINE FIRST | bmj.com page 1 of 8


2/8

confidence interval 1.3 to 2.1) comparing oral contra-ceptives containing drospirenone with those contain-inglevonorgestrel,and a study by vanHylckama Vliegetalyieldedanoddsratioof1.7(0.7to3.9)forthesamecomparison.

All four studies included at least some non-idiopathic cases of venous thromboembolismthatis, cases for which another cause or strong risk factorfor venous thromboembolism was present, whichcould result in attenuating an effect should it exist.12

By not restricting the study to idiopathic cases, onecannot calculate the risk attributableto oral contracep-tives among cases in the absence of other causes.

One study,9 instead of using the reference exposuremost commonly used in recent studies of oral contra-ceptives and venous thromboembolism (secondgeneration oral contraceptives containing levonorges-trel), used a reference category that included womenwho had taken third generation oral contraceptives.These third generation oral contraceptives have beenshown to increase the risk of venous thromboembo-lism compared with the secondgeneration oral contra-ceptives. Their inclusion in the reference categorywould therefore dilute the estimate of risk.

Given the potential serious clinical consequences ofvenous thromboembolisms and the growing popular-ity of oral contraceptives containing drospirenone,examining the association between venous throm-boembolism and oral contraceptives containingdrospirenone compared with those containing levo-norgestrelin studies that avoid the limitations of earlierpublicationsisimportant.Wethereforedidastudythatcompared the risk of non-fatal venous thromboembo-

lism in women using oral contraceptives containingdrospirenone with that in women using oralcontracep-tives containing levonorgestrel. This study wasrestricted to current users of one of these two oral con-traceptives and to idiopathic cases of non-fatal venousthromboembolism.

METHODS

Data resource

Data for this study came from the PharMetrics data-base. PharMetrics is a United States based, ongoinglongitudinal database with information on around 55million people going back as far as 1995. The database

is made up of data contributed by managed care plansthroughout the United States and contains informationon paid claims for drugs, medical diagnoses, and pro-cedures, as well as demographic information such asthe patients year of birth and sex and enrolmentdetails for each patient in the database. Prescriptionsfor drugs arecoded using theNationalDrug Code pro-vided by the US Food and Drug Administration. Eachdrug claim is entered as a separate entry and includesinformationonthespecificentitydispensed,thedateofdispensing, the quantity dispensed, and the length ofthe supply. All diagnoses are coded using the ICD-9(International Classification of Diseases, 9th revision)

coding system. Procedure codes are also included inthe database, coded using the Current Procedural

Terminology-4 system. All events described aboveare noted with the date on which the initial servicewas delivered. This database has been used for manyprevious studies of hormonal contraceptives in rela-tion to venous thromboembolism and other cardio-vascular outcomes.13-17

We designed this study to take into account the eva-luation of a relatively recently marketed drug and theuse of a comparison drug that has been marketed fordecades. We required that all cases and controls werecurrent users of either study drug after 1 January 2002;drospirenone oral contraceptives were first marketedin the United States in May 2001. Important variablesthat we controlled forin thedesign were age, as users ofthenew drug mayhavea differentage distribution thanusers of the older comparison drug, and calendar time(that is,the date of diagnosis), as thetwo contraceptiveswill have highly different usage characteristics in rela-tion to calendar time. We also explored duration ofuse, which may be correlated with both drug use andthe risk of venous thromboembolism.

Base population

We did a case-control study nestedin the population ofusers of oral contraceptives containing drospirenoneor levonorgestrel, aged 15 to 44, in the PharMetricsdatabase updated to the end of December 2008. Allpatients had to have filled at least one prescription fora study drug after 1 January 2002. We excludedwomen with risk factors for venous thromboembo-lism, such as any history of cancer (other than non-melanoma skin cancer), renal failure, chronic cardio-vasculardisease, or inflammatory or autoimmune con-

ditions, from the base population.

Cases

Cases were women aged 15 to 44 years who were cur-rent users of oral contraceptives containing drospire-none or levonorgestrel and who had a first timerecorded claim for a clinically diagnosed deep veinthrombosis or pulmonary embolism with a hospitaladmission, a visit to the emergency room, or a positiveindication of venous thromboembolism from diagnos-tic test results, and who subsequently received pro-longed anticoagulation treatment. We included in thestudy all cases with a first time diagnosis of venous

thromboembolism in 2002 or afterwards. Cases hadto have at least six months of medical history beforethe diagnosis of venous thromboembolism (indexdate) and had to be currently taking one of the studydrugs. We restricted the study to women currently tak-ing a study oral contraceptive because the effect of oralcontraceptives on the outcome, venous thromboem-bolism, is acute and diminishes rapidly after the drugis stopped. We determined oralcontraceptiveuse fromthe prescription claims data before the date of diagno-sis of venous thromboembolism and defined it as hav-ing a recorded claim for a prescription of a studycontraceptive whose filled use extended to within

30 days before the index date or beyond the indexdate. Long term anticoagulation must have been

RE S E ARC H

page 2 of 8 BMJ | ONLINE FIRST | bmj.com


3/8

started promptly,and no contraceptivecontainingoes-trogen could be prescribed after the date of diagnosis,indicating that the diagnosis of venous thromboembo-lism was considered to be confirmed.

To restrict thestudy to idiopathic cases,we excludedwomen from the case group if important clinical riskfactors for venous thromboembolism were present inthe 90 days before the index date. These includedsevere lower limb injury, major surgery, severetrauma, or pregnancy. An idiopathic case is one forwhich no other proximate cause (other explanation)exists for the venous thromboembolism. In cases forwhich another known cause for their venous throm-boembolism is present, such as recent surgery, the dis-tribution of drug use is expected to represent thedistribution in the base population, and inclusion ofthese non-idiopathic cases would dilute any trueincrease in risk because the risk in the non-idiopathiccases is expected to be most influenced by the otherproximate cause rather than by the oral contraceptive.By restricting the study to idiopathic cases, one cancalculate the risk attributable to oral contraceptivesamong cases in the absence of other causes.

To assess the eligibility of each potential case, wereviewed each patients computer record with theidentity of the study oral contraceptive masked. Weachieved agreement on inclusion of women as casesby consensus without knowledge of contraceptive use.

Controls

We matched four womenwho didnot have a diagnosisof venous thromboembolism to each case by using riskset sampling, by year of birth and the index date of the

case (calendar time). As with cases, all controls had tobe current users of one of the study contraceptives, tohave at least six months of enrolment in their healthplan before the index date (the event date of theirmatched case), and to have used a study contraceptiveafter 2002. We applied the same exclusion criteria tocontrols as to cases. One of the authors reviewed thecomputer record of each control.

Statistical methods

We generated descriptive characteristics of the casesand controls, as well as distributions of risk factors bycontraceptive use in the controls to assess potential

confounding. We used conditional logistic regressionto analyse the matched case-control data. When strati-fyingonvariablesotherthanthematchedfactorsofageand index year, we had to break the matching andtherefore calculated odds ratios adjusted for age andindex year (instead of conditional odds ratios). Weevaluated duration of contraceptive use before theindex date, switching from a different hormonal con-traceptive, obesity (if the record contained an ICDcode for obesity), other comorbidities, and number ofvisits to a physician or emergency room in the sixmonths before the indexdate as potentialconfounders.We used the 10% change in estimate rule to evaluate

confounding by comparing the crude odds ratio withthe odds ratio adjusted for each potential confounder

individually.18 Variables that resulted in a 10% orgreater change in the odds ratio would have been con-sidered to be material confounders; however, none ofthe risk factors evaluated were confounders accordingto this criterion.

We analysed the cohort data to estimate incidencerates and 95% confidence intervals. Current persontime was accumulated from the first prescription ofstudy drug to the last prescription plus 45 days. If agap of greater than 100 days existed in the prescriptionfill dates, the person time accumulation stopped at thelast prescription before the gap, plus 45 days; persontime accumulation then resumed at thenext recordof aprescription for a study drug. We estimated the inci-dence rates and rate ratios by using Stata version 11.1and Episheet.19

We stratified the case-control analysis on age cate-gory, index year, type of diagnosis (deep vein throm-bosis versus pulmonary embolism), levonorgestreldose, and new versus continuous use of the study oralcontraceptive. We considered a woman to have a newepisode of use if she had a previous episode for an oralcontraceptive with a gap of at least 100 days before thecurrent episode or no previous prescription for an oralcontraceptive and at least four months of recorded his-tory in her computer record. Among women who hada newepisode of useof thestudyoralcontraceptive,wefurther stratified the analysis on whether the womanhad a previous episode of oral contraceptive use. Weclassified all other women as users of unknown dura-tionthat is, those whose current episode of studydrugbeganwithinfourmonthsofthestartoftheircom-puter record. The four month period is based on the

finding that contraceptive prescriptions in the Phar-Metrics database are written for no longer than threemonthsatatime.Awindowofatleastfourmonthsthusprovided evidence that the first identified prescriptionwas a new prescription and not a refill of an existingprescription. We classified women with a prescriptionrecorded less than four months from the beginning oftheir record as users of unknown duration, because wecould not determine with any confidence that the firstprescription in the database was for a new episode ofuse. We defined duration of contraceptive use as thetime interval (in months) from the first use of an oralcontraceptive within the current episode (at the index

date) to the index date or as unknown. We defined awoman as a switcher if the patients record contained aprescription for a hormonal contraceptive productother than the one used at the index date in either thesix months or the 12 months beforethe index date. Weused SAS release 9.1 for analyses.

RESULTS

We identified 471 potentially eligible cases of venousthromboembolism, of whichwe determined 285 (61%)to be non-idiopathic after blinded review of eachpatients computer record. Our final study populationthus consisted of 186 cases of non-fatal venous throm-

boembolism and 681 controls (women without venousthromboembolism), matched by year of birth and

RE S E ARC H



4/8

index date. Table 1 shows the characteristics of thecases and controls and also shows characteristics byexposureamong controls only. Cases weremore likelyto have a diagnosis of obesity in their record and a visitto an emergency department or a physician in the sixmonths before the index date. Users of oral contracep-tives containing drospirenone were more likely thanusers of those containing levonorgestrel to be younger(aged under 30), to have a history of menstrual disor-ders, to have a shorter duration of use, and to have hada newepisode of use(morethan four months of historyin their record beforetheir first prescriptionfor a studyoral contraceptive within the current episode or had aprevious episode of use).

Among the 186 cases of idiopathic venous throm-boembolism, 121 (65%) women were currently usingan oral contraceptive containing drospirenone and 65(35%) were using one containing levonorgestrel.Among the controls, 313 (46%) were using an oral con-traceptive containing drospirenone and 368 (54%)were using one containing levonorgestrel. The unad-justed matched odds ratio and 95%confidence interval

for venous thromboembolism for oral contraceptivescontaining drospirenone compared with those

containing levonorgestrel was 2.3 (95% confidenceinterval 1.6 to 3.2) (table 2). After adjustment for dura-tion of exposure, the odds ratio was virtuallyunchanged(2.4, 1.7 to 3.4). Neither a history of obesitynor switching from another hormonal contraceptivehad an effect on the odds ratio. When we stratifiedcases on type of venous thromboembolism (deepvenous thrombosis compared with pulmonary embo-lism) the results did not materially differ(odds ratio 1.9(1.1 to 3.2) for deep venous thrombosis and 2.6 (1.6 to4.2) for pulmonary embolism). As users of oral contra-ceptives containing drospirenone were more likely tohave a new episode of use than were users of thosecontaining levonorgestrel, we stratified on new versusunknown duration of use to evaluate possible bias ormodification of effect. We also stratified on calendaryear to evaluate whether potential bias related to therecent marketing of oral contraceptives containingdrospirenone could explain our results. Neither analy-sis materially changed the results. The odds ratios forvenous thromboembolism, adjusted for age and indexyear, were 2.5 (1.7 to 3.8) for women with a new epi-sodeofuseand2.0(0.91to4.3)forthosewithunknownduration of use (table 2). The conditional odds ratiowas 2.1 (1.1 to 4.0) among women with an index datebetween 2002 and 2004 and 2.4 (1.6 to 3.6) amongthose with an index date between 2005 and2008, com-paring oral contraceptives containing drospirenoneand levonorgestrel.

As users of oral contraceptives containing drospire-none were more likely to be younger (less than 30)compared with women using oral contraceptives con-taining levonorgestrel, we investigated for modifica-

tion of effect by age. Women under age 30 who usedoral contraceptives containing drospirenone had ahigher risk of venous thromboembolism than didyoung women who used oral contraceptives contain-ing levonorgestrel (odds ratio 3.7, 2.0 to 6.9). Amongwomenaged 30 to 39, the odds ratio wasnot materiallydifferent from the effect in all women (1.9, 1.1 to 3.3).Among women aged 40 to 44, the odds ratio was 1.4(0.65 to 3.0); this age stratum contained fewer women.Young women taking oral contraceptives containingdrospirenone were more likely to have a shorter dura-tion of oral contraceptive use than were young womentaking oral contraceptives containing levonorgestrel;

however, when we adjusted for duration in the analy-sis, the conditional odds ratio increased to 4.0 (2.1 to7.7) in the younger age stratum.

Users of oral contraceptives containing drospire-none were slightly less likely to have a diagnosis ofobesity in this database (5.8% compared with 6.5% inusers of oralcontraceptives containinglevonorgestrel).When we added obesity to the model as a potentialconfounder, the conditional odds ratio increased to4.1 (2.1 to 7.9) among younger women. The condi-tional odds ratios were virtually identical in the unad- justed analyses and those adjusted for obesity for allwomen in the study: adjusted odds ratio 2.3 (1.6 to

3.3). Finally, drospirenone treated women were morelikely to have a record of a menstrual disorder;

Table 1 | Descriptive characteristics by cases and controls and by exposure among controls.

Values are numbers (percentages)

CharacteristicCases

(n=186)Controls(n=681)

Controls only

Drospirenone users(n= 313)

Levonorgestrel users(n= 368)

Age (years):


5/8

adjustment for this condition in the model yielded anodds ratio of 3.7 (2.0 to 7.0) for women aged under 30

using oral contraceptives containing drospirenonecompared with women under 30 using oral contracep-tivescontaininglevonorgestrel; the odds ratioadjustedfor menstrual disorder in the entire study populationwas 2.3 (1.6 to 3.2). When we restricted the analysis towomen with no history of menstrual disorders or obe-sity in the younger stratum, the odds ratio was 4.0 (1.8to 8.8), adjusted for index year (40 cases in drospire-none users and nine in levonorgestrel users).

Because of concern that the risk of venous throm-boembolism may be related to thedose of ethinylestra-diol in theoral contraceptive, we restricted theanalysisto cases and controls who received an oral contracep-

tive containing either drospirenone or levonorgestrelwith 30 g of ethinylestradiol (table 2). The durationadjusted odds ratio was not materially different fromthe main effect (2.2, 1.5 to 3.4). Among new users, wefurther restricted the analysis to women whohad a pre-vious episode of oral contraceptive use that precededthe episode at the index date to determine if the effectwas modifiedin re-starters comparedwith women withno previous use recorded in their computer record.The duration adjusted odds ratio in this stratum ofwomen was not materially different from the maineffect (2.6, 1.4 to 4.6). When we restricted the analysisto women with no previous episode of use recorded,

the odds ratio was slightly higher (2.8, 1.5 to 5.2)(table 2).

The study population contained 937 408 womenwho satisfied all the conditions for inclusion in thisstudy. These women contributed an estimated392 844 woman years for oral contraceptives contain-ing drospirenone and 521 824 woman years for oralcontraceptives containing levonorgestrel. The inci-

dence rates for venous thromboembolism in thestudy population were 30.8 (95% confidence interval25.6 to 36.8) per 100 000 woman years among users oforal contraceptives containing drospirenone and 12.5(9.61 to 15.9) per100 000 woman years amongusersoforal contraceptives containinglevonorgestrel. The ageadjusted incidence rate ratio for venous thromboem-bolism comparing current use of oral contraceptivescontaining drospirenone and levonorgestrel was 2.8(2.1 to 3.8). The incidence rate of venous thromboem-bolismincreased with increasing agefor both oral con-traceptives. The incidence rate per 100 000 womanyears among users of oral contraceptives containing

drospirenone was 24.8 (19.1 to 31.7) among womenaged 15-29 years, 39.0 (28.1 to 52.7) among thoseaged 30-39, and 51.2 (29.3 to 83.2) among those aged40-44. Among users of oral contraceptives containinglevonorgestrel, the corresponding incidencerates were5.39 (2.94 to9.05), 18.7 (13.0 to 26.0), and 21.3 (12.1 to34.5) (table 3).

DISCUSSION

These data provide evidence that current users of oralcontraceptives containing drospirenone have anincreased risk of non-fatal venous thromboembolismcompared with current users of oral contraceptives

containing levonorgestrel (adjusted odds ratio 2.4 (1.7to 3.4); incidence rate ratio adjusted for age 2.8 (2.1 to3.8)). We compared the risk of venous thromboembo-lism in users of oral contraceptives containing drospir-enone with that in users of oral contraceptivescontaining levonorgestrel, because these second gen-eration oral contraceptives have been shown to haveamong the lowest risk of venous thromboembolism oforal contraceptives on the market and have been usedin past studies of oral contraceptives in relation tovenous thromboembolism.5-7101120 Wealsodidanana-lysis restricted to women who were new users of thestudy oral contraceptive. We did this to ensurethat the

more recent availability of the oral contraceptives con-taining drospirenone did not bias the study findings, asall patients in the study had to have started the currentepisode of oral contraceptive use after 1 January 2002,soonafter oralcontraceptives containingdrospirenonewere first marketed. We also assessed the potential forconfounding or modification of effect by age and obe-sity. Although we found some differences in effect inyounger women compared with women aged 30 andolder, the main finding that oral contraceptives con-taining drospirenone conferred an increased risk ofvenous thromboembolism compared with those con-taining levonorgestrel remained. The increased risk

also remained when we adjusted for the effects of obe-sity and history of menstrual disorders.

Table 2 | Odds ratios for venous thromboembolism in users of oral contraceptives containing

drospirenone compared with those containing levonorgestrel

Exposure No (%) cases No (%) controlsCrude* odds ratio

(95% CI)Adjusted odds

ratio (95% CI)

Overall

Levonorgestrel 65 (15) 368 (85) 1.0 1.0

Drospirenone 121 (28) 313 (72) 2.3 (1.6 to 3.2) 2.4 (1.7 to 3.4)Levonorgestrel 20 users only

Levonorgestrel-20 20 (13) 131 (87) 1.0 1.0

Drospirenone 121 (28) 313 (72) 2.7 (1.6 to 4.7) 3.2 (1.8 to 5.5)

Levonorgestrel 30 users only

Levonorgestrel-30 45 (16) 237 (84) 1.0 1.0


New episodes of use only

Levonorgestrel 42 (14) 253 (86) 1.0 1.0


Users of unknown duration only

Levonorgestrel 23 (17) 115 (83) 1.0 1.0


New episodes of use with no previous episode

Levonorgestrel 20 (17) 97 (83) 1.0 1.0


New episodes of use with previous episode

Levonorgestrel 22 (12) 156 (88) 1.0 1.0


*For overall analysis, crude odds ratio is a conditional odds ratio; for stratified analyses, crude odds ratios areadjusted for age and index year.Also adjusted for duration.

RE S E ARC H



6/8

The results of the cohort analysis of these data areconsistent with those of earlier studies of oral contra-ceptives and venous thromboembolism. The rates

increased with increasing age and were of the samemagnitude as has been seen in previous studies.561314

The incidence rate ratio estimated from these data wassimilar to the odds ratio estimated from the case-con-trol analysis, although it was slightly higher. We werenot able to control as precisely for age and calendartime in the cohort study as in the case-control study,which could explain some of the small difference inthe effect measures.

Although women who received oral contraceptivescontaining drospirenone tended to have shorter dura-tionofusecomparedwiththosewhoreceivedoralcon-traceptives containinglevonorgestrel, both controlling

for duration of use and stratifying by duration of usedid not materially change the effect estimates, so dura-tion of use does not explain the increased risk in usersof drospirenone oral contraceptives. We also evalu-ated whether material bias existed related to the morerecent availability of oral contraceptives containingdrospirenone compared with the older ones contain-ing levonorgestrel. We found only a small differencein the effect when we looked at the first several yearsafter marketing compared with the later years; theeffect was greater in the later years, suggesting thatbias related to prescribing of a newly marketed oralcontraceptive did not explain our finding.

Strengths and limitations

This epidemiological study used a case-control designthat ensured comparability between cases and thecomparison group at the time of the case event. Ageand calendar time were closely controlledthat is,the controls were matched to cases on year of birth,and the date at which exposure was determined (theindex date) was identical in cases and controls. This isimportant, as oral contraceptives containing drospire-none have been on the market for around 10 yearswhereas those containing levonorgestrel have beenavailable for decades. We excluded patients with

chronic medical conditions such as cancer, coronaryartery disease, and autoimmune disease from the

study population. Although these conditions were notcommonly observed in this generally healthy youngpopulation of contraceptive users, the exclusion ofsuch patients from the study population limits con-cerns about selective prescribing of the study drug onthe basis of the presence of clinical risk factors (con-founding by indication). More than 937 000 womenwho used one of the study drugs in our study popula-tion provided information on therisk of venous throm-boembolism in relation to oral contraceptivescontaining drospirenone compared with levonorges-trel. Because of the prospective nature of data collec-tion, the information on contraceptive use wasrecorded before the outcome had occurred, all eligiblepatients with the outcome were included, and the like-lihood of correct diagnosis of venous thromboembo-lism was increased by the documentation of long termuse of anticoagulants. Although some of the cases thatwe included in the study may not have been true idio-pathic cases, the misclassification is highly unlikely tohave been associated with contraceptive use, particu-larly as allcaseswere currently using some oral contra-ceptiveattheindexdate.Mostwomeninthestudyhadbeen filling prescriptions for a studyoralcontraceptivefor several months and as much as several years beforethe index date, so misclassification of exposure is unli-kely to have been a material factor in this study.

This study does have some limitations. Because wecouldnotvalidatethecasesinthisstudythroughreviewof primary records, we may have included some casesthatwere nottruecasesof venous thromboembolism orthat were not idiopathic cases, although any such mis-classification would probably have been non-differen-

tial as we identified cases and controls withoutknowledge of the contraceptive that they had beenusing. Non-differential misclassification of a dichoto-mous variable tends to bias results toward the nulland thus would not explain the increased risk foundin users of the oral contraceptives containingdrospirenone.18 In addition, in the accompanyingstudy using the General Practice Research Database,in which we were able to validate many cases, wefound a similar magnitude of effect.20 We could notevaluate the effect of smoking in this study, as it is notregularly recorded in the PharMetrics database. How-ever, smoking has not been a material confounder in

previous studies on the association between oral con-traceptives and venous thromboembolism,56 nor was ita confounder of the relation between oral contracep-tives and venous thromboembolismin the accompany-ing General Practice Research Database study, whichwas able to control for smoking.20 It is thus not likely tobe a material confounder in this study. Neither heightnor weight was available in this study. Although bodymass index is independently associated with anincreased risk of venous thromboembolism, it has notconfounded the association between use of hormonalcontraceptives and venous thromboembolism in pre-vious studies.35-7 Furthermore, when we evaluated the

ICD-9 diagnosis for obesity, we found that obesity wasassociated with an increased risk of venous

Table 3 | Incidence rates and incidence rate ratios for venous thromboembolism in users of

drospirenone and levonorgestrel oral contraceptives

Exposure Cases (n=186) Person yearsIncidence rate (95% CI) per

100 000 person yearsIncidence rateratio (95% CI)

Drospirenone/ethinylestradiol

Age


7/8

thromboembolism(odds ratio 2.4,1.4 to 4.0),but inclu-sion of obesity in the model with exposure did notmaterially change the effect of contraceptive use, pro-viding additional reassurance that obesity is not likelyto be an important confounder in this study. We foundthatusers of theoral contraceptives containing levonor-gestrel were more likelyto be obese than were theusersof those containing drospirenone, so confounding byobesity would not explain the increased risk for oralcontraceptives containing drospirenone. So, althoughthe diagnosis of obesity could be subject to reportingbias, whereby only the most obese women have thediagnosis recorded in the data, it would not explainthe study result. Furthermore, we did have informationon body mass index in the accompanying GeneralPractice Research Database study, and it did not mate-rially confound the relation between the oral contra-ceptives and venous thromboembolism.20 As in thePharMetricsdata, women whoreceived oral contracep-tives containing levonorgestrel were more likely tohave a high body mass index than were users of thosecontaining drospirenone, but inclusion of body massindexinthelogisticregressionmodeldidnotmateriallychange the effect measure and did not account for theincreased risk of venous thromboembolism in users oforal contraceptives containingdrospirenone comparedwith levonorgestrel. Finally, we did not have informa-tion on family history of venous thromboembolism inthe database. Some selection biasamong women withafamily history of venous thromboembolism is possiblebut is unlikely to account for a material portion of theeffect, as idiopathic venous thromboembolism is notcommon.

These data are derived from the PharMetrics data-base, which contains health data for both insured andMedicaid patients, although most are insured.Although this study was carried out in women fromthe United States, other studies in other countrieshave provided no evidence that the association of oralcontraceptive withvenous thromboembolism differs indifferent populations.10112122 Also, this study evaluatednon-fatal venous thromboembolism and therefore didnot directly assess the risk of fatal venous thromboem-bolism, although the effect in fatal and non-fatal caseswould be unlikely to differ greatly. Finally, we studiedthe effect of twooral contraceptives on idiopathic cases

of venous thromboembolism, so this study does notconsider the risk in non-idiopathic cases.

Comparison with other publications

Among the four previously published studies thatexamined the risk of venous thromboembolism inwomen taking the newer oral contraceptives contain-ing drospirenone compared with those taking otheroral contraceptives,8-11 the studies by Dinger et al andSeeger et al found no associations between drospire-none and venous thromboembolism compared withlevonorgestrel and compared withother oralcontra-ceptives. In two more recently published studies by

Lidegaard et al and van Hylckama Vlieg et al, theauthors reported relative risks of 1.6 (1.3 to 2.1) and

1.7 (0.7 to 3.9) comparing oral contraceptives contain-ing drospirenonewiththosecontaining levonorgestrel.Women who were pregnant or postpartum wereexcluded from these two more recent studies, as werewomen with a previous venous thromboembolism.Women with previous cancer or cardiovascular dis-ease were further excluded from the Lidegaard study.However, inclusion of other proximate causes such asrecentsurgeryandinjurycouldexplainthelowereffectestimates in these studies compared with our study.The Seeger and Dinger studies analysed all venousthromboembolisms,includingpatients with a previousvenous thromboembolism, other risk factors, andother proximate causes. These factors could explainthe null association found in these studies. The inclu-sion of non-idiopathic cases of venous thromboembo-lism hasbeen shown to resultin attenuation of an effectshould it exist.1 As in our study, the risk of venousthromboembolism in users of oral contraceptives con-taining drospirenone was compared directly with therisk in users of oral contraceptives containing levonor-gestrel in the Lidegaard, van Hylckama Vlieg, andDinger studies. By contrast, the reference group inthe Seeger study included women using all other oralcontraceptives, including those containing cyproter-one and desogestrel, which have been found to havehigher risks of venous thromboembolism comparedwith the second generation oral contraceptives. Inclu-sion of these women would have led to a higher risk inthe reference group and a lower relative risk for thedrospirenone oral contraceptive. Our study includedonly current users of oral contraceptives containingeither levonorgestrel or drospirenone and avoided

including oral contraceptives carrying a higher risk inthe reference group.

Study implications

We found that, after adjustment for multiple potentialconfounders and biases, current users of oral contra-ceptives containing drospirenone were at around atwofold increased risk of non-fatal idiopathic venousthromboembolism compared with current users oforal contraceptives containing levonorgestrel. Thesefindings support more recent studies that suggest thatdrospirenone oral contraceptives are not as safe aslevonorgestrel oral contraceptives with respect to

venous thromboembolism and, in the absence ofother considerations, should not be the first choice inoralcontraception.A closecomparison of the methodsof the four previously published studies and the twonew studies show possible explanations for the differ-ing results.

Unanswered questions and future research

Additional studies with the same definitions of casesand contraceptive use should be done to see if theyreproduce the same results, particularly now that thedrospirenone oral contraceptive has been on the mar-ket for more than a decade. A systematic review of the

literature already published on this topic would be animportant addition to the literature at this time.

RE S E ARC H



8/8

Contributors: SSJ conceived and designed the study, obtained data andoversaw the conduct of the study, interpreted the results, and wrote the

manuscript. RKH conducted the study and the analyses, interpreted the

results, and contributed to the manuscript. B oth authors had full access toall of the data (including statistical reports and tables) in the study andcan take responsibility for the integrity of the data and the accuracy of the

data analysis. SSJ is the guarantor.

Funding: None.Competing interests: Both authors have completed the UnifiedCompeting Interest form at www.icmje.org/coi_disclosure.pdf (available

on request from the corresponding author) and declare: no support from

any organisation for the submitted work, no financial relationships with

any organisations that might have an interest in the submitted work in the

previous three years, and no other relationships or activities that could

appear to have influenced the submitted work.

Ethical approval: This study was exempt from review by the BostonUniversity Medical Center Institutional Review Board. Data from

PharMetrics are Health Insurance Portability and Accountability Act

compliant.

Data sharing: Open access to the data used in this study would be in

direct conflict with the data licensing agreement with the data vendorPharMetrics.

1 PorterJB, HunterJR, Jick H, Stergachis A. Oral contraceptivesandnonfatal vascular disease.Obstet Gynecol1985;66:1-4.

2 Kemmeren JM,AlgraA, Grobbee DE.Thirdgenerationoralcontraceptives and risk of venous thrombosis: meta-analysis. BMJ2001;323:131-4.

3 Gomes MP, Deitcher SR.Risk of venousthromboembolic diseaseassociated withhormonal contraceptives andhormonereplacementtherapy: a clinical review.Arch InternMed2004;164:1965-76.

4 Farmer RD, Lawrenson RA, ToddJC, WilliamsTJ, MacRaeKD, Tyrer F,et al.A comparison of therisks of venous thromboembolicdiseaseinassociation with different combined oral contraceptives.BrJ ClinPharmacol 2000;49:580-90.

5 Jick H, Jick SS,GurewichV, MyersMW, Vasilakis C. Risk of idiopathiccardiovascular death and nonfatalvenous thromboembolism in

women using oral contraceptives with differingprogestagencomponents. Lancet1995;346:1589-93.

6 Jick H, Kaye JA,Vasilakis-Scaramozza C, Jick SS. Risk of venousthromboembolism among usersof third generation oralcontraceptives compared with usersof oral contraceptives withlevonorgestrel before and after 1995: cohort and case-controlanalysis. BMJ2000;321:1190-5.

7 Vasilakis-ScaramozzaC, Jick H. Risk of venousthromboembolismwith cyproterone or levonorgestrel contraceptives [letter].Lancet

2001;358:1427-9.8 DingerJC, Heinemann LA, Khl-HabichD. Thesafety of a

drospirenone-containing oral contraceptive: finalresults fromtheEuropean Active Surveillance Studyon oralcontraceptives based on142,475 women-years of observation.Contraception2007;75:344-54.

9 SeegerJD, LoughlinJ, Eng PM, CliffordCR, CutoneJ, WalkerAM. Riskof thromboembolismin womentaking ethinylestradiol/drospirenone and other oral contraceptives.Obstet Gynecol2007;110:587-93.

10 Lidegaard , Lkkegaard E, Svendsen AL, Agger C. Hormonalcontraception andrisk of venous thromboembolism:national follow-up study.BMJ2009;339:b2890.

11 VanHylckama Vlieg A, Helmerhorst FM,Vandenbroucke JP,DoggenCJM, Rosendaal FR.The venousthromboticrisk of oralcontraceptives, effects of oestrogen dose and progestogen type:results of the MEGA case-control study. BMJ2009;339:b2921.

12 Rothman KJ,Poole C. A strengthening programme forweakassociations. Int J Epidemiol 1988;17(suppl):955-9.

13 JickS, KayeJ, RussmannS, JickH. Riskof nonfatalvenousthromboembolism with oral contraceptives containing norgestimateor desogestrel compared with oralcontraceptives containinglevonorgestrel. Contraception2006;73:566-70.

14 JickSS, KayeJA, RussmannS, JickH. Riskof nonfatalvenousthromboembolism in women using a contraceptive transdermalpatch and oral contraceptives containing norgestimate and 35microg of ethinyl estradiol.Contraception2006;73:223-8.

15 JickSS, KayeJA, LiL, Jick H.Further results onthe riskof nonfatalvenous thromboembolism in users of the contraceptive transdermalpatch comparedto users of contraceptives containing norgestimateand 35g ethinyl estradiol. Contraception 2007;76:4-7.

16 Jick SS, Jick H. Cerebral venoussinusthrombosisin users of fourhormonal contraceptives: levonorgestrel-containing oralcontraceptives, norgestimate-containing oral contraceptives,desogestrel-containing oral contraceptives and the contraceptivepatch.Contraception2006;74:290-2.

17 Jick SS,Jick H. Thecontraceptive patch in relationto ischemicstroke

and acute myocardialinfarction.Pharmacotherapy2007;27:218-20.18 Rothman KJ,Greenland S, Lash TL.Modern epidemiology. 3rded.

Lippincott Williams & Wilkins, 2008.19 Episheet. Spreadsheets for the analysis of epidemiologicdata

[program]. Episheet, 2005.20 Parkin L, Sharples K, Hernandez RK, Jick SS. Risk of venous

thromboembolism in users of oral contraceptives containingdrospirenoneor levonorgestrel: nested case-control study based onUK General Practice Research Database. BMJ2011;340:d2139.

21 World Health Organization. Venous thromboembolic disease andcombined oralcontraceptives: results of international multicentrecase-control study. Lancet1995;346:1575-82.

22 Spitzer WO,Lewis MA,Heinemann LA,Thorogood M, MacRae, KD.Third generation oral contraceptives and risk of venousthromboembolic disorders: an international case-control study.BMJ1996;312:83-8.

Accepted: 24 March 2011

WHAT IS ALREADY KNOWN ON THIS TOPIC

Previous studies that evaluated the risk of venous thromboembolism in users of oralcontraceptives containing drospirenone have found inconsistent results

Some studies have shown no increased risk compared with other oral contraceptives, andother studies have shown small increased risks with drospirenone

WHAT THIS PAPER ADDS

Users of oral contraceptives containing drospirenone had around a twofold increased risk ofidiopathic venous thromboembolism compared with users of those containinglevonorgestrel, although the overall risk was low

The effects remained when prescribing biases and confounding were taken into account

The incidence rates were 30.8 (95%confidence interval25.6 to 36.8) and 12.5 (9.61 to 15.9)per 100000 woman years among users of drospirenoneand levonorgestrel oralcontraceptives

RE S E ARC H

page 8 of 8 BMJ | ONLINE FIRST | bmj.com

Documents

drospirenona