Comments on the two papers by the Jick group
The two papers published this week in the BMJ by Parkin et al and
Jick et al, both from the same research group, purport to show an
increased risk of VTE in users of drospirenone (DRSP)-containing COCs
compared to levonorgestrel containing COCs. However, they suffer from the
same design faults as several other retrospective case controlled studies
before them. The study by Jick and Hernandez is based on data from a US
claims database. Database studies have been roundly criticised for their
many failings, which include lack of validation of VTE cases and lack of
information on important potential confounders such as duration of use,
family history, body mass index (BMI) and smoking (Grimes 2010, Shapiro
& Dinger, 2010). These criticisms apply to this study also, since none
of the cases were validated and there was no information on family
history, BMI or smoking. There are signs of prescriber bias towards
drospirenone COCs for women who were younger and therefore more likely to
be short-term users, both of which are associated with an increased risk
of VTE, compared to older, long term COC users. Matching of cases and
controls for important potential biases and confounders was therefore
clearly impossible. It is well documented that the risk of VTE is greatest
in the first three months of use and then falls as duration of use
increases (Dinger, et al, 2007, Reid et al 2010). From this paper duration
of use cannot accurately be determined and may lead to starters or
restarters being compared with long term users.
The retrospective case controlled study by Parkin et al is from the
UK General Practice Research Database (GPRD). Once again, the information
on potential confounders and biases is missing or incomplete; over a third
of cases were not validated and there is no data on family history of VTE.
It is interesting that when the analysis was restricted to women with
validated VTE, the results were not statistically significant. Information
on BMI was missing in 7% of cases and 14% of controls, and even when
present, the percentage of drospirenone and levonorgestrel COC users with
a BMI over 30 is not given, despite the fact that this is well documented
to cause a doubling or trebling of VTE risk (Lidegaard et al 2002, Dinger
et al, 2007). Imputation of missing values assumes these are randomly
distributed, but given the potential for prescriber bias, this is by no
means guaranteed. Matching was therefore again poor, and even more so when
one realises that while 4 controls were used per case for levonorgestrel
COC users (44/189), there were roughly 1.5 controls per case for
drospirenone COC users (17/26). It should be noted that in many areas of
the UK there are prescribing restrictions on the use of drospirenone COCs,
because of their higher price; this must result in a selection bias of
women who have some strong reason to be using these pills, rather than a
cheaper brand. This may also account for the considerably smaller numbers
of women in the DRSP group and the difficulty in finding enough controls.
Similarly it is noteworthy that the rates of observed VTE in both
studies, at 1 - 3 per 10,000 are lower than the rates of VTE found in non-
users of COCs in active surveillance studies (Reid et al, 2010). This may
be because they have chosen only to include what they consider to be
'idiopathic' cases. Such restriction has been criticised as failing to
account for unmeasured confounders and biases and leading to analyses
which are not representative of the population risk of VTE (Cushman 2010).
Finally, the authors state that there is no clear evidence for any
non-contraceptive benefits of DRSP containing COCs; however, such pills
have been granted a licence in the USA for both treatment of acne and
severe premenstrual syndrome, on the basis of their efficacy (Panay 2009).
Clinical Senior Lecturer, Wolfson Institute of Preventive Medicine, QMUL,
Consultant in Community Gynaecology and Reproductive Healthcare,
Clinical Director, Community Health Services,
Newcastle Hospitals Community Health
Newcastle upon Tyne
Cushman M. Patch instead of Pill: a safer menopausal estrogen?
Arterioscler Thromb Vasc Biol, 2010; 30: 136-7
Dinger JC, Heinemann LAJ, Kuhl-Habich D. The safety of a drospirenone
-containing oral contraceptive: final results from the European Active
Surveillance study on Oral Contraceptives based on 142,475 women-years of
observation. Contraception 2007; 75: 344-354.
Grimes D. Epidemiologic Research Using Administrative Databases:
Garbage In, Garbage Out.
Obstet Gynecol. 2010; 116: 1018-1019
Jick SS, Hernandez RK. Risk of non-fatal venous thromboembolism in
women using oral contraceptives containing drospirenone compared with
women using oral contraceptives containing levonorgestrel: case-control
study using United States claims data. BMJ 2011;340:d2151.
Lidegaard ?, Edstr?m B, Kreiner S. Oral contraceptives and venous
thromboembolism. A five-year national case-control study. Contraception
Panay N. Management of premenstrual syndrome. J Fam Plann Reprod
Health Care 2009; 35(3): 187-194
Parkin L, Sharples K, Hernandez RK, Jick S, Risk of venous
thromboembolism in users of oral contraceptives containing drospirenone or
levonorgestrel: nested case-control study based on UK General Practice
Research Database. BMJ 2011;340:d2139 doi:10.1136/bmj.d2139
Reid RL, Westhoff C, Mansour D, de Vries C, Verhaeghe J, Boschitsch
E, Gompel A, Birkh?user M, Krepelka P, Dulicek P, Iversen OE, Khamoshina
M, Dezman LV, Fruzzetti F, Szarewski A, Wilken-Jensen C, Seidman D, Kaaja
R, Shapiro S. Oral contraceptives and venous thromboembolism consensus
opinion from an international workshop held in Berlin, Germany in December
2009. J Fam Plann Reprod Health Care. 2010 Jul;36(3):117-22.
Shapiro S, Dinger J. Risk of venous thromboembolism among users of
oral contraceptives: a review of two recently published studies. J Fam
Plann Reprod Health Care 2010;36:33-8.
Competing interests: Both authors have received honoraria, conference sponsorship and consultancy fees from a number of pharmaceutical companies including Bayer HealthCare and MSD.