Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study, 2001-9
BMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d6423 (Published 25 October 2011) Cite this as: BMJ 2011;343:d6423
All rapid responses
Recent evidence from long term observations of hundreds of thousands of women, in 10 European Countries, clearly demonstrated that the use of oral contraceptives reduced mortality by roughly 10%. Despite all described thromboembolism risks, apparently.
Reference
doi: 10.1186/s12916-015-0484-3.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4627614/
Competing interests: No competing interests
(Continued from part 2)
Risk according to length of use.
Most previous studies have demonstrated a decreasing risk of VTE with increasing length of use, so that the risk is elevated about 50% the first year, and thereafter is almost constant.
We were also able to demonstrate such an overall trend by time. When you stratify current users into 16 different user groups, each with confirmed and non-confirmed events, and thereafter further subdivide your exposure time into four categories according to length of use, you cannot expect to find an equal decrease for all sub-groups by time. We nevertheless found a small although significant decrease for users of OCs with DRSP when comparing use for less than a year with use for a year or longer (2). The difference in the relative risk estimates by time of use for OCs with DRSP between our 2009 and 2011 publications are probably a result of the different censoring rules applied in the two analyses. Actually we have found no other study demonstrating a decreasing trend by time for specifically OCs with drospirenone. On the contrary, the FDA study (4) and an Israeli study (5) did not find any consistent trend in risk of VTE by duration of use for OCs with DRSP.
Surprisingly, Dinger did not report any risk estimates according to length of use in his EURAS publication or his German case-control study from 2010.
JD & SS write (page 3 second column) that “...the investigators stipulated rules based on several assumptions, in order to minimise the impact of the lack of precise exposure information..” This statement is wrong. The detailed set of allocation rules was a result of exceptionally precise exposure information. Therefore we had to decide how we would define e.g. continuous use, new use, re-started use and switched use, all made possible due to the very precise exposure information. These rules were not made on assumptions, but were established by the Steering Committee in order to ensure the highest possible validity of the risk estimates.
Confounding by BMI and family disposition.
Of nine existing studies examining the risk of VTE in users of OCs with DRSP, five had access to BMI. In none of these studies did confounder control for BMI change the risk estimates or rate ratios between OCs with DRSP versus LNG significantly, and in four of five not at all (5-9). Two of these studies were conducted by Dinger. The average BMI in his first study was 22.0 among users of OCs with LNG and 22.9 among users of OCs with DRSP (6). The proportion of women with BMI ≥30 was (read from Fig 1 in (6)) 5.3% in users of LNG OCs and 8.2% among users of DRSP OCs, but users of DRSP OCs were also slightly older than the users of LNG OCs. Dinger stated ““..the differences were small, and the preferential prescribing pattern identified here could only slightly increase the incidence of VTE … for the DRSP cohort” (6). As BMI did not influence the risk estimates materially in any of these five studies, it is difficult to understand why JD & SS continues to insist that our data should be invalid due the this missing information. The same studies have documented that users of DRSP are not selected according to BMI or other risk factors of VTE.
Similar arguments can be made concerning the postulated confounding influence from family disposition. Despite being a definite risk factor for VTE, in no study over the last 10 years was it found to be a confounder.
So despite being repeated by JD & SS again and again, there is no scientific evidence at all, suggesting that the lack of confounder control for BMI or family disposition distorted our results or rate ratios.
Audit
After we had delivered our EMA report, Bayer-Pharma asked us whether we were willing to participate in an external audit of our study. That request was not made by the Steering Committee. We accepted this audit on the condition that if it was used by Bayer-Pharma in any external connection, we should have the right to comment on the audit-report. With the statements about this audit made by JD & SS in their critique, Bayer-Pharma has violated this agreement. It should be noted, however, that the audit pertains to documentation of procedures, not the reliability of the results. Secondly we asked that the audit team not only analysed our scientific process but also our actual scientific results. This request was refused by Bayer-Pharma. The conclusion of the audit report was that certain formal recordkeeping procedures set forth in a set of accreditation rules, many of which are applicable in clinical trial settings rather than rules for use of a government national registry, were not followed, but that the auditors had no reason to doubt the qualifications of the investigator team or the validity of the results. We could add that no specific procedural rules were agreed to in advance of our study, and that we were working to meet specific time constraints making it difficult to fulfil some accreditation rules e.g. that two independent statisticians should have done all the analyses for comparison.
Several of the statements made by JD & SS concerning this audit are objectively wrong. E.g. there was a detailed signed protocol describing the statistical analysis strategy before the analysis was commenced. The audit team had access to all our data and all our analyses – so transparency was definitively present.
Concerning transparency, it is also important to be aware of the fact, that Danish registries are available for any (qualified) scientist who want to investigate a scientific issue. Thus any other researcher could get access to the same data as based our analyses, as these data belongs to the state and not to any particular scientist.
Finally the willingness to conduct a re-analysis of our 2009 study, and our agreement after the conclusion of the re-analysis to an external audit, prove more than anything else our scientific openness and wish to ensure transparency. Few researchers can claim to have accepted such oversight, nor have any of Bayer’s company sponsored studies been subjected to such scrutiny.
On the authors behalf
Øjvind Lidegaard
References
2. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard E. Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001-9. BMJ 2011; 343: d6423.
4. Food and Drug Adminidsration, Office of surveillance and epidemiology. Combined hormonal contraceptives (CHCs) and the risk of cardiovascular disease endpoints. FDA 2011: http://www.fda.gov/downloads/Drugs/DrugSafety/UCM277384.pdf.
5. Gronich N, Lavi I, Rennert G. Higther risk of venous thrombosis associated with drospirenone-containing oral contraceptives: a population-based cohort study. CMAJ 2011; DOI:10.1503/cmaj.110463.
6. Dinger JC, Heinemann LAJ, Kühl-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: 344-54.
7. Vlieg AVH, Helmerhorst FM, Vandenbroucke JP, Doggen CJ, Rosendaal FR. The venous thrombotic risk of oral contraceptives, effects of estrogen dose and progestagen type: results of the MEGA case-control study. BMJ. 2009; 339: b2921.
8. Dinger J, Assmann A, Möhner S, Minh TD. Risk of venous thromboembolism and the use of dienogest- and drospirenone-containing oral contraceptives: Results from a German case-control study. J Fam Plann Reprod Health Care 2010; 36: 123-9.
9. Parkin L, Sharples K, Hernandez RK, Jick SS. 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.
Competing interests: The primary investigator received no salary for his work with this study, the EMA report or the manuscript. OLi has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report. The other authors had nothing to declare
(Continued from part 1)
Validity of data
It is important to realise, that determining a reliable risk estimate of a certain pill demands valid exposure as well as valid end point data. Every time an exposure is misclassified or an end point is wrong, we will underestimate the real risk of VTE in current users of OCs.
Concerning the exposure data, it is difficult to imagine a more precise data source than a prescription registry. Not only does a prescription registry give precise information about the date a woman receives package of pills, but it also provides valid information about which type of pill is prescribed. As compared with retrospective studies, in which women are asked about use of OCs months or years back in time, or prospective studies, in which a certain exposure at a certain time is not up-dated, a prescription registry is by far the most reliable exposure data available.
The small uncertainty as to exactly which day a woman commences her use as compared with the date she buys the OC is minor as compared to misclassifications made from other data sources. And more important: Any such misclassification would underestimate both the risk of VTE with current use and the rate ratio between OCs with DRSP versus OCs with LNG.
Concerning the outcome data (VTE), the primary data source was discharge diagnoses from hospital wards. The fact that Denmark collects discharge diagnoses in a National registry does not make these diagnoses less reliable than diagnoses of VTE in other studies.
Moreover, we validated all 4,246 events by cross linking them with subsequent anticoagulation therapy, and restricted our analyses to these anticoagulation-confirmed events. This validation (not surprisingly) elevated the rate ratio estimates between users of OCs with DRSP versus LNG from 1.64 in our 2009 publication to 2.1 (1.6-2.8) in the new analysis (2).
Next, in a random sample of 200 women with a discharge diagnosis of VTE, through chart review we found a positive predictive value of confirmed VTE according to the registry data to be 99%. But again it is important to realise, that even when no record of anticoagulation therapy is present in the prescription registry, this is not the same as a false VTE diagnosis.
First, about 10% of women get the anticoagulation therapy for free from the departments, and they are therefore not included in the prescription registry. In addition, 5-10 per cent are diagnosed based upon clinical symptomes of VTE despite lack of confirmation in the paraclinical investigations, possibly because the clot was too small, or alternatively, that the clot might have dissolved spontaneously. In both cases no treatment may have been warranted, but the woman was found and told to have probably had a VTE. If we add the 10% ward treated women and those with clinical symptoms but without treatment, we approach the 88% achieved in our 2002 study, which was based upon information from departments and from questionnaires, and in which the women themselves confirmed their diagnosis.
Also the fact that some real events were not included in the group of confirmed events will not change the rate ratio estimates, and only marginally affect the confidence intervals. The fact that the proportion of confirmed events among users of OCs with DRSP (73.7%) and in users of OCs with LNG (73.2% and 74.2% for combined and cyclic products, respectively) was similar, strongly contradict a differential referral or a differential diagnosis of VTE among these two groups of OC users. And the number of confirmed and non-confirmed events was the same in the EMA report as in the BMJ publication. But JD & SS compared figures from the period 2001-2005 in the EMA report with figures from another period 2001-2009 in the BMJ publication.
In conclusion our study had well defined validation criteria for VTE, applied to all groups of users of OCs. The rate ratio estimate for confirmed events; 2.1 (1.7-2.7) was slightly higher than for the non confirmed events; 1.8 (1.2-2.6), indicating that the more valid the diagnoses are, the larger rate ratio estimates of VTE we find between users of OCs with DRSP versus LNG-users.
And last but not least. If anything, the inclusion of uncertain events will tend to decrease the risk estimates as demonstrated by the rate ratios described above, and cannot provide a basis to discount positive findings in our study.
Considering how carefully JD & SS have evaluated the smallest methodological details in our studies, it is surprising that they never made any reflection as to which direction their many proposals of bias would move the risk estimates. In fact, any of the alleged biases suggested by JD & SS if anything all tend to underestimate the rate ratios between users of OCs with DRSP versus OCs with LNG.
On the authors behalf
Øjvind Lidegaard
Continues in part 3
2. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard E. Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001-9. BMJ 2011; 343: d6423.
Competing interests: The primary investigator received no salary for his work with this study, the EMA report or the manuscript. OLi has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report. The other authors had nothing to declare.
Jürgen Dinger (JD) and Samiel Shapiro (SS) have published in the Journal of Family Planning and Reproductive Health Care a critique (1) of our new study published in BMJ online on October 25, 2011, in which we described the risk of venous thromboembolism (VTE) in users of different types of oral contraceptives (OCs)(2). Permit us to go through each of the critique points raised by JD and SS roughly in the same order as they appear in their critique.
Background
The BMJ publication was not an abbreviated version of the report made for the European Medicines Agency (EMA). On the one hand, the EMA report was a priori restricted in its aim, e.g. we were limited to include results defined in the protocol and by the Steering Committee. Where the EMA report analysed six different product groups, the BMJ publication reported 16 different product groups. Thus results on low-dose (20 µg oestrogen) OCs with drospirenone (DRSP) first published in the BMJ paper, were not included in the EMA report. On the other hand, the EMA report included many supplementary tables, not included in the BMJ publication. The BMJ publication was based on analyses conducted by the author team, with no influence from parties other than the authors. The authors had the aim of reporting all relevant results on all types of oral contraceptives, to calculate relative risks with non-users as a reference, and to present rate ratios between different types of OCs according to oestrogen dose, progestogen types and duration of use. It is our opinion that all relevant results were presented in the paper and its appendices.
Relevance and presentation of different analyses
First JD & SS state that users of OCs with DRSP could only have commenced their OC use in 2001. This is wrong. While DRSP was introduced to the market in 2001, the majority of DRSP users had taken other OCs before 2001. Therefore, users of OCs with either LNG or DRSP could have used OCs long before 2001. Therefore the “attrition of susceptibles” would be in effect for users in either group.
Secondly, it was decided in the protocol to conduct sub-analyses stratified into starters, new-users (defined as users with at least 12 weeks of pause before the new use), re-starters (4-12 weeks of pause) and switchers (less than 4 weeks of pause). The results demonstrated significantly increased rate ratios of VTE between users of OCs with DRSP and levonorgestrel (LNG) ranging from 1.96 to 2.69 for the different user categories. These results were reported in appendix 4 (2).
Third, no studies on OCs and VTE (to our knowledge), including the studies by Dinger et al. have ever required exclusion of every woman who used any type of OCs before the beginning of the study period. The reason that no studies use this criteria is obvious; such a requirement would not provide a sufficient number of exposed women or a sufficient number of events to ensure reliable estimates. Further, the pharmacologic effect of OCs on coagulation disappears within days rather weeks after cessation. Therefore, no scientific reason exists to support such an exclusion criteria
Nevertheless, SS (not the Steering Committee) insisted upon seeing a sub-analysis in which all users were not only starters after 2001 but also women who had never used OCs at any time prior to 2001. As expected the rate ratio estimates between OCs with drospirenone versus OCs with LNG were unstable and ranged from 0.5 to 1.9 for different duration categories, with an overall estimate of 1.0. We concluded this finding was due to chance as a result of the low number of events in users of OCs with LNG (n=11). This interpretation was confirmed by the rate ratio of 2.05 for all starters and new users as defined in the study.
Note that we conservatively required a pause of at least 12 weeks for a woman to be considered a new user in our study. No scientific evidence suggests that previous use before a pause of at least 12 weeks influences the risk of thrombosis with new use. Based upon the instability caused by the few cases contained in the sub-analysis of non-users before 2001, the lack of scientific basis for such an exclusion criteria and the inconsistency of these results when compared to our rationally designed comparison of starters and new users, we did not include the SS requested analysis in our BMJ paper. It simply did not add any reliable or relevant information to what already was included in Appendix 4. Further, as no previous study has adopted the requirement of no previous use of any OCs, all evidence suggests that the different results achieved in different studies have nothing to do with the speculations JD & SS have about this issue. We are confident that our inclusion of previous users of OCs before 2001 is a valid design decisions. It is also transparent as we do present in our paper rate ratios of the different user categories. As authors of a scientific paper it is our responsibility to present reliable data and not to report unsustainable results.
And fourth: If the speculations from JD & SS had the slightest relevance, you should expect a difference in risk between the older 3rd generation OCs with desogestrel or gestodene and the newer OCs with DRSP. However, the risk of VTE with use of OCs with DRSP was similar to the risk of VTE in users of 3rd generation pills. In addition, by allocating continuous users at start of study in the correct “duration category”, the study gained power in rate ratio analysis for longer duration of use categories.
Fifth, in Dingers own study (3), he stressed that the risk by length of use was constant after the first year, which contradicts the proposal of “attrition of susceptibles” in their critique.
And finally sixth, if previous use of OCs was a relevant issue, you should expect a differential rate ratio in risk of VTE according to comparable length of use categories. However, even in the group of more than four years of use, the rate ratio DRSP versus LNG was 2.31.
So for at least six good reasons there is no evidence of any differential influence from previous use of OCs on the risk estimates of VTE achieved in our new study among users of OCs with DRSP versus LNG.
On the authors behalf Øjvind Lidegaard
(continues in part 2 and part 3)
1. Dinger J, Shapiro S. Combined oral contraceptives, venous thromboembolism, and the problem of interpreting large but incomplete datasets. J Fam Plann Reprod Health Care 2011.doi:10.1136/fjprhc-2011-100260.
2. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard E. Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001-9. BMJ 2011; 343: d6423.
3. Dinger J, Thai DM, Moehner S. The Risk of Venous Thromboembolism in OC Users: Time Patterns after Initiation of Treatment. Pharmacoepidemiology and Drug Safety, 2010; 19: S214-215.
Competing interests: The primary investigator received no salary for his work with this study, the EMA report or the manuscript. OLi has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report. The other authors had nothing to declare.
Thanks to Dr. Alfred R Pauls for his rapid response to our paper on oral contraceptives (OC) and venous thrombosis (1).
First Dr. Pauls think I did wrong in using the term “accusations”. This term should be seen on the background of eight circumstances:
1. Samuel Shapiro and Jürgen Dinger published about half a year after our primary publication in 2009 (2) a critique in which they concluded that our study was invalid for mainly three reasons: a) left censorship, b) missing confounder control for BMI, c) lack of validation of our end points (VTE). We were never presented for this critique before publication, but refused all these three critique points as a possible explanation of the increased risk of VTE in users of OC with drospirenone as compared with users of OC with levonorgestrel (3).
2. Before publishing their critique, Shapiro and Dinger participated in so called “expert meeting” in Berlin, arranged by Bayer Pharma, in which the Danish and the Dutch studies were criticised according to the same points as in their paper. None of the authors of the Danish or Dutch studies were invited to ensure the possibility of defence of our study methods.
3. We accepted after a request from the European Medicines Agency to conduct a new analysis of Danish registry data. Shapiro planned as a member of the steering committee the protocol of this new analysis, and signed and approved the protocol. The new study a) eliminated left censorship, b) stated that BMI has not been a (significant) confounder in any study with access to this information and c) validated all end points. The new results demonstrated a rate ratio between users of OC with drospirenone versus users of OC with LNG of 2.1 (1.6-2.8)(1).
4. Nowhere in the protocol of the new study it was stated, that an analysis should be conducted in which the women were not allowed to have used hormonal contraception at any time before the study period. This wish was a tertiary wish from Shapiro, after he realised the results of the new analysis.
5. Contrary, it was clearly stated in the protocol, that we should make an analysis restricted to starters and new users, where new users were defined as women with at least 12 weeks pause before current use. These results were presented as planned in appendix 4 of our BMJ paper.
6. We conducted dozens of stratified supplementary analyses in the report to the European Medicines Agency. All these sub-analyses showed the same rate ratio of about 2. But there is a limit to how tiny sub-stratifications you can do, before you drown in stochastic noise. That is so few events that coincidence and chance are more prevalent than valid estimates. The request by Shapiro was such an example. With exclusion of 94% of the exposure time and 93% of the VTE events in users of OC with levonorgestrel you simply have too little power to make valid and reliable estimates. I could add that I could deliver several other small samples demonstrating larger rate ratios than two. As author you have the responsibility to present reliable results, not stochastic noise. This has nothing to do with censoring results for you or anybody else.
7. No other study to my knowledge has limited included women to women who have never used any type of hormonal contraception before current use. Neither the EURAS study by Dinger et al. in which Shapiro was also in the steering committee.
8. All these things have we explained Samuel Shapiro in writing and verbally several times. So he is and was fully aware of these scientific facts.
Perhaps “accusations” was not the most appropriate word, - I take the full responsibility for that – but that choice could be influenced by an approaching saturation point of an apparently endless critique from Samuel Shapiro, despite how many times we prove his critique points wrong.
Next Dr. Pauls talks about “true first-ever users” with reference to two studies by Suissa et al. (4) and by Dinger et al. (5), respectively. However, in the study by Suissa, as I read it, the inclusion criteria was that 2nd and 3rd generation users of OC were first time users of these products, not first time users of any hormonal contraception. In the study of Dinger et al. (which by the way was just an abstract) they defined starters as women with a pause of at least 4 weeks before current use. By way of comparison we had at least 12 weeks of pause. So nothing in the two references Dr. Pauls refers to indicates that our starters were less “true” than starters in previously conducted studies.
And further. If Dr. Pauls think women should have been without hormonal contraceptive use for a longer period than 12 weeks before current use, what should then be the scientific argument for that? I have seen no study demonstrating or even suggesting that previous use more than 12 weeks ago should have any influence on risk estimates of VTE with new use. If such study exists, I would be grateful for a reference.
Finally concerning John Tukey: Why not choose an exact answer to a right question?
Øjvind Lidegaard
1. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard E. Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001-9. BMJ 2011; 343: d6423.
2. Lidegaard O, Lokkegaard E, Svendsen Al, Agger C. Hormonal contraception and risk of venous thromboembolism: national follow-up study. BMJ 2009; 339: b2890.
3. Lidegaard Ø. Critique of a Danish cohort study on hormonal contraception and VTE. J Fam Plann Reprod Health Care 2010;36:103-104 doi:10.1783/147118910791069303
4. Suissa S, Blais L, Spitzer WO, Cusson J, Lewis M, Heinemann L. First-time use of newer oral contraceptives and the risk of venous thromboembolism. Contraception 1997; 56(3):141-6.
5. Dinger J, Thai DM, Moehner S. The Risk of Venous Thromboembolism in OC Users: Time Patterns after Initiation of Treatment. Pharmacoepidemiology and Drug Safety, 2010; 19: S214-215
Competing interests: All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare that: Bayer Schering Pharma is thanked for covering the expenses of the analysis. All funding was given to Rigshospitalet, and the primary investigator received no salary for his work with this study, the EMA report or this manuscript. OL has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report.
The authors of the Danish cohort study (1) would do well by using adequate style and content in their reply to the rapid responses. Serious arguments that challenge the validity of their results should not be dismissed in an apodictic and impolite style: The use of “accusations” for serious scientific concerns is not what the readers of the BMJ are entitled to.
As regards respect for your readers: I prefer to decide for myself whether information is relevant, rather than have it censored by authors. In the case of Shapiro’s request for a separate analysis comparing true first-ever users, the relevance of which is supported by the results of several scientific groups (2, 3), I concur with the principle attributed to John Tukey: “Far better an approximate answer to the right question, than an exact answer to the wrong question”.
1. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard E. Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study 2001-9. BMJ 2011; 343: d6423.
2. Suissa S, Blais L, Spitzer WO, Cusson J, Lewis M, Heinemann L. First-time use of newer oral contraceptives and the risk of venous thromboembolism. Contraception 1997; 56(3):141-6.
3. Dinger J, Thai DM, Moehner S. The Risk of Venous Thromboembolism in OC Users: Time Patterns after Initiation of Treatment. Pharmacoepidemiology and Drug Safety, 2010; 19: S214-215
Competing interests: The author has consulted and worked for various scientific institutions, pharmaceutical companies, agricultural and medical device industry. Ongoing consultancy in sexual medicine, drug research methodology and epidemiology.
Although repeated indefinitely, a risk factor is not the same as a
confounder. There is no evidence so far in any of the studies having
access to BMI, that this variable is a confounder despite how many times
it has proved to be a risk factor. A dose response relationship is an
argument for a variable being a risk factor, not a confounder. So far, no
study have demonstrated any confounding influence from BMI, as adjustment
for BMI in studies with this information did change the rate ratio between
COC with different progestogens only marginally or not at all1-5.
Therefore, it is rather unlikely, that especially Danish women in
this decade who smoke or are adipose or are predisposed for venous
thrombosis should preferentially be prescribed a certain pill rather than
another. Thus, we have good reasons to anticipate that confounding cannot
explain the Danish findings.
It is true that adjustment for age changed the Danish estimates much more
than age adjustment in the study of Dinger et al2. The reason for this
difference is quite simple: There were large age differences between users
of oral contraceptives with levonorgestrel and drospirenone, respectively
in our study, an age difference not found in the study by Dinger.
About preferential prescribing of third generation oral
contraceptives to women with family predisposition early after their
introduction, we dealt with that issue in our last answer to Szarewski
after our 2009 publication
(http://www.bmj.com/content/339/bmj.b2890.full/reply#bmj_el_270693). Our
opinion has not changed since then, except that we now have further new
studies confirming that no preferential prescribing is in effect according
to the risk factors mentioned by Anne Szarewski3, 5.
Therefore the conclusion by Szarewski et al. that our results should
be due to confounding is without any empirical support and against
substantial consistent epidemiological findings over the last decade.
1. Lidegaard O, Edstrom B, Kreiner S. Oral contraceptives and venous
thromboembolism. A five-year national case-control study. Contraception
2002; 65: 187-96.
2. 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: 344-54.
3. Vlieg AVH, Helmerhorst FM, Vandenbroucke JP, Doggen CJ, Rosendaal
FR. The venous thrombotic risk of oral contraceptives, effects of
oestrogen dose and progestagen type: results of the MEGA case-control
study. BMJ. 2009; 339: b2921.
4. Dinger J, Assmann A, Mohner S, Minh TD. Risk of venous
thromboembolism and the use of dienogest- and drospirenone-containing oral
contraceptives: Results from a German case-control study. J Fam Plann
Reprod Health Care 2010; 36: 123-9.
5. Parkin L, Sharples K, Hernandez RK, Jick SS. 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.
Competing interests: All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare that: Bayer Schering Pharma is thanked for covering the expenses of the analysis. All funding was given to Rigshospitalet, and the primary investigator received no salary for his work with this study, the EMA report or this manuscript. OL has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report.
Why were most women in Professor Lidegaard's Danish study using
combined oral contraceptives containing desogestrel, gestodene,
drospirenone, or cyproterone which each increased risk of venous
thromboembolism by 6-7 times compared with non-current use?1 Is it
because women usually feel better if they use more oestrogenic
progestogens?2
In Denmark 91.4% of 3 049 560 recorded women years of current use
between 1991 and 2009 was with COCs containing desogestrel, gestodene,
drospirenone, or cyproterone. Only 8.6% of current use was with
levonorgestrel containing COCs which had a 3 fold increased risk of venous
thromboembolism. Levonorgestrel has extra androgenic activity but no
inherent oestrogenic activity.
Lidegaard and colleagues believe that venous thromboembolism is the
greatest health risk for current users but this is not necessarily true.
Levonorgestrel is the active half of norgestrel which was tested,
along with other progestogens, in a range of doses in the 1960s in London.
John Pryse-Davies and I discovered that strongly progestogenic/low dose
oestrogen COCs increased monoamine oxidase activities throughout use and
were more likely to cause depression and loss of libido than more
oestrogenic COCs.2
Violence, like in an extended premenstrual syndrome, was noticeable
in such combinations when there were also marked arteriolar and venous
changes in the endometrium. Venous effects were commonest with more
oestrogenic COCs and increased with longer use. All cases of VTE or
thrombophlebitis were reported after the first 12 months of use.3
Lidegaard has recorded similar increased risks of VTE with both short and
longer use of the more oestrogenic progestogens.
Common reasons for early discontinuations or for women reqesting
different hormonal contraceptives are irregular bleeding, migraine
headaches, rapid weight gain or depressive mood changes.4
Violence was the commonest cause of death in current users the RCGP
pill study was from 1968 to 1972 when most current pill taking happened.
(The average use was recorded as only 44 months by 2010). There were 13
deaths from suicides accidents and violence compared with 6 in controls
and 7 deaths from arterial vascular diseases but only 2 deaths from venous
thromboembolism.5 By 2010 there were 156 deaths from violence in ever
takers of the pill compared with 51 in never pill users - adjusted
relative risk 1.92 (1.22 to 3.01).6
Such extended studies of pill ever users are usually confounded by
older age use of fertility drugs or HRT which prevent realistic and
accurate risk estimates. However in the RCGP study by 2010 the commonest
causes of death in ever users were cancers (1312), especially breast
cancer, and cardiovascular (353) and cerebrovascular (227) diseases.
Venous thromboembolism was not listed separately but presumably as a part
of other circulatory disease (126).
Thromboembolism is dramatic and life threatening but changing
progestogens types or using long-acting progestogen-only contraceptives
can also increase a range of serious health risks. Further details of
biochemical changes are freely available at www.harmfromhormones.co.uk
Very important increases in cancers and arterial and mental side
effects have been underestimated or ignored for 50 years to the detriment
of family health, presumably due to the overwhelming need for
contraception.
1 Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard
E. Risk of venous thromboembolism from use of oral contraceptives
containing different progestogens and oestrogen doses: Danish cohort study
2001-9. BMJ 2011; (http://www.bmj.com/content343/bmj.d6423/suppl/DC1).
2 Grant EC, Pryse-Davies J. Effect of oral contraceptives on
depressive mood changes and on endometrial monoamine oxidase and
phosphatases. BMJ 1968 Sep 28;3(5621):777-80.
3 Grant ECG. Venous effects of oral contraceptives. BMJ 1969;4:73-7.
4 Anon Editorial. Changing oral contraceptives. BMJ 1969;4:789-791.
5 Royal College of General Practitioners. Oral contraceptives and
health. Pitman Medical, 1974
6 Hannaford PC, Iversen L, Macfarlane TV, Elliott AM, Angus V, Lee
AJ. Mortality among contraceptive pill users: cohort evidence from Royal
College of General Practitioners' oral contraception study. BMJ
2010;340:c927.
Competing interests: No competing interests
Thanks to Professor emeritus Samuel Shapiro (SS) for his
considerations on our new study on venous thrombosis in users of different
types of oral contraceptives (1).
Already after we published our first report in BMJ in 2009 (2) SS and
Dinger published a six-page long critique of our (and a Dutch) study, in
which they postulated that we underestimated the risk of venous thrombosis
for levonorgestrel resulting in a systematic overestimation of the rate
ratio between other oral contraceptives and levonorgestrel due to left
censoring bias, and concluded that "the increased risk of VTE in OC users
is .... independent of the progestogen used" (3).
In our new study we eliminated left censoring bias by putting current
users of OC at the beginning of the study period 2001-2009 into the
correct length of use category. This was possible due to information about
use of hormonal contraception back to 1995. Thus women who had used OC
with levonorgestrel for more than four years before and up to 2001 were
placed in the category use of >4 years use already from January 1,
2001. Use even longer back in time would not change the category as by
January 2001. The secular change in diagnostic precision was accounted for
by adjusting for calendar year.
We also restricted the new study to confirmed events of venous
thromboembolism by assessing anticoagulation therapy after the diagnosis.
Finally we added four new study years (2006-2009) in our new study.
The results of the new study rejected the accusations brought in
print by SS and Dinger. First the rate ratio estimates
of venous thrombosis between users of OC with drospirenone versus OC with
levonorgestrel (LNG) increased from 1.6 (1.3-2.1) to 2.1 (1.6-2.8) in the
new study, proving that we underestimated the risk of OC with drospirenone
versus OC with LNG in our primary study and not the opposite.
Next, we made indeed sub-analyses restricted to starters and new
users of different types of OC. New users were defined as women who had a
pause of at least 12 weeks without use of hormonal contraception prior to
current use. These results were reported with the new publication as
appendix 4. Restricted to starters in the period 2001-2009 brought a rate
ratio between users of OC with drospirenone versus OC with LNG of 2.69
(1.76-4.10) and restricted to starters + new users of 2.05 (1.56-2.70). In
other words, results quite consistent with the overall results.
We also elaborated these results to EMA in an in some respect more
restricted analysis (e.g. we did not report results on combined pills with
20 microgram ethinylestradiol in the EMA report). In other respects the EMA
report included dozens of additional tables, requested by the Steering
Committee. Having seen the results of these successively elaborated
supplementary tables, SS requested a further analysis, in which he wanted
the starters and new users during the study period 2001-2009 not to have
used any type of hormonal contraception before 2001. The other members of
the steering committee did not find this analysis relevant considering the
fact that such a restriction eliminated 93.6% of the exposure time and
93.4% of the events in users of OC with LNG in the study period 2001-2009.
SS insisted, however, in getting this table, and got it.
As expected the rate ratio estimates between OC with drospirenone
versus OC with LNG were unstable and ranged from 0.5 to 1.9 for different
duration categories, with an overall estimate of 1.0. We concluded this
was due to chance, due to the low number of events (n=11), an
interpretation that was confirmed by the rate ratio of 2.05 by inclusion
of all starters and new users. After a pause of at least 12 weeks nobody
believes that previous use influences the risk of thrombosis with new use.
For these reasons we did not include this unreliable and according to
different duration groups inconsistent results in our BMJ paper, it simply
did not add any further relevant information in addition to what already
is included in Appendix 4.
Surprisingly, SS instead of acknowledging the improvements in our new
analysis founders a single unreliable estimate based on a tiny fraction of
our exposure time and thrombotic events.
After our on-line publication, the FDA published last week the results of
an American historical cohort study including 898,251 women years and 625
venous thromboses in users of hormonal contraception. They found a
significantly higher risk of hospitalized venous thrombosis in users of OC
with drospirenone as compared with users of OC with LNG; Rate ratio 1.49
(1.11-2.01) and if restricted to only new users of 1.72 (1.14-2.59),
results further confirming a differential risk of venous thrombosis with
different progestogen types (4).
On the authors' behalf, Ojvind Lidegaard
1. Lidegaard O, Nielsen LH, Skovlund CV, Skjeldestad FE, Lokkegaard
E. Risk of venous thromboembolism from use of oral contraceptives
containing different progestogens and oestrogen doses: Danish cohort study
2001-9. BMJ 2011; 343: d6423.
2. Lidegaard O, Lokkegaard E, Svendsen Al, Agger C. Hormonal
contraception and risk of venous thromboembolism: national follow-up
study. BMJ 2009; 339: b2890.
3. 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.
4. FDA Office of Surveillance and Epidemiology. Combined hormonal
contraceptives (CHCs) and the risk of cardiovascular disease endpoints.
http://www.fda.gov/downloads/Drugs/DrugSafety/UCM277384.pdf
Competing interests: All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare that: Bayer Schering Pharma is thanked for covering the expenses of the analysis. All funding was given to Rigshospitalet, and the primary investigator received no salary for his work with this study, the EMA report or this manuscript. OL has within the last three years received honorariums for speeches on pharmacoepidemiological issues, including fees from Bayer Pharma Denmark and Novo Nordisk, and will be an expert witness for plaintiffs in a legal US case in 2011-2; FES received compensation for his work in the steering committee of the European Medicines Agency report.
Re: Risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses: Danish cohort study, 2001-9
In this study on risk of venous thromboembolism from use of oral contraceptives containing different progestogens and oestrogen doses (a Danish cohort study, 2001-9), the length of use of oral contraceptives was adjusted during analysis. Obesity, increasing age, prolonged immobilization (long distance travel or bed rest) are other risk factors (confounders) for venous thromboembolism and should also have been adjusted during analysis in this population based study. This study is done in women aged 15-49 years whereas, age above 40 years is a risk factor for VTE.
Reference:
1. BMJ 2011;343:d6423
Competing interests: No competing interests