{"paper_id":"cba255ef-8841-466f-a0c3-aae2e71ddb28","body_text":"ISSN: 2398-2799\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\nCommentary \nFrontiers in Women’s Health\n Volume 5: 1-6\nCOCs containing dienogest and 30 µg ethinylestradiol \nmay carry a higher VTE risk compared to corresponding \npreparations with levonorgestrel: A meta-analysis of four \nlarge cohort studies\nJürgen Dinger*\nZEG-Berlin Center for Epidemiology and Health Research, Invalidenstrasse 115, 10115 Berlin, Germany \nAbstract\nBackground: The European Medicines Agency requested a meta-analysis of four large, multi-national cohort studies on hormonal contraceptives to clarify whether \ndienogest/ethinylestradiol-containing combined oral contraceptives (DNG/EE) carry a different risk of venous thromboembolic events (VTE) compared to \nlevonorgestrel/ethinylestradiol-containing preparations (LNG/EE). The primary objective of the meta-analysis was to assess VTE risk in a study population that is \nrepresentative for the actual users of the individual preparations.\nMethods: All four studies were prospective, observational cohort studies. Cohorts consisted of new users of hormonal contraceptives: starters, switchers and restarters. \nStudy participants were followed up for up to 10 years. The analysis was restricted to preparations containing 30 µg of ethinylestradiol. Primary risk measure: VTE \nhazard ratio (HR) in the European study population for DNG/EE versus LNG/EE adjusted for age, BMI, duration of current use, family history of VTE and data source.\nResults: The analysis set included data from 228,122 users of hormonal contraceptives. The European study participants had used DNG/EE and LNG/EE (WY) \nfor 38,708 and 45,359 woman years, respectively. The meta-analysis includes 102 VTEs: DNG/EE, 56 cases and 14.5 VTE/10,000 WY; LNG/EE, 46 cases and 10.1 \nVTE/10,000 WY. The primary analysis showed an adjusted HR for DNG/EE versus LNG/EE of 1.6 (95% confidence interval 1.1-2.3). Four alternative analyses \nshowed similar results although only one of these analyses reached statistical significance.\nConclusion: DNG/EE is probably associated with a slightly higher risk of VTE compared to LNG/EE. However, some uncertainty regarding the validity of this \nresult remains.\n*Correspondence to: Jürgen Dinger, ZEG-Berlin Center for Epidemiology and \nHealth Research, Invalidenstrasse 115, 10115 Berlin, Germany, Tel: +49 30 \n9451010; E-mail: j.dinger@zeg-berlin.de\nKey words: meta-analysis, combined oral contraceptives, dienogest, levonorgestrel, \nvenous thromboembolism\nReceived: January 02, 2020; Accepted: January 17, 2020; Published: January 20, \n2020\nAbbreviations: BMI: Body Mass Index; CI: Confidence Interval; \nCOC: Combined Oral Contraceptive; DNG: Dienogest; DNG/\nEE: Combined Oral Contraceptive containing Dienogest and 30 \nµg of Ethinylestradiol; EE: Ethinylestradiol; HR: Hazard Ratio; \nLNG: Levonorgestrel; LNG/EE: Combined Oral Contraceptive \ncontaining Levonorgestrel and 30 μg of Ethinylestradiol; VTE: Venous \nThromboembolic Event; WY: Woman Years\nIntroduction\nThe safety of combined oral contraceptives (COCs) has improved \nover the years with the reduction in doses of estrogen and progestogen. \nHowever, concerns about COC safety have remained, peaking in \nthe mid-1990s and early 2010s with discussion on whether COCs \ncontaining so-called “third” and “fourth generation” progestogens \n(desogestrel/gestodene and drospirenone, respectively) have a higher \nrisk of cardiovascular side effects - especially venous thromboembolc \nevents (VTE) - than older formulations [1-4].\nAlthough the specific combination of 2 mg of dienogest (DNG) and \n30 μg of ethinylestradiol (EE) has a substantial market share in Europe, \npublished data on VTE risk are limited. It is currently scientifically \nunclear whether this specific combination (DNG/EE) is associated \nwith a different risk of VTE compared to levonorgestrel (LNG)/EE-\ncontaining COCs which are often used as the reference standard for \nthe VTE risk associated with combined hormonal contraceptives. \nThe Berlin Center for Epidemiology and Health Research conducted \nseveral large prospective cohort studies on the risk of VTE associated \nwith the use of hormonal contraceptives. Four of these studies included \na substantial number of women using DNG/EE or LNG/EE-containing \nCOCs. The European Medicines Agency requested a meta-analysis \nof these four prospective cohort studies to clarify whether DNG/\nEE carries a different VTE risk compared to LNG/EE. Therefore, the \ndata on DNG/EE and LNG/EE from the following four prospective \ncohort studies were combined: i) “Long-term Active Surveillance \nStudy for Oral Contraceptives” (LASS) [5]; ii) “International Active \nSurveillance Study of Women Taking Oral Contraceptives” (INAS-OC) \n[6]; iii) “Transatlantic Active Surveillance on Cardiovascular Safety of \nNuvaring” (TASC) [7]; and iv) “International Active Surveillance Study - \nSafety of Contraceptives: Role of Estrogens” (INAS-SCORE) [8]. \n\nDinger J (2020) COCs containing dienogest and 30 µg ethinylestradiol may carry a higher VTE risk compared to corresponding preparations with levonorgestrel: \nA meta-analysis of four large cohort studies\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\n Volume 5: 2-6\nMaterials and methods\nDetails of design and methodology of the four cohort studies are \ndescribed elsewhere [1-4]. All studies were conducted in accordance \nwith the ethical principles of the Declaration of Helsinki. The primary \nethical approvals in Europe for all four studies were provided by the \nethical committee of the physicians' association in Berlin, Germany. \nThe study outlines were published at ClinicalTrials.gov prior to the \nrecruitment phase of the individual studies. Each study was governed \nby an independent Safety Monitoring and Advisory Council to ensure \nits scientific independence.\nThe primary objective of the meta-analysis was to assess the risk \nof VTE associated with the short and long-term use of DNG/EE and \nLNG/EE in a study population that is representative for the actual \nusers of the individual preparations. The secondary objective was to \ncharacterize the baseline risk of users of the two formulations (lifetime \nhistory of co-morbidity, prognostic factors for VTE, co-medication, \nsocio-demographic and life-style data).\nThe LASS study was conducted in Europe only; the other three \nstudies were transatlantic studies that included a large proportion of \nEuropean women. Study participants were recruited via large networks \nof hormonal contraceptive prescribing health care professionals in \nthe United States and a total of 12 European countries. DNG/EE-\ncontaining COCs are not available in the United States. Thus, only \nEuropean data were used for the meta-analysis presented here. The \nmethodology used in the four studies was almost identical. All studies \nwere large, prospective, observational, active surveillance studies \nthat focused on the risk of VTE associated with the use of hormonal \ncontraceptives. Cohorts consisted of new users of COCs: starters, \nswitchers and restarters. A ‘non-interference’ approach was used to \nprovide standardized, comprehensive, reliable information under \nroutine medical conditions: i.e., all patients who were new users of an \nOC were eligible for enrolment if they gave their informed consent, \nand the physicians' prescribing behaviour was not influenced by \nquotas for specific OCs. Study participants were followed up for up \nto 10 years. All outcomes of interest were captured by direct contacts \nbetween the investigator team and the study participants. Inclusion \nand exclusion criteria, methods of patient recruitment, follow-up and \ndata documentation (including prognostic factors for VTE) were almost \nidentical. The data from these studies could therefore be combined for the \nplanned meta-analysis without any substantial methodological problems. \nOverall, the analysis set included data from 228,122 users of \nhormonal contraceptives with a follow-up of 736,793 woman years \n(WY) of observation. The European study participants had used DNG/\nEE and LNG/EE (preparations with 30 µg EE only) for 38.708 WY and \n45.359 WY , respectively. The proportions of starters, switchers and \nrestarters in the DNG/EE user group were 21%, 31% and 48%. The \ncorresponding values for LNG/EE were 26%, 27% and 47%.\nA low “loss to follow-up rate” was essential for the validity of all \nfour studies. In order to minimize loss to follow-up the same multi-\nfaceted, four-level follow-up process was established in all studies. \nLevel1 activities included mailing the follow-up questionnaire and in \ncase of no response two reminder letters. If level1 activities did not \nlead to a response, multiple attempts were made to contact the woman, \nher friends, relatives, and gynaecologist/primary care physician \nby phone. In parallel to these level2 activities, searches in national \nand international telephone and address directories as well as social \nnetworks were started (level3 activities). If this was not successful, an \nofficial address search via the respective governmental administration \nwas conducted (in some countries centralized, in others decentralized \nat community level). This level4 activity usually yielded information on \na new address (or information that the respondent had moved abroad \nor died). Overall, the loss to follow-up rate was 3.3% or lower in each \nof the four studies. \nIn all four studies the same procedures were used for the validation \nof reported adverse events. All serious adverse events and particularly \nVTEs were validated via the diagnosing and/or treating physician. All \nVTEs were checked at the end of the studies by three independent \nmedical experts specializing in radiology/nuclear medicine, cardiology, \nand internal medicine/phlebology. For the blinded adjudication \nprocess, the brand names, dose, regimen and composition of the \nOC(s) used by the reporting woman were rendered anonymous. The \nadjudicators conducted their reviews independently of each other \nand without knowing the judgement of the other adjudicators or the \ninvestigators.\nThe primary risk measure was the VTE hazard ratio (HR) in the \nEuropean study population for DNG/EE versus LNG/EE. In general, it \nis very difficult to interpret a relative risk of two or less in observational \nresearch [9,10]. Therefore, the author focused the analysis on excluding a \ntwofold risk. Accordingly, the null hypothesis prior to the meta-analysis \nwas: HRVTE>2 (i.e., the adjusted VTE hazard ratio for DNG/EE vs. LNG/\nEE is higher than 2). The alternative hypothesis was: HRVTE ≤ 2. However, \nregulatory authorities often request the exclusion of a 1.5-fold risk - \neven in non-experimental studies. The a priori power of the pooled \nanalysis to exclude a twofold and 1.5-fold VTE risk for DNG/EE \ncompared to LNG/EE was about 94% and 58%, respectively [11,12]. The \ncorresponding values for the individual studies are shown in table 1. \nThese power calculations confirm that the meta-analysis is sufficiently \npowered to show non-inferiority of DNG/EE compared to LNG/EE if a \nnon-inferiority hazard ratio of 2 is accepted. If a non-inferiority hazard \nratio of 1.5 is requested, the power of the pooled analysis is limited. \nInferential statistics were based on Cox proportional hazard \nmodels. Crude and adjusted HRs between the two cohorts of interest \n- DNG/EE and LNG/EE - were calculated. Four prognostic factors for \nVTE - age (continuous variable), BMI (continuous variable), current \nduration of use (continuous variable), and family history of VTE \n(binary variable) -  were included as covariates in the Cox models. \nThis selection was based on those factors that had consistently shown \na substantial impact on VTE risk estimates in the regular statistical \nanalyses of the individual studies. Furthermore, the data source (i.e. \nthe study the women were participating in: LASS, INAS-OC, TASC or \nINAS-SCORE) was included in the Cox models for the pooled analysis. \nTo assess the robustness and validity of the primary statistical \nmodel, four alternative Cox models were used for sensitivity analyses: i) \noverall information on 20 potential confounders for VTE was available \nin all 4 studies (e.g., concomitant medication, smoking, acne), and the \n20 potential confounders were included as covariates in a saturated Cox \nmodel; ii) starting with the 20 potential confounders for VTE a backward \nStudy Power to exclude \na twofold risk\nPower to exclude \na 1.5-fold risk\nLASS 84% 46%\nINAS-OC 31% 16%\nTASC 14% 9%\nINAS-SCORE 31% 16%\nPooled analysis 94% 58%\nTable 1. Power calculations for the individual studies and the pooled analysis\n\nDinger J (2020) COCs containing dienogest and 30 µg ethinylestradiol may carry a higher VTE risk compared to corresponding preparations with levonorgestrel: \nA meta-analysis of four large cohort studies\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\n Volume 5: 3-6\nstepwise procedure was used to reduce the number of covariates; all \ncovariates that did not change the point estimate of the hazard ratio \nby more than 10% or that had no statistically significant impact (p > \n0.05) were removed from the model in a stepwise procedure; iii) a Cox \nmodel with acne included in the primary model - in addition to the \nprognostic factors included in the primary model: this model was used \nto investigate the impact of the baseline differences between the two \ntreatment groups; iv) a Cox model selected by the Akaike information \ncriterion (AIC) as an estimator of the relative quality of different Cox \nmodels [13]: i.e. during a stepwise backward procedure (see model ii) \nthe model with the lowest AIC value was chosen.\nRegulatory authorities requested that the statistical analyses be \nconducted based on the \"as treated\" (AT) population as well as the \n\"intention to treat\" (ITT) population. For the AT analyses, data on \noutcomes of interest were assigned to the product actually used by \nthe respective study participant at the time of the event. For the ITT \nanalyses, all data from individual participants were assigned to the \ntreatment they used at study entry, regardless of any switching (or \nstopping) or of any different (or no) product being used at the time \nof the event. For studies on efficacy, the “intention to treat” (ITT) \napproach is often preferred because it is conservative with respect to the \nsuperiority of a new treatment. For an analysis of drug safety, however, \nthe ITT approach dilutes differences between treatments. Therefore, \nthe “as treated” analysis was designated as the primary analysis for \nassessing the data.\nThe primary analysis was based on the pooled cohorts. As requested \nby regulatory authorities HRs were calculated per study as well as per \nuser status (starter/switcher/restarter) for exploratory reasons. It should \nbe noted that the total number of statistical tests is 25: 5 (4 studies \nplus pooled analysis) times 4 (complete cohorts, starters, switchers, \nrestarters) plus 5 ITT analyses (complete cohorts in the pooled analysis \nand four studies). Accordingly, the likelihood of incorrectly rejecting a \nnull hypothesis was substantial. Therefore, 24 out of the 25 tests were \ncalculated for exploratory reasons only. Furthermore, a sub-analysis \nwas only conducted if a minimum of three VTE was available for each \nof the two comparison groups.\nResults\nOverall, baseline characteristics were similar for the DNG/EE and \nLNG/EE users (Table 2). The similarities include age, weight, height, \nBMI, cardiovascular risk factors (e.g. family history of VTE), medical \nhistory and concomitant medication. A substantial difference was \nfound for the prevalence of acne (30.0% and 9.1% for DNG/EE and \nLNG/EE, respectively). Slight to moderate differences were also found: \ni) a higher proportion of LNG/EE users had delivered a child or had \nbeen pregnant prior to study entry; ii) a higher proportion of DNG/\nEE users had switched OCs prior to study entry; and iii) DNG/EE users \nhad a higher educational level. \nAcne is associated with polycystic ovary syndrome (PCOS) which \nis associated with an approximately twofold risk of VTE [14-16]. \nTherefore, it was important that the sensitivity analyses included a Cox \nmodel that adjusted for acne (alternative model iii). A more detailed \nanalysis of the age profile showed that the similar mean age of the two \nexposure groups is slightly misleading as the LNG/EE group included \nboth, high proportions of teenagers and women age 30. Given the \nexponential increase of the VTE risk with age the LNG/EE users already \nhad a slightly higher VTE risk prior to their enrolment. Therefore, it \nwas to be expected that age adjusted VTE hazard ratios for DNG/EE \nversus LNG/EE would be slightly higher compared to crude hazard \nratios. Given the opposing effects of adjustment for acne and age, it was \nnot expected that adjustment for the differences discussed above would \nresult in a substantial change of the unadjusted VTE hazard ratio.\nThe meta-analysis is based on 102 VTEs: 56 and 46 VTEs \noccurred in the DNG/EE and LNG/EE exposure groups, respectively. \nOnly preparations with the same EE-content (30 µg) were used for \nthis comparison. The VTE incidence rates were higher for DNG/\nEE compared to LNG/EE (14.5 vs. 10.1 VTE/10,000 WY). The \ncorresponding overall incidence rate ratio was 1.4; the 95% confidence \ninterval (CI) included unity: 1.0 - 2.1. The results for the individual \nstudies are shown in figure 1. Breaking down VTE into deep venous \nthrombosis (DVT) and pulmonary embolism (PE) showed similar \nresults: DVT, incidence rate ratio 1.4 (95% CI, 0.9 - 2.2); PE, incidence \nrate ratio 1.4 (95% CI 0.6 - 3.1).\nThe primary analysis (pooled dataset for all users of DNG/EE or \nLNG/EE) resulted in an adjusted HR of 1.6 (95% CI, 1.1 - 2.3). The result \nDNG/EE LNG/EE\nNumber of women 9.593 (100.0%) 13.309 (100.0%)\nPatient characteristics\nAge at study entry (years) 25.1 (± 7.98) 25.5 (±8.52)\nAge < 20 2.679 (27.9%) 3.868 (29.1%)\n20 ≤ 30 4.444 (46.3%) 5.690 (42.8%)\n30 ≤ 40 1.811 (18.9%) 2.605 (19.6%)\n40 ≤ 50 634 (6.6%) 1.100 (8.3%)\n50+ 25 (0.3%) 46 (0.3%)\nWeight at study entry (kg) 62.6 (±11.80) 63.5 (±12.43)\nHeight at study entry (cm) 167 (±6.2) 167 (±6.4)\nBMI at study entry 22.4 (±3.95) 22.8 (±4.17)\nGynaecological history\nAge at menarche (years) 12.9 (±1.39) 13.0 (±1.40)\nGravidity (ever) 3.177 (33.1%) 5.317 (40.0%)\nParity (ever) 2.860 (29.8%) 4.838 (36.4%)\nNumber of live births 1.5 (±0.7) 1.6 (±0.8)\nOC history\nOC user type: Starter 1.971 (20.5%) 3.194 (24.0%)\n                       Switcher 3.016 (31.4%) 3.708 (27.9%)\n                       Restarter 4.606 (48.0%) 6.407 (48.1%)\nEver switched OC 4.883 (50.9%) 5.284 (39.7%)\nDuration of OC-use at study \nentry (years) 6.4 (± 6.08) 6.9 (± 6.79)\nAcne\nCurrent acne 2.874 (30.0%) 1.207 (9.1%)\nCardiovascular risk factors\nHigh blood pressure 287 (3.0%) 420 (3.2%)\nHigh cholesterol 204 (2.1%) 206 (1.5%)\nFamily history of ATE 131 (1.4%) 182 (1.4%)\nFamily history of VTE 567 (5.9%) 737 (5.5%)\nSmoker 3.345 (34.9%) 5.245 (39.4%)\nHeavy smoker (>15 cigarettes) 574 (6.0%) 903 (6.8%)\nMedical history\nDiabetes 42 (0.4%) 10 (0.1%)\nCancer 46 (0.5%) 45 (0.3%)\nAny surgery 3.042 (31.7%) 3.819 (28.7%)\nConcomitant medication\nRegular use of medication 1.397 (14.6%) 1.784 (13.4%)\nEducation\nHigher than university \nentrance level 3.868 (40.3%) 4.691 (35.2%)\nTable 2. Baseline characteristics for the European users of DNG/EE* and LNG/EE*.\n*preparations with 30 µg ethinylestradiol only\n\nDinger J (2020) COCs containing dienogest and 30 µg ethinylestradiol may carry a higher VTE risk compared to corresponding preparations with levonorgestrel: \nA meta-analysis of four large cohort studies\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\n Volume 5: 4-6\nof the ITT analysis was similar: 1.5 (95% CI, 1.0 - 2.4). Also, the results \nfor starters, switchers and restarters showed similar results: starters, 1.6 \n(95% CI, 0.6 - 4.6); switchers, 1.9 (1.0 - 3.4); and restarters 1.3 (95% CI, \n0.7 - 2.5). A comparison of the individual study results shows consistent \nresults (Figure 2). The huge LASS dataset had the strongest impact on \nthe overall results. However, the data are sufficiently consistent across \nstudies to justify combining of all four study databases.\nThe results of the four alternative Cox models are shown in table \n3. The point estimates of the VTE hazard ratio were only slightly lower \ncompared to the estimate from the primary model. The lower limits of \nthe 95% confidence interval were close to one in all cases.. Formal statistical \nsignificance for DNG/EE versus LNG/EE was only reached for the primary \nmodel and alternative model ii (backward stepwise procedure).\nDiscussion and conclusion\nThe incidence rate for VTE was higher for DNG/EE compared \nto LNG/EE and the primary statistical analysis yielded an adjusted \nVTE hazard ratio for the comparison of DNG/EE versus LNG/EE of \n1.6. The corresponding 95% CI did not include unity and suggested \na slightly higher VTE risk for DNG/EE compared to LNG/EE. This \nresult is supported in principle by the results of four alternative analyses \nalthough only one of these analyses reached statistical significance.\nIn non-experimental studies like LASS, INAS-OC, TASC and INAS-\nSCORE the possibility of bias and residual confounding can never be \nentirely eliminated, and the ability to infer causation is correspondingly \nlimited [17]. Valid information on potential sources of confounding, \nand sophisticated statistical and epidemiologic methodology help to \nreduce the impact of bias and residual confounding [18]. However, the \ndifficulty remains unresolved when all that exists is a weak association \n[19,20]. Relative risk estimates that are close to unity may not allow \ndifferentiation between causation, bias and confounding [21,22]. In \ngeneral, it is very difficult to interpret a relative risk of two or less in \nobservational research [9,10]. \nSelection and misclassification bias were probably not a major \nissue in any of the four cohort studies because i) their participants are \nrepresentative for adult COC users [23]; and ii) reliable information \non exposure and duration of OC use was available. Furthermore, the \nlow loss to follow-up rates of 3.3% or less in all studies is noteworthy. \nIn theory, a disproportionately high percentage of VTE could have \noccurred in those patients who were lost to follow-up, because VTEs \ncould be the reason for the break in contact with the investigators. An \nadvantage of the design of the four included studies, however, is that the \ninvestigator teams had direct contact with the participants; contact was \nnot lost if the women changed their gynaecologists (e.g. due to change \nof residence or dissatisfaction with treatment).\nIn contrast, it was impossible to exclude diagnostic bias. Clinical \nsymptoms of VTE cover the spectrum from a complete absence \nor unspecific, slight symptoms to dramatic, acute, life-threatening \nsymptoms [24-26]. A high awareness of potential cardiovascular risks \nof combined oral contraceptive use might have led to more diagnostic \nprocedures and therefore to more detected VTEs. However, this is a \ngeneral consideration and there is no evidence that diagnostic bias \ninfluenced the results of this meta-analysis. Another issue is the fact \nthat information on specific gene mutations was only available for VTE \ncases but not for the vast majority of study participants. This limitation \nwas mitigated by information on family history of VTE which has a \nhigher predictive value for VTE compared to gene mutations [27].\nThe studies included in this meta-analysis combine several \nmethodological strengths that are substantial for the validity of the \nresults such as: i) prospective, comparative cohort design; ii) availability \nof important confounder information (e.g. BMI and family history \nof VTE); iii) validation of outcomes of interest and exposure for the \nrelevant cases; iv) comprehensive follow-up procedure and very low \nloss to follow-up to minimize underreporting; v) independent, blinded \nadjudication of VTE cases; vi) study population representative for \noral contraceptive users under routine clinical conditions; vii) quite \ndifferent statistical approaches resulting in similar risk estimates and \n95% CIs support the validity and robustness of the primary statistical \nmodel; and viii) supervision by an independent Safety Monitoring and \nAdvisory Council as well as scientific independence from the study \nfunder.\nFigure 1. VTE incidence rates and 95% CI for the individual studies and the pooled dataset: \nDNG/EE* versus LNG/EE*\nFigure 2. VTE hazard ratios and 95% CI for the individual studies and the pooled dataset: \nDNG/EE versus LNG/EE\nCox Model Hazard Ratio 95% Confidence Interval\nPrimary model 1.57 1.07 - 2.30\nAlternative models\ni) Saturated model 1.47 0.98 - 2.20\nii) Backward stepwise 1.54 1.05 - 2.26 \niii) Primary with acne included 1.44 0.97 - 2.13\niv) AIC selected model* 1.44 0.97 - 2.13\nTable 3. VTE hazard ratios: Comparisons between the primary Cox model and 4 alternative \nCox models\n*identical with alternative model iii because the AIC selection process ended in a model  \nthat was identical with model iii.\n\nDinger J (2020) COCs containing dienogest and 30 µg ethinylestradiol may carry a higher VTE risk compared to corresponding preparations with levonorgestrel: \nA meta-analysis of four large cohort studies\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\n Volume 5: 5-6\nThe validity of synthetic meta-analysis as applied to non-\nexperimental (observational) studies has been challenged [28-30]. \nReasons why this approach is questioned include issues such as: \nvariation in quality among studies; variation in methodology including \nvariable definitions of exposure and outcome; and variable precision in \nthe recording, measurement, and control of confounding factors. These \nissues do not apply to this meta-analysis. However, the possibility that \na series of studies may tend to share the same biases and sources of \nconfounding cannot be rejected for the studies included in this meta-\nanalysis. The latter consideration is of special relevance when it comes \nto considering small associations [19].\nIn the author’s judgment, the results of this meta-analysis are valid \nwithin the general limitations of observational research and meta-\nanalysis of observational studies. The impact of residual confounding \nand bias could be limited to an extent that would allow causal \ninterpretation of statistically significant results with hazard ratios that \nare equal to or higher than 2. Statistically significant results with hazard \nratios of 1.5 to 2 can be cautiously interpreted as risk. However, some \nuncertainty remains in these cases. \nThe thresholds of 1.5 and 2 also reflect the limitations of the statistical \npower of this meta-analysis. The analysis was sufficiently powered to \ndetect a twofold risk of VTE but had limited power to detect smaller \nrisks. Nevertheless, the primary statistical analysis yielded a statistically \nsignificant increased risk of VTE for DNG/EE compared to LNG/EE. \nThe adjusted hazard ratio of 1.6 is clearly below 2, and therefore some \nuncertainty regarding a causal interpretation of this result remains. \nThe fact that the alternative analysis that adjusted for acne - the only \nbaseline characteristic with a substantial difference between the \nexposure groups - showed no statistically significant difference adds to \nthis uncertainty. It is also conceivable that adjustment for acne did not \ncompletely adjust for the likely differences in the prevalence of PCOS, \nand that some residual confounding remained unadjusted. However, \nit should be noted that primary and alternative analyses showed \nquantitatively similar risk estimates and the lower limits of the 95% CIs \nwere always close to one. \nGiven the methodological strengths of the individual studies, the \nsimilarity of their study designs, and the quantitative consistency of the \nanalysis results, the investigator considers it likely - but not definitively \nproven - that in 30 µg EE preparations DNG carries a higher risk of \nVTE compared to LNG. \nAuthorship\nThe meta-analysis was planned, conducted and reported by the \nauthor based on requests of the European Medicines Agency. \nAcknowledgements\nThe author’s special thanks are due to Sabine Möhner and Thai Do \nMinh for the data management and statistical analysis of the individual \nstudies, and to Anja Bauerfeind for the statistical analysis of the pooled \ndataset. The author also wants to thank Marlene Schoofs for editorial \nsupport in preparing the manuscript. Furthermore, the author would \nlike to highlight the contributions of numerous colleagues who were \nresponsible for the field work in the four cohort studies. They clarified \ndata inconsistencies and missing data, validated patient-reported \nadverse events with patience, care and tenacity, and their untiring \ncommitment enabled a remarkably low loss to follow-up rate. \nFunding information\nThe costs for the meta-analysis were covered by the manufacturer \nof dienogest containing COCs, Bayer AG, Germany.\nCompeting interests\nThe meta-analysis was requested by the European Medicines \nAgency. The costs of the meta-analysis were covered by Bayer AG. \nBayer AG was not involved in the planning, conduct and reporting of \nthe results.\nReferences\n1. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid \nHormone Contraception (195) Effect of different progestogens in low oestrogen oral \ncontraceptives on venous thromboembolic disease. Lancet 346: 1582-1588.\n2. Spitzer WO, Lewis MA, Heinemann LA, Thorogood M, MacRae KD (1996) Third \ngeneration oral contraceptives and risk of venous thromboembolic disorders: an \ninternational case-control study. Transnational Research Group on Oral Contraceptives \nand the Health of Young Women. BMJ 312: 83-88. [Crossref]\n3. Lidegaard Ø, Løkkegaard E, Svendsen AL, Agger C (2009) Hormonal contraception and \nrisk of venous thromboembolism: national follow-up study. BMJ 339: b2890. [Crossref]\n4. Jick SS, Hernandez RK (2011) Risk of non-fatal venous thromboembolism in women \nusing oral contraceptives containing drospirenone compared with women using oral \ncontraceptives containing levonorgestrel: case-control study using United States claims \ndata. BMJ 340: d2151. [Crossref]\n5. Dinger J, Möhner S, Heinemann K (2016) Cardiovascular risks associated with the use \nof drospirenone-containing combined oral contraceptives. Contraception 93: 378-385. \n[Crossref]\n6. Dinger J, Bardenheuer K, Heinemann K (2014) Cardiovascular and general safety \nof a 24-day regimen of drospirenone-containing combined oral contraceptives: final \nresults from the International Active Surveillance Study of Women Taking Oral \nContraceptives. Contraception 89: 253-263. [Crossref]\n7. Dinger J, Möhner S, Heinemann K (2013) Cardiovascular Risk Associated with the Use \nof an Etonogestrel-Containing Vaginal Ring. Obstet Gynecol 122: 800-808. [Crossref]\n8. Dinger J, Do Minh T, Heinemann K (2016) Impact of Estrogen Type on Cardiovascular \nSafety of Combined Oral Contraceptives. Contraception 94: 328-339. [Crossref]\n9. Taubes G (1995) Epidemiology faces its limits. Science 269: 164-169. [Crossref]\n10. Hill AB (1965) The environment and disease: association or causation? Proc R Soc Med \n58: 295-300. [Crossref]\n11. Chow SC, Shao J, Wang H (2008) Sample Size Calculations in Clinical Research. (2nd \nEdtn), Chapman & Hall/CRC. pp: 174-78. \n12. Schoenfeld DA (1983) Sample Size Formula for the Proportional-Hazards Regression \nModel. Biometrics 39: 499-503. [Crossref]\n13. Akaike H (1974) A new look at the statistical model identification. IEEE Transactions \non Automatic Control 19: 716-723. \n14. Bird ST, Hartzema AG, Brophy JM, Etminan M, Delaney JAC (2013) Risk of venous \nthromboembolism in women with polycystic ovary syndrome: a population-based \nmatched cohort analysis. CMAJ 185: 115-120. [Crossref]\n15. Okoroh EM, Hooper WC, Atrash HK, Yusuf HR, Boulet SL (2012) Is polycystic ovary \nsyndrome another risk factor for venous thromboembolism) United States, 2003-2008. \nAm J Oststet Gynecol 207: 377. e1-377.e8. [Crossref]\n16. Seaman HE, de Vries CS, Farmer RD (2003) Differences in the use of combined oral \ncontraceptives amongst women with acne and without acne. Hum Reprod 18: 515-521. \n[Crossref]\n17. Susser M (1991) What is a cause and how do we know one? A grammar for pragmatic \nepidemiology. Am J Epidemiol 133: 635-648. [Crossref]\n18. Rothman KJ, Poole C (1988) A strengthening programme for weak associations. Int J \nEpidemiol 17: 955-959. [Crossref]\n19. Shapiro S (2000) Bias in the evaluation of low-magnitude associations: an empirical \nperspective. Am J Epidemiol 151: 939-945. [Crossref]\n20. Khoury MJ. James LM. Flanders WD. Erickson JD (1992) Interpretation of recurring \nweak associations obtained from epidemiologic studies of suspected human teratogens. \nTeratology 46: 69-77. [Crossref]\n\nDinger J (2020) COCs containing dienogest and 30 µg ethinylestradiol may carry a higher VTE risk compared to corresponding preparations with levonorgestrel: \nA meta-analysis of four large cohort studies\nFront Womens Healt, 2020         doi: 10.15761/FWH.1000175\n Volume 5: 6-6\nCopyright: ©2020 Dinger J. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, \ndistribution, and reproduction in any medium, provided the original author and source are credited.\n21. Shapiro S (2008) Causation, bias and confounding: a hitchhiker’s guide to the \nepidemiological galaxy. Part 2. Principles of causality in epidemiological research: \nconfounding, effect modification and strength of association. J Fam Plann Reprod \nHealth Care 34: 185-190. [Crossref]\n22. Shapiro S (2008) Causation, bias and confounding: a hitchhiker’s guide to the \nepidemiological galaxy. Part 3: principles of causality in epidemiological research: \nstatistical stability, dose- and duration response effects, internal and external \nconsistency, analogy and biological plausibility. J Fam Plann Reprod Health Care 34: \n261-264. [Crossref]\n23. Dinger JC, Heinemann AJ, Kuehl-Habich D (2007) The safety of a drospirenone-\ncontaining oral contraceptive: final results from the European Active Surveillance study \non oral contraceptives based on 142,475 woman-years of observation. Contraception  \n75: 344-354. [Crossref]\n24. Scurr JH, Machin SJ, Bailey-King S, Mackie IJ, McDonald S, et al. (2001) Frequency \nand prevention of symptomless deep-vein thrombosis in long-haul flights: a randomised \ntrial. Lancet 357: 1485-1489. [Crossref]\n25. Belcaro C, Geroulakos G, Nicolaides AN, Myers KA, Winford M (2001) Venous \nthromboembolism from air travel: the LONFLIT Study. Angiology  52: 369-374. \n[Crossref]\n26. Schwarz T, Siegert G, Oettler W, Halbritter K, Beyer J, et al. (2003) Venous thrombosis \nfollowing long-haul flights. Arch Intern Med 163: 2759-2764. [Crossref]\n27. Bezemer ID, van der Meer FJ, Eikenboom JC, Rosendaal FR, Doggen CJ (2009) The \nvalue of family history as a risk indicator for venous thrombosis. Arch Intern Med 169: \n610-615. [Crossref]\n28. Greenland S (1994) Can meta-analysis be salvaged? Am J Epidemiol  40: 783-787. \n[Crossref]\n29. Shapiro S (1994) Meta-analysis/Shmeta-analysis. Am J Epidemiol  40: 771-778. \n[Crossref]\n30. Shapiro S (1997) Is meta-analysis a valid approach to the evaluation of small effects in \nobservational studies? J Clin Epidemiol 50: 223-239. [Crossref]","source_license":"CC0","license_restricted":false}