Remember me
REVIEW ARTICLE
Year : 2021 | Volume
: 5
| Issue : 3 | Page : 154-160
Comparison of risks of thromboembolism of drospirenone-containing and nondrospirenone -containing combined oral contraceptive use: A meta-analysis
Wan-Lin Zhang1, Zhe Dong2, Jun-Yan Zhang3, Min-Zhi Lyu4, Wei Zhang5, Jian-Lei Huang1, Xin Yang6
1 Reproductive Medical Center, Department of Obstetrics and Gynecology, Tangdu Hospital, the Airforce Military Medical University, Xi'an 710038, China
2 Department of Obstetrics and Gynecology, Peking University International Hospital, Beijing 102206, China
3 Bothwin Clinical Study Consultant, Redmond, WA 98053, USA
4 Department of Biostatistics, Zhongshan Hospital, Fudan University, Shanghai 200032; Center of Evidence-Based Medicine, Fudan University, Shanghai 200433, China
5 Department of Biostatistics, School of Public Health, Fudan University, Shanghai 200433, China
6 Department of Obstetrics and Gynecology, Peking University People's Hospital, Beijing 100044, China
Correspondence Address:
Xin Yang
Department of Obstetrics and Gynecology, Peking University People's Hospital, 11th Xizhimen South Street, Xicheng District, Beijing 100044
China
Source of Support: None, Conflict of Interest: None
DOI: 10.4103/2096-2924.324822
This study aimed to estimate the risk of venous thromboembolism (VTE), arterial thromboembolism (ATE), and other side effects following the use of drospirenone (DRSP)-containing combined oral contraceptives (COCs). When compared with non-DRSP-containing COCs, DRSP-containing COCs decreased the risk of VTE by 15% in the overall study population, although this was not statistically significant (adjusted hazard ratio/risk ratio [95% confidence interval] 0.85 [0.69, 1.04]). DRSP-containing COCs also showed significant benefits in terms of ATE risk. The body mass index of the subjects significantly decreased by 0.64 kg/m2 after taking the DRSP-containing COCs for 6 months. We concluded that DRSP-containing COCs were safe for use and could be broadly recommended.
Keywords: Arterial Thromboembolism; Combined Oral Contraceptives; Drospirenone; Venous Thromboembolism
Since the introduction of combined oral contraceptives (COCs) in the early 1960s,[1] they have remained a widely used contraceptive method. Approximately 104 million sexually active women aged 15–49 years use COCs worldwide. However, general practitioners must be provided with evidence-based information about the risks and benefits of contraceptives to accurately address patient questions and concerns.
COCs contain various types and doses of estrogen and progestogen. To maximize efficacy and minimize side effects, the pharmaceutical industry has been committed to modifying this class of medicines. COCs have been developed to the fourth generation, represented by drospirenone (DRSP)-containing COCs. Lynestrenol and norethisterone, referred to as the first-generation progestogens, have both estrogen and androgen activities. Second-generation progestogens, such as levonorgestrel (LNG) and norgestrel, and third-generation progestogens, such as desogestrel and gestodene, demonstrate antiestrogenic activity; however, androgen activity is retained in both. In contrast to the previous three generations of progestogens, the most recent progestogen generation has both antiestrogen and antiandrogenic activities.[2] The estrogen contained within these COCs is ethinylestradiol at doses ranging from 20 to 50 μg.[2]
All COCs can effectively prevent unwanted pregnancies.[3] However, these drugs have various side effects, such as breakthrough bleeding, acne, body weight gain, and hirsutism. Venous thromboembolism (VTE) is a rare, but potentially serious side effect of COCs.[4] In addition, the estrogen component and its dose levels within COCs have been found to be associated with an elevated incidence of VTE, and accumulating evidence from international studies indicates that progestogens are also responsible for VTE occurrence.[2],[5] To reduce the side effects of COCs, the estrogen dose has been considerably reduced, and the progestogen compound has been modified in recent decades. Unexpectedly, several studies have shown an elevated risk of VTE with the newest type of DRSP progestogen when compared with previous-generation progestogens, such as LNG.[6],[7],[8],[9],[10] However, no consensus has been reached, since other studies have reported no difference.[11],[12],[13],[14],[15] It is still a hotspot of scientific debate that whether the DRSP-containing COCs is truly related with the risk of VTE, or it was just because of the methodologic bias that made them look like to be associated.
These circumstances raise the need for a systematic review of the available literature to allow a combined assessment of published findings. Regrettably, few studies have specifically focused on the risk of VTE associated with DRSP-containing COCs.[16],[17],[18] One meta-analysis concluded that DRSP/30–40 μg of ethinylestradiol showed a significantly increased VTE risk when compared to LNG with 30–40 μg of ethinylestradiol.[18] Notwithstanding the heterogeneous study design claimed by the authors, we considered four reasons for updating the meta-analysis. First, some referenced primary articles traced the same study cohort;[8],[19] second, one study used placebo as the comparator, which was not suitable for comparing the VTE risk for DRSP-containing COCs with non-DRSP-containing COCs; third, the previous meta-analysis did not compare other side effects of DRSP-containing and non-DRSP-containing COCs, including ATE, change in body mass index (BMI)/body weight/blood pressure, breakthrough bleeding, or method failure; and fourth, retrospective studies have considerable intrinsic bias, and therefore, a meta-analysis involving only prospective studies would be more powerful in assessing the COC-related risks of VTE and ATE.
Our present meta-analysis aimed to estimate the risks of VTE, ATE, and other side effects associated with DRSP-containing COC use compared with non-DRSP-containing COC use. Furthermore, we aimed to provide evidence to optimize medicine prescriptions for reproductive-aged women choosing an oral contraception method.
Materials and MethodsData sources and searches
Our meta-analysis was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement, in which the Medline (OvidSP), EMBASE (OvidSP), and the Cochrane Central Register of Controlled Trials (CENTRAL) databases were primarily searched for randomized controlled trials (RCTs) on the drug Yasmin from January 2000 to January 2020. The keywords were as follows: (YAZ or YASMIN or DROSPIRENONE AND ETHINYLESTRADIOL or DROSPIRENONE AND ETHINYL ESTRADIOL or JASMIEL or ZUMANDIMINE or NIKKI or ZARAH or OCELLA or GIANVI or SYEDA or LORYNA). Only RCTs and prospective observational studies were included to calculate the pooled outcomes. The retrieval strategy for RCTs was formulated according to the Cochrane Manual. A subject-heading search was performed. Conference papers and reference lists of the selected studies were screened. The publication language was restricted to English.
Inclusion and exclusion criteria
Studies were included if they met the following criteria: (1) they compared any below-mentioned outcome between groups of DRSP-containing and non-DRSP-containing COCs and (2) prospective studies. Studies were excluded if they met the following criteria: (1) the placebo was used as the comparator; (2) the data involved in the study were not extractable; and (3) the data were the post hoc analysis of RCTs or re-analysis of the same data.
Study selection and data extraction
Two authors (Jun-Yan Zhang and Min-Zhi Lyu) independently performed the study screening, data extraction, and bias assessment. In the initial screening, studies that were clearly irrelevant (e.g. studies focusing on other conditions and reviews) were removed by reading their titles and abstracts. Full articles of the potentially eligible studies were then screened twice for inclusion by the same two review authors. Studies that appeared to meet the predetermined inclusion criteria (stated above) were selected. Irrelevant papers were excluded according to the selection criteria. Any disagreements were resolved through group discussion with the first author (Wan-Lin Zhang) throughout the process of data selection.
Outcome measures
The primary outcomes of the study were the risk of developing venous thrombosis (VTE) and arterial thromboembolism (ATE). Definitions of VTE or ATE varied significantly among articles; therefore, strict definitions of VTE or ATE were not critical in this study. Idiopathic VTE (a diagnosis of venous thromboembolism in the absence of identifiable clinical risk factors[20]) was not included. With regard to the risks of VTE and ATE, only prospective studies were included in the analysis.
The secondary outcomes were method failure, Pearl index, weight loss (weight at 6-month baseline), change in BMI, change in systolic blood pressure (SBP)/diastolic blood pressure (DBP), and mortality related to COCs. All these data were analyzed in this study.
Assessment of heterogeneity
Statistical heterogeneity among studies was assessed using the Higgins I2 statistic and divided into four grades according to the I2 value (low: I2< 40%; moderate: 40%–60%; substantial: 50%–90%; considerable: 75%–100%).
Quality assessment and publication biases
The quality of the included studies was assessed using the GRADE Pro/GDT system (www.gradepro.org). Explicit judgments were made regarding the study design, risk of bias, inconsistency of results, indirectness of evidence, imprecision, and other considerations for data reporting.
Publication bias, including method of randomization, allocation concealment, blinding, and incomplete data reporting were analyzed following the guidelines of the Cochrane Handbook for Systematic Reviews of Intervention.
Statistical analysis
Adjusted hazard ratios (HRs), risk ratios (RRs), and odds ratios (ORs) were used to assess the risk of VTE and other COC-related complications in populations exposed to DRSP or other progestins. HR, RR, and OR were transformed to their natural logarithms (ln), and the corresponding 95% confidence intervals (95% CIs) were used to calculate standard errors. Continuous data, such as changes in body weight, BMI, and the Pearl index, were measured with an inverse variance. Statistical heterogeneity was explored and quantified using the I2 statistic and Cochran's Q test. In cases with no inconsistencies in the risk estimate (I2 ≤ 50%), the fixed model of Mantel–Haenszel was applied. In cases of statistically significant heterogeneity (I2> 50%), the random-effects DerSimonian Laird model was used to obtain the pooled estimate. Publication bias was assessed only when the number of studies was ≥6, using Egger's regression asymmetry test. Statistical significance was defined as a two-sided P ≤ 0.05. STATA software (Version 13.1, STATA Corporation LP, College Station, TX, USA) was used to calculate the pooled estimates for observational studies. Review Manager 5.3 software (Copenhagen: The Nordic Cochrane Center, The Cochrane Collaboration, 2014) was used to perform statistical analyses of the pooled outcomes of RCTs.
ResultsResults of searches
Eight hundred and fifty articles were found in an electronic search. Cross-references were manually searched, and 10 additional eligible articles were obtained. In total, 860 articles were identified. One hundred and seventy-three duplicate articles were removed. Five hundred and eighty-two articles were excluded after the initial screening of titles and abstracts. After reading the full manuscripts, an additional 85 articles were removed because of the reasons listed in the flowchart [Figure 1], showing the screening procedure following the standards suggested by the PRISMA statement. Finally, 11 articles were included in the meta-analysis.
Study characteristics
These articles included four prospective cohort studies[7],[12],[13],[14] and seven RCTs.[21],[22],[23],[24],[25],[26],[27] Among these, three prospective cohort studies[12],[13],[14] compared the primary outcome of VTE and ATE between groups of DRSP-containing COCs and other progestin-containing OCs. Four of the seven RCTs reported method failure, while one reported a change in body weight, another reported a change in BMI, and one RCT reported a change in both body weight and BMI. None of the RCTs reported the primary outcomes of interest. The study characteristics are summarized in [Supplementary Table 1]. The quality values of the observational studies are summarized in [Supplementary Table 2], and the risks of bias of RCTs are presented in [Supplementary Figure 1].
Primary outcomes of the meta-analysis
VTE
Three prospective studies[12],[13],[14] reported the occurrence of VTE [see funnel plots in [Supplementary Figure 2] for heterogeneity assessment]. Two publications[12],[13] appeared twice in the overall and subgroup analyses because they separately focused on comparisons between DRSP-containing and non-DRSP-containing COCs and a sub-cohort of LNG-containing COCs[12],[13] or among DRSP-containing COCs and other COCs. In overall comparison, DRSP-containing COCs decreased the risk of VTE (adjusted HR/RR [95% CI] 0.85 [0.69, 1.04]) compared with non-DRSP-containing COCs, although the difference did not reach the level of statistical significance [Figure 2]a. The outcomes of subgroup analysis based on different comparators showed that DRSP-containing COCs demonstrated a similar risk for VTE compared with LNG-containing COCs (adjusted HR [95% CI], 1.02 [0.73, 1.42]), while DRSP-containing COCs significantly decreased the risk for VTE when compared with other COCs (adjusted HR [95% CI], 0.76 [0.59, 0.98]) [Figure 2]b.
ATE
Two publications, including four prospective cohort studies, have reported the outcomes of ATE.[12],[13] These studies focused on comparisons between DRSP-containing and non-DRSP-containing COCs and a subcohort of LNG-containing COCs. The results indicated that DRSP-containing COCs produced a significantly overall lower risk for ATE (adjusted HR [95% CI], 0.45 [0.20, 0.71]). Subgroup analysis indicated that there was a significantly lower risk of ATE in the DRSP-containing COC group than in the LNG-containing COCs group (adjusted HR [95% CI], 0.42 [0.22, 0.81]). The risk for ATE was also significantly lower in the DRSP-containing COC group than in the other COC group (adjusted HR [95% CI], 0.48 [0.26, 0.89]) [Figure 3].
Figure 3: Forest plot of the pooled outcomes regarding to the combined oral contraceptive-related risk of arterial thromboembolism. (a) Comparisons of risks of arterial thromboembolism among DRSP-containing and non-DRSP-containing combined oral contraceptives and a subcohort of levonorgestrel-containing combined oral contraceptives. Note: (a) These two comparisons were obtained from the same study (Dinger et al.[12]). (b) These two comparisons were obtained from the same study (Dinger et al.[13]). (1) The comparison was the DRSP cohort with the other combined oral contraceptives (excluding LNG) cohort. (2) The comparison was the DRSP cohort with the LNG cohort. (b) Subgroup comparison of arterial thromboembolism risk between DRSP-containing and non-DRSP-containing combined oral contraceptives based on the type of comparator. Note: (a) These two comparisons were obtained from the same study (Dinger et al.[12]). (b) These two comparisons were obtained from the same study (Dinger et al.[13]). DRSP: Drospirenone; LNG: Levonorgestrel.Secondary outcomes of the meta-analysis
Four RCTs reported method failure.[21],[22],[25],[26] Because not all studies followed the intention-to-treat (ITT) principle to report the data, the outcome was compared using three models: Model 1: the number of total events for the pooled estimate calculation used the number that actually adhered to the medication method; Model 2: the total number for pooled estimate calculation followed the ITT principle, and the event number was the number reported in the articles; and Model 3: the total number for pooled estimate calculation followed the ITT principle, and the missing data were added to the event number reported in the articles. The rates of method failure in the DRSP and non-DRSP groups were 0.55% (11/2,015) and 0.47% (18/3,834), respectively, in Model 1; 0.52% (11/2,125) and 0.45% (18/3,963), respectively, in Model 2; and 5.7% (121/2,125) and 3.7% (147/3,963), respectively, in Model 3. The results of all three models indicated that the method of failure in the population taking DRSP-containing COCs was not different from that in the population taking non-DRSP-containing COCs (RR [95% CI], Model 1: 1.55 [0.74, 3.22], Model 2: 1.55 [0.75, 3.22], Model 3: 1.01 [0.78, 1.31]) [Supplementary Figure 3]a, [Supplementary Figure 3]b, [Supplementary Figure 3]c.
The outcome showed that DRSP-containing COCs significantly decreased BMI by 0.64 kg/m2 after the pills were taken for 6 months, compared to non-DRSP-containing COCs (mean difference [95% CI] −0.64 [−1.24, −0.04]) [Supplementary Figure 3]d. However, the change in body weight between the two groups was not significantly different (mean difference [95% CI] −1.63 [−5.44, 2.19]) [Supplementary Figure 3]e. Other secondary outcomes, such as the Pearl index, SBP, DBP, and COC-related mortality between the groups, were also comparable [Supplementary Figure 3]f, [Supplementary Figure 3]g, [Supplementary Figure 3]h, [Supplementary Figure 3]i.
DiscussionWe performed a systematic review and meta-analysis based on 20 studies, of which five prospective studies in three publications reported the primary outcome of VTE, and four prospective studies in two publications reported the primary outcome of ATE. The results (although not statistically significant) suggested that DRSP-containing COCs decreased the risk of VTE by 15% in the overall study population when compared with non-DRSP-containing COCs. This benefit was statistically significant when compared with other COCs, excluding LNG-containing COCs. The results also indicate that DRSP-containing COCs resulted in a nearly 60% lower risk of ATE than the LNG-containing COCs and more than 50% lower ATE than the other COCs. Another advantage of DRSP-containing COCs was that they significantly decreased the BMI by 0.64 kg/m2 after pills were taken for 6 months, compared to non-DRSP containing COCs. However, body weight loss between the two groups was not significantly different. The method failure, Pearl index, SBP and DBP change, and COC-related mortality between groups of DRSP-containing and non-DRSP-containing COCs were comparable.
Several studies have investigated the side effects of COCs according to the types and doses of progestins and estradiols they contain; some of these results support the findings of this study. For example, the results of a network meta-analysis from a Cochrane review conducted by de Bastos et al.[5] indicate that the COC-containing DRSP and 30 μg ethinylestradiol significantly decreased the risk of VTE by 40% when compared to COC-containing LNG and 30 μg ethinylestradiol but showed similar VTE risk when compared to COC-containing GSD and 30 μg ethinylestradiol (RR [95% CI] 1.0 [0.7–1.3]) or COC-containing DSG and 30 μg ethinylestradiol (RR [95% CI] 1.1 [0.8–1.5]). However, the conclusions remain controversial. A meta-analysis by Bateson et al. indicated a comparable risk of VTE between DRSP-containing COCs and other progestin-containing COCs in prospective cohort studies (aOR [95% CI] 0.95 [0.74, 1.21]).[18] Oedingen et al.[2] conducted a meta-analysis based largely on retrospective studies comparing the risk of VTE and considering both progestogen type and estrogen dose. The results showed that, compared to LNG with 30–40 μg ethinylestradiol, DRSP/30–40 μg ethinylestradiol showed a significantly increased VTE risk (adjusted RR [95% CI] 1.40 [1.26–1.56]).
Within the search scope, we found only one meta-analysis that specifically focused on the risk of VTE associated with DRSP-containing COCs.[18] This study included 15 studies that reported on DRSP-containing COCs related to VTE. To compare the study selection of this meta-analysis with that of Bateson et al., full texts of the 15 articles were screened, including the meta-analysis by Bateson et al.[18] Two articles,[12],[14] including three prospective studies that reported the primary results of interest, remained in our study. Dinger et al.[28] updated their previously published study (of the same cohort reported in 2007[29]) in 2016; thus, their newer report[13] was selected for this study. The conclusion obtained, however, implied a comparable or decreased risk of VTE in populations taking DRSP-containing and non-DRSP-containing COCs. In addition, Bateson et al. did not report on the risk of ATE, which created a severe complication. Our findings suggest lower ATE risks in DRSP-containing COC takers. Two other meta-analyses[2],[5] contrarily suggested a higher risk of VTE in the DRSP-containing COC-taking population than in the non-DRSP-containing COC-taking population; however, assessing the risk of VTE related to DRSP-containing COCs was only one small part of the whole study. Moreover, four RCTs and three models were included in this study to assess the method of failure of the DRSP-containing COCs, because this study should not only focus on the side effects of COCs, such as VTE and ATE, but also highlight how they have protected women from the harm caused by unplanned pregnancies since their invention. In addition to VTE, ATE, and method failure, this study compared other COC-associated side effects between DRSP-containing COCs and non-DRSP-containing COCs, such as Pearl index, weight loss (weight at 6-month baseline), change in BMI, change in SBP/DBP, and mortality related to COCs. In summary, this study not only updated the meta-analysis conducted by Bateson et al.[18] but also comprehensively assessed the side effects related to DRSP-containing COCs and in a further step separately compared DRSP-containing COCs and LNG-containing COCs in subgroup analysis, which may support best-practice prescription for medically eligible women choosing an oral contraception method.
Three major limitations emerged in the present study.
First, although we identified five eligible RCTs in our meta-analysis, none of these reported the primary outcomes of interest but only observational studies including VTE as one of the outcomes, which compromised the grade of the evidence in our study.
Second, the heterogeneity among the studies was significant. To minimize the risk of reporting bias, three models were applied to calculate the pooled risk of method failure for the study group and the comparator group. Breakthrough bleeding is considered a treatment failure for contraceptive drugs. Unfortunately, the parameters in each RCT report were not handled in the same way. Foidart et al.[21] have reported intermenstrual bleeding; Fathizadeh et al.[30] have reported days of bleeding after 3 and 6 months. Bitzer et al.[31] have reported the number of days of bleeding but had not mentioned when this bleeding occurred.
Third, the present study failed to perform stratification analysis based on the estrogen dose contained in the COCs. It has been well established that estrogen increases the risk of both arterial and venous thrombosis.[32] However, very few articles involved in the present study separately reported the risk of VTE related to different doses of estrogen contained in DRSP-containing COCs. With more published data, we would like to perform a better study in the future.
ConclusionIn conclusion, DRSP-containing COCs did not increase the risk of VTE and significantly decreased the risk of ATE when compared with non-DRSP-containing COCs. Thus, it is safe to recommend DRSP-containing COCs. However, women should also be sufficiently informed of the merits and demerits of any type of COC before use.
Supplementary information is linked to the online version of the paper on the Reproductive and Developmental Medicine website.
Acknowledgments
We thank Bo-Zheng Zhang, Bothwin Clinical Study Consultant, WA, for English editing.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References
Comments (0)