Background The coronavirus disease COVID-19 is associated with an increased risk of thrombotic events. Individuals with COVID-19 using hormonal contraception could be at additional risk for thromboembolism, but evidence is sparse.
Methods We conducted a systematic review on the risk of thromboembolism with hormonal contraception use in women aged 15–51 years with COVID-19. We searched multiple databases through March 2022, including all studies comparing outcomes of patients with COVID-19 using or not using hormonal contraception. We applied standard risk of bias tools to evaluate studies and GRADE methodology to assess certainty of evidence. Our primary outcomes were venous and arterial thromboembolism. Secondary outcomes included hospitalisation, acute respiratory distress syndrome, intubation, and mortality.
Results Of 2119 studies screened, three comparative non-randomised studies of interventions (NRSIs) and two case series met the inclusion criteria. All studies had serious to critical risk of bias and low study quality. Overall, there may be little to no effect of combined hormonal contraception (CHC) use on odds of mortality for COVID-19-positive patients (OR 1.0, 95% CI 0.41 to 2.4). The odds of hospitalisation for COVID-19-positive CHC users may be slightly decreased compared with non-users for patients with body mass index <35 kg/m2 (OR 0.79, 95% CI 0.64 to 0.97). Use of any type of hormonal contraception may have little to no effect on hospitalisation rates for COVID-19-positive individuals (OR 0.99, 95% CI 0.68 to 1.44).
Conclusions Not enough evidence exists to draw conclusions regarding risk of thromboembolism in patients with COVID-19 using hormonal contraception. Evidence suggests there may be little to no or slightly decreased odds of hospitalisation, and little to no effect on odds of mortality for hormonal contraception users versus non-users with COVID-19.
- hormonal contraception
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. No data are available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
WHAT IS ALREADY KNOWN ON THIS TOPIC
Individuals with COVID-19 have an elevated risk of thrombotic events. However, it is unknown whether or not concurrent use of hormonal contraception further increases this risk.
WHAT THIS STUDY ADDS
Not enough evidence exists to draw conclusions regarding the risk of thromboembolism or other surrogates for severe disease in reproductive-aged patients with COVID-19 using hormonal contraception.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This review identifies a large gap in the literature, though it is a topic of paramount importance. More robust evidence related to hormonal contraception, COVID-19, and thromboembolism risk could inform international guidelines regarding the appropriate use of hormonal contraceptive in COVID-19-positive women.
The coronavirus disease COVID-19—caused by infection with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)—has affected millions worldwide and led to significant mortality and morbidity, including a high incidence of related thrombotic events.1 The prothrombotic effects of COVID-19 may be related to increased inflammatory cytokine release, platelet activation, endothelial dysfunction, upregulation of the renin–angiotensin–aldosterone system, and blood flow abnormalities.1–3 It is not yet completely understood how disease pathogenicity may be modulated by various individual-level characteristics, including the influence of endogenous or exogenous sex hormones.
Oestrogen and progesterone may play a protective role in the pathogenicity of COVID-19. There are well-documented sex differences in COVID-19 outcomes, with increased mortality seen in males4 and a protective effect from death in post-menopausal women treated with oestrogen.5 Among a cohort of hospitalised COVID-19-positive people in China, the proportion of non-menopausal women with severe COVID-19 disease was significantly lower than the proportion with severe COVID-19 disease among age-matched men.6 Estradiol levels have been shown to be negatively correlated with disease severity as well as interleukin (IL) IL-6 and IL-8 levels.6 In both humans and mouse models, estradiol suppresses production of pro-inflammatory cytokines while stimulating the anti-inflammatory cytokine response.7 Additionally, estradiol may decrease gene expression of angiotensin-converting enzyme 2 (ACE2) receptors in bronchial epithelial cells,8 which are the means of cell entry for SARS-CoV-2.
It is unclear whether hormonal contraception use increases or attenuates the known risk of thromboembolism in those with COVID-19. Use of combined hormonal contraception (CHC)—which includes combined oestrogen and progestin-containing methods such as pills, patches, and rings—confers a two- to threefold increased risk of venous thromboembolism (VTE) compared with non-use.9 Ethinyl estradiol (EE) in CHCs leads to increased levels of coagulation factors II, VII, VIII, X and fibrinogen, and decreased plasma levels of anticoagulant factors, including antithrombin and tissue factor pathway inhibitor.10 This effect is dose dependent, with higher levels of EE affording increased risks of thromboembolism. Progestin-only contraception (POC)—which includes systemic progestin-only methods such as pills, injectables, implants, and intrauterine devices—do not appear to increase risk of VTE in most populations, though some studies have shown increased risk of VTE with depot medroxyprogesterone acetate use.11
There is currently no international consensus regarding the influence of hormonal contraceptive use on thrombosis risk among COVID-19-positive women. At present, the WHO supports the use of all forms of contraception during the COVID-19 pandemic.12 The Society of Family Planning recommends that CHCs be discontinued for all hospitalised women infected with COVID-19 given the theoretical increased risk of thromboembolism, but progestin-only and non-hormonal methods may be continued.13 Specific recommendations for CHC use in COVID-19-positive women vary across the globe, with some guidelines recommending cessation of CHCs or transition to POCs depending on disease severity. Synthesising the evidence related to CHCs, COVID-19 and thromboembolism risk could bring clarity to this space and affect international guidelines.
This review aims to determine whether hormonal contraception increases the risk of venous and arterial thrombosis as well as other markers of COVID-19 severity among COVID-19-positive women, and if this risk differs by type of hormonal contraception or other individual characteristics. It is likely that the conclusions of this review will change as new evidence is generated.
We conducted this systematic review according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.14 We searched for all published, unpublished, and ongoing studies, without restrictions on language or publication status. We searched the following databases from their inception to March 2022: Cochrane Central Register of Controlled Trials via EBM Reviews (Ovid), MEDLINE ALL (Ovid), Embase.com, CINAHL (EBSCOhost), LILACs, Global Health (Ovid), and Scopus. For greater detail, please see the full Cochrane Review publication.15
Studies employing the following designs were included: randomised trials (clustered or individually randomised); quasi-experimental designs, such as non-randomised studies of interventions (NRSIs); and cohort studies with a control group. As these data are emerging, we also included non-comparative studies and case series if they had at least five cases. While randomised controlled trials represent the most rigorous type of study for addressing questions of efficacy and safety, we included other types of studies for this topic because we did not expect to find adequate trial evidence to address our review objectives. It is extremely unlikely that a hormonal contraception method would be randomised in this clinical situation. Additionally, the efficacy and safety outcomes of interest are very rare and the number of participants willing to be randomised to hormonal contraceptive methods would likely be limited. This reduces the feasibility and likelihood of adequately powered randomised trials. NRSIs are likely to provide the best available data for observing differences in outcomes associated with different hormonal contraceptive methods among women with COVID-19.
We included studies of women of reproductive age (ages 15–51 years) who were COVID-19-positive (or presumed positive). We excluded women who were pregnant or less than 3 weeks postpartum given the elevated risk of VTE during this time-period.16
We sought studies comparing COVID-19-positive women using CHC with similar non-pregnant women using either no contraception, non-hormonal contraception, or POC. We planned the following intervention comparisons for this review: CHC versus no contraception; CHC versus non-hormonal contraception; CHC versus POC; POC versus no contraception; and POC versus non-hormonal contraception. Our primary outcome was diagnosis of VTE or arterial thromboembolism during the study period. Our secondary outcomes were mortality, critical illness requiring hospitalisation, diagnosis of acute respiratory distress syndrome (ARDS), and intubation.
Titles and abstracts were screened independently by two authors. We retrieved the full-text study reports and two authors independently screened the full-text and identified studies for inclusion. We resolved any disagreement through discussion or, when required, we consulted a third review author. The study characteristics and outcome data from included studies were extracted, and information was entered into a standard data collection form.
We assessed the risk of bias for key outcomes from NRSIs using the Risk of Bias in Non-randomised Studies of Interventions (ROBINS-I) instrument.17 We considered the following factors to be possible confounding factors for this topic: age, personal history of VTE, recent pregnancy, obesity, severity of COVID-19, ethinyl estradiol dose, and progestogen type.
We used Cochrane GRADE methods and GRADEpro18 19 to assess the certainty of evidence and to prepare ‘Summary of findings’ tables (table 1 and table 2). Given that we used the ROBINS-I tool to assess risk of bias for included NRSIs, we designated the evidence for each NRSI to start at ‘high certainty’ and we then downgraded the certainty of the evidence as appropriate. Two review authors worked independently to assess evidence certainty (eg, high, moderate, low or very low) and resolved inconsistencies through discussion or through a third author.
Had we obtained multiple, comparable studies, we would have calculated intervention effectiveness in a meta-analysis, to produce pooled OR, RR, or mean difference effect estimates with 95% CI. Narrative synthesis was conducted for outcomes lacking adequate data to combine studies.
Our study search yielded 8220 studies, of which 2119 remained for title and abstract screening after removal of duplicates. Of these, 31 full-text articles were reviewed, and five studies met inclusion criteria. The study selection process was recorded in a PRISMA flow diagram14 (figure 1).
Three studies—Seeland et al,20 Mujumdar et al,21 and Costeira et al 22—were comparative NRSIs with 314 704 participants in total (table 3). Seeland et al 20 and Costeira et al 22 stratified outcomes for premenopausal and postmenopausal patients, and we included only information from the premenopausal women. Seeland et al 20 and Mujumdar et al 21 ascertained current use of contraceptives as well as COVID-19 positivity using diagnostic codes within medical records.
Of note, Costeira et al 22 determined contraceptive use as well as presumed COVID-19 positivity based on self-reported data from reproductive-aged women in the UK who used the COVID-19 Symptom Study Smartphone Application from 7 May to 15 June 2020. Users were not required to be COVID-19 positive, so this may represent a different population than outlined in our protocol. However, we anticipate that users of the application who were tracking symptoms were doing so due to concern for having COVID-19. The study also represented some of the best data available to date. Therefore, we included the study for analysis.
One study, Mujumdar et al,21 included all users of any type of hormonal contraception as their exposure group while the others compared CHC users to people without any hormonal therapy. No comparative studies directly assessed venous or arterial thromboembolism as an outcome. Seeland et al 20 measured our secondary outcomes of mortality, while Mujumdar et al 21 and Costeira et al 22 measured hospitalisation rates and intubation. No studies reported data on our secondary outcome of ARDS.
We included two case series that reported COVID-19-positive patients who experienced VTE, Chima et al 23 and Hameed et al.24 Chima et al reported on adolescent patients with pulmonary embolism (PE), and Hameed et al reported on patients with cerebral venous thromboembolism (CVT).
Risk of bias
We used the ROBINS-I tool to assess risk of bias in the three included NRSIs. We judged all three NRSIs to be at serious or critical risk of bias. We judged the two case series to be at high risk of bias given the nature of the study design. There was also no information presented in the case series regarding the temporality of contraceptive use in relation to the thromboembolic outcomes.
We noted serious risk of bias due to confounding for all three NRSIs. We judged Seeland et al 20 to be at serious risk of bias from confounding as they did not identify which covariates were used for propensity score matching. Similarly, Mujumdar et al 21 did not report on variables used for confounding assessment. Costeira et al 22 did adjust for body mass index (BMI) and age, but did not include all prespecified confounders, and we judged there to be additional risk that variables were not validly measured as they were self-reported. No study included personal history of thromboembolism, estradiol dose, or progestogen type as confounding variables, which we pre-specified as likely confounders.
Seeland et al 20 and Mujumdar et al 21 derived their data from electronic health records, but they were retrospective cohorts, so we rated them as moderate risk of bias due to selection of participants. As Costeira et al 22 relied on patient use of their mobile electronic tracking application, we judged there to be serious risk of selection bias as individuals using contraception may be more conscious of potential health risks and more likely to use a health and symptom mobile tracking application.
We judged Seeland et al 20 and Mujumdar et al 21 to be at critical risk of bias due to deviations from intended intervention, as they did not ascertain whether patients were actively using the forms of contraception documented in the medical record at the time of the outcome. We judged Costeira et al 22 to be at moderate risk of bias for deviations from intended intervention as they utilised patient self-report for determining active use of contraception.
Both Mujumdar et al 21 and Costeira et al 22 performed analyses on a smaller subsample due to missing data. We deemed Mujumdar et al 21 to have serious risk of bias due to its small sample size, which was further reduced due to missing data for the outcome variable (hospitalisation). Seeland et al 20 reported no information on missing data or how they were handled.
Hospitalisation (Mujumdar et al 21) and mortality (Seeland et al 20) are definite outcomes unlikely to be measured incorrectly in electronic health record data. We thus judged these to have low risk of bias. The outcome of hospitalisation for Costeira et al,22 however, was derived from self-reported data, so we judged this to be at moderate risk of bias as conceivably some people could have misclassified the outcome; for example, if a patient only had an emergency department visit but reported this as a hospitalisation.
Effects of interventions
Four studies reported on outcomes for use of CHC compared with no use of contraception in COVID-19 patients (Chima et al,23 Hameed et al,24 Seeland et al 20) or patients who were at risk of having COVID-19 (Costeira et al 22) (table 1). Only one study (Mujumdar et al 21) reported on outcomes for use of any type of hormonal contraception compared with no use of contraception in COVID-19 patients (table 2). The included studies reported outcomes on mortality, hospitalisation rates, intubation, and thromboembolism. No studies reported data on ARDS.
Seeland et al 20 measured mortality of COVID-19-positive patients who were users of CHC versus contraception non-users with data derived from electronic health records from healthcare organisations in 17 countries. Given the limitations of the database, the authors could not ascertain current contraceptive use at the time of the outcome. Based on results from this NRSI, there may be little to no effect of combined hormonal contraception use on odds of mortality for COVID-19-positive patients (adjusted OR 1.0, 95% CI 0.41 to 2.4), but the evidence is very uncertain.
Costeira et al 22 found that CHC users may have a slight decrease in their odds of hospitalisation compared with non-users, after adjusting for BMI, age, and smoking status (adjusted OR 0.79, 95% CI 0.64 to 0.97). This study evaluated hospitalisation for COVID-19 in individuals self-reporting use or non-use of CHC, where COVID-19 disease status was not confirmed through testing but via symptom reporting through a mobile tracking application; thus we deemed this evidence to be of low certainty, downgraded for serious risk of bias and for indirectness. Mujumdar et al 21 found little to no effect on risk of hospitalisation for COVID-19-positive patients based on exposure to any hormonal contraception (adjusted OR 0.99, 95% CI 0.68 to 1.44). They obtained their data from electronic health records from one tertiary care organisation, but could not ascertain current contraceptive use at the time of the outcome. The evidence from this study is very uncertain.
Mujumdar et al 21 found that no COVID-19 patients in their study required intubation, regardless of whether or not they were exposed to hormonal contraception. In one case series of seven COVID-19 patients, two required intubation. Of these, one patient was on CHC, while the other was not.24 Thus, there may be little to no effect of combined hormonal contraception use on odds of requiring intubation in patients with COVID-19, but the evidence is very uncertain.
Venous or arterial thromboembolism
The two case series together reported on 13 patients with a VTE concurrent with COVID-19 infection. Of these 13 patients, two were reportedly using CHC. One of the six female paediatric patients (<18 years in age) with COVID-19 who developed a PE was taking hormonal contraception.23 One of seven reproductive-aged women with COVID-19 and CVT was taking oral contraceptive pills.24 Notably, this individual was also diagnosed with antiphospholipid antibody syndrome, a thrombophilia. Of note, no comparative studies reported on arterial or venous thromboembolic outcomes. The evidence for any effect of CHC on the risk of developing VTE is very uncertain. There was no evidence assessing the risk of arterial embolism.
There is an increased risk of thromboembolism in patients with COVID-19.1 2 1 2 While CHC use is an independent risk factor for thrombosis, evidence describing risks of hormonal contraception use during the COVID-19 pandemic is sparse.9 Our primary objective was to assess the risk of venous or arterial thromboembolism in patients with COVID-19 disease using CHC. Our secondary objectives were to investigate other markers of COVID-19 severity such as ARDS, intubation, hospitalisation, and mortality for those using CHC or other forms of hormonal contraception. We identified only five studies addressing these objectives.20–24
We found no comparative studies assessing thromboembolism risk among COVID-19-positive individuals using hormonal contraception compared with non-users. Two case series23 24 reported on a total of 13 individuals with VTE who were COVID-19 positive, of whom only two individuals were taking combined contraception or oral contraceptive pills. The evidence for any effect of CHC use on the risk of developing VTE is very uncertain, and we found no evidence assessing risk of arterial embolism.
Two observational studies20 22 and one case series24 assessed markers of COVID-19 severity for users of CHC versus non-users. We found little to no effect of combined hormonal contraceptive use on odds of mortality among COVID-19 patients, but the evidence is very uncertain.20 Combined hormonal contraception use may slightly decrease the odds of hospitalisation for individuals with a BMI <35 kg/m2; however, the study population was not confirmed to be COVID-19-positive and the evidence is very uncertain.22 Use of CHC among COVID-19-positive patients appears to have little to no effect on the odds of intubation, but again this evidence is very uncertain.24
One observational study21 assessed markers of COVID-19 severity among users of hormonal contraception (including both CHC and POC) versus non-users of contraception. There may be no effect of any hormonal contraception use on odds of hospitalisation for COVID-19-positive patients, but the evidence is very uncertain. We could not measure the relative effect of hormonal contraceptive use on intubation as no intubations occurred in either group. The quality of evidence for risk of thrombosis for CHC users versus non-users who are COVID-19-positive is extremely low.
This review has identified a large gap in the literature, though it is a topic of paramount importance. Future studies would benefit from collecting pertinent information on confounders. These include: patient age; BMI; history of prior thromboembolism; medical comorbidities associated with increased risk of VTE; reason for hormonal contraception use (contraception versus treatment of medical condition); recent pregnancy or other thrombophilia; contraception formulation including type of oestrogen and dose of oestrogen for combined hormonal contraception; and duration of contraception use. No studies reported indication for hormonal contraceptive use, which is important because individuals who use hormonal management for medical conditions such as heavy menstrual bleeding may have different risk profiles compared with individuals using hormones for contraceptive purposes. Additionally, several studies included were downgraded due to failure to ascertain actual contraceptive use and adherence at the time of the outcome. Certainty of the evidence would improve if current use or recent use of contraception were regularly ascertained for individuals at time of the outcome of interest. As COVID-19 continues to evolve and new variants emerge, reporting of variants as well as therapeutics used for treatment may also be important for analysis, but were not reported in any studies. Additionally, there were no data for populations of differing COVID-19 severity, that is, ambulatory versus hospitalised patients, which is needed before evidence-based recommendations can be provided to hormonal contraceptive users who contract COVID-19.
Although the evidence is of very low certainty and there is heterogeneity among studies in exposures, populations and outcomes, the current available evidence suggests there may be little to no or slightly decreased odds of hospitalisation and little to no effect on odds of mortality for hormonal contraception users versus non-users who are COVID-19-positive. There is not enough evidence to draw conclusions regarding risk of venous or arterial thromboembolism in patients with COVID-19 who are using hormonal contraception.
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. No data are available.
Patient consent for publication
We would like to acknowledge and thank the Cochrane Fertility Regulation Group for their support in adapting the original Cochrane Review on which this publication is based and the authorship team of the primary publication.
Contributors Conceiving and designing the review: Megan A Cohen (MAC), Alison Edelman (AE), Jillian Henderson (JH). Guarantor: JH. Co‐ordinating the review: MAC. Designing search strategies: Robin Paynter (RP), MAC, AE. Searching and selecting studies for inclusion in the review: Katie Hansen (KH), Shaalini Ramanadhan (SR), MAC, AE, JH. Collecting data for the review: MAC, AE. Assessing the risk of bias in the included studies: MAC, AE, JH. Analysing the data: MAC, AE, JH. Assessing the certainty in the body of evidence: MAC, AE, JH. Interpreting the data: MAC, AE, JH. Writing the review: KH, SR, MAC, AE, JH.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests KH, MC, SR have declared that they have no conflict of interest. AE: reports PI of an implant study for Merck, paid to institution; PI of an oral contraceptive study for HRA Pharma, paid to institution; reports royalties or licenses from Up to Date, personal payment. These are not COVID-19 related, but are hormonal contraception related. JH and RP have declared that they have no conflict of interest.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Author note This review article is based on a Cochrane Review published in the Cochrane Database of Systemic Reviews (CDSR) 2023, Issue 1, DOI: https://doi.org/10.1002/14651858.CD014908.pub2 (see www.cochranelibrary.com for more information). As Cochrane Reviews are regularly updated as new evidence emerges, the CDSR should be consulted for the most recent version of this review.