After consulting with a librarian from the University of British Columbia, search terms were developed to reflect an interest in COVID-19, menstruation, and vaccination. The COVID-19 search included the following terms: COVID-19, SARS-CoV-2, nCoV-2019, coronavirus infection, viral pneumonia, and pandemics. The menstruation search terms included menstruation, menstruation disturbance, menses, menstrual flow, menstrual discharge, menorrhea, menarche, and monthlies. Search terms that encompassed vaccination included vaccination, vaccines, and COVID-19 vaccines. The full search terms are available in Supplementary method 1.

The search terms were deployed on July 20, 2022, to identify peer-reviewed observational articles published between December 1, 2019, and July 1, 2022, from MEDLINE, Embase, and Web of Science (17–19). Following the database search, the reference lists of relevant eligible articles were manually searched to identify additional eligible articles.

Following the article search, all potential articles were compiled and uploaded into the Covidence tool for systematic reviews, where duplicates were automatically removed (20). The articles were filtered according to the inclusion and exclusion criteria during this process. Subsequently, the article review phase consisted of a title and abstract screening, and a full-text review. This process was completed in a blind, independent manner by MC and ME to avoid selection bias. Disagreements were discussed until a consensus was reached.

The inclusion criteria were: (i) articles published from December 1, 2019, to July 1, 2022, found in MEDLINE, Embase, and Web of Science, (ii) original articles that described observational studies, (iii) articles that had full-text available in English, (iv) articles that discussed menstruating women of reproductive age, and (v) articles specifically reporting on menstrual cycle irregularities for unvaccinated and COVID-19 vaccinated women of reproductive age (17–19). However, articles were not excluded based on geographical location, patient age, COVID-19 status, pregnancy status, ovulatory status, reproductive history, vaccine type, other vaccination-related symptoms, or any other patient factors.

The exclusion criteria were: (i) articles that did not have abstracts, (ii) articles that did not have a full-text available in English, (iii) publication types that were opinion pieces, letters, commentaries, guidelines, or simulations/models, (iv) articles that did not discuss women of reproductive age, and (v) articles that did not discuss menstrual cycle irregularities for COVID-19 vaccinated and unvaccinated women of reproductive age.

The 14-item, NIH Study Quality Assessment Tool for Observation Cohort and Cross-Sectional Studies was used to critically appraise the cross-sectional studies included in this systematic review and meta-analysis (16). This tool was specifically designed to critically appraise cross-sectional studies for a systematic review. Thus, in this systemic review, this tool was used to assess the internal validity of the study design of each included cross-sectional study (16).

The NIH assessment tool consists of 14 questions that were applied to each study. Specifically, the tool assessed the clarity of the research question, the clarity of the study population, the participation rate of eligible persons, if the participants were recruited from the same population, the justification for sample size, if the exposures of interest were measured before the outcome of interest was measured, the sufficiency of the timeframe, how variable amounts of exposure impact the outcome, the validity of the exposure measurements if the exposures were measured multiple times, the validity of the outcome measured, if the assessment was blinded if the attrition rate to follow-up was <20%, and how the impact of the potential confounding variables was measured (16).

For each of the 14 items, each cross-sectional study was scored as Yes, No, or Not applicable (16). For each article, the number of items that reduced the bias was totaled. The articles were then categorized into quartiles to represent their risk of bias. The first quartile consisted of articles that met 11–14 of the NIH criteria, which represented low risk of bias. The second quartile consisted of articles that met 7–10 of the NIH criteria, which represented some risk of bias. The third quartile consisted of articles that met 4–9 of the NIH criteria, which represented moderate risk of bias. The fourth quartile consisted of articles that only met 0–3 of the NIH criteria, which represented high risk of bias. Only cross-sectional studies that fell within the first quartile were included in this systematic review. The appraisal tool was completed by MC and ME in a blind and independent manner, and any disagreements were discussed until resolved.

The NIH Study Quality Assessment Tool for Observation Cohort and Cross-Sectional Studies was also used to appraise the cohort studies included in this systemic review and meta-analysis (16). This tool was specifically designed to critically appraise cohort studies for a systematic review. Thus, in this systemic review, this tool was used to assess the internal validity of the study design of each included cohort study (16).

As previously noted, this 14-item quality assessment tool consists of 14 questions applied to each study. Specifically, the tool assessed the clarity of the research question, the clarity of the study population, the participation rate of the eligible persons, if the participants were recruited from the same population, the justification for sample size, if the exposures of interest were measured before the outcome of interest was measured, the sufficiency of the timeframe, how the variable amounts of exposure related to the outcome, the validity of the exposure measurements, if the exposures were measured multiple times, the validity of the outcome measured, if the assessment was blinded, if the attrition rate to follow-up was <20%, and how the impact of potential confounding variables was measured.

For each of the 14 items, each cross-sectional study was scored as Yes, No, or Not applicable (16). For each article, the number of items that reduced the bias was totaled. The articles were then categorized into quartiles to represent their risk of bias. The first quartile consisted of articles that met 11–14 of the NIH criteria, which represented low risk of bias. The second quartile consisted of articles that met 7–10 of the NIH criteria, which represented some risk of bias. The third quartile consisted of articles that met 4–9 of the NIH criteria, which represented moderate risk of bias. The fourth quartile consisted of articles that only met 0–3 of the NIH criteria, which represented high risk of bias. Only cross-sectional studies that fell within the first quartile were included in this systematic review. The appraisal tool was completed by MC and ME in a blind and independent manner, and any disagreements were discussed until resolved.

During the data extraction phase, the reviewers generated a 30-item Covidence data collection form, populated each of the 30 items for all included articles, and exported the populated form into Microsoft Excel (20, 21). The data collection form captured the study characteristics and the outcomes of interest. Outcomes of interest were extracted from both the included article and its corresponding Supplementary information files.

The outcome of interest was the incidence of irregularities in menstrual cycles. The research question was to investigate the incidence of menstrual irregularities in women of reproductive age following COVID-19 vaccination compared to unvaccinated women of reproductive age. In this context, irregular menstrual cycles consisted of self-reported changes in menstrual volume, menstrual duration, cycle length, ovulation timing, and/or pain level.

Discrete numerical data were collected on the incidence of menstrual irregularities for women of reproductive age following COVID-19 vaccination in comparison to the rate for unvaccinated women of reproductive age. To manage the unpopulated data fields, the number of women was imputed based on the given values. If, for example, the number of vaccinated women with menstrual irregularities was missing, while the total number of women with menstrual irregularities and the number of unvaccinated women with menstrual irregularities were given, the missing value was calculated using subtraction.

Data collection was performed irrespective of dose numbers, such as first dose, second dose, or booster dose(s). For articles that did include the dose number, the first dose was used because many patients received the first dose but not subsequent doses. Furthermore, this systematic review did not provide information on the resolution of menstrual disruptions based on the vaccine manufacturer.

The odds ratio (OR) at the 95% confidence interval (CI) was calculated for each included article to determine the odds that the outcome is associated with the exposure. The Mantel-Haenszel fixed-effect method was employed to assess the discrete data. Subsequently, the I²-, τ²-, and χ²-test were employed to quantify the level of heterogeneity between studies in the meta-analysis, and a p-value was calculated to determine the level the significance of the heterogeneity.

Weighted mean differences were used to analyze statistical data effectiveness, and a 95% CI was calculated. Once the population mean OR was determined, the Z-test statistic was employed to evaluate the reduction of uncertainty in past events, and the p-value was calculated.

A total of 84 articles were identified through a search of three databases (14 articles from MEDLINE, 68 articles from Embase, and two articles from Web of Science) (17–19). After 16 duplicate articles were removed, the abstracts and titles of the remaining 71 articles were screened, resulting in 16 articles for full-text review. Following the full-text review, one article remained for data extraction. All the included articles were hand-searched for references; thus, three additional articles were identified for review.

Consequently, four articles were included in this systematic review and meta-analysis, as described in Figure 1. They described 19,019 women of reproductive age in the vaccinated (exposed) group and 6,045 women of reproductive age in the unvaccinated (unexposed) group, for a total of 25,054 women.

All four articles included in this systematic review fell into the first quartile during the screening process, where cohort studies were reviewed using the NIH Study Quality Assessment Tools (16). These articles had different observational study designs: two cross-sectional and two cohort studies. Of the four articles, there is low representation from both high-income countries (two articles from the United States) and middle-income countries (one from Jordan and one from China); there is no representation from low-income countries.

The Study Quality Assessment Tool scores for all the included articles are presented in Table 1. The full quality assessment is presented in Supplementary Tables 1, 2.

Overall, 25,054 women of reproductive age were included in the studies in this systematic review, consisting of 19,019 vaccinated women and 6,035 unvaccinated women of reproductive age (Figure 2). The OR with the accompanying 95% CI was calculated for each of the values ranging from 0.74 to 1.38. Only one study had an OR value <1 (23).

The pooled OR is 1.91 (CI: 1.76–2.07), indicating an association between vaccination and menstruation changes. The statistical analysis results indicate heterogeneity between the studies and the dispersion of values is statistically significant (χ² = 195.0 with df = 3, p < 0.00001, I² = 98%). The overall effect of the mean was statistically significant (Z = 16.01, p < 0.0001).

These findings are relevant to the worldwide trend of vaccine hesitancy. There is concern that post-vaccination menstrual irregularities are a source of vaccine hesitancy for this population and may provide fuel for anti-vaccination campaigns (26). The findings of this meta-analysis are relevant to the greater context of vaccine acceptance and designing strategies to increase uptake.

High heterogeneity in vaccination uptake rates exists between countries. The health system is reported to be a strong determinant of vaccine uptake as they are responsible for delivery and effective communication to the public (35). Other major determinants (36) include:

Interestingly, Asian countries have strong trust in government recommendations and high vaccine uptake, while middle-income countries tend toward acceptance of vaccines (35).

There is a mixed response to encouraging vaccination uptake through vaccine mandates (37). Public health agencies must balance the goal of herd immunity with the freedom to choose vaccination (37). In 2021, the European Union produced principles that state that vaccination must occur according to individual choice, so an individual must consider the benefits against the harms but face no discrimination for refusing to be vaccinated (37). Considering their principles, the policies across the European Union vary in terms of vaccination requirements, passports, mask mandates, and testing requirements (37).

Significant gains globally will occur by targeting populations of low income and low education (35, 36). Hesitancy can stem from marginalization, social exclusion, negative experiences with healthcare, misinformation, mistrust in government, and mistrust in health authorities (38). Strategies to address these populations include providing clear communication using a trusted local source, addressing historical issues of distrust, and being sensitive to religious or cultural beliefs (35). Interestingly, many cultures are more accepting of getting vaccinated when recommended by an employer but are less likely to get vaccinated if mandated by an employer; such attempts are perceived as limiting the employee's freedom or expressing employers' self-interest (35).

Many personal factors are associated with high vaccine uptake, including strong COVID-19 knowledge, high perceived seriousness of the pandemic, and good preventative practices; these patterns were consistent in many low- and middle-income countries, including Congo, Ethiopia, and some middle eastern populations (36).

There is a vital role for health workers in building confidence and acceptance of vaccines in public; clear communication about the safety and efficacy of the COVID-19 vaccine, as well as personal reasons and recommendations from a trusted health worker, can increase vaccine uptake in many populations (38).

Populations with specific health concerns, like pregnant women, often receive poor information regarding the benefits and safety of vaccinates (39). They are more likely to accept a vaccine from a health worker when health workers educate patients about it and recommend it to them, so vaccine counseling should be part of prenatal and pregnancy care. Similarly, health workers have a crucial role in communicating the safety and efficacy of vaccines despite the short-term menstrual side effects to women of reproductive age; they may attempt to provide personal recommendations to encourage vaccination to this population (31).

These strategies can help limit the consequences of the COVID-19 pandemic and improve vaccine uptake. Vaccines remain the most effective manner to limit the morbidity and mortality associated with SARS-CoV-2 infection (38). Research has shown that mass vaccination to achieve herd immunity is the most effective manner to overcome the COVID-19 pandemic (37). It is worth reiterating that despite short-term menstrual changes, there is no research to support the fear that COVID-19 vaccinations impact fertility, and vaccines are recommended for women of reproductive age (13, 31).

There are several limitations associated with this systematic review and meta-analysis. The primary limitation is the lack of randomized control trials, which restricts the investigation of a potential causal link (6). As such, the included articles discussing observational studies may only provide support for a correlative link.

Second, the increased frequency of menstrual irregularities reported in the media creates a heightened awareness of health, especially menstrual health, due to the increase in reporting during the pandemic and anecdotal reports of vaccination side effects (6). This has caused many women of reproductive age to monitor their health more closely, which may result in false attributions or overreporting.

Third, menstrual characteristics are often subjective, and data are often collected using self-reporting methods, leaving room for miscalculation and systemic error (6). Many methods exist to monitor a woman's menstrual cycle, so it is unclear what methods the women in the included articles used to quantify their irregularities. In particular, the symptom of pain is a subjective measurement, and heightened awareness of symptoms due to media communication could produce higher pain scores (6).

Fourth, there were limited amounts of population data so subgroup analyses could not be performed. Due to the lack of reporting, the type of vaccine manufacturer could not be correlated with menstrual irregularities, which may reduce the applicability in real-world decisions to select one vaccine over another. Similarly, this systematic review could not account for the number of vaccine doses that a subject received, which may be further complicated as boosters continue to be administered at rapid rates. Data were further limited in the study's inability to report when a menstrual irregularity was observed relative to when the vaccine was administered. Such factors may include the length of time after menstruation in which the woman received the vaccine, the phase of the woman's menstrual cycle when she received the vaccine, and if multiple doses were taken in the same cycle, which may provide further clarity into the temporal aspect of the vaccine's impact on menstrual regularity.

Fifth, this systematic review's strict inclusion criteria limit the generalizability of its findings to the global population. Searches were limited to published articles available in English, which primarily produced articles from high-income countries and middle-income countries, with no representation from low-income countries. This may favor reporting specific subpopulations (including some racial groups) that may have variations not captured in this review, so our results may not represent the global population.

Finally, there were reporting biases in the selected articles because all the articles were observational studies, which are known to overreport cases in the exposed group and underreport normal cases, causing an overestimating of the effect size.

To overcome the lack of resolution by vaccine type, future articles may investigate the relationship between the vaccine manufacturer, the number of vaccine doses, and menstrual irregularities. As booster shots become more prevalent, it is crucial to investigate their impact. Additional studies may record the time it takes for menstrual irregularities to normalize, such that subsequent review articles may synthesize the results to inform practitioners of the full scope of these side effects.

Future studies may investigate the co-occurring impacts of SARS-CoV-2 infection with COVID-19 vaccination to fully characterize the impact on women of reproductive age. Future studies should also consider how menstrual cycle irregularities may have downstream impacts on women's reproductive and sexual health.

To overcome reporting bias, future studies should investigate unpublished, non-English literature from a representative selection of countries. Moreover, understanding the heterogeneity between studies will help elucidate how a broader context may bias the results.