COVID-19 vaccination and menstrual cycle changes: A United Kingdom (UK) retrospective case-control study

34 Background. There has been increasing public concern that COVID-19 vaccines cause 35 menstrual cycle disturbances, yet there is currently limited data to evaluate the impact of 36 vaccination on menstrual health. Our objectives were (1) to evaluate the prevalence of 37 menstrual changes following vaccination against COVID-19, (2) to test potential risk factors 38 for any such changes, and (3) to identify patterns of symptoms in participants’ written accounts. 39 Methods. We performed a secondary analysis of a retrospective online survey titled “The 40 Covid-19 Pandemic and Women's Reproductive Health”, conducted in March 2021 in the UK 41 before widespread media attention regarding potential impacts of SARS-CoV-2 vaccination on 42 menstruation. Participants were recruited via a Facebook ad campaign in the UK and eligibility 43 criteria for survey completion were age greater than 18 years, having ever menstruated and 44 currently living in the UK. In total, 26,710 people gave consent and completed the survey. For 45 this analysis we selected 4,989 participants who were pre-menopausal and vaccinated. These 46 participants were aged 28 to 43, predominantly from England (81%), of white background 47 (95%) and not using hormonal contraception (58%). 48 Findings. Among pre-menopausal vaccinated individuals (n=4,989), 80% did not report any 49 menstrual cycle changes up to 4 months after their first COVID-19 vaccine injection. Current 50 use of combined oral contraceptives was associated with lower odds of reporting any changes 51 by 48% (OR = 0.52, 95CI = [0.34 to 0.78], P<0.001). Odds of reporting any menstrual changes 52 were increased by 44% for current smokers (OR = 1.44, 95CI = [1.07 to 1.94], P<0.01) and by 53 more than 50% for individuals with a positive COVID status [Long Covid (OR = 1.61, 95CI = 54 [1.28 to 2.02], P<0.001), acute COVID (OR = 1.54, 95CI = [1.27 to 1.86], P<0.001)]. The 55 effects remain after adjusting for self-reported magnitude of menstrual cycle changes over the 56 year preceding the survey. Written accounts report diverse symptoms; the most common words 57 include “cramps”, “late”, “early”, “spotting”, “heavy” and “irregular”, with a low level of 58 clustering among them. 59 Conclusions. Following vaccination for COVID-19, menstrual disturbance occurred in 20% of 60 individuals in a UK sample. Out of 33 variables investigated, smoking and a previous history 61 of SARS-CoV-2 infection were found to be risk factors while using oestradiol-containing 62 contraceptives was found to be a protective factor. Diverse experiences were reported, from 63 menstrual bleeding cessation to heavy menstrual bleeding. 64 65 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 3

66 67 68 There has been increasing public concern that COVID-19 vaccines cause disruption of 69 menstrual cycles [1–3], leading to problematic menstrual symptoms, vaccine hesitancy [4] and 70 fears about the impact of vaccination on fertility [5–7]. There are currently limited data [8] for 71 investigating the relationship between the COVID-19 vaccines and menstrual cycles [1,9,10]. 72 This is despite rising awareness among clinicians that the menstrual cycle should be used as a 73 vital sign of female health [11,12], that sex is a biological variable which should be considered 74 in immunological studies [13] and that there have been reports of heavy, infrequent or irregular 75 menstrual bleeding following vaccination [1,8–10]. Quantitative evidence for any such 76 relationship between COVID-19 vaccination and menstrual cycle disturbance, as well as the 77 factors mediating this relationship, are crucial for evaluating how female health has been 78 impacted by the pandemic. 79 80 The first published study on the topic of vaccine effects on menstrual cycles dates back to 1913, 81 when a medical doctor at the Presbyterian Hospital, New York, concluded that there was a 82 striking relationship between the prophylactic typhoid vaccine and menstrual disturbances 83 among one hundred cases [14]. After ruling out all other apparent causes, he found that 53% 84 showed some type of disturbance, including increased or decreased frequency, increased or 85 decreased volume and dysmenorrhoea [14]. These disturbances disappeared within 6 months 86 of the vaccine, suggesting that any such vaccine side-effect was temporary. There has also been 87 a report of menstrual disturbances following inoculation with the hepatitis vaccine in a Japanese 88 study conducted in 1982. Among 16 hospital employees, 7 reported various menstrual 89 abnormalities including decreased volume of menstruation, infrequent or too frequent menses 90 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 4 [15]. The changes were attributed to the use of human plasma to make the vaccine (antigens 91 were derived from human plasma, containing hormonal impurities). More recently, large-scale 92 studies on the effects of vaccination on menstrual disturbances reported mixed results. A 2018 93 study of 29,846 female residents of Nagoya City, Japan, found that none of the 24 symptoms 94 investigated, including menstrual symptoms, were associated with increased odds of occurring 95 after administration of the HPV vaccine. However, age-adjusted odds of hospital visits were 96 increased for “abnormal amount of menstrual bleeding” (OR=1.43, 95%CI=[1.13 to 1.82]), 97 “irregular menstruation” (OR=1.29, 95%CI=[1.12 to 1.49]) and chronic, persisting “abnormal 98 amount of menstrual bleeding” (OR 1.41, 95% CI: 1.11–1.79)[16]. Although retrospective and 99 sensitive to recall bias among those receiving the vaccine, the study suggests a possible link 100 between the HPV vaccine and menstrual irregularities. Another study applying a signal 101 detection analysis on the FDA Vaccine Adverse Event Reporting System (VAERS) shows a 102 disproportionate number of reports of premature ovarian insufficiency, amenorrhea, irregular 103 menstruation, increase in FSH and premature menopause following administration of the HPV 104 vaccine [17]. However, the evidence is non-causal, and relationships might depend on the type 105 of vaccine. With regards to COVID-19, the UK’s Medicine and Healthcare products Regulatory 106 Agency (MHRA) is closely monitoring reports of menstrual disorders [18], with more than 107 30,000 reports made to its yellow card surveillance scheme by 2 September 2021 for both 108 mRNA and adenovirus-vectored COVID-19 vaccines [19]. Recent data from a gender-diverse 109 sample receiving COVID-19 vaccination in the US suggests that changes in the form of heavy 110 and breakthrough bleeding affect many people. However, there has been no quantitative 111 assessment of the risk factors for menstrual disturbances following COVID-19 vaccination 112 prior to widespread media attention ([8], Box 1). 113 114 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 5

of the study 115 The objectives of this study are three-fold: (1) to evaluate the incidence of reports of menstrual 116 changes of any kind following COVID-19 vaccination in a sample broadly representative of 117 those who menstruate in the UK, (2) to investigate the risk factors for reporting any menstrual 118 changes following COVID-19 vaccination, and (3) to capture the types and breadth of menstrual 119 disturbances by analysing the text written by participants. We build on a large retrospective 120 cross-sectional study on menstruation during the pandemic conducted in the UK, launched 121 before UK media coverage of concerns over menstrual vaccine side-effects and including both 122 quantitative and textual data on menstrual cycle changes perceived to be induced by the 123 COVID-19 vaccines. 124 125

126 127 128 Study design 129 The online survey was initially designed to evaluate whether and how the COVID-19 pandemic 130 influenced menstrual health. Retrospective and self-reported data on menstrual cycles, 131 behaviour, life circumstances and health before and during the pandemic as well as SARS-132 CoV-2 infection and vaccination status were collected using an online survey hosted on the 133 Qualtrics platform (www.qualtrics.com). All survey responses were anonymized using 134 randomly generated IDs. The study, titled “The Covid-19 Pandemic and Women's Reproductive 135 Health” has been reviewed by, and received ethics clearance through, the Oxford University 136 School of Anthropology and Museum Ethnography Departmental Research Ethics Committee 137 [SAME_C1A_20_029]. 138 139 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 6 Patient and Public Involvement 140 During the design of survey questions, input from a panel of women suffering from Long Covid, 141 referred to us by the Long Covid Support online group (https://www.longcovid.org/), was 142 incorporated. The results were discussed with panel members who were also invited to co-143 author the paper and co-design dissemination plans. 144 145 Study population 146 The online survey was launched on March 8, 2021. The title of the survey was kept general 147 (“female reproductive health and the COVID pandemic”) so as not to oversample individuals 148 with specific interest in menstrual cycles and COVID infection or vaccination. The survey was 149 disseminated through a Facebook advertising campaign, and included images of women of 150 diverse ethnicities, ages, and abilities, as well as images of breastfeeding and pregnant women 151 (SI1); we fine-tuned the ad targeting (to the extent that Facebook allows) throughout the 152 campaign to ensure even geographical and socio-economic spread. As explained in the 153 information page (SI2), participants could only complete the survey if they were over 18, had 154 ever menstruated, currently lived in the UK, and gave informed consent to the use of their data. 155 The survey included a maximum of 105 questions depending on individual circumstances (SI3) 156 and took an average of 24 minutes to complete. Of the eligible participants who started the 157 survey, 61% answered all questions after giving their consent (on average participants 158 completed 80% of the questionnaire). In case of survey fatigue, progress could be saved for up 159 to 14 days to allow participants to resume later. The survey was disseminated through a 160 Facebook advertisement campaign targeting all menstruators in the UK, from 08/03/21 to 161 01/06/21, at which point there had been no new entries for a week. During the campaign, we 162 used a stratified sampling strategy to ensure that subgroups of the UK population in terms of 163 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 7 age, income and ethnicity were represented in the final sample. In total, 695,543 people viewed 164 the survey ad on their Facebook page and 26,710 with eligible criteria gave consent and 165 completed it (there were no duplicates), leading to a 3.8% response rate. The data, data 166 dictionary and scripts are available on the Open Science Framework Platform 167 (https://osf.io/pqxy2/). 168 169 Outcome: vaccine side-effects on menstrual cycles 170 While the survey did not initially aim to evaluate the impact of vaccination on menstrual cycles 171 specifically, a question was included to assess participants’ perception of their menstrual cycles 172 following vaccination at the end of the survey. Specifically, participants who indicated that they 173 had been menstruating in the past 12 months, received 1 or 2 doses of the COVID-19 vaccines 174 and were not involved in a clinical trial were asked “ Have you noticed any changes to your 175 menstrual cycles since you got vaccinated?”, to which 1 of 4 possible answers could be given: 176 “No”, “Yes, my menstrual cycles are MORE disrupted”, “Yes, my menstrual cycles are LESS 177 disrupted”, “Other (please state)”. Although “disruption” per se was not defined, by the time 178 participants answered this question, they had already completed many questions on menstrual 179 cycle regularity, duration, and symptoms. At the time of the survey design, anecdotal reports of 180 menstrual effects of the vaccine were only just beginning to circulate, while people with Long 181 Covid were reporting either improvement or worsening of their symptoms in general after 182 vaccination. This question was included with the intention of investigating the latter effects. 183 Participants could select the answer “Other”, which in some cases may not have been a different 184 decision from choosing either “more disrupted” or “less disrupted”. For analysis, we thus 185 transformed these variables to represent a binary outcome (“No changes” vs. “Any other 186 changes”). 187 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 8 188 Exposures 189 A total of 33 variables were extracted for this analysis. In addition to socio-demographic 190 variables (age, income, education, gender, ethnic group, marital status), and standard proxies 191 for health (BMI, smoking status, physical activity, regular use of vitamins/supplements, regular 192 use of medicine), the dataset included vaccine-related, COVID and pandemic-related, and 193 reproductive variables (See SI4 for the operationalization of variables). First, data on the type 194 of vaccine received, of which only two had been approved for use in the UK at the time (Pfizer 195 BioNTech/Oxford-AstraZeneca/Not sure), and the timing of the first vaccination (month/year) 196 were included. Second, COVID status was operationalized in two ways: (i) based on whether 197 people thought they had had COVID, as widespread testing had not been available in the UK 198 in the early months of the pandemic which fell within the survey period, leading to three 199 categories: No COVID, acute COVID (symptoms lasting less than 28 days) and Long Covid 200 (symptoms lasting more than 28 days) as well as (ii) based on a combination of testing and self-201 diagnosis, leading to three categories: No COVID (no tests or negative tests), COVID tested + 202 (positive test) and “ Self-diagnosed positive” (referring to individuals who had a suspected or 203 clinically diagnosed COVID infection but had not obtained positive PCR, antigen or antibody 204 tests). We included this last category due to the unavailability of widespread testing in the UK 205 in the first wave of the pandemic in 2020 and ongoing questions about the accuracy and optimal 206 timing of antigen and antibody tests. In addition, variables indicative of changes in both life 207 satisfaction and menstrual cycle symptoms compared to before the pandemic were also included 208 to adjust for changes experienced because of the pandemic and/or the infection rather than 209 vaccination. Third, reproductive variables indicative of menstrual health before the pandemic 210 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 9 (age at menarche, cycle length, period length, cycle irregularity, heavy bleeding), reproductive 211 history (number of deliveries) and contraceptive use were included. 212 213 Statistical analysis 214 The aim of the quantitative analysis was two-fold: (1) to quantify the extent to which individuals 215 answered “No changes” when asked about any perceived changes to their menstrual cycle 216 following COVID-19 vaccination, and (2) to evaluate potential risk and protective factors for 217 selecting any other answer. The original outcome variable is nominal (two or more categories 218 with no intrinsic order) but violates the IIA assumption (Independence or Irrelevant 219 Alternatives) as options were not independent, thus we dichotomized the variable into two 220 mutually exclusive categories (“No changes”, “Any other changes”) and performed logistic 221 regressions. We first conducted a series of exploratory univariable analyses, investigating each 222 of 33 variables as potential risk factors for reporting changes in menstrual cycles following 223 vaccination. We then retained all variables significant at the false discovery rate (FDR) 224 threshold (FDR-corrected P<0.05) [20] for consideration in multivariable analyses. We then 225 conducted separate multivariable analyses with each of the variables identified in the 226 univariable analyses as exposures variables. Each multivariable model was adjusted for 227 potential confounders, which were defined as variables significant at the FDR threshold in the 228 univariable analyses and with a potential confounding (but not mediating) effect according to 229 hypothesized directed acyclic graphs (DAG, SI5). Estimates and confidence intervals on the 230 log-odds scale were converted to odds-ratios for reporting. To test the significance of individual 231 coefficients, p-values were derived from Wald χ2 statistics. For all models, we plotted a receiver 232 operating characteristic curve (ROC) and computed a measure of the accuracy of the chosen 233 model in predicting the outcome using the area under the curve (AUC). As an alternative way 234 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 10 of selecting covariates for the multivariable models, and to improve model prediction accuracy, 235 we also performed LASSO regression using the “glmnet” package in R [21]. As the range and 236 scale of variables can influence the penalization for having too many variables in elastic net 237 models, all ordinal variables were coded numerically and re-classed as continuous, and all 238 continuous variables were centered and standardized. Nominal categorical variables were 239 broken out into individual binary dummy variables for all response levels except for the 240

level. 241 242 Missing data 243 The analysis of complete cases only can introduce bias and lead to a substantial reduction of 244 statistical power [22], especially if it is plausible that the data are missing at random or not 245 completely at random. An evaluation of the missing data suggested that multiple imputation 246 was advisable (SI6). The average proportion of missing values across all variables in the dataset 247 was 3.8%, which was mostly accounted for by the variable BMI (38% of missing data, SI6). To 248 handle missing data, we used a multiple imputation approach using the R package ‘missRanger’ 249 [23], which combines random forest imputation with predictive mean matching [23]. Prior to 250 all analyses, we imputed 5 datasets, with a maximum of 10 iterations specified for each 251 imputation. Each imputation was also weighted by the degree of missing data for each 252 participant, such that the contribution of data from participants with higher proportions of 253 missingness was weighted down in the imputation. We set the maximum number of trees for 254 the random forest to 200 but left all other random forest hyperparameters at their default. The 255 average out-of-bag (OOB) error rate for multiple imputation across all imputed datasets was 256 0.08 in women (range: 0 to 0.77) and 0.08 in men (range: 0 to 0.69). Parameter estimates for 257 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 11 all five datasets were pooled to provide more accurate estimates. A sensitivity analysis was also 258 performed on the complete cases without missing data imputation (n=1,548 (SI7)). 259 260 Text analysis 261 We first built a custom text cleaning function using the ‘textclean’ [24] and ‘ tidytext’ [25] R 262 packages to analyse the text written by participants selecting the “Other” category in the 263 outcome variable (n=574). The resulting corpus was tokenized (broken into individual units) 264 and lemmatized (words derived from others, such as “vaccine” and “vaccination” were grouped 265 by their stem version “vaccine” (SI8). The corpus was analysed to answer the following 3 266 questions: (i) which single words (unigrams) and pairs of adjacent words (bigrams) are most 267 frequent? (ii) which words co-occur in the same sentence? (iii) Are there clusters of symptoms? 268 To investigate the commonality of words, we explored the frequency of unigrams and bigrams 269 within all responses. We performed a correlation analysis on the most important words for 270 menstrual cycle descriptions to measure the association between words using the correlation 271 index (phi coefficient (φ)). To explore patterns of symptoms we examined the words that 272 commonly occur together (though not necessarily adjacent) to visualize groups of words that 273 cluster together. Clusters were visualized by arranging correlated words into a combination of 274 connected nodes (network graph) using the ‘igraph’ package [26]. 275 276

277 278 279 Out of the 26,710 individuals who completed the survey, 8,539 (32%) reported having been 280 vaccinated, with either 1 (n=7,270) or 2 doses (n=1,269). In the final sample, we only included 281 individuals living in the UK who knew about their vaccination status, who had a period in the 282 last 12 months and who were also pre-menopausal and not pregnant. We also excluded 283 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 12 participants who selected “Other changes” and contributed text to the effect of “too early to 284 say” when describing menstrual disturbances following COVID-19 vaccination (n=369, 64% 285 of those selecting the answer “Other changes)” (Fig. 1) 286 287 288 289 290 291 Figure 1. Flowchart of the study population selection 292 293 The final sample size of vaccinated individuals is 4,989, of which 53% received the Oxford-294 AstraZeneca and 47% the Pfizer BioNTech vaccine (Table 1). The median age is 35 (IQR: 28 295 to 43) years old, with most participants living in England (81%), self-reporting as white (95%) 296 and self-identifying as women (99%). We then grouped categories for the variables gender 297 (women vs. other) and ethnic group (white vs. other). Although the UK vaccination campaign 298 targeted older and at-risk populations to begin with, there does not seem to be an over-299 representation of over 40-year-olds. Note that 54% of participants had no deliveries and 49% 300 had a university or college degree. 301 302 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 13 Characteristic N = 4,989 Age, Median (IQR) 35 (28 – 43) Education level, n (%) Higher or secondary or further education (A-levels, BTEC, Baccalaureate) 851 (17) Primary & Secondary 303 (6.2) Post-graduate degree 1,324 (27) College or University 2,395 (49) Unknown 116 Place of residence, n (%) UK-England 4,031 (81) UK-Northern Ireland 159 (3.2) UK-Scotland 542 (11) UK-Wales 257 (5.2) Ethnic group, n (%) White 4,734 (95) Asian 113 (2.3) Black 21 (0.4) Mixed 101 (2.0) Other 18 (0.4) Unknown 2 Net income before pandemic, n (%) Between £13,682 and £22,140 656 (15) Between £22,140 and £29,254 614 (14) Between £29,254 and £39,397 795 (18) Between £39,397 and £76,144 1,453 (33) Less than £13,682 430 (9.8) More than £76,144 427 (9.8) Unknown 614 Smoking status before pandemic, n (%) I have never smoked 3,327 (67) No, but I have smoked in the past 1,157 (23) Yes, I usually smoked fewer than 10 cigarettes/day 334 (6.7) Yes, I usually smoked more than 10 cigarettes/day 170 (3.4) Unknown 1 Marital status, n (%) Separated 348 (7.2) Married/partnered 2,033 (42) Nevermarried/partnered 2,449 (50) Widowed 27 (0.6) Unknown 132 Gender, n (%) Man 1 (<0.1) Non Binary 24 (0.5) Other (please state) 22 (0.4) Woman 4,939 (99) Unknown 3 Number of deliveries, n (%) 0 2,694 (54) 1 693 (14) 2 1,017 (20) 3+ 584 (12) Unknown 1 Contraceptive use at the time of the survey, n (%) Combined estradiol-progestin 441 (11) Copper IUD 225 (5.4) None 2,421 (58) Other 84 (2.0) Progestin only 854 (21) Sterilization 130 (3.1) Unknown 834 COVID status (type), n (%) COVID - 3,377 (75) Long COVID 462 (10) Short COVID 687 (15) Unknown 463 COVID status (diagnosis), n (%) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 14 Characteristic N = 4,989 Negative 3,377 (76) Self diagnosed + 395 (8.9) Tested + 671 (15) Unknown 546 Number of vaccination shots, n (%) Yes, one shot 4,096 (82) Yes, two shots 893 (18) Vaccine type, n (%) Oxford-AstraZeneca 2,600 (53) Pfizer-BioNTech 2,335 (47) Unknown 54 Timing of 1st dose, n (%) Before 2021 331 (6.7) February 2021 1,469 (30) January 2021 1,497 (30) March 2021 1,659 (33) Unknown 33 Table 1. Summary of the sample characteristics 303 304 305 Risk factors for COVID-19 vaccine-related changes in menstrual cycles 306 Most individuals reported no changes to their menstrual cycles following COVID-19 307 vaccination (80%). Only 6.1% reported more disruption, 1.5% reported less disruption and 308 11.5% reported “Other changes”, which, based on the previous questions participants were 309 exposed to, could be interpreted as any changes in cycle length and regularity, period duration 310 and volume of menstrual bleeding as well as premenstrual symptoms. 311 312 The univariable analyses show that the odds of reporting any changes to menstrual cycles after 313 COVID-19 vaccination is associated with contraceptive type, smoking behaviour, COVID 314 status and menstrual cycle changes over the last year (Fig. 2). All univariable models offered 315 poor discriminative utility (AUC below 0.65, SI9). There were no differences associated with 316 age, body mass index, ethnic group, gender, marital status, physical activity, income, education, 317 place of residence, cycle length, period length, irregular cycles, heavy bleeding, vaccine type, 318 vaccine timing, parity, life satisfaction changes, medication use, use of vitamins/supplements, 319 endometriosis, PCOS, thyroid disease, uterine polyps, uterine fibroids, inter cystitis and eating 320 disorders (Fig. 2; SI10). 321 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 15 322 Figure 2. Outputs of univariable models for the odds of reporting any menstrual cycle changes 323 following COVID-19 vaccination. The figure depicts odds-ratio and 99%CI for 33 variables. **: 324 FDR P-value < 0.01; *** FDR P-value < 0.001. 325 326 The multivariable analyses show that the usage of combined oral contraceptives is associated 327 with lower odds of reporting any changes by 48% (OR=0.52, 95CI=[0.34 to 0.78], P<0.001) 328 while the odds of reporting any changes is increased by 44% (OR=1.44, 95CI=[1.07 to 1.94] 329 for current smokers, P<0.01) and by 49 to 70% for individuals with a positive COVID status 330 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 16 [Long Covid (OR=1.61, 95CI=[1.28 to 2.02], P<0.001), acute COVID (OR=1.54; 95CI=[1.27 331 to 1.86], P<0.001); self-diagnosed positive (OR=1.70, 95CI=[1.34 to 2.16], P<0.001), tested 332 positive (OR=1.49, 95CI=[1.20 to 1.84], P<0.01), Figs 3 & 4, SI11]. The effects remain after 333 adjusting for self-reported overall magnitude of menstrual cycle changes over the year 334 preceding the interview (pandemic-related changes in menstrual cycle (PRCM)), which is 335 positively associated with the risk of reporting any changes (OR=1.16, 95CI=[1.06 to 1.26], 336 P<0.01). The findings were replicated when using complete cases data (SI7), indicating that the 337

are not an artefact of the missing data imputation process. 338 339 340 Figure 3. Outputs of multivariable models for the odds of reporting any menstrual cycle changes 341 following COVID-19 vaccination. Each of the 5 exposures associated with the outcome at FDR-342 adjusted P<0.05 in the univariable analysis (i.e., pandemic-related menstrual changes (PRMC), 343 contraceptive use, COVID-19 type, COVID-19 diagnosis, smoking behaviour) was entered in a 344 multivariable model together with potential confounding (but not mediating) effects where appropriate 345 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 17 (see SI5 for DAGs). Model I: Smoking behaviour; Model II: Contraceptive use; Model III: COVID-19 346 type adjusted for contraceptive use and smoking behaviour; Model IV: COVID-19 diagnosis adjusted 347 for contraceptive use and smoking behaviour; Model V: PRMC adjusted for COVID-19 type; Model VI: 348 PRMC adjusted for COVID-19 diagnosis. 349 350 351 352 353 Figure 4. Predicted probability of reporting any menstrual changes following COVID-19 354 vaccination. Predicted values and 95 confidence intervals given contraceptive use, COVID status 355 (based on type and certainty of diagnosis) and menstrual cycle changes over the last year. Most 356 individuals (80%) reported no menstrual disturbances following COVID-19 vaccination. This 357 probability was lower for users of combined (including oestradiol) contraceptives and higher for current 358 smokers and those who had had a positive COVID status. 359 360 The type of contraceptive used and the history of COVID infection, while correlated, did not 361 offer good predictive value for whether an individual will report changes to their menstrual 362 cycle. Each exposure alone contributed an increase of only 1 to 3% of explained variance. The 363 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 18 AUCs for the multivariate models were low across the imputed datasets (0.57 to 0.61) and the 364 complete case dataset (0.63): the variables considered are not sufficient for predicting 365 accurately whether an individual will report menstrual changes after vaccination. To improve 366 the prediction accuracy of our models, we also performed a LASSO regression considering all 367 33 variables, but no improvement in AUC was obtained (SI12), suggesting that key variables 368 are missing from our dataset and/or that the subjective outcome is not defined specifically 369 enough for accurate prediction, especially if experiences of menstrual changes are diverse. 370 371 Description of menstrual cycle changes following COVID-19 vaccination 372 Most common changes reported. The analysis of text written by participants who selected 373 “Other changes” (n= 574, 57% of those reporting any changes) rather than “MORE disruption” 374 or “LESS disruption” showed concerns over cycle length and menstrual bleeding patterns. The 375 most common unigrams (individual words) were “late”, “bleed”, “early”, “long”, "heavy”, 376 “spotting”, “short”, “pain” and “stop” and the most common bigrams (pairs of adjacent words) 377 were “day late”, “period start”, “heavy bleed”, and “late period” (Fig. 5). While many reported 378 menstrual cycle changes that entailed heavier bleeding/period, there was no one single pattern 379 of symptoms, with changes including both early and late period, and diverse experiences 380 reported (from “miss period” to “heavy bleed”). 381 382 383 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 19 384 Figure 5. Most common words used to describe menstrual cycle changes following COVID-19 385 vaccination (n = 574). (A) Most common words. (B) Most common pairs of adjacent words. 386 387 388 Associations between symptoms. Only a few symptoms are correlated (φ 0.2). 389 “Cramps” positively correlate with “pain” and “heavy” and “bleed” negatively correlates with 390 “late”. Further, “lighter” positively correlates with “normal”, as participants report that “period 391 was two days late, and lighter than normal ”. However, “lighter” and “late” do not co-occur 392 more than expected by chance (Fig. 6). 393 394 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 20 395 Figure 6. Correlation matrix between key words within sentences describing menstrual cycle 396 changes following COVID-19 vaccination . The size and colour of the dots indicates the strength of 397 the correlation (phi coefficient) between words. 398 399 Clusters of words . Different clusters of symptoms emerge from the text, such as irregular 400 periods, heavy cramps, and pain. However, the “pain” cluster encompassed many words that 401 are weakly correlated, suggesting a diversity of pain experience. There was also some 402 uncertainty regarding which changes do occur, with participants finding it “ hard to say if the 403 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 21 irregular periods are still due to covid or the vaccination”. When only correlations >0.20 were 404 considered (Fig. 7), 4 clusters emerged: “heavy, painful, cramps”, “irregular, disruption”, “lot, 405 clot”, and an experiential cluster “symptom, experience, pain, increase, feel”. Notably, various 406 pain experiences that do not directly relate to menstrual cramps were reported in the main text, 407 including stomach pain and headache. 408 409 410 411 Figure 7. Network of words describing menstrual cycle changes following vaccination with 412 COVID-19. Words have been lemmatised to the root of their words, for example “light” can represent 413 both “lighter” and “light. Node size represents degree centrality (the commonality of words, only words 414 with more than 5 occurrences are included). Edge thickness is a measure of correlation between words. 415 416 417 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 22

418 419 420 Using data collected in the UK prior to widespread media attention to menstrual disturbances 421 following COVID-19 vaccination, this study found that among pre-menopausal vaccinated 422 individuals who menstruated in the 12 months preceding the survey, 20% reported any changes 423 to their menstrual cycles up to 4 months after receiving their first injection. In this sample, there 424 was an association between a history of COVID infection and an increased relative risk of 425 reporting changes of menstrual cycles following vaccination against COVID-19, independently 426 of how COVID status was determined, i.e., using COVID type (Acute vs. Long Covid) or 427 certainty of diagnosis (tested vs. self-diagnosed positive). This study also found that using 428 contraceptives containing oestradiol (e.g., the pill, the vaginal ring, and the patch) is associated 429 with a 50% lower odds of reporting menstrual cycle changes post-COVID-19 vaccination. 430 Beyond smoking, none of the other variables investigated including age, BMI, socio-economic 431 status, or vaccine type were associated with post-vaccination menstrual disturbances. 432 Descriptive accounts point to diverse menstrual disturbances including “late” and “early” 433 periods as well as “heavy bleeding” (Box 1). 434 435 Meaning of the study: Most menstruating people in our sample did not experience menstrual 436 changes following COVID-19 vaccination. This provides reassuring data when counselling 437 reproductive-aged women about COVID-19 vaccination and menstrual changes. However, one 438 in five did report menstrual disturbance following COVID-19 vaccination, a proportion that is 439 above the threshold for a "very common" adverse reaction according to international 440 pharmacovigilance standards. Clinicians should consider counselling women about these 441 possible menstrual effects following COVID-19 vaccination, while emphasising the need to 442 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 23 seek medical advice if they are severe, last more than one cycle or involve "red flag" symptoms 443 such as inter-menstrual bleeding, post-coital bleeding, or post-menopausal bleeding. This study 444 also suggests that current smoking and having had COVID-19 may make one more likely to 445 experience menstrual disturbance following COVID-19 vaccination and that those on the 446 COCP are less likely to experience menstrual disturbance. Knowledge of risk factors may help 447 tailor advice to individuals who menstruate prior to COVID-19 vaccination. 448 449 Strengths and weaknesses of the study: The analysis is drawing upon a survey not specifically 450 designed to investigate the impact of COVID-19 vaccination on menstruation. It is retrospective 451 in nature as well as sensitive to selection, recall and report biases and does not systematically 452 assess the full spectrum of menstrual disturbance defined by the International Federation of 453 Gynecology and Obstetrics Abnormal Uterine Bleeding System 1 [27]. We took several steps 454 to limit selection bias during sampling (see methods) and the initial survey is broadly 455 representative of people infected with COVID (8.9% with a positive PCR test compared to a 456 national proportion of 6.6% at the time [28]). However, approximately 45% of the sample had 457 received at least one dose of the vaccine, as compared to the national proportion of 59% by the 458 time of the last survey entry [29]. In addition, menstrual changes may manifest later, and our 459 study does not have the time depth to evaluate this possibility. However, among the studies of 460 other vaccines conducted on a longer timescale, no effect was found by 6-9 months [14,30]. 461 462 Strengths and weaknesses of the study in relation to other studies: While the survey is also 463 sensitive to recall bias, it is limited as compared to more recent surveys [8] as the issue of 464 menstrual disturbances was not reported by the British Broadcasting Corporation until May 13, 465 2021 [31], as compared to a flurry of attention in US media throughout April [1–3]. 466 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 24 Reassuringly, reporting bias would be expected to affect all sections of the sample similarly, 467 and thus it would not explain specific associations such as with contraceptive type. 468 469 Unanswered questions and future research: The association between a history of SARS-470 CoV-2 infection and menstrual disturbances post-vaccination in this study may be partly due to 471 the effect of prior infection with SARS-CoV-2 on the immune response to vaccination, which 472 has been found to be heightened [32]. Biological data would be needed to verify this hypothesis. 473 The findings also suggest that exogenous oestrogen may reduce post-vaccination menstrual 474 disturbances through anti-inflammatory or anti-viral effects. This is consistent with the recent 475 suggestion that an “inflammatory” rather than an “ovulatory” route might explain menstrual 476 disturbances following COVID-19 vaccination given the high prevalence of breakthrough 477 bleeding among users of long-acting reversible contraceptives (LARC) [8]. A protective effect 478 of oestrogen [33]. and oestradiol [34] has been suggested in relation to the severity of COVID-479 19, and randomized control trials on unbiased samples would be needed to establish causality 480 between oestrogen and the reduced risk of menstrual disturbances following COVID-19 481 vaccination. Finally, the diversity of menstrual responses to COVID-19 vaccination might be 482 partly explained by the timing of vaccination in relation to the menstrual cycle. The findings 483 thus call for routine menstrual data collection in COVID-19 and vaccination studies as well as 484 research into the mechanisms of menstrual disturbance following vaccination. 485 486 487 488 489 490 491 492 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 25 493 494

495 1. McShane J. Can the vaccine make your period worse? These women say yes. The Lily. 496 2021. 497 2. Saar T. Women Say COVID Vaccine Side Effects Impact Their Periods, So Why Don’t 498 Doctors Care? Haaretz. 499 3. Efrati I. Some Vaccinated Israeli Women Report Irregular Menstrual Cycles, Bleeding. 500 Haaretz. 2012. 501 4. Butler K. Social Media Influencers Are Spreading Wild Rumors About COVID-19 502 Vaccines and Periods. In: motherjones.com [Internet]. Available: 503 https://www.motherjones.com/politics/2021/04/social-media-influencers-are-504 spreading-wild-rumors-about-covid-19-vaccines-and-periods/ 505 5. Shimabukuro TT, Kim SY, Myers TR, Moro PL, Oduyebo T, Panagiotakopoulos L, et 506 al. Preliminary Findings of mRNA Covid-19 Vaccine Safety in Pregnant Persons. N 507 Engl J Med. 2021;384: 2273–2282. doi:10.1056/NEJMoa2104983 508 6. Joi P. Do COVID-19 vaccines affect menstruation and fertility? In: Gavi, the Vaccine 509 Alliance [Internet]. 2021. Available: https://www.gavi.org/vaccineswork/do-covid-19-510 vaccines-affect-menstruation-and-fertility 511 7. Shanes ED, Otero S, Mithal LB, Mupanomunda CA, Miller ES, Goldstein JA. Severe 512 Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Vaccination in Pregnancy. 513 Obstet Gynecol. 2021;Publish Ah. doi:10.1097/AOG.0000000000004457 514 8. Lee KMN, Junkins EJ, Fatima UA, Cox ML, Clancy KBH. Characterizing menstrual 515 bleeding changes occurring after SARS-CoV-2 vaccination. medRxiv. 2021; 516 2021.10.11.21264863. doi:10.1101/2021.10.11.21264863 517 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 26 9. Gunter J. The COVID-19 vaccine and menstrual irregularities. In: The Vajenda 518 [Internet]. 2021. Available: https://vajenda.substack.com/p/the-covid-19-vaccine-and-519 menstrual 520 10. Sharp G, Fraser A, Sawyer G, Kountourides G, Easey K, Ford G, et al. The COVID-19 521 pandemic and the menstrual cycle: research gaps and opportunities. Int J Epidemiol. 522 2021;In press. doi:10.31219/osf.io/fxygt 523 11. ACOG Committee Opinion No. 651: Menstruation in Girls and Adolescents: Using the 524 Menstrual Cycle as a Vital Sign. Obstet Gynecol. 2015;126: e143-6. 525 12. Li K, Urteaga I, Wiggins CH, Druet A, Shea A, Vitzthum VJ, et al. Characterizing 526 physiological and symptomatic variation in menstrual cycles using self-tracked mobile-527 health data. npj Digit Med. 2020;3: 79. doi:10.1038/s41746-020-0269-8 528 13. Klein SL, Flanagan KL. Sex differences in immune responses. Nat Rev Immunol. 529 2016;16: 626–638. doi:10.1038/nri.2016.90 530 14. Lamb AR. Experiences with prophylactic Typhoid vaccination. Arch Intern Med. 531 1913;XII: 565. doi:10.1001/archinte.1913.00070050082008 532 15. Shingu T, Uchida T, Nishi M, Hayashida K, Kashiwagi S, Hayashi J, et al. Menstrual 533 Abnormalities after Hepatitis B Vaccine. Kurume Med J. 1982;29: 123–125. 534 16. Suzuki S, Hosono A. No association between HPV vaccine and reported post-535 vaccination symptoms in Japanese young women: Results of the Nagoya study. 536 Papillomavirus Res. 2018. doi:10.1016/j.pvr.2018.02.002 537 17. Gong L, Ji H-H, Tang X-W, Pan L-Y, Chen X, Jia Y-T. Human papillomavirus 538 vaccine-associated premature ovarian insufficiency and related adverse events: data 539 mining of Vaccine Adverse event Reporting System. 123AD. doi:10.1038/s41598-020-540 67668-1 541 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 27 18. Products M& H, Agency R. Coronavirus vaccine - weekly summary of Yellow Card 542 reporting (1st July 2021). 2021. Available: 543 https://www.gov.uk/government/publications/coronavirus-covid-19-vaccine-adverse-544 reactions/coronavirus-vaccine-summary-of-yellow-card-reporting 545 19. Male V. Menstrual changes after covid-19 vaccination. BMJ. 2021; n2211. 546 doi:10.1136/bmj.n2211 547 20. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and 548 Powerful Approach to Multiple Testing. J R Stat Soc Ser B. 1995. doi:10.1111/j.2517-549 6161.1995.tb02031.x 550 21. Friedman J, Hastie T, Tibshirani R. Regularization paths for generalized linear models 551 via coordinate descent. J Stat Softw. 2010. doi:10.18637/jss.v033.i01 552 22. Sterne JAC, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple 553 imputation for missing data in epidemiological and clinical research: potential and 554 pitfalls. BMJ. 2009;338: b2393–b2393. doi:10.1136/bmj.b2393 555 23. Mayer M. missRanger: Fast Imputation of Missing Values. 556 24. Rinker TW. textclean: Text Cleaning Tools version 0.9.3. Buffalo, New York; 2018. 557 Available: https://github.com/trinker/textclean 558 25. Silge J, Robinson D. tidytext: Text Mining and Analysis Using Tidy Data Principles in 559 R. J Open Source Softw. 2016;1: 37. doi:10.21105/joss.00037 560 26. Csardi G, Nepusz T. The Igraph Software Package for Complex Network Research. 561 InterJournal, Complex Syst. 2006;1695: 38. 562 27. Munro MG, Critchley HOD, Fraser IS. The two FIGO systems for normal and 563 abnormal uterine bleeding symptoms and classification of causes of abnormal uterine 564 bleeding in the reproductive years: 2018 revisions. Int J Gynecol Obstet. 2018;143: 565 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 28 393–408. doi:10.1002/ijgo.12666 566 28. Government U. Coronavirus Cases in the United Kingdom. Available: 567 https://coronavirus.data.gov.uk/details/cases 568 29. UK Health Security Agency. Coronavirus (COVID-19) in the UK: Vaccinations in 569 United Kingdom. Available: https://coronavirus.data.gov.uk/details/vaccinations 570 30. Shahani S, Patel KL, Merchant P. Evaluation of endocrine parameters in clinical trials 571 with β-hCG vaccine. Contraception. 1991;43: 67–75. doi:10.1016/0010-572 7824(91)90127-2 573 31. Robinson O, Schraer R. Covid vaccine: Period changes could be a short-term side 574 effect. BBC. 2021. Available: https://www.bbc.com/news/health-56901353 575 32. Kelsen SG, Braverman AS, Patel P, Aksoy MO, Hayman J, Rajput C, et al. Heightened 576 COVID-19 Vaccine Response Following SARS-CoV-2 Infection. medRxiv. 2021; 577 2021.03.18.21253845. doi:10.1101/2021.03.18.21253845 578 33. Haitao T, Vermunt J V., Abeykoon J, Ghamrawi R, Gunaratne M, Jayachandran M, et 579 al. COVID-19 and Sex Differences. Mayo Clin Proc. 2020;95: 2189–2203. 580 doi:10.1016/j.mayocp.2020.07.024 581 34. Costeira R, Lee KA, Murray B, Christiansen C, Castillo-Fernandez J, Lochlainn MN, 582 et al. Estrogen and COVID-19 symptoms: Associations in women from the COVID 583 Symptom Study. PLoS One. 2021. doi:10.1371/journal.pone.0257051 584 585 586 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 29 Box 1 587 What is already known on this topic? 588  Menstrual disturbances including changes in frequency and/or dysmenorrhoea 589 following vaccination have been reported as early as 1913 for the typhoid vaccine (1). 590 Since then there have only been a few studies investigating this topic, using small 591 sample sizes (hepatitis vaccine (2)) or reporting mixed results (HPV vaccine (3,4)). 592  The UK’s Medicine and Healthcare products Regulatory Agency (MHRA) is closely 593 monitoring reports of menstrual disorders, with more than 30,000 reports made to its 594 yellow card surveillance scheme by 2 September 2021 following vaccination with both 595 mRNA and adenovirus-vectored COVID-19 vaccines (5). 596  In a recent preprint of a retrospective case-control study of 21,380 pre-menopausal 597 participants living in the US, 45.8% of 9,579 people with regular menstrual cycles 598 experienced heavier bleeding after COVID-19 vaccination. In addition, 70.5% of 1,545 599 non-menstruating people using long-acting reversible contraceptives (LARC) 600 experienced breakthrough bleeding after COVID-19 vaccination (6). This informative 601 study may be affected by selection bias and may not be generalisable. 602 603 What this study adds 604  In a large sample of participants vaccinated against COVID-19 surveyed in the UK 605 before widespread media attention to related menstrual changes, the prevalence of 606 menstrual changes was 1 in 5. 607  Out of 33 socio-demographic, health, vaccine, COVID- and pandemic-related and 608 reproductive variables, the odds of reporting any menstrual changes following COVID-609 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 30 19 vaccination were associated with a history of SARS-CoV-2 infection, smoking 610 behaviour and the type of contraceptives used. 611  Menstrual changes that were reported were diverse, ranging from increased bleeding to 612 the cessation of bleeding. 613  The study highlights the need for greater consideration of the menstrual cycle in health 614 interventions. 615 616 617 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint 31 Supporting Information Caption 618 619 SI1: Recruitment facebook ads 620 SI2 : Information sheet 621 SI3 : Survey questions 622 SI4 : Operationalization of variables 623 SI5 : DAG 624 SI6: Missing data evaluation 625 SI7: Complete cases analysis 626 SI8 : Text analysis 627 SI9 : AUC univariable models 628 SI10: Table univariable models 629 SI11 : Table multivariable models 630 SI12: AUC Lasso 631 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 6, 2021. ; https://doi.org/10.1101/2021.11.23.21266709doi: medRxiv preprint