A Prospective Study of the Relationship of COVID-19 Vaccination to Menstrual Cycle Characteristics in Adolescent Girls.

Booster group participants (n=47) consisted of participants who had received an initial series of any COVID-19 vaccine at least 6 months prior to receiving their booster dose. Control group participants (n=18) included those who had never received any COVID-19 vaccine doses (n=2, 11.1% of group), those who had received a complete initial series at least six months prior to participating but had not received a booster at the time of participation (n=4, 22.2% of group), and those who had received a complete initial series and one booster at least six months prior to participating (n=12, 66.7% of group). The mean age of the total sample was 17.3 years (SD = 1.94), with no significant differences (two-sample t -test p-value = 0.24) between the booster (M = 17.1; SD = 1.98) and control (M = 17.8, SD = 1.78) groups. The majority of the total sample identified as White, and there was no significant difference in the proportion identifying as White versus another race (chi-square test I-value = 0.42) between booster (55.3%) and control (66.7%) groups. The mean age at menarche of the total sample was 12.2 years (SD = 1.32), with no significant differences (two-sample t -test p -value = 0.11) between the booster (M = 12.1, SD = 1.09) and control (M = 12.7, SD = 1.75) groups. All participants received an mRNA vaccine, with the majority of participants receiving Pfizer (n=37, 78.7%) and others receiving Moderna (n=10, 21.3%). Note that since the sample size of our study is relatively small, the tests we performed for the demographic characteristics might be under-powered. The p-values presented here were mainly intended as a summary of the level of uncertainty in the estimated between-group differences. The marginal means of cycle length pre- and post-booster for each of the four subject groups, defined by cycle phase at vaccination (follicular vs. luteal) and booster status (booster group vs. control), range from 28 days (luteal and booster group, pre-booster) to 34.1 days (follicular and booster group, pre-booster). See Supplement Table 1 for additional details on cycle length means and standard deviations. The results for menstrual cycle length based on the mixed models are summarized in Table 1 . Specifically, variables in Table 1 are covariates in a single mixed model, meaning that the regression coefficients are the effects of the corresponding variable adjusting for all other variables in the table. Post Booster, Cycle Phase and Control are dummy variables, and we used a saturated model where all their two-way and three-way interactions are included to allow for arbitrary mean cycle lengths in different subject groups defined by these three variables. Therefore, a particular contrast between two subject groups of interest can be tested using the corresponding regression coefficients in the model. For example, the main effect of Post Booster is the pre-post difference in cycle length when fixing Cycle Phase and Control at their reference levels, instead of the marginal difference in all patients regardless of their values of Cycle Phase and Control. The post-booster difference in cycle length between control and booster group when fixing the Cycle Phase at its reference is estimated by combining the main effect of Control and the interaction Control×Post Booster. The average baseline cycle length (pre-booster) is 4.86 days (CI: [−8.47, −1.24], p-value = 0.01) shorter for subjects who received their booster at the luteal phase compared to those who received their booster during the follicular phase, so cycle phase was included as a variable in all analytical models. When including all pre- and post-booster cycles in the GLMM (see Table 1 ), we identified a significant pre-post difference of cycle length for those in the booster group, with post-booster cycles an average of 5.35 days (CI: [−10.24, −0.47], p-value = 0.03) shorter than the pre-booster cycles. Additionally, we found that the baseline PSE score had a negative association with cycle length (CI: [−2.73, −0.13], p-value = 0.03) adjusted for vaccination status, cycle phase, and pre-post booster status. None of the interaction terms were statistically significant. This suggests the menstrual cycle phase in which a subject received the booster might have an impact on the post-booster menstrual cycle characteristics. We also did a separate analysis with each individual post-booster cycle included in the GLMM to pinpoint the drivers of the pre-post differences we identified in the overall analysis (see Table 2 ). Results suggest the most significant change is observed for Cycle B+2, where the change is estimated to be −8.13 days (CI: [−15.52, −0.73], p-value = 0.03), indicating that the overall shortening of cycle lengths after booster is mainly attributable to the change in cycle length observed two cycles after receiving the booster. To illustrate the results of the pre-post comparisons in different subgroups (booster vs. control and follicular vs. luteal phase), we also visualized the estimated marginal means of cycle length for four subgroups pre- and post-booster in Figure 3 . The only statistically significant pre-post comparison is in subjects who received boosters in the follicular, such that receiving the booster in the follicular phase was associated with significantly shorter mean post-booster cycle length (reduction in mean cycle length = 2.487, CI: [0.488,4.486], p-value = 0.0157). We did not find any significant pre-post changes for the other three cycle-related measures (menstrual pain, menstrual flow, and menstrual symptoms; see Supplement Tables 2 – 4 ). Of note, there is a significant association between PSE score and menstrual symptoms in that one unit increase in PSE scores are associated with 9.21 (CI: [4.21, 14.21], p-value < 0.001) increase in the self-reported menstrual symptom scores.

The study began enrollment on 11/8/2021 and data collection was completed on 12/1/2022. Potential participants were recruited from social media advertisements, research recruitment website, and through other ongoing studies in the United States. Individuals who expressed interest in the study were screened by telephone to assess eligibility and to obtain cycle dates for the previous three menstrual cycles. Eligible participants completed the online informed consent/assent/parent permission survey. A Research Electronic Data Capture (REDCap) 13 , 14 survey link was sent via email and completed within 24 hours prior to the booster dose appointment, such that this survey was completed prospectively and before any booster vaccination was received. Research team members later confirmed receipt of the vaccine by self-report with booster group participants within one day of its scheduled administration. Control group participants were those who had not received a COVID-19 vaccine for at least 6 months and were not planning to receive a vaccine during study participation. We used a 6-month cutoff to define the control group because of an existing body of evidence that demonstrates that neutralizing antibodies wane significantly 6 months post COVID-19 vaccination (for a review, see Lapuente et al. 15 ). Questionnaire links for completion of measures online were sent monthly for three months following the date of initial survey completion. Participants were required to complete the survey within 4 weeks. Study procedures were identical for the booster and control groups, with the control group being assigned a designated artificial “booster day” during which data collection began, which allowed us to easily compare the two groups in terms of specific study timepoints. See Figure 1 for a timeline of study participation. The sequence of the menstrual cycles was identified with Cycle B reflecting the menstrual cycle in which the booster vaccine was received, Cycles B−1, B−2, and B−3 reflecting the three cycles prior to the booster vaccine, and Cycles B+1, B+3, and B+3 reflecting the three cycles after the booster vaccine. Participants were compensated up to $100 for completion of the entire study, and the study was approved by the Mass General Brigham IRB (protocol 2021P002543). Inclusion criteria were: 1) females ages 13–20 years, which was generally considered an adolescent population based on the World Health Organization criteria for adolescents (ages 13–19) 16 ; 2) self-reported “regular” menstrual cycle averaging 22–35 days; 3) menarche at least 12 months prior to participation and having regular cycles for at least 6 months; 4) ability to provide verbal informed consent or assent, and parental permission for minors (under 18 years); 5) able to read and understand English; and 6) access to a smartphone or computer with video conferencing capacity. Exclusion criteria were: 1) use of oral contraceptives or any exogenous hormones in the 3 months prior to participation; 2) presence of factors indicative of secondary dysmenorrhea (e.g., self-reported diagnosis of endometriosis, presence of persistent pelvic pain throughout the month); 3) diagnosis of chronic pain condition; 4) currently pregnant or breastfeeding; 5) history of pelvic inflammatory disease or sexually transmitted disease; 6) developmental delay, diagnosis of autism, or significant cognitive impairment that may preclude understanding of study procedures. Five hundred and two individuals expressed interest in participating. Seventy-five individuals were enrolled; of these, 10 (13.3% of enrollees) were withdrawn for non-compliance or failure to receive a vaccine before study closure (see Figure 2 for flow diagram of participant enrollment). Enrolled participants were recruited from previous study enrollees (37.3%; the study included a community sample of adolescents with regular menstrual cycles and no symptoms associated with secondary dysmenorrhea), social media advertisements (24.0%), research recruitment website (20.0%), word of mouth/referral (8.0%), previous studies (8.0%), and other/unknown (2.7%). Demographic variables (e.g., date of birth, race, and ethnicity) and menstrual history variables (age at menarche, dates of the three most recent menstrual periods, etc.) were assessed using an instrument designed for this study. Menstrual pain without taking medication for treatment was assessed using an 11-point numeric rating scale (NRS) from 0 (no pain) to 10 (worst pain possible). 17 , 18 Participants were asked to rate their average level of menstrual pain prior to vaccination and for each subsequent period. Usual level of menstrual flow was assessed by a categorical question with 3 answer choices (light, moderate, and heavy). Menstrual flow questions completed for periods occurring after Cycle B (see Figure 1 ) focused on perceived change in menstrual flow and was assessed using a 5-point categorical variable from “a lot lighter” to “a lot heavier” than usual. Assessment of menstrual symptoms included fourteen items rating the severity of various menstrual symptoms (e.g., bloating, nausea, low back pain, etc.) over the prior six months. 19 , 20 The severity of each symptom was rated on an NRS from 0 (not present) to 10 (extremely severe). The sum of these items ranges from 0 to 140 and was used as the measure of menstrual symptom severity for the current analyses. Menstrual pain, menstrual flow, and menstrual symptoms were assessed at baseline and at each subsequent monthly survey. Menstrual cycle phase at the time of vaccination was determined by calculating the number of days elapsed between the date of booster administration and the first day of the following cycle (i.e., first day of menstrual bleeding). Participants for whom this value was 14 days or fewer were categorized as having received the booster during the luteal phase, others were categorized as follicular. For cycles 15 days and fewer, cycle phase was considered follicular due to the likelihood that the cycle was anovulatory. Stress was assessed 0–24 hours at baseline using the Patient-Reported Outcomes Measurement Information System (PROMIS) Pediatric Psychological Stress Experiences (PSE) scale, which is an 8-item self-report measure assessing dimensions of psychological stress (e.g., feeling overwhelmed, having trouble concentrating, etc.) over the prior 7 days. Items are rated on a 5-point Likert scale from “Never” to “Always.” Higher scores indicate higher levels of psychological stress. The PSE has demonstrated consistency, validity, and reliability in pediatric populations as young as 8 years old, including across varied socio-demographics and child health statuses. 21 , 22 We used generalized linear mixed models (GLMMs) to estimate the pre-post changes of the set of four menstrual cycle-related outcome measures (cycle length, menstrual pain, menstrual flow, and menstrual symptoms) based on the longitudinal data collected both pre- and post-vaccine, and their relationships with vaccination status (booster v. control) and cycle phase (follicular vs. luteal) at the time of the vaccine. For continuous outcomes (cycle length, menstrual pain, and menstrual symptoms), we used an identity link with normal distribution for the residuals in GLMM whereas a cumulative link was employed for the ordinal outcome (changes in menstrual flow after booster). In all models, we specified a subject-specific random intercept to capture the intra-subject correlations among the repeated measures and included baseline PSE score as a covariate. For all outcomes except for the menstrual flow variable, we analyzed booster and control groups jointly. We only analyzed data from the booster group for menstrual flow due to the high rate of missing data in control group. We standardized all predictors and added interactions between the covariates and the main predictors of interest (group and pre-post booster indicator), as long as the resulting design matrix was not rank-deficient. GLMMs for cycle length, menstrual pain, and menstrual symptoms were fitted using the R package lme4 and the GLMM for changes in menstrual flow after booster was fitted using a Bayesian procedure implemented in R package brms to ensure its numerical stability.

In this prospective study comparing menstrual cycle length and associated menstrual characteristics before and after the COVID-19 vaccine booster, we found that adolescent girls who received the booster reported significantly shorter menstrual cycles, compared to pre-vaccine cycles, while this effect was not observed in the control group. This change was primarily driven by shorter cycles in the second cycle post-booster, although baseline stress scores also contributed to shorter cycles. Additionally, receiving the vaccine in the follicular phase was associated with significantly shorter cycles post-vaccine, compared to pre-vaccine cycles. We found no evidence of changes in menstrual flow, menstrual pain, or menstrual symptoms, compared to pre-booster, in either the booster or the control group. Only baseline ratings of psychological stress were associated with report of menstrual symptoms, both pre- and post-booster, in both groups. These data provide additional support for previous findings demonstrating changes in the menstrual cycle after vaccination with the COVID-19 vaccine. However, the changes in cycle length observed in the present study are in the opposite direction as those found in adult women. This is not entirely surprising, as the HPO axis which controls menstrual cyclicity, is less mature in younger girls. 23 , 24 Previous vaccination may have impacted these findings, as our booster group had all had prior vaccine exposure. Our data also confirmed the impact of vaccination during the follicular phase on menstrual cycle length, even though the findings are different than what has been shown in adults. 5 , 7 Baseline (pre-booster) stress was also associated with shorter cycle length, regardless of vaccination status, suggesting the possible role of psychological and biological mechanisms involved in menstrual changes. Other menstrual characteristics (menstrual flow, pain, and symptoms) did not differ in pre- versus post-booster in either group. While a recent study found higher levels of pain and menstrual disturbances in adult women following first and second doses of the COVID-19 vaccine, 25 to our knowledge, only one study has examined menstrual pain, and those data did not show any differences in menstrual pain or menstrual symptoms in vaccinated women compared to women who were not vaccinated. 6 Data is accruing that an increase in menstrual bleeding may occur, with one study by Darney et al. 26 showing a small and temporary increase in the likelihood of more total bleeding quantity following the initial dose of the COVID-19 vaccine. Mechanisms involved in the observed menstrual cycle changes following COVID-19 vaccination remain unclear, although these data, along with previously published reports indicate that menstrual cycle phase at the time of vaccination is a mechanism involved in menstrual changes. These data implicate disruption of follicle stimulating hormone (FSH) as a key driver of these effects. Extant research has clearly documented the impact of estrogens on immune function 27 – 29 impact immune function that alters response to vaccines and increases risk for auto-immune disorders. 30 , 31 Yet, less is known about the impact of immune response on ovarian hormones. Stressors such as an intense immune and inflammatory response to an mRNA vaccine or psychological stress could potentially disrupt this delicate balance and thus lead to menstrual cycle disturbances. 32 To our knowledge, only limited data from animal models exist to support this notion. 33 , 34 Identifying these mechanisms is the clear next step to better understanding menstrual cycle disturbances following COVID-19 vaccination in girls and women and providing appropriate clinical recommendations for individual patients. Future research is greatly needed to 1) better understand the impact of COVID-19 vaccination on menstrual cyclicity and menstrual symptoms in subpopulations (e.g., adolescent girls using exogenous hormones, etc.) in order to improve counseling for patients and providers and 2) clarify potential mechanism involved in menstrual changes post-vaccination. While some have hypothesized that medications that elevate and stabilize estrogens may have a beneficial effect on immune function, 35 no research to date has specifically answered this question. Additionally, our finding that primarily the second post-booster cycle was significantly shorter than pre-booster cycles warrants further investigation. This suggests that the impact of the vaccine on menstruation may be somewhat delayed in the adolescent population, and the changes appear to be temporary, although we did not follow participants for longer than three post-booster cycles so these data must be interpreted with caution. The present study has several limitations. First, the sample size for the two study groups (booster and control) were small and included a relatively homogenous sample. Additionally, the findings from this study may not generalize to populations, including those with chronic health conditions or those using oral contraceptives. Second, because the timing of the study necessitated that we evaluate only responses to booster vaccines, the control group included girls who had received an initial vaccine series, and some who had received a booster. It is possible the initial doses or boosters impacted responses to the vaccinations in this study. Furthermore, we are not able to clearly determine who had been infected with the SARS-CoV-2 virus during the menstrual cycle tracking, and an infection in either of the groups may have impacted menstrual cycle characteristics in unknown ways. 9 , 36 We did not obtain height/weight of study participants, and it is possible that some individuals with extreme BMI scores may have altered menstrual cycle functioning. Additionally, stress was not measured throughout the post-booster timeframe, so we cannot draw conclusions about how changes in stress may have affected post-booster menstrual cycles. The measure of menstrual symptoms was summed for a total score, which does not take into account level of functioning or the use of analgesics. Also, due to small sample size in both groups, the operating characteristics (such as Type 1 error and power) of the model may be compromised. Finally, the mean age of study participants was 17, so it is possible that this sample of slightly older adolescents reflects individuals with a more mature HPO-axis, resulting in a greater likelihood of ovulatory cycles. However, the likelihood of anovulatory cycles is still higher in younger populations, which makes identification of cycle phase difficult if ovulation has not been confirmed with a serum progesterone test. Results may therefore be different in younger populations evaluated closer to the onset of menarche or if ovulation was confirmed. The COVID-19 vaccine booster was associated with shorter menstrual cycle length in adolescent girls, whereas no evidence of changes in cycle length was observed in adolescent girls who did not receive a booster. No evidence of changes in menstrual flow, menstrual pain, or menstrual symptoms was observed in either group.

In early 2021, online magazine reports of menstrual cycle changes (increased bleeding, longer/shorter menstrual cycles) following widespread administration of the Coronavirus Disease 2019 (COVID-19) vaccine began to emerge. 1 , 2 Because the clinical trials of the COVID-19 vaccines did not capture these data and initial reports were scarce due to the more passive official reporting systems like the US CDC Vaccine Adverse Event Reporting System (VAERS), individuals and families did not receive any information about whether the vaccine would have an impact on menstrual health. As a result, there were concerns about the impact of vaccination, with approximately 1%−2% of individuals indicating possible concerns about infertility from the COVID-19 vaccine, 3 which led to the National Institutes of Health (NIH) requesting proposals to investigate these anecdotal reports. 4 Research has now demonstrated a clear and relatively consistent relationship between administration of the COVID-19 vaccine and a slight, temporary increase in cycle length in adult women of reproductive age. 5 – 10 Yet, relatively little is known about changes to menstrual characteristics in adolescent girls. One small, prospective study of 39 adolescent girls ages 12–16 found that 8 of 23 girls reported menstrual irregularity in the 3 months following vaccination, although no abnormalities in anti-mullerian hormone levels were noted, suggesting ovarian reserve did not change. 11 However, this study did not include a non-vaccinated control group, so it is possible that these changes observed are attributable to natural variations in the menstrual cycle. A larger retrospective study of 7565 menstruating adolescent girls ages 12–15 in Norway asked mothers to report on menstrual cycle disturbances in their daughters and included both girls who had and had not been vaccinated at the time of the questionnaire. 12 The authors found an elevated risk of menstrual disturbances, including heavier bleeding, prolonged bleeding, shorter cycles, longer cycles, and increased pain during periods following vaccination, compared to pre-vaccine cycles. Evaluating the impact of the COVID-19 vaccine in adolescents is particularly important given the rapid physical and hormonal changes occurring across the period of pubertal development. Maturation of the hypothalamic-pituitary-ovarian (HPO) axis is incomplete in adolescents, thereby potentially resulting in different menstrual cycle changes following administration of the COVID-19 vaccine than those observed in adult women. Additionally, these data are potentially important for counseling parents regarding potential vaccine refusal in the future for their teen daughters. The purpose of the current study was to prospectively evaluate the impact of the COVID-19 booster vaccine on menstrual cycle characteristics in adolescent girls who received the COVID-19 booster, compared to a group of adolescent girls who did not receive a COVID-19 booster vaccine in the study timeframe. Additionally, we aimed to tease apart the role of stress at the time of vaccination to determine if stress influenced menstrual cycle changes. Consistent with data from the adult literature, we hypothesized that girls who received the COVID-19 booster would report longer menstrual cycles following the booster, whereas girls who did not receive the booster would report no changes in length of menstrual cycles. We also hypothesized girls who received the booster would report increased menstrual flow, menstrual pain, and menstrual symptoms, controlling for self-reported stress level at baseline.