COVID-19 vaccination and menstrual cycle characteristics: A prospective cohort study

Abstract

We prospectively examined the association between COVID-19 vaccination and menstrual cycle characteristics in an internet-based prospective cohort study. We included a sample of 1,137 participants who enrolled in Pregnancy Study Online (PRESTO), a preconception cohort study of couples trying to conceive, during January 2021-August 2022. Eligible participants were aged 21–45 years, United States or Canadian residents, and trying to conceive without fertility treatment. At baseline and every 8 weeks for up to 12 months, participants completed questionnaires on which they provided information on COVID-19 vaccination and menstrual cycle characteristics, including cycle regularity, cycle length, bleed length, heaviness of bleed, and menstrual pain. We fit generalized estimating equation (GEE) models with a log link function and Poisson distribution to estimate the adjusted risk ratio (RR) for irregular cycles associated with COVID-19 vaccination. We used linear regression with GEE to estimate adjusted mean differences in menstrual cycle length associated with COVID-19 vaccination. We adjusted for sociodemographic, lifestyle, medical and reproductive factors. Participants had 1.1 day longer menstrual cycles after receiving the first dose of COVID-19 vaccine (95 % CI: 0.4, 1.9) and 1.3 day longer cycles after receiving the second dose (95 % CI: 0.2, 2.5). Associations were attenuated at the second cycle post-vaccination. We did not observe strong associations between COVID-19 vaccination and cycle regularity, bleed length, heaviness of bleed, or menstrual pain. In conclusion, COVID-19 vaccination was associated with a ∼1 day temporary increase in menstrual cycle length, but was not appreciably associated with other menstrual cycle characteristics.

1. Introduction

Anecdotal reports of menstrual cycle irregularities after COVID-19 vaccination began appearing on social media and through vaccine surveillance systems in early 2021 [1], [2]. Reported menstrual irregularities have included changes in the length, frequency, regularity, and heaviness of menstrual bleeds [3]. In addition, some individuals who do not typically menstruate (e.g., postmenopausal individuals) have reported breakthrough uterine bleeding following vaccination [4], [16]. These reports have contributed to public concern about the safety and side effects of COVID-19 vaccines [1].

A link between vaccination and menstrual function is biologically plausible. Vaccination triggers an immune response that could temporarily affect the hypothalamic-pituitary-ovarian (HPO) axis, which regulates hormones involved in the menstrual cycle [5]. The immune response to vaccination could also affect immune cells in the uterine lining responsible for the build-up and breakdown of endometrial tissue, which could increase menstrual flow [6]. Other types of vaccines have previously been found to temporarily alter menstrual function without lasting effects on menstruation or fertility, including vaccines for typhoid [7], hepatitis B [8], and human papillomavirus [9], [10].

Several studies have examined changes in menstrual cycle characteristics after receipt of the COVID-19 vaccine [11], [12], [13], [14], [15], [16], [17], [18], [19], [20]. However, most studies ascertained menstrual change subjectively (i.e., “Did you notice a change to your menstrual cycles after vaccination?”), which could introduce reporting bias [14], [15], [16], [18], [19]. Other studies lacked an unvaccinated comparison group to account for time trends and normal variability in menstrual function, had limited ability to control for potential confounders, and were not able to assess the extent to which menstrual cycle changes persisted over time [14], [15], [16], [17], [18], [19]. The four studies that assessed menstrual function and included an unvaccinated comparison group found an association between COVID-19 vaccination and a small, temporary increase in menstrual cycle length, but little association with cycle irregularity [11], [12], [13], [20]. There has been limited assessment of COVID-19 vaccination and other menstrual cycle characteristics, including bleed length, heaviness of flow, and menstrual pain.

We examined the association between COVID-19 vaccination and menstrual cycle characteristics using prospectively collected data from an ongoing preconception cohort study.

2. Materials and methods

2.1. Study design and population

Pregnancy Study Online (PRESTO) is an internet-based preconception cohort study of couples trying to conceive (June 2013-present) [21]. The study was originally designed to identify risk factors for subfertility. Eligible participants self-identified as female, were 21–45 years old, resided in the United States or Canada, and were trying to conceive without the use of fertility treatment. Participants were recruited on the internet, primarily through social media (e.g., Facebook, Instagram), but also through advertisements on pregnancy and health-related websites [21]. Participants completed a baseline questionnaire on sociodemographics, lifestyle factors, and medical and reproductive history. Participants then completed shorter follow-up questionnaires every 8 weeks for up to 12 months to update information on menstrual cycle characteristics, pregnancy status, and lifestyle or medical factors. We randomized participants with 50 % probability to receive a premium subscription to Kindara.com, a fertility tracking app [22]. All U.S. residents were offered six Clearblue pregnancy tests [23] and couples that enrolled together (i.e., both completed their baseline questionnaires) were entered into a lottery to win a gift card.

The institutional review board at the Boston University Medical Campus approved the study protocol and all participants provided informed consent.

2.2. Assessment of COVID-19 vaccination

Beginning in January 2021, we collected information on COVID-19 vaccination on the baseline and follow-up questionnaires (Fig. S1). For vaccinated participants, we asked for the brand of vaccine and the date of each dose. We added questions on booster doses (including brand and date) in November 2021. There were insufficient numbers of participants reporting a booster dose during the study period to examine the association between booster doses and menstrual characteristics in this analysis.

For vaccinated participants, we used vaccination date(s) to identify vaccinated and unvaccinated menstrual cycles. For unvaccinated participants, we assigned a “treatment date” (i.e., a date on which they could have been vaccinated but were not) using a spatiotemporal matching approach [24]. Specifically, for each unvaccinated participant, we identified the vaccinated participant who lived in the same geographic region (Northeastern, Southern, Midwestern, Western U.S.; Canada) and who enrolled in PRESTO on the date closest to the unvaccinated participant. We then assigned the unvaccinated participant the “treatment date” of their vaccinated match. This approach allowed us to truncate follow-up time of the unvaccinated in a similar way to the vaccinated, which prevented over-representation of unvaccinated participants at later time points.

2.3. Assessment of menstrual cycle characteristics

On the baseline questionnaire, participants reported the date of their last menstrual period (LMP) and whether their cycles were regular within the past couple of years (i.e., “regular in a way that you could usually predict about when the next period would start”). Response options included “yes”, “no”, and “cannot say because I was taking hormones most of the time.” Participants who responded “yes” or “cannot say…” reported their typical cycle length (i.e., “number of days from the first day of one menstrual period to the first day of your next menstrual period”). Participants with irregular cycles reported how many menstrual periods they typically have in one year. All participants reported how long their menstrual period usually flowed (hereafter “bleed length”), the total amount of menstrual flow (“light”, “moderate”, “moderate/heavy”, or “heavy”, corresponding to ≤10, 11–20, 21–30, and >30 pads or tampons per period, respectively; hereafter “heaviness of bleed”), and the presence and severity of pain during their menstrual period (none, “mild cramps with medication seldom needed”, “moderate cramps with medication usually needed”, “severe cramps with medication and bedrest required”; hereafter “menstrual pain”).

On follow-up questionnaires, participants reported their LMP date and pregnancy status. We asked participants whether their menstrual periods had been regular since their previous questionnaire and the length of their most recent menstrual cycle. Beginning in April 2021, participants reported all LMP dates since their previous questionnaire. Beginning in June 2021, participants reported on bleed length, heaviness of bleed, and menstrual pain of their most recent menstrual cycle. On all questionnaires, the menstrual cycle questions preceded the questions about COVID-19 vaccination and appeared in distinct sections of the survey.

2.4. Statistical analysis

We restricted our analysis to participants who completed at least one questionnaire after January 25, 2021 (i.e., had the opportunity to provide information on COVID-19 vaccination; n = 2663). We then excluded participants who received their first vaccine dose before study enrollment (n = 1374) and participants with no new LMP dates during follow-up (e.g., conceived immediately or were experiencing amenorrhea; n = 152), leaving 1,137 participants eligible for analysis (Fig. 1 ).

Fig. 1.

Flow chart showing how the analytic sample was derived.

Our sample for analysis of COVID-19 vaccination and irregular cycles included 1,082 participants with non-missing outcome data (Fig. 1). In primary analyses, we compared risk of irregular cycles on unvaccinated follow-up questionnaires with risk on the first and second follow-up questionnaires after vaccination (Fig. S2 a). Thus, the follow-up questionnaire was the unit of analysis. We estimated adjusted risk ratios (RRs) and 95 % confidence intervals (CIs) by fitting generalized estimating equation (GEE) models with a log link, Poisson distribution, and exchangeable correlation matrix to account for correlated observations within individuals. We examined the first and second dose separately and grouped all vaccine brands together. We conducted a sensitivity analysis restricting to participants without irregular cycles at baseline.

For analyses of COVID-19 vaccination and cycle length, we excluded participants with irregular, short (<24 days), or long (>38 days) menstrual cycles [25] at baseline for an analytic sample of 825 participants (Fig. 1). In primary analyses, we assessed the mean difference in cycle length during unvaccinated menstrual cycles with the first and second cycle after vaccination. Therefore, the menstrual cycle was the unit of analysis. We estimated adjusted regression coefficients (β) and 95 % CIs using GEE models with an identity link, normal distribution, and exchangeable correlation matrix. We examined the first and second dose separately and grouped all vaccine brands together.

We next stratified results by vaccine brand (Pfizer-BioNTech® Comirnaty vs. Moderna® Spikevax; too few participants received other vaccines). We hypothesized that any potential association would be due to a vaccine-related immune response rather than vaccine components and that associations would be similar across vaccine brands. Next, we restricted to participants who had never tested positive for the SARS-CoV-2 virus to limit confounding by infection. Finally, to assess the extent to which reporting bias could have influenced our results, we stratified by calendar time of questionnaire completion (January-March 2021 vs. April 2021-August 2022). We hypothesized that associations would be stronger after April 2021, when a potential link between COVID-19 vaccination and menstrual function became widely publicized.

In secondary analyses, we compared pre-post vaccination differences in menstrual cycle characteristics in vaccinated and unvaccinated groups, analogous to a difference-in-differences analysis [26]. An advantage of this analysis is that it compares within-person changes in menstrual cycle characteristics across vaccination groups. To do this, we fit a GEE model, but instead of using all outcome data during follow-up, we compared outcome data in the time points immediately before and after the first dose (or assigned treatment date in the unvaccinated; Fig. S2 b). In another variation of the difference-in-differences analysis, we compared menstrual cycle characteristics in the time point immediately after vaccination with typical characteristics reported at baseline across vaccination groups (Fig. S2 c).

We selected potential confounders by identifying factors associated with COVID-19 vaccination and menstrual cycle characteristics. Final models for irregular cycles and cycle length adjusted for age (<25, 25–29, 30–34, ≥35 years), educational attainment (<college degree, college degree, graduate school), private health insurance (yes vs. no), parity (yes vs. no), geographic region of residence (Northeastern, Southern, Midwestern, Western U.S.; Canada), pregnancy attempt time at study entry (continuous), and calendar month of menstrual cycle. We also adjusted for baseline menstrual cycle characteristics (e.g., for the analysis of cycle regularity, we adjusted for cycle regularity at baseline). History of SARS-CoV-2 infection, occupation, and Perceived Stress Scale score [27] were not appreciably associated with vaccination status or menstrual cycle characteristics, so we did not adjust for these variables.

In June 2021, we added questions on bleed length, heaviness of bleed, and menstrual pain to the follow-up questionnaires. Given the smaller sample size for these analyses (n = 518), we performed exploratory analyses comparing the unadjusted prevalence of the following outcomes on vaccinated and unvaccinated follow-up questionnaires: bleed length ≥7 days, menses requiring ≥20 pads/tampons, and menstrual pain requiring medication.

3. Results

Overall, 1,137 participants contributed a median of 2 follow-up questionnaires (range = 1–6) and 5 menstrual cycles (range = 1–14). Participants either completed 6 follow-up questionnaires (13.8 %), conceived before 12 months (64.6 %), initiated fertility treatment (11.1 %), stopped trying to conceive (2.0 %), or were lost to follow-up/withdrew (8.5 %). By design, all participants were unvaccinated for COVID-19 at study entry. Of these, 437 (38.4 %) received at least one COVID-19 vaccine dose during the study period (32.3 % Moderna® Spikevax, 60.9 % Pfizer-BioNTech® Comirnaty, 6.6 % Janssen [Johnson & Johnson®], and 0.2 % AstraZeneca®). Participants who were vaccinated during the study period had higher educational attainment and income, and were less likely to have a previous birth compared with those who were unvaccinated (Table 1 ). Among vaccinated participants, 83.6 % were vaccinated between February and May 2021.

Table 1.

Characteristics of 1,137 participants by receipt of at least one COVID-19 vaccine dose during study period, PRESTO, 2021–2022.

Characteristica	Received first dose during study (n = 437)	Unvaccinated at end of study (n = 700)
Age (years), mean	31.4	31.4
Attempt time at study entry (cycles), mean	4.0	3.9
Educational attainment (years), %
≤High school	1.0	5.2
Some college	11.5	19.2
College graduate	35.9	33.0
Graduate school	51.6	42.7
Household income (USD/year), %
<$50,000	10.5	15.9
$50,000-$99,999	34.6	37.8
$100,000-$149,999	31.9	26.9
≥$150,000	23.1	19.5
Race/ethnicity, %
Non-Hispanic white	85.4	84.4
Non-Hispanic Black	1.6	2.1
Non-Hispanic Asian	2.0	1.4
Non-Hispanic other	5.3	5.6
Hispanic	5.8	6.5
Geographic region, %
Northeastern U.S.	19.4	18.0
Southern U.S.	20.2	22.8
Midwestern U.S.	18.6	20.3
Western U.S.	19.2	16.5
Canada	22.5	22.4
Current smoker, %	3.2	4.9
Parous, %	44.9	61.2
History of infertility, %	14.3	15.9
Pap smear in past three years, %	91.6	89.9
Private health insurance, %	87.1	81.7
Occupation in healthcare industry,b %	19.9	20.2
Rotating shift work, %	10.3	10.3
Night shift work, %	8.3	8.3
Ever tested positive for SARS-CoV-2, %	12.2	11.6
Perceived Stress Scale score, mean	16.7	16.6
Major Depression Inventory score, mean	12.2	11.7
Body mass index (kg/m2), mean	27.4	27.5
Calendar month of vaccination, %
December 2020 – January 2021	13.3	–
February – March 2021	39.7	–
April – May 2021	43.9	–
June 2021 – August 2022	3.1	–

Abbreviations: PRESTO = Pregnancy Study Online.

^a Values were missing for attempt time at study entry (n = 2), household income (n = 25), rotating shift work (n = 21), night shift work (n = 71), occupation in health care industry (n = 66), body mass index (n = 1), Major Depression Inventory (n = 4), and pap smear in past three years (n = 9). No variable was missing for>6 % of observations.

^b Occupation in health care industry defined based on United States Census Industry codes 8190 (Hospitals), 8180 (Other health care services), 8170 (Home health care services), 8080 (Offices of other health practitioners), 8070 (Offices of optometrists), 8090 (Outpatient care centers), 8270 (Nursing care facilities), 8290 (Residential care facilities, without nurses), 7970 (Offices of physicians) and 7980 (Offices of dentists).

3.1. Irregular menstrual cycles

Fifteen percent of participants reported irregular menstrual cycles at baseline. On follow-up questionnaires when participants were unvaccinated, the prevalence of irregular cycles was 24.4 %. On the first and second follow-up questionnaires following the first dose, the prevalence of irregular cycles was 22.7 % and 20.4 %, respectively. After adjustment for confounders, the risk of irregular cycles following the first (adjusted RR = 1.02, 95 % CI: 0.88, 1.18) and second (adjusted RR = 1.02, 95 % CI: 0.83, 1.25) follow-up questionnaires after the first dose were similar to the unvaccinated time period (Table 2 ). The results were similar for the second vaccine dose (Table 2). Results remained consistent when we restricted to follow-up questionnaires contributed by participants with regular cycles at baseline (data not shown), when we stratified by vaccine brand (Pfizer-BioNTech® Comirnaty vs. Moderna® Spikevax; Table S1), when we restricted to participants who had never tested positive for the SARS-CoV-2 virus (Table S2), and when we stratified by calendar time (January 2021-March 2021 vs. April 2021-August 2022; Table S3). Finally, the difference in the prevalence of irregular cycles in the follow-up questionnaires before and after vaccination (or for the unvaccinated, assigned treatment date) was similar by vaccination status (Table S4).

Table 2.

Association between COVID-19 vaccination, irregular menstrual cycles, and menstrual cycle length, PRESTO,^a 2021–2022.

	Unvaccinated	First vaccine dose		Second vaccine dose
		First time pointb after first dose	Second time pointb after first dose	First time pointb after second dose	Second time pointb after second dose
Irregular cycles (n = 1,082)
No. of follow-up questionnaires	1199	428	299	310	204
No. ( %) with irregular cycles	292 (24.4)	97 (22.7)	61 (20.4)	77 (24.8)	36 (17.7)
Unadjusted RR (95 % CI)	Reference	0.96 (0.82, 1.13)	0.98 (0.80, 1.21)	1.09 (0.92, 1.30)	0.89 (0.70, 1.14)
Adjusted RR (95 % CI)	Reference	1.02 (0.88, 1.18)	1.02 (0.83, 1.25)	1.13 (0.95, 1.33)	0.91 (0.70, 1.19)
Menstrual cycle length (n = 825)
No. of menstrual cycles	1519	351	286	264	210
Mean (SD) menstrual cycle length	29.6 (6.6)	30.9 (7.5)	30.3 (9.5)	31.1 (6.7)	29.9 (6.3)
Unadjusted mean difference (95 % CI)	Reference	1.1 (0.4, 1.9)	0.5 (-0.5, 1.6)	1.4 (0.2, 2.5)	0.1 (-0.8, 0.9)
Adjusted mean difference (95 % CI)	Reference	1.1 (0.4, 1.9)	0.6 (-0.5, 1.6)	1.3 (0.2, 2.5)	0.2 (-0.7, 1.1)

Abbreviations: CI = confidence interval; PRESTO = Pregnancy Study Online; RR = risk ratio; SD = standard deviation.

^cAdjusted for age, educational attainment, type of health insurance, parity, geographic region of residence, cycles of attempt time at study entry, and month of follow-up questionnaire. Models for irregular cycles adjust for cycle regularity at baseline; models for cycle length adjust for typical cycle length at baseline.

^aAnalysis of irregular cycles restricted to participants with non-missing data on cycle regularity during follow-up. Analysis of menstrual cycle length restricted to participants with regular cycles at baseline and typical cycle length 24–38 days.

^bFor analysis of irregular cycles, “time point” refers to follow-up questionnaires. For analysis of cycle length, “time point” refers to menstrual cycles.

3.2. Menstrual cycle length

The mean typical menstrual cycle length reported at baseline was 28.6 days (standard deviation [SD] = 2.3) among those with regular cycles. Mean cycle length in unvaccinated menstrual cycles was 29.6 days (SD = 6.6) compared with mean cycle length in the first (30.9 days, SD = 7.5) and second (30.3 days, SD = 9.5) cycles after the first vaccine dose. After adjustment for confounders, there was a 1.1-day increase in mean cycle length immediately following the first dose (adjusted β = 1.1 day, 95 % CI: 0.4, 1.9; Table 2). These results were slightly attenuated during the second cycle after the first dose (adjusted β = 0.6, 95 % CI: −0.5, 1.6). This corresponds to the first cycle after the second dose for 78.8 % of vaccinated participants. The results were similar when examining the first cycle after the second vaccine dose, but were fully attenuated for the second cycle after the second dose (Table 2). Results were consistent across strata of vaccine brands (Table S5) and when we restricted to participants who had never tested positive for the SARS-CoV-2 virus (Table S6). The association was slightly stronger during April 2021-August 2022 (adjusted β for first cycle after first dose = 1.4, 95 % CI: 0.3, 2.5) than during January 2021-March 2021 (adjusted β for first cycle after first dose = 0.8, 95 % CI: −0.2, 1.9; Table S7).

Among the unvaccinated, mean cycle length before and after the assigned treatment date was identical (29.8 days), whereas among the vaccinated, mean cycle length was 1.3 days longer after the first vaccine dose compared with before (30.9 vs. 29.6; Table S8). After adjustment for confounders, the association for post vs. pre-treatment date was different among the unvaccinated (adjusted β = −0.07, 95 % CI: −0.88, 0.74) and vaccinated (adjusted β = 1.19, 95 % CI: 0.32, 2.05) (interaction term = 1.26, 95 % CI: 0.08, 2.44). In other words, among the vaccinated, the pre-post vaccine difference in menstrual cycle length was 1.3 days longer than the pre-post difference among the unvaccinated. We observed similar findings when comparing the cycle after vaccination (or assigned treatment date) with typical cycle length at baseline (Table S8).

When we examined cycle length categorically, we found that the prevalence of short cycles (<24 days) did not vary appreciably by vaccination status (Table 3 ). The prevalence of long cycles (>38 days) increased from 5.9 % (95 % CI: 4.7 %, 7.1 %) in unvaccinated cycles to 11.1 % (95 % CI: 7.8 %, 14.4 %) in the first cycle following the first dose, then declined to 7.3 % (95 % CI: 4.3 %, 10.3 %) in the second cycle following the first dose (Table 3). Due to sample size constraints, we could not perform multivariable regression for these analyses.

Table 3.

Prevalence of other menstrual cycle characteristics by vaccination status, PRESTO, 2021–2022.

Outcome	No. of follow-up questionnaires	No. of follow-up questionnaires with outcome	Percentage with the outcome (95 % CI)
Menstrual cycles < 24 days
Unvaccinated follow-up questionnaires	1519	73	4.8 (3.7, 5.9)
First dose
First follow-up questionnaire	352	11	3.1 (1.3, 4.9)
Second follow-up questionnaire	286	13	4.6 (2.2, 7.0)
Menstrual cycles >38 days
Unvaccinated follow-up questionnaires	1519	89	5.9 (4.7, 7.1)
First dose
First follow-up questionnaire	352	39	11.1 (7.8, 14.4)
Second follow-up questionnaire	286	21	7.3 (4.3, 10.3)
Bleed length ≥7 days
Unvaccinated follow-up questionnaires	245	19	7.8 (4.4, 11.2)
First dose
First follow-up questionnaire	159	9	5.7 (2.1, 9.3)
Second follow-up questionnaire	236	9	3.8 (1.4, 6.2)
Menstrual periods requiring ≥20 pads/tampons
Unvaccinated follow-up questionnaires	245	48	19.6 (14.6, 24.6)
First dose
First follow-up questionnaire	159	29	18.2 (12.2, 24.2)
Second follow-up questionnaire	236	29	12.3 (8.1, 16.5)
Menstrual pain requiring medication
Unvaccinated follow-up questionnaires	245	73	29.8 (24.1, 35.5)
First dose
First follow-up questionnaire	158	55	34.8 (27.4, 42.2)
Second follow-up questionnaire	232	74	31.9 (25.9, 37.9)

Abbreviations: CI = confidence interval; PRESTO = Pregnancy Study Online.

3.3. Bleed length, heaviness of bleed, and menstrual pain

The prevalence of bleed lengths ≥7 days and menses that required ≥20 tampons/pads was similar by vaccination status (Table 3). The prevalence of menstrual pain requiring medication was 29.8 % (95 % CI: 24.1 %, 35.5 %) on unvaccinated follow-up questionnaires and 34.8 % (95 % CI: 27.4 %, 42.2 %) and 31.9 % (95 % CI: 25.9 %, 37.9 %) on the first and second follow-up questionnaires after the first dose, respectively. Given the limited precision for these analyses, we did not adjust for potential confounders.

4. Discussion

4.1. Main findings

In this prospective cohort study of North American non-contracepting pregnancy planners, we found a small and temporary increase in cycle length following COVID-19 vaccination, but little difference in cycle regularity, bleed length, heaviness of bleed, or menstrual pain. The first menstrual cycle after each COVID-19 vaccine dose was approximately 1 day longer on average than menstrual cycles before vaccination, but returned to the pre-vaccination length at the second cycle. The prevalence of long cycles (>38 days) was also slightly higher immediately after vaccination, but returned to baseline by the following cycle.

4.2. Strengths and limitations

Our internet-based study design allowed us to continue enrolling and following participants throughout the pandemic and to rapidly add new questions as they became relevant. Because participants were all trying to conceive, no participants used contraception during the study period, and all participants had at least 8 weeks since contraceptive discontinuation at the time of outcome assessment. We intentionally separated our questions on menstrual cycle characteristics from those on COVID-19 vaccination to reduce reporting bias. We also collected repeated measures on a range of menstrual cycle characteristics over time and included an unvaccinated comparison group to account for natural variability in menstrual cycle characteristics over time. We adjusted for a wide range of potential confounders.

Our study had several limitations. Because this analysis was nested in an ongoing preconception cohort study designed to identify risk factors for subfertility, our study population was restricted to individuals who were trying to conceive. Therefore, a large proportion of participants conceived during the study period, which precluded the examination of associations beyond 1–2 cycles post-vaccination, given the selected nature of participants who contributed multiple follow-up questionnaires. PRESTO participants were aged 21–45 years; therefore, our results may not generalize to older individuals, including those who are perimenopausal. Results also may not generalize to individuals using exogenous hormones. The majority of participants in our analysis identified as non-Hispanic white and were college graduates. However, we do not anticipate that any biologic mechanisms linking COVID-19 vaccination with menstrual cycle characteristics would vary by sociodemographic characteristics; therefore, we expect that our results extend to more racially and socioeconomically diverse populations.

Furthermore, although we asked questions about menstrual cycle characteristics separately from COVID-19 vaccination, the potential link between vaccination and menstrual function was highly publicized beginning in April 2021. Therefore, reporting bias remains possible, and we cannot exclude the possibility that vaccinated participants may have more carefully monitored their menstrual cycles after vaccination. Our finding of a stronger association with cycle length for cycles after April 2021 (compared with January-March 2021) supports the premise that some reporting bias exists. Our study size was limited; therefore, some of our findings lack precision. For example, the results for irregular cycles are consistent with no association or a weak association in either direction, based on the confidence interval. Finally, because bleed length, heaviness of bleed, and menstrual pain were not added to the follow-up questionnaires until June 2021, we had limited sample size for these analyses and could not control for confounding.

4.3. Interpretation

The timing of the menstrual cycle is tightly regulated by the HPO axis, which can be influenced by the environment, stress, and other exposures [28], [29]. Individual variability in cycle length is common [30], [31], and a change in cycle length of <8 days is considered normal [25]. However, even changes considered clinically normal, when unexplained, can be concerning and adversely affect quality of life [32], [33], [34]. Menstrual health is understudied in basic and translational research [35]. leading to gaps in our understanding of factors that can influence menstrual cycle characteristics. In the context of the COVID-19 pandemic, these knowledge gaps, combined with the lack of data on menstrual function from pre-clinical COVID-19 vaccine trials and the spread of anecdotal reports on social media, have led to lower vaccination rates among reproductive-aged individuals [36].

Our finding of slightly longer menstrual cycles immediately after COVID-19 vaccination is largely consistent with previous studies. Among users of the menstrual-charting application “Natural Cycles,” including those from the U.S. [11] and worldwide [13], the adjusted difference in cycle length before and after vaccination was ∼0.6 days longer in vaccinated compared with unvaccinated cycles. Two other studies that used menstrual charting apps also showed slightly longer cycles after COVID-19 vaccination, particularly when the vaccine was administered during the follicular phase [14], [20]. Finally, in the Nurses’ Health Study 3, vaccinated individuals had 72 % higher odds of longer cycle length (≥32 days) within the first six months after vaccination compared with unvaccinated participants; results were driven by participants who had short, long, or irregular cycles at baseline [12]. Importantly, our study and those described above all show that the increase in cycle length is temporary, with cycle lengths returning to normal within 1–2 cycles following vaccination.

We found no appreciable association between COVID-19 vaccination and changes in bleed length. This finding is consistent with the two other studies that have addressed this topic, both among users of the “Natural Cycles” app [11], [13]. We also found that the prevalence of heavy bleeds and severe menstrual pain was similar by vaccination status. Heavier flow has been reported after COVID-19 vaccination [16]; however, to our knowledge, ours is the first study to compare heaviness of flow and menstrual pain among vaccinated and unvaccinated individuals. Previous studies restricted to vaccinated participants were likely enriched with participants who noticed a change in their cycles; our data do not support a strong association between COVID-19 vaccination and heavy bleeding or pain requiring medication.

PRESTO was not designed to identify mechanisms underlying the association between COVID-19 vaccination and menstrual cycle characteristics. However, other studies have shown similar results across COVID-19 vaccine types (mRNA, adenovirus-vector, and inactivated), indicating that the most likely mechanism is through activation of the immune system, rather than any specific component of the vaccine itself [12], [13], [14]. Indeed, cytokine production as part of the immune response to vaccination may temporarily interfere with the HPO axis and production of ovarian hormones responsible for controlling menstrual function [5], [37], [38], [39]. This hypothesis is supported by evidence from a previous study, in which the delay in menses following COVID-19 vaccination was shorter among participants currently taking combined hormonal contraceptives (which suppress normal regulation of endogenous hormones), than among those not using hormonal contraceptives [14]. A role for cytokine production is also supported by the observation that vaccines for other diseases, including typhoid [7], hepatitis B [8], and human papillomavirus [9], [10] are associated with temporary menstrual changes. Finally, SARS-CoV-2 infection, which likewise produces a strong immune response, has been associated with menstrual changes [40], [41], although PRESTO had too few participants reporting an infection during the study period to assess this hypothesis.

Importantly, although menstrual function may be reflective of overall reproductive health [35], our previous study [42] and others [43], [44], [45], [46], [47], [48] have demonstrated no meaningful association between COVID-19 vaccination and fertility, either with spontaneous conception or conception through assisted reproductive technology. Taken together, these results indicate that short-term changes in menstrual cycle characteristics likely do not translate into meaningful differences in fertility.

5. Conclusions

In this prospective cohort study, we found a one-day average delay in menses and a higher prevalence of long menstrual cycles following COVID-19 vaccination, which resolved by the next menstrual cycle. Other menstrual cycle characteristics, including cycle regularity, bleed length, heaviness of bleed, and menstrual pain, were not strongly associated with COVID-19 vaccination.

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: LAW serves as a fibroid consultant for AbbVie, Inc. She also receives in-kind donations for primary data collection in Pregnancy Study Online (PRESTO) from Swiss Precision Diagnostics (home pregnancy tests), Kindara.com (fertility apps), and FertilityFriend.com (fertility apps). All of these relationships are for work unrelated to this manuscript. The remaining authors report no conflicts of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

The data that has been used is confidential.