Considerations and practical recommendations for identifying perimenopause in longitudinal research.

Over the past one hundred years, there have been several landmark longitudinal studies focused on female reproductive aging and associations with health and well-being (see Supplemental Table 1 ). An understanding of these older studies is required to understand how the definition of menopausal stages has been operationalized for research over time. The Tremin Research Program on Women’s Health (TREMIN) was groundbreaking in that Treloar was the first to describe the perimenopause as the stretch of time prior to the final menstrual period characterized by alterations in menstrual bleeding patterns ( Treloar, 1981 ; Treloar et al., 1967 ). TREMIN’s first cohort was enrolled in the 1930s, and participants prospectively tracked menstruation on physical calendar cards distributed and collected each year ( Mansfield and Bracken, 2003 ). The sheer size of the data collected through TREMIN is remarkable: in 2003, the average TREMIN participant had contributed 21 years of calendar data ( Mansfield and Bracken, 2003 ). The Massachusetts Women’s Health Study (MWHS) recruited a random sample of women aged 45–55, with telephone interviews every nine months for five years and annual menstrual cycle calendars ( Avis and McKinlay, 1995 ; McKinlay et al., 1992 ). The MWHS investigators identified a series of retrospective questions that could predict with high sensitivity and specificity (>70 %) that a participant would become postmenopausal within three years ( Brambilla et al., 1994 ). Specifically, individuals who reported either amenorrhea of 3–11 months or increased menstrual cycle irregularity in the absence of amenorrhea had reached perimenopause and were within a few years of the FMP ( Brambilla et al., 1994 ). The Melbourne Women’s Midlife Health Project (MWMHP) was modeled on the MWHS and was a population-based longitudinal study of Australian women aged 45–55 at enrollment. MWMHP participants classified their menstrual cycle frequency and flow retrospectively at annual interviews, selecting from a list of predetermined categories (e.g., less often, about the same, more often, or changeable for cycle frequency over the past year), in addition to completing annual menstrual cycle calendars ( Guthrie et al., 2005 ; Taffe and Dennerstein, 2002 ). Early menopause transition status was defined as menstruation within the three months prior to the interview but with changes in menstrual cycle frequency; late menopausal transition was defined as a self-report of 3–11 months of amenorrhea ( Guthrie et al., 2005 ). The Seattle Midlife Women’s Health Study (SMWHS) recruited women from the Seattle area aged 35–55 who had experienced at least one menstrual cycle in the year prior to enrollment. Annual retrospective questionnaires captured data on regularity of menstrual cycles, and participants also filled out annual menstrual cycle calendars, recording all bleeding/spotting days. Uniquely, stage assignments were based on prospective menstrual cycle pattern data, rather than both retrospective and prospective data, due to poor agreement between the two forms of reporting (kappa = 0.192) ( Smith-DiJulio et al., 2005 ). “Early menopause transition” was defined as any perceived changes in frequency or flow despite regular menstrual cycles (consecutive cycle lengths differed by less than seven days according to the calendar data or no difference in irregularity compared to self-reported patterns between ages 20–34). The “middle menopause transition” was characterized by at least one instance of cycle irregularity (i.e., consecutive cycle lengths differing by at least seven days, as per annual calendar data), and the “late menopausal transition was assigned when participants had at least one skipped menstrual cycle (i.e., a bleed-free interval of at least twice the modal cycle length for that year). The Study of Women’s Health Across the Nation (SWAN) was a multi-site longitudinal study that collected data from multiple ethnic groups across the United States for over 25 years. Initially, SWAN used stage definitions based on MWHS using the date of the most recent menstrual period and self-reported changes in cycle length in the past 12 months (i.e., becoming further apart, closer together, occurred at more variable intervals, became more regular, no change, or don’t know) ( Luetters et al., 2007 ). Therefore, “late perimenopause” was assigned when no menses had occurred for the past 3–11 months, based on the time since the date of the last menstrual period. “Early perimenopause” was established when a participant reported decreased predictability ( Luetters et al., 2007 ). SWAN participants also returned menstrual cycle calendars each month ( Paramsothy et al., 2013 ). This data was later used to determine reproductive stage based on the ReSTAGE definitions (see 2.2 below) ( Paramsothy et al., 2013 ). Finally, the Penn Ovarian Aging Study (POAS) recruited a slightly younger cohort than MWHS, MWMHP, SWAN, and SMWHS (aged 35–47 years) who reported regular menstrual cycles at enrollment, defined as 22–35 days apart. During in-home interviews that occurred every nine (first five years of the study) or twelve months thereafter, participants reported the start dates for their three most recent menstrual cycles. Cycle lengths were calculated and compared to cycle length at enrollment to menopausal status ( Freeman and Sammel, 2016 ). “Late premenopausal” was defined as a difference in cycle lengths of at least seven days relative to that participant’s cycle length at enrollment. “Early transition” was defined as a difference in cycle lengths of at least seven days relative to cycle length at enrollment that persisted for at least two consecutive cycles or a period of amenorrhea of up to 60 days. “Late transition” was assigned if a participant reported a period of amenorrhea of more than 60 days but less than 12 months. Also, it is important to highlight how retrospective methods occasionally misclassified menopausal status. MWHS, MWMHP, and SWAN categorized women who reported a period within the past three months at the interview but had previously experienced a period of amenorrhea of ≥ 90 days as “early menopause transition” whereas in fact they had already met criteria for the “late menopause transition.” The staging criteria selected at the Stages of Reproductive Aging Workshop (STRAW) in 2001 was based on expert consensus, not empirical findings ( Soules et al., 2001 ). Therefore, to validate the proposed markers for the early and late menopausal transition, the ReSTAGE Collaboration examined menstrual cycle calendar and hormonal data from four large, longitudinal studies: TREMIN, MWMHP, SMWHS, and SWAN ( Harlow et al., 2006 ; Harlow et al., 2007 ; Randolph et al., 2006 ). ReSTAGE found that a persistent cycle length difference of at least seven days generally occurred before any of the other candidate criteria put forth by SMWHS, MWMHP, and TREMIN and therefore was potentially the best marker for entering the early menopausal transition ( Harlow et al., 2008 ). For the late menopausal transition, at least 60 days of amenorrhea was determined to be the most reliable and easiest to calculate bleeding criterion to mark the start of the late menopausal transition ( Harlow et al., 2006 ). Additionally, an FSH level of ≥ 40 IU/l during the early follicular phase is a reliable indicator of proximate FMP when coupled with candidate bleeding criteria ( Harlow et al., 2007 ). This finding highlighted the potential to incorporate a quantitative FSH marker into an update to the STRAW staging criteria ( Harlow et al., 2007 ). The current gold standard for reproductive staging is the consensus of the 2011 Stages of Reproductive Aging Workshop + 10 (STRAW+10) ( Harlow et al., 2012 ). As a follow-up to STRAW in 2001, STRAW+ 10 incorporated the empirical evidence from the ReSTAGE Collaboration ( Harlow et al., 2007 ; Harlow et al., 2008 ; Soules et al., 2001 ) ( Fig. 2 ). The STRAW+ 10 staging criteria primarily rely on observable bleeding patterns, as a low-cost, noninvasive marker of ovarian function. Levels of hormones such as FSH, Anti-Müllerian hormone (AMH), and inhibin B serve only as secondary, supportive criteria. The development of international pituitary standards and published reference ranges across the menstrual cycle for FSH permitted the specification of a minimum concentration during the late menopausal transition (25 IU/L) ( Randolph et al., 2006 ; Stricker et al., 2006 ), but the lack of widely available, standardized clinical assays for AMH and inhibin B at the time prevented incorporation of reference intervals for these additional hormonal biomarkers. STRAW+ 10 stages also included a subdivided stage −3, with subtle changes in vaginal bleeding flow and/or duration occurring in stage −3a as the conclusion of the late reproductive period. “Perimenopause” refers to stages −2, −1, and + 1a, encompassing the menopausal transition and the first year of postmenopause. Stage −2 is accompanied by persistent changes in menstrual cycle patterns, defined as a difference in consecutive cycle lengths of at least seven days that recurs within ten cycles. Stage −1, the end of the menopausal transition, is marked by a period of amenorrhea of at least 60 days in length. Postmenopause begins when the FMP occurs and is split into stages + 1a, + 1b, + 1c, and + 2. Stages + 1a and + 1b are the first and second years following the FMP, respectively. Variability in FSH stabilizes in stage + 1c, and then individuals remain in stage + 2 for the rest of their lifespan. Notably, STRAW+ 10 expanded STRAW’s reproductive staging scheme to include individuals who smoke and with a higher body-mass index (BMI) ( Harlow et al., 2012 ). Adipose tissue is a steroid conversion and synthesis site and estrogen plays a role in adipose tissue distribution, very low and very high BMI can be accompanied by altered menstrual patterns and reproductive disorders ( Aladashvili-Chikvaidze et al., 2015 ; Kurylowicz, 2023 ). Thus, STRAW had advised against applying their staging criteria to individuals with BMI less than 18 or greater than 30 kg/m 2 ( Soules et al., 2001 ). As the prevalence of obesity increases across menopause ( Davis et al., 2012 ), the applicability of STRAW+ 10 criteria to individuals regardless of BMI allows for greater inclusion in research studies and a better understanding of biological and psychological changes across the menopausal transition. Overall, the STRAW+ 10 consensus statement provides straightforward criteria about the corresponding observable changes in vaginal bleeding patterns from the reproductive phase into early postmenopause and emphasizes that the staging system should “allow for prospective classification of women.” However, several significant challenges remain regarding how to implement this in clinical research studies. STRAW+ 10 did not include a consensus on specific recommendations for data collection strategies to implement its staging scheme in longitudinal research studies nor has a standardized retrospective questionnaire been validated yet ( Harlow, 2018 ).

As mentioned in 3.3 , STRAW+ 10 consensus did not recommend applying its reproductive stages in individuals pre-existing oligomenorrhea or amenorrhea from PCOS, HIV/AIDS, hysterectomy without oophorectomy, uterine ablations, or undergoing cancer treatment. Recent work has revealed that moderate to severe vasomotor symptoms can identify the onset of perimenopause, and emergence of these symptoms could be used to identify stage −2 (early menopausal transition) among this cohort that doesn’t menstruate or whose menstruation is affected by factors above and beyond ovarian aging ( Islam et al., 2025 ). However, it is important to note that vasomotor symptoms of any severity were reported by many participants in all studied menopausal stages, ranging from 43.1 % of premenopausal participants to three-quarters of participants in perimenopause, indicating that the presence of vasomotor symptoms is not specific to a particular stage (or stages). Even one-third of late postmenopausal (aged 60–69 years) participants reported the presence of vasomotor symptoms. While the development of moderately or severely bothersome vasomotor symptoms may help identify the onset of the late menopausal transition, many individuals never experience this severity of vasomotor symptoms. A related topic of future research interest is assessing the influence of glucagon-like peptide 1 receptor agonists (GLP-1 RA) on menstrual cycles on pre- and peri-menopausal individuals ( Rubino and Schon, 2025 ). As a potential treatment for PCOS, use of these medications has been associated with regulation of menstrual cycle patterns among individuals with PCOS and obesity whose weight was not sufficiently reduced with lifestyle modifications ( Carmina and Longo, 2023 ). Specifically monitoring for use of GLP-1 RAs in future studies of perimenopause will provide the opportunity to examine the effect of these medications on reproductive aging. The development and validation of a standardized questionnaire to assign STRAW+ 10 stage cross-sectionally or at enrollment for longitudinal studies would be a major asset in women’s midlife health research. Recent papers have put forth practical recommendations for how research studies should capture data on sex and gender ( Peters et al., 2023 ), the menstrual cycle ( Schmalenberger et al., 2021 ), hormonal contraceptive use ( Beltz, 2024 ), and sample open-ended questions to assess menopausal status among women at risk for developing eating disorders ( Klump et al., 2025 ). Additionally, the Ann S. Bowers Women’s Brain Health Initiative is in the process of developing a new gold “STANDARD” instrument to collect comprehensive self-report data about research participants’ reproductive health. In conjunction with the recommendations in the present paper, the validation of a standardized retrospective tool to accurately identify STRAW+ 10 stage would improve confidence in stage assignment at the time of data collection, facilitate the application of stage-based inclusion and exclusion criteria, and allow for cross-investigation comparisons. Wearable technology (e.g., Oura ring, Apple Watch) can retrospectively detect ovulation based on physiological biomarkers ( Thigpen et al., 2025 ; Wang et al., 2025b ). The menopausal transition is characterized by an increase in anovulatory cycles ( O’Connor et al., 2009 ). Additionally, advances in wearable monitoring devices now allow for continuous hormone measurement that will permit a more precise understanding of how hormonal changes during perimenopause influence health outcomes in vivo ( Ye et al., 2024 ). By leveraging these developments in in wearable technology, future studies will be able to examine the granular associations between hormone levels, anovulation, reproductive stage, menopausal symptoms, and other aspects of health.

In the nearly 15 years since STRAW+ 10, there has been noteworthy progress in hormone assay and retrospective questionnaire development (e.g., individuals with HIV/AIDS). In 2011, the STRAW+ 10 workshop participants highlighted the lack of highly sensitive, standardized, and low-cost assays as a major barrier for efforts to incorporate hormonal changes into reproductive staging ( Harlow et al., 2012 ). At the time of the workshop, clinical assays for AMH with low enough limits to assess ovarian reserve in perimenopausal women were not widely available ( Iwase et al., 2016 ). However, an ultrasensitive enzyme-linked immunosorbent assay (ELISA), the MenoCheck picoAMH ELISA, was approved by the United States Food and Drug Administration in 2018 for diagnostic use in the determination of menopausal status based on data from SWAN ( U.S. Food and Drug Administration, 2018 ). This assay has a limit of detection of 1.85 pg/mL, far lower than previous AMH assays, and can accurately predict when an individual will experience the FMP ( Finkelstein et al., 2020 ). As an ovarian hormone (rather than FSH, which is released by the pituitary), AMH levels are a more direct assessment of ovarian reserve ( Cedars, 2022 ). Additionally, AMH levels remain relatively stable across a menstrual cycle, unlike FSH and inhibin b levels which fluctuate ( Streuli et al., 2008 ). Thus, the timing of a blood sample for AMH assay does not have the same scheduling considerations. For many studies, being able to accurately determine reproductive stage at enrollment with a questionnaire would be invaluable. This has historically been challenging, as there is no validated retrospective questionnaire to gauge STRAW+ 10 stage (although there have been several recent efforts, described below). There are often discrepancies between perceived and actual changes in bleeding patterns that make retrospective self-assessment of cycle variability, and therefore assignment of reproductive stage, unreliable ( Paramsothy et al., 2013 ; Taffe and Dennerstein, 2000 ). In another example, a recent systematic review illustrated the challenges when trying to apply STRAW (2001) criteria to retrospective, self-reported cycle variability ( Ambikairajah et al., 2022 ). Most of the studies included in the review, which were published 2005–2018 and focused on changes in fat mass across reproductive aging, used definitions consistent with STRAW’s criteria for postmenopause (i.e., at least 12 months of amenorrhea). However, the authors found heterogeneity in the definitions for “premenopause,” which allowed for the inclusion of individuals in this group who would likely meet criteria for STRAW’s early or late menopausal transition stages ( Ambikairajah et al., 2022 ; Soules et al., 2001 ). Additionally, the exact wording used in a questionnaire can influence stage assignment (see the example of misclassification due to the prioritization of bleeding patterns in the past three months over bleeding patterns in the past year in SWAN, MWHS, and MWMHP described above). In the Study of Women Entering and in Endocrine Transition (SWEET), STRAW+ 10 stages were assigned based on answers to both close- and open-ended questions about their menstrual cycles ( Jaff et al., 2014 ). Participants provided the date of their most recent menstrual period. If participants endorsed a recent “skipped” period, they were asked if their most recent period occurred in the (a) past three months, (b) past six months, (c) past year, or (d) over a year ago and then were asked follow-up questions about any cycle changes they were experiencing. However, this approach may not have accurately (or at least not methodically) assessed the criterion for the late menopausal transition (amenorrhea of over 60 days), underscoring the critical need for a standardized structured interview or questionnaire to assess menstrual cycle characteristics. Two recent studies have attempted this. First, investigators from the Australian Women’s Midlife Years (AMY) Study used a retrospective menstrual cycle questionnaire to assign STRAW+ 10 stage. Participants who endorsed “My menstrual cycle has changed in length (by at least one week) and I now get my period more often or less often” were assigned to the early menopausal transition (stage −2) and “I have not had a period for more than 3 months but within the last year” were assigned to the late menopausal transition (stage −1) ( Islam et al., 2025 ). However, this approach did not query for amenorrhea of over 60 but less than 90 days. Second, the Epidemiological Investigation of Menopause Status among Chinese Women (EIM-CW) used the time since self-reported most recent period and self-reported menopausal status (binary) to assign STRAW+ 10 stage ( Wang et al., 2025a ). Of note, individuals whose recent menstrual period was over 60 days ago and self-reported that they were not menopausal were assigned to stage −1, consistent with the STRAW+ 10 criterion for amenorrhea length. However, neither of these retrospective approaches (SWEET, EIM-CW) asked about bouts of amenorrhea of over 60 days among those who recently menstruated and therefore potentially misclassify stage −1 individuals into stage −2, just like SWAN, MWHS, and MWMHP. While the STRAW staging criteria were originally only applicable to generally healthy women, the consensus at STRAW+ 10 was that the updated staging criteria could be applied to individuals who smoke and individuals with BMI > 30 or < 18 kg/m 2 . Still, the attendees at STRAW+ 10 concluded that applying these criteria to individuals with oligomenorrhea or amenorrhea due to reasons other than reproductive aging, such as polycystic ovary syndrome (PCOS), human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS), uterine ablations, and chemotherapy, would likely not be appropriate. As such, perimenopause is not well characterized in these populations ( Millan-de-Meer et al., 2023 ), and health care providers often lack specialized training and confidence in menopause management for these individuals ( Chirwa et al., 2017 ). Over the past nearly 15 years, STRAW+ 10 staging has successfully been applied to individuals who are HIV-positive ( Jaff et al., 2014 ; Jalil et al., 2021 ). Updating the staging criteria or proposing an alternative staging scheme based directly on measures of ovarian aging (e.g., AMH values) rather than bleeding patterns would permit future stage-based research in other populations with altered menstrual patterns.

Building on the long history of reproductive staging in clinical research, assigning STRAW+ 10 stages using prospectively tracked vaginal bleeding data across a longitudinal study can provide valuable insights into potential associations between perimenopause and psychiatric symptoms ( Brown et al., 2024 ). This approach could extend to examine quality of life, changes in drug efficacy, and structural and functional neural integrity across the menopausal transition ( Gonzalez-Rodriguez et al., 2022 ; Matthews and Bromberger, 2005 ; Than et al., 2021 ). In clinical settings, identification of reproductive stage using an observable, low-cost method like prospectively tracked vaginal bleeding patterns collected between appointments could alert healthcare providers to periods of increased vulnerability to major depressive disorder recurrence and other serious mental health conditions such as menopause-associated psychosis ( Bromberger et al., 2015 ; Brown et al., 2024 ). Following the effort to standardize terminology by STRAW+ 10, validated methods to collect, process, and analyze prospective bleeding data will facilitate more precise and reliable reproductive staging that are critically needed to increase our understanding of how perimenopause increases risk for adverse health outcomes and develop novel interventions to increase quality of life during reproductive aging. However, while this manuscript provides recommendations for prospective reproductive staging, several critical research directions must be pursued to validate and implement these approaches effectively. Priority areas include: (1) validation studies comparing digital methods against established approaches and demonstrating predictive validity for health outcomes, (2) open-source algorithms and inter-laboratory reliability protocols, (3) a standardized, validated retrospective questionnaire to assign reproductive stage, (4) equity research addressing technology access barriers and cultural adaptation needs for prospective digital staging methods, (5) clinical translation studies evaluating cost-effectiveness and healthcare integration, and (6) population-specific staging validation in diverse and underrepresented groups. These validation and implementation studies will be essential for transforming the current recommendations from theoretical frameworks into evidence-based, widely adopted methodologies that can advance our understanding of reproductive aging and improve health outcomes for women worldwide.

Approximately half of the global population will experience the menopausal transition: the decline in ovarian reserve and eventual cessation of menses or, for a minority of individuals, “surgical” menopause through the removal of ovaries ( Price et al., 2021 ). Bothersome symptoms, such as hot flashes, sleep difficulties, and mood and cognitive disturbances, peak during the perimenopause—the period of time between the end of the late reproductive years and the beginning of the postmenopausal stage ( Ambikairajah et al., 2022 ; Harlow et al., 2012 )— and cost up to approximately $1.8 billion in lost work productivity in the United States every year ( Faubion et al., 2023 ). However, a major gap in knowledge about the menopausal transition exists due to research and funding inequities ( Gilmer et al., 2023 ; Ledford, 2023 ; Smith, 2023 ). Even tracking funding support for menopause-related research has been challenging, as “menopause” was only added to the National Institutes of Health (NIH) Research, Condition, and Disease Categories (RCDC) in 2023. Along these lines, the 2024–2028 NIH-Wide Strategic Plan for Research on the Health of Women emphasizes the urgent need for a specific research focus on menopause and understanding women’s health and disease across reproductive stages. The menopausal transition is accompanied by increased incidence of many chronic health conditions, including hypertension, stroke, arthritis, osteoporosis, diabetes, and coronary heart disease ( Ghosh et al., 2014 ). Peri- and early postmenopause are also associated with exacerbation or new onset of psychiatric disturbances such as mood and psychotic disorders and new onset of cognitive complaints ( Bromberger et al., 2001 ; Culbert et al., 2022 ; Epperson et al., 2013 ; Freeman et al., 2007 ; Michopoulos et al., 2023 ; Page et al., 2023 ; Schmidt et al., 2004 ; Shitomi-Jones et al., 2024 ). Cognitive and mood concerns have been shown to be more prominent in specific stages of the menopausal transition, emphasizing the importance of “fine tuning” how stages of menopause are determined and defined ( Metcalf et al., 2024 ; Metcalf et al., 2022 ). Perimenopause is characterized by alterations in the function of the hypothalamic-pituitary-ovarian (HPO) axis. More specifically, diminishing numbers of follicles result in decreasing levels of inhibin B, increasing follicle-stimulating hormone (FSH) due to the reduction of negative feedback from inhibin B, and fluctuating levels of (E2) and progesterone (P4) ( Fig. 1 ). For an individual, these changes in hormones result in increasing menstrual cycle irregularity until the final menstrual period (FMP) occurs. Precise identification of reproductive stage is critical to understand the risk and deploy effective interventions for mid-life health conditions, as new onset or exacerbation of psychiatric, cognitive, neurologic, and cardiovascular conditions appear to be related to reproductive aging rather than chronological age ( Bromberger et al., 2001 ; Culbert et al., 2022 ; El Khoudary and Nasr, 2022 ; Michopoulos et al., 2023 ; Mosconi et al., 2021 ; Schmidt et al., 2004 ; Shitomi-Jones et al., 2024 ). Unfortunately, the nomenclature related to the menopausal transition has often been inconsistently defined and operationalized in clinical research ( Ambikairajah et al., 2022 ) and can conflict with colloquial usage ( Cunningham et al., 2025 ). This has led to confusion for patients and hindered focused translational and clinical research. Over the past several decades, experts have made many notable advancements in refining terminology and identifying reproductive stage demarcations based on observable bleeding patterns and hormone levels, yet more work is needed to understand how these reproductive stages relate to risk for physical and mental health conditions, which is the focus of many ongoing longitudinal studies. A critical methodological gap remains in that research studies must identify where their participants are in the process of reproductive aging at enrollment and apply the current gold standard for reproductive aging staging in longitudinal data in a reproducible, standardized manner. Therefore, the purpose of the current review is to examine historical methods and the current gold standard of determining reproductive stage, discuss important considerations in prospective methods for tracking bleeding patterns, provide recommendations for future studies ( Box 1 ), and discuss future research directions.

Many individuals have inaccurate retrospective recall of most recent menstrual period, FMP, and menstrual cycle characteristics ( Bean et al., 1979 ; Wegienka and Baird, 2005 ). For example, the kappa coefficient between reported and observed cycle lengths among American women in their late 30’s was only 0.33, suggesting moderate agreement ( Jukic et al., 2008 ). In a contraceptive clinical trial, only 53 % of participants had prospectively tracked menstrual cycle lengths that were within seven days of their retrospectively reported cycle lengths ( Steiner et al., 2001 ). Thus, we recommend prospective data collection of vaginal bleeding patterns to eliminate/minimize recall bias and inter-individual differences in perceptions of “variability.” This is especially important for longitudinal studies that are examining changes in, for example, psychiatric and cognitive symptoms and their neural correlates across midlife. Perimenopause is a life stage characterized by dramatic changes in hormonal levels and reproductive function, metabolism, cardiovascular health, neurological health, sleep quality, and psychological health ( Brinton et al., 2015 ; Marlatt et al., 2022 ; Santoro, 2016 ). However, pinpointing the transitions from late reproductive (stage −3a) to early (−2) and late (−1) menopause transition and early postmenopause (+1a and onwards), is extremely challenging, if not impossible, to do retrospectively. We recommend collecting as much data on vaginal bleeding patterns as reasonably possible, given the study’s resources. Ideally, prospective tracking would start two or three months prior to the first study assessment visit and continue for at least two or three months after the final study assessment. The additional data on either end of the visit timeline will likely provide more reliable stage estimates for the beginning and end of the study. For studies focused on the early menopausal transition, collecting data for at least 10 consecutive menstrual cycles would be necessary to confirm if a participant is in stage −3a or stage −2, based on the STRAW+ 10 definitions. However, we acknowledge that this ideal length of data collection can be prohibitive in practice, given the challenges of engaging participants for long periods of time and financial constraints from most grant mechanisms. For studies that enroll participants who are already experiencing persistent variability in cycle length or short bouts of amenorrhea, prospective data collection should continue until stage −1 (e.g., at least 60 days of amenorrhea) or stage + 1a (e.g., 365 days of amenorrhea) can be confirmed, depending on study goals. Lengthy prospective staging methods may not be feasible for studies that are shorter in duration or even cross-sectional; careful consideration of the stage(s) of interest will facilitate the selection of the most appropriate data collection timeline. As STRAW+ 10 stages can only be assigned using bleeding patterns, frequent data collection is vital for accurate staging. A prospective sampling design that requires daily recording poses a significant participant burden. Although previous studies have been able to successfully collect daily vaginal bleeding data for multiple years, there is a concern that participants generally do not record data for non-events ( Taffe and Dennerstein, 2002 ; Treloar et al., 1967 ). Thus, there may be a problematic overlap between missing daily data and daily data in which bleeding did not occur. Weekly reports, in which participants report bleeding patterns for the past seven days, are likely more feasible. In other contexts, weekly recall of objective events like physical activity tends to have high correspondence with daily diaries ( Dishman and Steinhardt, 1988 ). Even weekly recall of subjective events like pain is consistent with daily diaries ( Jamison et al., 2006 ). Asking about menstrual cycle patterns on a monthly or quarterly basis likely opens more opportunities for bias, including recall bias ( Wegienka and Baird, 2005 ), digit preference bias ( Jukic et al., 2008 ; Savitz et al., 2002 ; Small et al., 2007 ; van Oppenraaij et al., 2015 ; Waller et al., 2000 ), overestimation of cycle length ( Jukic et al., 2008 ), and stereotyped thinking about variability ( Small et al., 2007 ). Notably, recall accuracy declines dramatically over the month following vaginal bleeding, with the most accurate recall occurring within a week of recent vaginal bleeding and the least accurate recall when asked three or more weeks after recent vaginal bleeding ( Wegienka and Baird, 2005 ). Therefore, we recommend that bleeding data should be recorded on at least a weekly basis to minimize bias. Due to the length of data collection required to capture stage transitions, as mentioned in the section above, weekly data collection is likely more acceptable to participants, while still minimizing recall bias. For shorter studies that focus on more granular changes in mood or menopause-related symptoms and the association with vaginal bleeding, daily reporting of both symptoms and bleeding patterns may be desirable. In that case, a daily survey could be completed in the evening to document any vaginal bleeding that day. Ask participants to describe the flow amount for days on which bleeding occurred. A simple validated tool such as the Mansfield-Voda-Jorgensen Menstrual Bleeding Scale (MVJ) could be used to capture this data ( Mansfield et al., 2004 ). This 6-point Likert scale asks about how often the respondent would need to change a sanitary pad/tampon for a given flow amount, rather than how often the respondent prefers to change it. The correlation between MVJ score and actual blood volume loss is very high (r = 0.683) ( Mansfield et al., 2004 ). This level of detail allows researchers to examine correlates of heavy flow, changes in flow across perimenopause, and identify spotting outside of menstrual periods. Currently, flow quantity and spotting patterns are not included in STRAW+ 10 staging criteria; however, high-quality prospective data collection that includes these factors will improve our understanding of the complex changes in vaginal bleeding patterns that occur during perimenopause. For example, the SWAN menstrual calendar substudy collected and analyzed data on flow amounts ( Paramsothy et al., 2014 ). “Spotting” was defined as not requiring or filling a regular-sized sanitary product, “light-to-moderate bleeding” as only needing to change sanitary products a few times during the day or every 3–4 h, and “very heavy bleeding” as needing to change sanitary products every 1–2 h for at least a four-hour period during the day. Results from this analysis highlighted the prevalence of menses with prolonged spotting and very heavy bleeding during the menopausal transition. Conversely, the SMWHS menstrual calendar cards did not include specific definitions but rather asked participants to score the flow amount for bleeding days on a scale from 1 to 4 (1 = light, 2 = moderate, 3 = heavy, 4 = very heavy/flooding) ( Woods and Mitchell, 2016 ). Due to interindividual differences in relative flow amounts and the lack of accepted standardized definitions, definitions of different flow amount categories should be provided to both participants and included in publications. Hormonal medications are commonly used by perimenopausal individuals for contraception, management of menopause-related symptoms, gender-affirming care, and the treatment of gynecological conditions such as endometriosis, adenomyosis, uterine fibroids, PCOS, abnormal uterine bleeding, and endometrial hyperplasia ( National Academies of Sciences and Medicine, 2024 ; Imai et al., 2014 ; Rousseau, 1999 ). Common hormonal medications include combined hormonal contraceptives (CHCs), progestin-only contraceptives, hormone replacement therapy (HRT), hormone-releasing intrauterine devices (IUDs), selective estrogen receptor modulators (SERMs), gonadotropin-releasing hormone agonists/antagonists, aromatase inhibitors, and gender-affirming hormone therapy (e.g., testosterone for transmasculine individuals). These treatments exert their effects via modulation of the HPO axis and direct actions on the endometrium ( Amiri et al., 2018 ; Imai et al., 2014 ; Pinkerton and Stanczyk, 2014 ; Rousseau, 1999 ). For example, CHCs and progestin-based therapies suppress ovulation and stabilize endometrial tissue, often resulting in amenorrhea or more regular, lighter periods ( American College of Obstetricians and Gynecologists’ Committee on Clinical Consensus–Gynecology, 2022 ; Amiri et al., 2018 ). The levonorgestrel-releasing IUD is particularly effective for menstrual suppression and is commonly used to treat dysmenorrhea and endometriosis-associated pain ( Imai et al., 2014 ). HRT, used for menopausal symptoms, can cause irregular or heavy bleeding, especially when regimens are adjusted, and the pattern depends on whether therapy is sequential/cyclical or continuous ( Harper-Harrison et al., 2025 ). Gender-affirming hormone therapy with testosterone typically leads to amenorrhea due to endometrial atrophy, though breakthrough bleeding can occur, particularly early in treatment or with sub-physiologic dosing ( American College of Obstetricians and Gynecologists’ Committee on Clinical Consensus–Gynecology, 2022 ). Failure to account for hormonal medication use and related alterations in bleeding patterns can confound the classification of menstrual patterns and reproductive stage, as both exogenous hormones and underlying pathologies may independently affect bleeding patterns ( Rousseau, 1999 ; Schmalenberger et al., 2021 ). Study designs should exclude cycles during which hormonal medications were used, create separate analytical categories, or at minimum, collect detailed information on current and recent hormonal medication use (type, dosage, duration) as potential confounders ( Schmalenberger et al., 2021 ). Rigorous screening and documentation will enhance the accuracy and interpretability of research findings and ensure recommendations are robust and generalizable across diverse populations ( Rousseau, 1999 ; Schmalenberger et al., 2021 ). Throughout the study, screening for STRAW+ 10 contraindications is recommended to ensure that STRAW+ 10 staging is not inappropriately applied or at least can be interpreted with caution ( Harlow et al., 2012 ). Ideally, this is conducted at the beginning of the study, as some conditions and medications described below may be exclusion criteria. Screening should be repeated at regular intervals (e.g., quarterly or biannually) until the conclusion of data collection. Medical conditions that can affect vaginal bleeding patterns and therefore render STRAW+ 10 criteria inappropriate to apply include HIV/AIDS, PCOS, primary ovarian insufficiency, premature ovarian failure, and cancer ( Harlow et al., 2012 ). For example, women living with HIV/AIDS are more likely to experience periods of amenorrhea and oligomenorrhea compared to HIV-negative women ( Chirgwin et al., 1996 ; Ezechi et al., 2010 ). The majority of perimenopausal individuals undergoing chemotherapy experience alterations in vaginal bleeding, although pre-treatment vaginal bleeding patterns occasionally resume ( Kabirian et al., 2023 ; Sukumvanich et al., 2010 ). Additionally, regular screening for common medical procedures in this age group is vital. Hysterectomies, double oophorectomies, and uterine ablations and embolizations disrupt menstruation (or cause it to cease entirely). In particular, hysterectomies are the most common non-obstetric surgical procedure among female patients aged 18–64 ( National Hospital Discharge Survey, 2012 ). Data collected after these procedures should not be used to perform reproductive staging, as per the STRAW+ 10 recommendations. In the future, hormonal levels (e.g., estradiol, FSH, and AMH) could be used to stage these individuals once standardized criteria are established. Many states in the U.S. and countries have laws restricting and/or criminalizing access to reproductive healthcare. Research participants may be concerned about sharing data about their menstrual cycle patterns and hesitant to participate in reproductive health-related research studies ( Weinmeyer et al., 2023 ). For example, many users of period tracking apps worry about how their data is stored, who has access to it, and if it would be sold to third parties ( Mohan and Jenkins, 2025 ). Asking individuals to track bleeding patterns for reasons beyond self-monitoring and share the data with a research team may increase anxiety about data privacy. Therefore, we recommend the use a secure, digital platform (e.g., REDCap, Qualtrics) that prioritizes ease of use for the participant—ideally a platform that is accessible from a phone, tablet or computer. Smartphone applications that require downloading and installation, regular updates, and/or rely on the individual’s personal cell phone notification settings may not be ideal, as research staff may need to troubleshoot these issues regularly. This is a particularly critical consideration as capturing the menopausal transition and the FMP could require many months or even years of prospective data collection for large cohorts. A user-friendly platform that is straightforward to deploy and maintain for many participants is critical for the success of a long-term, longitudinal study design. However, it is important to note that relying on this approach to collect data could introduce disparities as individuals of lower socioeconomic status may not have a device and/or internet access. Research teams should consider ways to address these barriers, such as providing participants with internet-enabled devices. To reduce burden on the research team and enhance completion rates, participants should receive automated reminders to submit their data. The reminder notification should be personalized and delivered as per the participant’s communication preferences (e.g., text, email, or both). Drawing on recommendations for ecological momentary assessments (EMA) and other frequent reporting methodologies, compensation for completing vaginal bleeding reports should be commensurate with the required time and effort and sufficiently incentivize high completion rates. Businelle and colleagues found that compensating participants with a fixed amount per completed EMA versus with a variable amount based on the percentage of completed EMAs at the end of the assessment period did not result in a significant difference in compliance rates ( Businelle et al., 2024 ). A combination of these compensation structures (a fixed amount per completed report plus a bonus for perfect or near-perfect completion per month) may be superior to either method alone. For weekly bleeding pattern reports, this could be a nominal amount per completed survey plus a monthly bonus for 100 % completion. Beyond this, other optimal retention strategies employed by successful longitudinal studies should be leveraged to promote retention. These strategies include prioritizing excellent communication skills when recruiting research staff, fostering a strong relationship between the research team and the participant by designating a primary research team member to consistently contact each participant, and maintaining personal engagement (e.g., sending birthday and holiday cards, sharing relevant resources) and even creating a sense of community within the cohort (e.g., small group discussions, annual events) ( Abshire et al., 2017 ). Encouraging enthusiastic participation is vital for long, longitudinal studies with frequent data collection time points. Rather than calculating cycle length manually, use a computerized set of algorithms to apply the STRAW+ 10 criteria to the data and identify changes in menstrual cycle patterns across months and years. Assessing bleeding characteristics manually is impractical for smaller research groups, or in the case of very large sample sizes, and introduces opportunities for human error. In response to this challenge, our research team has developed and published R code to apply STRAW+ 10 stages to prospectively collected vaginal bleeding data among a cohort of Black women aged 40–55 years ( https://github.com/meganhuibregtse/peri-code ). Our code requires a comma-separated values (csv) file of bleeding data collected weekly—participants reported if they had experienced any menstrual bleeding within the past week and, if so, on which days—into long data of one row per participant per day with variables for participant identification number, observation date, and bleeding (binary). Our algorithms could be applied to data already in this long format with some modifications to a wrapper function that applies the set of staging-related functions (e.g., identification of bleeding cycles, calculation of cycle length, comparison of consecutive cycle lengths, identification of criteria for each STRAW+10 stage). Our operational definitions for key terms and stages are summarized in Table 1 . In particular, “menstrual cycle” (also referred to a “bleeding segment” or a “period”) has been inconsistently defined in the literature, such as (1) at least one day of bleeding between any number of bleeding-free days, (2) at least one day of bleeding followed by at least three bleeding-free days, or (3) at least two days of bleeding which could be interspersed with up to two bleeding-free days followed by an interval of at least four consecutive bleeding-free days ( Belsey and Carlson, 1991 ; Belsey and Farley, 1988 ; Belsey and Pinol, 1997 ; Harlow et al., 2008 ; Paramsothy et al., 2013 ; Paramsothy et al., 2014 ; Rodriguez et al., 1976 ; Woods and Mitchell, 2016 ). The definition used by the ReSTAGE collaboration was based on the recommendation from the World Health Organization (WHO) and required at least three bleed-free days given bleeding patterns observed in the late reproductive and menopausal transition stages ( Harlow et al., 2006 ; Harlow et al., 2008 ). In our algorithm, we used a relatively inclusive definition for a menstrual cycle of at least one day of vaginal bleeding (preceded by at least two days without any vaginal bleeding) followed by a bleeding-free interval of at least two days. For example, four days of vaginal bleeding (which were preceded by more than two bleeding-free days) followed by one bleeding-free day, one day of bleeding, and 23 bleeding free days would be considered one menstrual cycle with a length of 29 days ( Fig. 3 ). The single day of bleeding would not be considered the start of the next menstrual cycle as it was not preceded by at least two bleeding-free days. This approach is in line with the Apple Women’s Health Study, a massive digital cohort study that began in 2019 and has already enrolled over 120,000 participants ( Li et al., 2023 ). For research teams interested in applying different operational definitions of a bleeding segment/cycle, we are adding a feature to our algorithm that would allow users to select an alternative (e.g., ReSTAGE operational definitions). The selection of the number of bleeding-free days required to precede the start of a new menstrual cycle may ultimately depend on the reproductive stage(s) of interest. Missingness can be minimized at the data collection stage by incentivizing high completion rates, yet some amount of missingness is to be expected in lengthy, prospective vaginal bleeding tracking data-sets. Bleeding cycles with missing data should not be used for staging. In practice, if any weekly reports are missing, then the days between the start of bleeding prior to the missing data and the start of the next complete menstrual cycle should be censored from the staging algorithm ( Fig. 3 ). For individuals who are reporting a long period of amenorrhea, the number of missing days of data can be added to the minimum days to meet criteria for Stage −2 (at least 60 days of amenorrhea) or Stage + 1a. For example, if a participant has been reporting amenorrhea but has missed three weekly surveys after their last episode of vaginal bleeding, then 81 bleeding-free days would be required to meet criteria for Stage −2 (or 386 bleeding-free days for Stage +1a). Further, incorporating a minimum compliance threshold (e.g., 60–70 %) or a compliance grading system (e.g., <60 % = low, 61–75 % = fair, 76–90 % = acceptable, ≥91 % = excellent) may be considered to increase confidence in the stage assignments. Furthermore, the collection and storage of biospecimens (e.g., blood, urine, and saliva) can provide future opportunities to validate stage assignments with circulating hormone levels (see Table 2 for reference ranges by stage), if not already part of the primary study aims. Levels of AMH, FSH, and inhibin B were included in the STRAW+ 10 staging system but only as supportive criteria ( Harlow et al., 2012 ). The collection of biospecimens permits testing of novel hypotheses about the associations between hormones levels during a specific reproductive stage and biological markers of inflammation, the immune response, cardiovascular health, and neurological health. Thus, we recommend biospecimen collection whenever possible to assess how hormones influence mechanisms that confer risk for adverse health outcomes during reproductive aging. Serum and salivary levels of E2, which is primarily synthesized in the ovaries and is a neuroactive steroid hormone, drop considerably by early postmenopause ( Stanczyk and Clarke, 2014 ; Tivis et al., 2005 ), and serial urine sample collection may be particularly advantageous instead. This non-invasive approach can be employed by participants at home, as urinary metabolite concentrations correlate strongly with serum concentrations yet exceed than those observed in serum. Additionally, urine metabolites are relatively stable at room temperature for up to 1–2 days and are resilient to multiple freeze-thaw cycles ( O’Connor et al., 2003 ). The timing of urine sample collection should be scheduled with the understanding that they reflect hormone levels from the day prior.