Increase in BMI is associated with increasing cycle length
Changes in body composition and fat distribution have been the focus of numerous epidemiological studies. A longitudinal study of 1000 girls 11 reported that elevated BMI causes early onset of menarche, to more severe premenstrual symptoms, and PCOS. This may be due to the impact of nutritional status sensing through leptin and kisspeptin action influencing the activity of gonadotropin releasing hormone (GnRH) 12,13. In addition, adipose tissue that is responsible for maintenance of body fat depots is estrogen-sensitive and subject to changes in hormonal availability 14. In normo-typical scenarios, estrogen signaling regulates the glucose and lipid metabolism within adipose cells, however, in obese conditions, the link is believed to be severed due to adipose tissue inflammation.
Within our globally-sourced user base, we found cycle length range between 28-30 days for users with BMIs between 18-25 (Figure 1) across the 25-40 years spectrum with appreciable fluctuations (error bars depicting interquartile range, Figure 1d) appearing either pre-25 or post-45 years of age. As BMI increased, though the cycle lengths ranged between 28-30 days, the fluctuations were much higher in the younger age group (<25 years) and around 40 years (Figure 1b). In the greater than 30 BMI segment (termed obese), barring there were small sub-groups where the cycle length spanned 28-30 days, with the general median trends indicated shortening of cycles with increasing age (Figure 1c).
Figure 1: Graphical representation of median menstrual cycle length across normal, overweight and obese BMI women, aged 20 to 55 (N(normal)=1,642, N(overweight)=654, N(obese)=449). Error bars represent IQR.
We refined our analysis by plotting the interquartile intervals across all BMI categories (Figure 1d). Ring AIR women users with >25 BMI (blue and green lines) showed higher variability in cycle range than those within the 18-25 BMI range (red line). We compared the absolute median cycle length of women in all three categories and compared the variation using a Levene’s test. We found a trend level difference with increasing BMI within the 20-25 year age group (p=0.1), while a statistically significant extension in cycle length (p=0.04) in women of the 40-45 year age group.
Figure 1d: Interquartile ranges of menstrual cycle length across normal, overweight and obese BMI women, aged 20 to 55. ‘*’ denotes p<0.05.
Stress has a high impact on cycle length variability but has age specificity as well
Stress can be caused by a myriad of factors and activates an ancient response pathway in our bodies. That stress impacts cycle function is well known for the past 100 years with the precise mechanism of action emerging recently 15. The Ultrahuman stress rhythm score provides an insightful proxy into the daily stress processing of women Ring AIR users. We looked at the average stress level of women in this group across four months to represent the chronic stress levels.
Surprisingly, in the low stress levels group, younger women had the most cycle length variability, with fluctuations dissipating with age (Figure 2a). Women reporting medium levels of stress, seemed to have the most consistent cycle lengths in this entire dataset (Figure 2b). Women who reported consistently high levels of stress showed the maximum variability in either direction on cycle length (Figure 2c). Cycle lengths in this segmentation analysis seemed to decline with age.
Figure 2: Graphical representation of median menstrual cycle length across low, mid and high stress women, aged 20 to 55 (N(low)=1,204, N(mid)=1,372, N(high)=169). Error bars represent IQR.
IQR analysis (Figure 2d) that within the high stress group, cycles varied between 2-8 days across the age span. Interestingly, we did obtain some 0 day variability in IQR in the high stress group, which may be reporting artifacts. In the medium stress group the IQR values were higher in the younger age groups (2-10 days), with a standalone peak at 50. In contrast, low-stress members showed a widely varying age profile with a more modest fluctuation range of 4-6 days. IQR segmentation yielded a statistically significant difference between the groups for the 20-25 range. A Kruskal-Wallis test to examine variations in IQR across all ages among the three stress-level groups revealed a statistically significant difference, with a Dunn’s post hoc (p=0.0086, high vs low stress).
Figure 2d: Interquartile ranges of menstrual cycle length across low, mid and high stress women, aged 20 to 55. ‘*’ denotes p<0.05.
Daily activity levels have mild impact but are directionally associated with shorter cycles with age
While directed exercise has been a cornerstone of longevity, intense exercise among women athletes has been associated with luteal phase shortening and missed periods 16. We looked to see if there was a case for increased average daily activity with cycle synchronicity. We found no apparent trend of cycle length variability with increasing daily activity ranges. However, within groups based on activity, we observed a gradual trend of small cycles with age (difference of Figure 3a-c). Plotting IQR values to check whether increased activity did show increased cycle variation across all age groups, but this effect was not statistically significant (Figure 3d).
Figure 3: Graphical representation of median menstrual cycle length across sedentary, moderately active and active women, aged 20 to 55 (N(sedentary)=1,086, N(moderately active)=1,034, N(active)=625). Error bars represent IQR.
Figure 3d: Interquartile ranges of menstrual cycle length across sedentary, moderately active, and active women, aged 20 to 55.
Conclusions, Limitations and Future Directions
Cycle length is one of the most noticeable aspects of a woman's reproductive health and serves as a key indicator for tracking the onset of menarche, using the calendar method for fertility, and monitoring the transition into menopause. This study was an effort to provide longitudinal insights into how menstrual cycle length changes with age in the backdrop of metabolic differences (BMI), stress and daily activity and is a preliminary analysis of trends observed within our user base.
While it was expected that inter-cycle variability would increase with increasing age after 40 years as a group effect, we registered an intriguing variability in young women of 20-25 years in BMI and stress stratification groups. We found that sample size was not a contributing factor for 20-25 years and 40-45 years were similar (N(20-25)=426, N(40-45)=405). It is therefore likely that our representative sample among young women may have other reasons for variable cycling. This additional information is not collected on the Ultrahuman application at the moment and hence merits future in-depth analysis. We also had sparse data sets in the 50+ years age range when segmented by high stress. Hence there were discontinuous representations in Figure 2c. It was noteworthy that in general, across any segmental analysis, the most frequent cycle length remained 28-30 days, further supporting the global consensus on this number.
It is well known that a woman’s gynaecological function is impacted by several key lifestyle factors, such as diet, activity, stress, and sleep and the trends at the epidemiological level have been largely replicable. However, multi-decade longitudinal studies are challenging to execute and track, which makes initiatives like SWAN 17 and PREDICT 18 cohorts critical for guidance values. Wearable tracking information tagged with remote at-home testing for hormonal panels provides a powerful decentralized method of gathering baseline information of a woman’s changing physiology. Wearables like the Ring AIR that gather cumulative data on heart rate, temperature, and oxygen saturation may be in a position to uncover other associational trends that can help anticipate/predict ovulation or menopause symptoms like vasomotor symptoms (hot flashes) better. Hence we hope that this initial survey can form a launching pad for community stakeholders in their study design and hypothesis building.
The Ultrahuman Ring AIR, along with it’s connected ecosystem (M1 CGM Sensor, Home Health, and Blood Vision), can leverage emerging trends to offer more precise and personalised tracking of menstrual cycle phases. We are committed to supporting women across their healthy aging journey and provide such high level group patterns for better interpreting individual data.
Reach out to support@ultrahuman.com for commercial queries and science@ultrahuman.com for scientific queries.
