The association of activity patterns on female reproductive diseases: a prospective cohort study of UK biobank

Table  1 shows the differences in the baseline characteristics of 49,540 participants in terms of demographic, anthropometric, lifestyle and disease histories of participants according to activity patterns. Compared to participants who engaged in MVPA during the midday-afternoon, those with a mixed-timing pattern of MVPA were more likely to be non-white, have higher Townsend deprivation index scores, engage in more total daily exercise, and accumulate more MET minutes per week across all activities. They also tended to have lower BMI, a higher proportion of college graduates or above, a higher moderate drinking rate and current smoking rate, and a lower prevalence of cancer, hypertension, diabetes, and COPD. Furthermore, we compiled a baseline characteristics table for participants who were included and excluded due to missing or invalid accelerometer data ( n  = 502,411), as shown in Supplementary Tables 1 , to demonstrate the generalizability of the study sample. In addition, a baseline characteristics table excluding participants with baseline female reproductive diseases ( n  = 42,099) is presented in Supplementary Table 2 . The baseline characteristics of this group are consistent with those reported in Table  1 . Table 1 Baseline characteristics according to activity patterns Characteristics activity patterns Morning Evening Mixed Midday-afternoon P -value Number of participants 8096 3726 15,142 22,576 Age (years) 56.7(7.7) 53.2(7.8) 53.6(7.7) 56.6(7.4) <0.001 White [N, (%)] 7532(93.0) 3343(89.7) 13,615(89.9) 20,800(92.1) 0.001 College graduate or above [N, (%)] 3001(37.1) 1659(44.5) 7000(46.2) 9331(41.3) 0.019 Townsend deprivation index -1.9(2.7) -1.6(2.9) -1.4(2.9) -1.8(2.8) <0.001 BMI (kg/m²) 26.6(4.9) 26.6(5.0) 25.6(4.5) 26.2(4.7) <0.001 Moderate drinking [N, (%)] 6425(79.4) 3016(80.9) 12,548(82.9) 18,378(81.4) 0.001 Current smoking [N, (%)] 450(5.6) 232(6.2) 862(5.7) 1254(5.6) 0.079 Total daily exercise time (minutes) 58.5(14.5) 56.5(14.0) 64.5(15.7) 58.9(14.1) 0.018 Summed MET minutes per week for all activity (minutes) 2570.0(2211.3) 2273.9(1999.1) 2485.2(2126.1) 2479.8(2140.8) 0.025 Current cancer [N, (%)] 556(6.9) 255(6.8) 916(6.0) 1618(7.2) 0.016 Current hypertension [N, (%)] 1599(19.8) 596(16.0) 2197(14.5) 4182(18.5) 0.543 Current diabetes [N, (%)] 63(0.8) 25(0.7) 89(0.6) 149(0.7) 0.181 Current COPD [N, (%)] 34(0.4) 14(0.4) 61(0.4) 116(0.5) 0.991 Continuous variables are presented as mean (SD, standard deviation). Categorical variables are presented as numbers (%, percentage). BMI, body mass index. Moderate drinking was defined as thecriterion: ≤once/month. MET, Metabolic Equivalent Task. COPD, chronic obstructive pulmonary disease Baseline characteristics according to activity patterns Continuous variables are presented as mean (SD, standard deviation). Categorical variables are presented as numbers (%, percentage). BMI, body mass index. Moderate drinking was defined as thecriterion: ≤once/month. MET, Metabolic Equivalent Task. COPD, chronic obstructive pulmonary disease The results of the multivariate Cox proportional hazard regression models examining the association between activity patterns and hazard of female reproductive diseases are presented in Table  2 . After adjustment for age, race, education, Townsend deprivation index, BMI, smoking status, moderate drinking, total daily exercise time, summed MET minutes per week for all activity, and histories of cancer, diabetes, hypertension, COPD, we found that women engaging in morning or mixed-timing exercise were associated with significantly lower risks for female reproductive diseases (HR morning =0.81, 95% CI: 0.67–0.98; HR mixed =0.79, 95% CI: 0.69–0.91, P-trend < 0.05), compared with the reference group. These associations remained consistent across models 2 to 4. Meanwhile, sensitivity analyses also confirmed the stability of the relationship between activity patterns and female reproductive diseases (Table  3 ). Similar findings were also observed after excluding participants diagnosed with cancer. Furthermore, after adjusting for additional covariates, including dietary factors and biochemical indicators, the conclusions remained virtually unchanged. After excluding women diagnosed with reproductive diseases within one year of follow-up, the association between mixed-timing activity and female reproductive diseases remained stable. The protective trend of morning exercise on female reproductive diseases remained unchanged, although it lost statistical significance at the marginal level, possibly due to the smaller sample size of women with reproductive diseases. To further assess the robustness of our findings, we performed data imputation using R 4.2.2 (package “mice”) and re-conducted Cox regression analysis on the imputed data. As shown in the results of Supplementary Table 3 , these findings were consistent with our initial results. Table 2 Hazard ratios and 95% confidence intervals for activity patterns with the hazard of female reproductive diseases Activity patterns Morning Evening Mixed Midday-afternoon P for trend Cases/N 132/6906 121/3071 350/12,743 441/19,379 M1 0.84(0.69,1.02) 1.75(1.43,2.14) 1.21(1.05,1.39) Ref <0.001 M2 0.81(0.69,0.99) 1.05(0.86,1.28) 0.77(0.66,0.88) Ref 0.001 M3 0.81(0.67,0.98) 1.02(0.83,1.25) 0.79(0.69,0.91) Ref 0.005 M4 0.81(0.67,0.98) 1.02(0.83,1.25) 0.79(0.69,0.91) Ref 0.005 Data are HRs and 95%CI; Model 1 was not adjusted; Model 2 was adjusted for age, race, education, Townsend deprivation index; Model 3 was model 2 with additional adjustments for BMI, smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity; Model 4 was model 3 with additional adjustments for histories of cancer, diabetes, hypertension, COPD. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis,infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) Hazard ratios and 95% confidence intervals for activity patterns with the hazard of female reproductive diseases Data are HRs and 95%CI; Model 1 was not adjusted; Model 2 was adjusted for age, race, education, Townsend deprivation index; Model 3 was model 2 with additional adjustments for BMI, smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity; Model 4 was model 3 with additional adjustments for histories of cancer, diabetes, hypertension, COPD. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis,infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) Table 3 Sensitivity analyses for association between activity patterns and hazard of female reproductive diseases Activity patterns Morning Evening Mixed Midday-afternoon P for trend Excluding participants diagnosed with cancer 0.79(0.65,0.97) 1.00(0.81,1.24) 0.79(0.69,0.92) Ref 0.007 Adjusting for covariates dietary factors 0.81(0.67,0.99) 1.03(0.84,1.26) 0.79(0.68,0.91) Ref 0.005 Adjusting for covariates biochemical indicators 0.81(0.67,0.98) 1.02(0.83,1.24) 0.79(0.69,0.91) Ref 0.005 Excluding participants diagnosed with female reproductive diseases within one year of follow-up. 0.84(0.68,1.05) 0.11(0.89,1.39) 0.81(0.69,0.96) Ref 0.038 Results were adjusted for age, race, education, Townsend deprivation index, BMI, smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity, histories of cancer, diabetes, hypertension, COPD at baseline. Dietary factors include fresh fruit intake, dried fruit intake, oily fish intake, salt added to food, cereal intake, processed meat intake, mineral and other dietary supplements. Biochemical indicators include albumin, triglyceride, glucose, LDL, cholesterol, total bilirubin. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis,infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) Sensitivity analyses for association between activity patterns and hazard of female reproductive diseases Adjusting for covariates dietary factors Adjusting for covariates biochemical indicators Results were adjusted for age, race, education, Townsend deprivation index, BMI, smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity, histories of cancer, diabetes, hypertension, COPD at baseline. Dietary factors include fresh fruit intake, dried fruit intake, oily fish intake, salt added to food, cereal intake, processed meat intake, mineral and other dietary supplements. Biochemical indicators include albumin, triglyceride, glucose, LDL, cholesterol, total bilirubin. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis,infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) We also conducted stratified analyses according to the potential risk factors including age at baseline, BMI, Townsend deprivation index, smoking status, moderate drinking, as shown in Table  4 . The protective effect of mixed-timing exercise on hazard of female reproductive diseases was not observed in obese participants (BMI ≥ 30 kg/m 2 ) (P for interaction =0.030). In addition, we found that the association between activity patterns and female reproductive diseases was no longer significant in current-smokers (P for interaction =0.001). No significant interactions were found between other potential confounders and activity patterns on hazard of female reproductive diseases. Table 4 Stratified analyses for association between activity patterns and hazard of female reproductive diseases Activity patterns Morning Evening Mixed Midday-afternoon P trend P for interaction Age at baseline 0.059 <50 years old 0.81(0.65,1.03) 0.93(0.74,1.18) 0.78(0.66,0.92) Ref 0.006 ≥ 50 years old 0.79(0.55,1.14) 1.76(1.19,2.61) 1.06(0.80,1.41) Ref 0.306 BMI 0.030 <25 kg/m 2 0.81(0.61,1.09) 0.97(0.71,1.31) 0.80(0.65,0.98) Ref 0.045 25–29.9 kg/m 2 0.94(0.68,1.29) 1.13(0.80,1.59) 0.74(0.57,0.95) Ref 0.040 ≥ 30 kg/m 2 0.61(0.38,0.97) 0.96(0.62,1.51) 0.86(0.63,1.19) Ref 0.513 Townsend deprivation index 0.251 < median 0.75(0.56,1.00) 1.01(0.75,1.37) 0.76(0.61,0.94) Ref 0.028 ≥median 0.86(0.66,1.12) 1.03(0.78,1.35) 0.82(0.68,0.99) Ref 0.069 Smoking status 0.001 Never 0.84(0.66,1.07) 0.92(0.71,1.20) 0.83(0.70,0.99) Ref 0.047 Ever 0.80(0.56,1.16) 1.25(0.87,1.79) 0.74(0.56,0.97) Ref 0.086 Current 0.50(0.19,1.32) 1.13(0.48,2.66) 0.78(0.42,1.45) Ref 0.587 Moderate drinking 0.139 Yes 0.85(0.62,1.16) 1.14(0.83,1.57) 0.86(0.67,1.09) Ref 0.360 No 0.79(0.61,1.01) 0.97(0.75,1.27) 0.76(0.64,0.91) Ref 0.007 Results were adjusted for age,race, education, Townsend deprivation index, BMI smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity, histories of cancer, diabetes, hypertension, COPD at baseline. Moderate drinking was defined as thecriterion: < once/week. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis, infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) Stratified analyses for association between activity patterns and hazard of female reproductive diseases Results were adjusted for age,race, education, Townsend deprivation index, BMI smoke status, alcohol intake frequency, total daily exercise time, summed MET minutes per week for all activity, histories of cancer, diabetes, hypertension, COPD at baseline. Moderate drinking was defined as thecriterion: < once/week. Female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis, infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia) The results of the multivariate Cox proportional hazard regression models between activity patterns and hazard of PCOS, HMB, endometriosis, stillbirth, spontaneous miscarriage, or termination are shown in Table  5 . After adjustment for potential confounders, including age, race, education, Townsend deprivation index, BMI, smoking status, moderate drinking, total daily exercise time, summed MET minutes per week for all activity, and histories of cancer, diabetes, hypertension, COPD, we found that women with morning exercise and mixed-timing exercise had significantly lower risks of developing PCOS (HR morning =0.38, 95% CI: 0.15–0.97; HR mixed =0.27, 95% CI: 0.13–0.57, P-trend<0.001), compared with the reference group. Furthermore, mixed-timing exercise was associated with a reduced risk for HMB (HR mixed =0.81, 95% CI: 0.70–0.95, P-trend < 0.05), compared with the midday-afternoon exercise group. However, the multivariate Cox regression models did not reveal statistically significant associations between activity patterns and endometriosis, stillbirth, spontaneous miscarriage, or termination. Meanwhile, Sensitivity analyses further confirmed that these findings were robust, with minimal impact on the observed relationships between activity patterns and specific reproductive diseases (Supplementary Table 4 ). Table 5 Hazard ratios and 95% confidence intervals for activity patterns with the hazard of specific female reproductive disease Diagnosis Activity patterns Morning Evening Mixed Midday-afternoon P for trend PCOS Cases/N 5/8048 3/3687 9/15,020 35/22,434 M1 0.40(0.16,1.02) 0.52(0.16,1.70) 0.38(0.19,0.80) Ref 0.007 M2 0.39(0.15,0.99) 0.33(0.10,1.09) 0.25(0.12,0.53) Ref <0.001 M3 0.38(0.15,0.96) 0.33(0.10,1.09) 0.27(0.13,0.57) Ref <0.001 M4 0.38(0.15,0.97) 0.34(0.10,1.10) 0.27(0.13,0.57) Ref <0.001 Heavy menstrual bleeding Cases/N 131/7337 106/3308 327/13,743 368/20,550 M1 0.99(0.82,1.22) 1.80(1.45,2.24) 1.33(1.15,1.55) Ref <0.001 M2 0.95(0.77,1.15) 1.03(0.83,1.27) 0.79(0.68,0.92) Ref 0.004 M3 0.94(0.77,1.15) 0.99(0.80,1.24) 0.81(0.70,0.95) Ref 0.011 M4 0.94(0.77,1.15) 0.99(0.81,1.24) 0.81(0.70,0.95) Ref 0.011 Endometriosis Cases/N 38/7840 36/3606 92/14,683 127/21,926 M1 0.84(0.58,1.20) 1.73(1.19,2.50) 1.08(0.83,1.42) Ref 0.249 M2 0.83(0.58,1.19) 1.31(0.90,1.91) 0.86(0.66,1.13) Ref 0.469 M3 0.82(0.57,1.18) 1.25(0.86,1.81) 0.92(0.70,1.21) Ref 0.794 M4 0.82(0.57,1.18) 1.24(0.86,1.81) 0.92(0.70,1.21) Ref 0.792 Stillbirth, spontaneous miscarriage, or termination Cases/N 2/7947 6/3634 16/14,755 10/22,135 M1 0.56(0.12,2.54) 3.66(1.33,10.06) 2.40(1.09,5.29) Ref 0.010 M2 0.52(0.11,2.35) 1.76(0.64,4.84) 1.24(0.56,2.73) Ref 0.419 M3 0.51(0.11,2.35) 1.70(0.62,4.72) 1.30(0.58,2.91) Ref 0.347 M4 0.50(0.11,2.29) 1.74(0.63,4.83) 1.28(0.57,2.85) Ref 0.365 Data are HRs and 95%CI; Model 1 was not adjusted; Model 2 was adjusted for age, race,education, Townsend deprivation index; Model 3 was model 2 with additional adjustments for BMI, drink, smoke status, total daily exercise time, summed MET minutes per week for all activity; Model 4 was model 3 with additional adjustments for cancer, diabetes, hypertension, COPD. PCOS, polycystic ovary syndrome Hazard ratios and 95% confidence intervals for activity patterns with the hazard of specific female reproductive disease Data are HRs and 95%CI; Model 1 was not adjusted; Model 2 was adjusted for age, race,education, Townsend deprivation index; Model 3 was model 2 with additional adjustments for BMI, drink, smoke status, total daily exercise time, summed MET minutes per week for all activity; Model 4 was model 3 with additional adjustments for cancer, diabetes, hypertension, COPD. PCOS, polycystic ovary syndrome

The data used in this study came from the UK Biobank study and a brief overview of the study design is provided below. The UK Biobank is a large, population-based cohort study that recruited 502,411 participants aged 37 to 73 years old (with a 5.5% response rate), who attended one of the 22 assessment centers throughout England, Wales, and Scotland from 2006 to 2010 [ 31 ]. From 2013 to 2015, a total of 240,000 invitations were randomly sent to participants, with physical activity (PA) measured using accelerometers. The response rate was 44%. Devices were sent to 106,053 participants, and data from 92,091 participants (49,540 female) were included in the analysis [ 32 ]. Among them, data from 42,099 female participants on activity patterns were available for further analyses in our present study after removing participants diagnosed with female reproductive diseases at baseline. Moreover, we constructed separate cohorts to analyze several different items of specific female reproductive diseases (PCOS, HMB, endometriosis, and stillbirth, spontaneous miscarriage, or termination) and activity patterns, where participants diagnosed with corresponding disease at baseline were removed from analyses of that item (Fig.  1 ). The study was approved by the North West Multi-Centre Research Ethics Committee and the Tulane University (New Orleans, LA, USA) Biomedical Committee Institutional Review Board, and all participants provided written informed consent. Fig. 1 Flow chart of study population. female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis, infertility, stillbirth, spontaneous miscarriage, or termination, and pre-eclampsia or eclampsia. The International Classification of Diseases-10th revision (ICD-10) coding system was used to record diagnoses, including PCOS (E28), endometriosis (N80), HMB (N92), infertility (N97), stillbirth, spontaneous miscarriage, or termination (O03), pre-eclampsia (O14), eclampsia(O15) Flow chart of study population. female reproductive diseases include polycystic ovary syndrome (PCOS), heavy menstrual bleeding (HMB), endometriosis, infertility, stillbirth, spontaneous miscarriage, or termination, and pre-eclampsia or eclampsia. The International Classification of Diseases-10th revision (ICD-10) coding system was used to record diagnoses, including PCOS (E28), endometriosis (N80), HMB (N92), infertility (N97), stillbirth, spontaneous miscarriage, or termination (O03), pre-eclampsia (O14), eclampsia(O15) PA data from 49,540 female participants in the UK Biobank were collected through a 24-hour wrist-worn accelerometer (Axivity AX3) [ 32 ]. Participants who underwent accelerometry measurement and those who refused the measurement had similar baseline demographic and health-related characteristics [ 33 ]. The accelerometer was worn continuously on the dominant wrist for seven days, during which participants maintained their usual daily activities [ 34 ]. Periods of continuous inactivity lasting more than one hour were classified as non-wear time [ 32 ], and data from these periods were imputed based on corresponding wear time data from similar times on different days. Additionally, participants without PA data for any one-hour period within the 24-hour cycle were excluded. Detailed procedures for data processing and analysis have been published elsewhere [ 32 ]. Consistent with previous research [ 21 , 22 ], we payed attention to PA at higher intensities, specifically MVPA, to identify clear and robust activity patterns. MVPA has been extensively validated as an effective substitute for PA. It is typically defined as requiring moderate to vigorous effort, accompanied by a significant increase in heart rate [ 33 ]. Light PA was not included in this study, as it is predominantly associated with low-intensity movements, such as walking or sitting, which may obscure the timing distribution of more effective PA [ 22 ]. Additionally, the UK Biobank accelerometer expert working group processed the raw accelerometer data (Field ID 90001) and generated physical activity intensity data, represented as the mean vector magnitude in milligravity units for 5-second intervals (Field ID 90004). The raw accelerometer signals were calibrated to gravity. Moderate-intensity physical activity was identified based on an average acceleration of 100 to 400 milligravities for more than 80% of the 5-second intervals during a 5-minute period [ 33 ]. Vigorous-intensity activity was defined as a 5-second period in which the average acceleration exceeded 400 milligravities [ 35 ]. As in previous studies [ 21 , 22 ], individuals with frequent evening activity (> 10% of PA accumulated between 01:00 and 04:00) were excluded, as we payed attention to those with a diurnal lifestyle, and frequent evening activity is often associated with sleep disorders. Moreover, we selected 2-hour time window intervals, with a 3-hour interval applied only for the 21:00–24:00 period, to balance sample size and accuracy. Previous research attempting to categorize PA timing using average acceleration data found that only 0.74%, 10.1%, and 0.70% of participants were assigned to the morning, midday-afternoon, and evening groups, respectively, while 88.5% placed in the mixed-timing group [ 22 ]. Our approach, utilizing a 50% allocation method, is similar to previous research [ 21 , 22 ]. This method ensures each participant is assigned to a single timing group, preventing multiple group assignments. Referring to the previous study [ 21 , 22 ], if they spent ≥ 50% of total daily MVPA in the same time window, we assigned participants to the corresponding timing groups: morning (05:00–11:00), midday-afternoon (11:00–17:00), and evening (17:00–24:00) groups; if < 50% of the total daily MVPA occurred in all three time windows, participants would be placed to the mixed-timing group. Moreover, we used the midday-afternoon group as the reference group. Our exposure was the exercise patterns, as we wanted to account for the time-of-day when exercise occurred. Thus, we computed the total daily exercise time over the wear time by summing the minutes of moderate-intensity and vigorous-intensity PA between 05:00 and 24:00. The total daily exercise time of MVPA over the wear time was not included in the exposure and was adjusted in analyses. In summary, besides excluding participants without PA data within any one hour of the 24-hour cycle and those with excessive evening activity (> 10% of PA accumulated between 01:00 and 04:00) as mentioned above, we also excluded participants with unreliable or invalid acceleration measurement data. The standards information regarding unreliable or invalid acceleration measurement data, as well as further details on the exposure, are provided in Supplementary material. Female reproductive diseases in our study include PCOS, heavy menstrual bleeding (HMB), endometriosis, infertility, stillbirth, spontaneous miscarriage, or termination, and pregnancy-related disorders (pre-eclampsia or eclampsia), according to previous study [ 4 ]. Outcomes were ascertained using hospital inpatient records containing data on admissions and diagnoses obtained from the Hospital Episode Statistics for England, the Scottish Morbidity Record data for Scotland, and the Patient Episode Database for Wales. And the International Classification of Diseases-10th revision (ICD-10) coding system was used to record diagnoses, including PCOS (E28), endometriosis (N80), HMB (N92), infertility (N97), stillbirth, spontaneous miscarriage, or termination (O03), pre-eclampsia (O14), eclampsia(O15). The follow up person-years were counted from the date of assessment center visit until the event of interest (the date of female reproductive disease first reported) or censored event (the date of death or February 1, 2022), whichever came first. We also controlled a series of potential confounders, including age (years), race (White, Mixed, Asian or Asian British, Black or Black British, Chinese or Other ethnic group), education (college or university degree, A levels/ AS levels or equivalent, O levels/ GCSEs or equivalent, CSEs or equivalent, NVQ or HND or HNC or equivalent, Other professional qualifications e.g.: nursing, teaching, none of the above), Townsend Deprivation index, the body mass index (BMI), smoking status (never, past and current), drinking (once or more a month, special occasions only or never), total daily exercise time (minutes), summed Metabolic Equivalent Task (MET) minutes per week for all activity (minutes), and history of diseases including cancer, diabetes, hypertension, and chronic obstructive pulmonary disease (COPD). Details of the assessment of these covariates are provided in Supplementary material. The baseline characteristics in terms of demographics, anthropometric, lifestyle and health conditions were presented as mean (standard deviation, SD) and numbers (percentage). General linear models and a chi-square test were used to compare the differences for baseline characteristics by each activity pattern. Cox proportional hazards models were performed to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association of activity patterns with the prevalence of female reproductive diseases, with the follow-up time as the time scale. In addition, we assessed the relationship between activity patterns and the risk of specific female reproductive disease using cox proportional hazards models, but certain diseases (infertility, pre-eclampsia, and eclampsia) had an outcome number of 0 for a given exercise pattern so they were not analyzed further. Moreover, we performed stratified analyses by following factors: age at baseline (<50 or ≥ 50 years old), Townsend deprivation index (< median or ≥ median), BMI (<25, 25–29.9, or ≥ 30 kg/m 2 ), smoking (never, ever, current), Alcohol intake frequency (< once/week or ≥ once/week). To evaluate interactions between activity patterns and these factors, multiplicative interaction was assessed by adding interaction terms to the cox models. Four sensitivity analyses were conducted to assess the robustness of our study findings. The first analysis excluded participants diagnosed with cancer to evaluate whether a history of cancer modified the association between activity patterns and the risk of female reproductive diseases. The second analysis included additional adjustment for dietary covariates (including fresh fruit intake, dried fruit intake, oily fish intake, salt added to food, cereal intake, and processed meat intake) to examine whether dietary factors influenced the relationships. The third analysis further adjusted covariate biochemical indicators (including albumin, triglyceride, glucose, LDL, cholesterol, and total bilirubin) for participants, to remove the impact of certain biochemical indicators unrelated to production on outcomes. The forth analysis excluded participants diagnosed with reproductive diseases within one year of follow-up, to examine whether an early diagnosis of female reproductive diseases during the follow-up period might influence the results. All statistical analyses were conducted by R 4.2.2, and p -values < 0.05 were considered statistically significant.

This study examined the association of activity patterns with the female reproductive diseases. This study found that women with morning exercise and mixed-timing exercise were substantially associated with decreased risks for female reproductive diseases, compared with midday-afternoon exercise group. Further, in this prospective study of 4 separate cohorts, compared with the reference group, we found that morning exercise and mixed-timing exercise had significantly lower risks of PCOS, and mixed-timing exercise was associated with a lower risk for HMB, but not endometriosis and stillbirth, spontaneous miscarriage, or termination. Moreover, these associations were independent of confounding factors included in the model such as biological factors, lifestyle, socioeconomic level, and pre-existing diseases. Similarly, the relationships between exercise patterns and female reproductive diseases were also observed in sensitivity analyses, demonstrating the robustness of the results. Previous human trials have focused primarily on the short-term effects of activity pattern on metabolic function and have yielded conflicting results [ 26 – 28 ]. The majority of these studies only used structured exercise to investigate the effects of MVPA timing on health [ 26 , 27 ]. However, not all MVPA occurs during organized physical activity, which may lead to confounding bias, thereby potentially resulting in the the large discrepancies in the previous evidence. Conversely, our study captured the full spectrum of moderate to vigorous exercise objectively measured in free-living settings [ 22 ], which is essential for defining the MVPA patterns. To date, only a few studies using accelerometers have explored the long-term health effects of MVPA timing, and to our knowledge, studies using accelerometers to explore the long-term effects of exercise patterns on female reproductive diseases have not been reported. In addition, a cross-sectional study found that mixed-timing exercise was associated with an 11% reduced risk of all-cause mortality and a 26% reduced risk of cardiovascular disease mortality [ 21 ], which is consistent with mixed-timing exercise having beneficial long-term effects on female reproductive health in our study. Our study found that morning MVPA may be associated with a 19% reduced risk of female reproductive diseases, with a more substantial 68% reduced risk of PCOS specifically. This association may be related to its positive effects on enhancing metabolism and promoting weight reduction. The metabolic impact of morning exercise involves several factors, including lipid metabolism, glucose metabolism, body temperature, basal metabolic rate, and effects on behavioral patterns. First, morning exercise may be more conducive to lipid oxidation. In particular, pre-breakfast exercise has been shown to significantly enhance lipid oxidation [ 36 ], as the period before breakfast is the longest fasting time of the day, allowing the body to directly utilize stored fat as an energy source. This effect is thought to result from the depletion of liver and muscle glycogen, which enhances lipid oxidation [ 37 ]. Studies have also demonstrated that morning exercise was more effective in reducing abdominal fat in women than evening exercise [ 38 ], and abdominal (central) adiposity is more directly related to the female reproductive system through insulin resistance and hyperandrogenemia [ 6 , 39 ]. Second, morning exercise may be effective in improving glucose metabolism [ 40 ]. For instance, it has been reported that morning exercise has a greater metabolic impact by increasing branched-chain amino acids and muscle acylcarnitine compared to evening exercise [ 41 ]. A study on participants with type 1 diabetes showed that morning exercise reduced the risk of delayed hypoglycemia at night and improved metabolic control the next day compared to evening exercise [ 42 ]. Since impaired glucose metabolism in granulosa can lead to damage to PCOS oocytes through mechanisms involving Sirtuin 3 [ 43 ], morning exercise may help improve PCOS outcomes by promoting favorable metabolic changes. Hyperglycemia, which disrupts ovarian function and promotes inflammatory responses, is known to negatively affect women’s reproductive health [ 44 – 46 ]. Third, individuals with a morning chronotype tend to report healthier lifestyle behaviors, such as lower smoking rates and healthier diets [ 47 ], a characteristic observed in our study participants as well. Lifestyle modifications, including smoking cessation and dietary control, are key to improving women’s reproductive health [ 1 , 5 , 48 ]. Notably, individuals who exercise in the morning tended to be healthier, with lower rates of psychiatric disorders [ 49 ], gastrointestinal diseases [ 50 ], breast cancer [ 51 ], and prostate cancer [ 52 ]. Furthermore, circadian regulation of hormone secretion may play a key role in the metabolic improvements induced by exercise patterns. For instance, the secretion of growth hormone (GH) and cortisol follows a circadian rhythm, with their levels typically peaking in the early morning [ 53 – 55 ]. Morning exercise, particularly high-intensity physical activity, can further stimulate the secretion of these hormones, which contributes to the regulation of lipid and glucose metabolism [ 56 ]. Moreover, human trials showed that the greatest vagal nerve withdrawal occurs around 9:00 [ 55 ], with inflammatory markers peaking in the morning [ 57 ]. Thus, morning exercise could activate the vagus nerve, which in turn induces dopamine release and alleviates systemic inflammation [ 58 ]. Although our results show a statistically significant relationship between morning exercise and the risk of PCOS, the wide confidence intervals suggest that the small sample size of PCOS cases may have influenced the findings. This limitation highlights the need for future studies with larger sample sizes to better define the relationship between morning exercise and PCOS risk. Moreover, our results indicate that mixed-timing MVPA may be related to a 21% reduction in the risk of female reproductive system diseases. We speculate several possible reasons that may explain the beneficial impact of mixed-timing exercise on female reproductive health. First, potential differences in exercise modalities may play a role in modulating the association between MVPA timing and female reproductive health risks, as mixed-time exercise is more likely to involve combined modalities. Research have reported that combined training can reduce body fat more than aerobic or resistance training alone [ 59 ]. In addition, the circadian rhythm changes of other behavioral/environmental factors, such as diet and light exposure, may also influence the relationship between exercise patterns and female reproductive health. For instance, postprandial walking has been shown to be more effective in improving glucose control than walking in the morning or afternoon [ 28 ]. However, the current study lacks specific time-of-day data regarding exercise modalities and behavioral/environmental factors, and we are unable to validate these hypotheses. Therefore, the underlying mechanisms remain unclear and warrant further investigation. Second, the timing of exercise may influence insulin sensitivity, contributing to better blood glucose regulation. Studies have shown that morning exercise is more effective in controlling type 1 diabetes [ 41 ], while middy-afternoon exercise or evening exercise appears to be more beneficial in controlling blood glucose and improvement in insulin sensitivity, which is more effective in improving type 2 diabetes [ 27 ]. Moreover, a randomized trial of obese men excluding type 2 diabetes suggested that evening exercise might be associated with improved glycemic control [ 26 ]. Finally, mixed-timing exercise may play a role in regulating the biological clock and maintaining circadian stability. The timing of exercise can influence the sleep-wake cycle and circadian rhythms [ 60 ]. Research has shown that compared to morning exercise, evening exercise is more likely to cause sleep complaints [ 61 , 62 ], including difficulties with sleep onset and maintaining sleep [ 63 , 64 ]. In fact, it has been reported that evening exercise also leads to more frequent nightmares and insomnia symptoms than morning exercise [ 65 ]. Furthermore, research suggests that sleep disorders are linked to menstrual irregularities, infertility, and adverse pregnancy and delivery outcomes [ 66 ]. The mechanisms behind these associations are likely multifaceted and complex. In addition to genetic influences, disruptions in circadian rhythms may impact reproductive health through dysfunction of the hypothalamic-pituitary-adrenal axis, insulin resistance, oxidative stress, and systemic inflammation [ 66 ]. Therefore, we speculate that evening exercise may negatively affect sleep quality, and optimizing the timing of exercise may improve sleep and, in turn, enhance metabolic regulation [ 67 ]. However, this study did not have specific data on sleep quality and related factors, so these hypotheses cannot be conclusively verified, and the underlying mechanisms still need further clarification. Overall, distributing exercise over different times of the day may optimize the metabolic benefits of PA. Exercise remains a cornerstone treatment for female reproductive system diseases, such as PCOS, and is crucial for managing these conditions. We recommend that women with reproductive health concerns or those seeking disease prevention consider adopting a morning exercise routine or a mixed-timing exercise schedule to maximize potential benefits. The results of the subgroup analysis indicated that the negative correlation between mixed-timing exercise patterns and lower risk of female reproductive diseases disappeared among participants who were obese or smoked. This suggests that obesity and current smoking may obscure the relationship between activity patterns and female reproductive health. One possible explanation is that obesity and smoking can reduce the beneficial effects of exercise by altering metabolic processes or inflammation levels [ 68 – 70 ], thereby masking the protective effect of varied exercise timing on reproductive diseases. Obesity [ 3 , 4 , 6 – 8 ] and smoking [ 43 , 71 , 72 ] can induce a range of physiological changes, leading to a state that is less responsive to exercise. For instance, metabolic function is often impaired in obese individuals [ 69 , 73 ], while smokers experience elevated oxidative stress levels [ 70 , 74 ], both of which can undermine the positive metabolic regulation typically promoted by exercise. As a result, the beneficial impact of exercise timing on health outcomes may be less pronounced in obese individuals and smokers compared to non-obese or non-smokers. Furthermore, a Mendelian analysis had demonstrated that obesity is linked to an increased risk of female reproductive diseases, as shown in large-scale genetic studies examining etiological relationships [ 4 ]. Obesity has also been significantly associated with conditions such as irregular menstrual cycles [ 3 , 7 ], PCOS [ 6 ], and the severity of endometriosis [ 8 ]. These associations are notably influenced by hormonal factors, particularly sex hormone-binding globulin (SHBG), testosterone, and insulin, which can lead to insulin resistance and hyperandrogenemia [ 3 , 6 ]. Consequently, in addition to promoting exercise, we recommend that women with reproductive health issues consider quitting smoking and pursuing weight loss. Furthermore, the lack of significant findings regarding evening exercise is noteworthy, and there remains no consensus on its impact on female reproductive diseases. Several studies have indicated that evening or afternoon exercise can improve blood glucose control and enhanced insulin sensitivity, proving particularly effective in managing type 2 diabetes [ 26 , 27 ]. Additionally, evening PA have been shown to significantly increase levels of IL-6, adrenaline, and free fatty acids [ 75 – 77 ], which may indicate an acceleration of lipolysis. Conversely, a previous observational case-control study indicated that an evening chronotype was linked to poorer hormonal and metabolic outcomes in women with PCOS, suggesting that evening exercise might exacerbate PCOS symptoms by disrupting sleep patterns [ 78 ]. Research has also demonstrated that the timing of exercise influences bedtime, with evening workouts correlating with reduced sleep quality and lower sleep efficiency compared to morning sessions [ 79 ]. Sleep disturbances can increase food intake, potentially leading to overweight conditions and heightened visceral adipose tissue [ 80 ], which in turn contributes to chronic low-grade inflammation [ 81 ]. Despite these findings, the relationship between evening exercise and sleep quality remains inconsistent. For instance, reviews examining the effects of high-intensity nighttime exercise on sleep in healthy adults found that such exercise performed 2–4 h before bedtime does not disrupt sleep in healthy young and middle-aged individuals [ 82 , 83 ]. Although we controlled for exercise intensity as MVPA, we were unable to regulate other potential confounding factors such as duration, sleep interruptions, or behavioral patterns that may influence the results, representing a limitation of our study. Consequently, we cannot draw definitive conclusions about the true impact of evening activity patterns on female reproductive diseases. Additionally, our research indicates that mixed-timing exercise may be linked to a 73% reduction in the risk of PCOS and an 18% reduction in the risk of HMB. Similar relationships between activity patterns and PCOS or HMB were observed in the sensitivity analysis, which further supports the robustness of the results. However, it is important to note that the sample size concerning specific female reproductive diseases is relatively small, which may introduce statistical errors. PCOS is the most common female gynecological endocrinopathy characterized by polycystic ovaries, oligo-ovulatory or anovulatory cycles, biochemical or clinical hyperandrogenemia and hyperinsulinemia [ 1 , 5 ]. HMB, a leading cause of anemia and reduced quality of life in adolescents [ 84 ], is mainly attributed to immaturity of the hypothalamic-pituitary-ovarian (HPO) axis or hormonal disorders such as PCOS leading to anovulatory cycles [ 84 , 85 ]. Exercise timing may influence hormone secretion through circadian rhythms, thereby regulating the metabolism and hormonal functions associated with PCOS. For instance, a review of circadian clock function in mammalian ovaries indicates that temporal disturbances can adversely affect reproductive function and fertility in rodent models and women with shift work schedules. Modulating clock function within the HPO axis may offer therapeutic benefits for female reproductive disorders stemming from chronic circadian rhythm disruptions [ 86 ]. A zoological study demonstrated a significant correlation between night shift work and PCOS, revealing notable differences in the diurnal patterns of corticotropin-releasing hormone, adrenocorticotropin, and prolactin in PCOS model rats [ 87 ]. Furthermore, oocyte maturation relies on the pentose phosphate pathway, pyruvate metabolism, and the tricarboxylic acid cycle [ 88 ], but genome-wide temporal disturbances in ovarian granulosa cells may lead to metabolic dysfunction of carbohydrate and amino acid metabolism and the tricarboxylic acid cycle [ 89 ]. Moreover, obesity, a typical clinical manifestation of PCOS, intensifies the hormonal and clinical features of PCOS, with multiple potential mechanisms involved [ 90 ]. Lifestyle modifications, particularly increased exercise, are considered first-line treatments measure for women with PCOS [ 91 ], as they can can reduce body weight, insulin, and testosterone levels, thus restoring ovulation and fertility [ 1 , 5 , 48 ]. Lifestyle interventions are essential, especially when obesity and/or insulin resistance (IR) coexist with PCOS [ 92 ]. Therefore, we recommend women with PCOS and HMB engage in mixed-timing exercise, as it may effectively improve both conditions by promoting metabolism. Furthermore, our findings suggest no significant association between activity patterns and endometriosis, stillbirth, spontaneous miscarriage, or termination. Taking endometriosis as an example, it is a benign estrogen dependent gynecological disease characterized by the growth of endometrial tissue outside the uterine cavity [ 93 ]. Evidence suggested that local inflammatory reactions, caused by ectopic endometrial implants, contribute to symptoms of the disease [ 94 ]. Moreover, oxidative stress contributes to the development and maintenance of inflammatory processes related to endometriosis, as reactive oxygen species in the peritoneal fluid of women with endometriosis appear to be elevated [ 95 ]. Currently, PA has been widely used in treating diseases involving inflammatory processes [ 96 ], since it has been shown to increase the systemic levels of various anti-inflammatory cytokines [ 16 , 97 ]. However, the evidence regarding the effect of PA timing on oxidative stress remains inconsistent. On the one hand, some studies suggest that evening exercise offers greater benefits for inflammation and lipid metabolism compared to morning exercise [ 98 ]. On the other hand, other research indicates that evening exercise may induce sleep disturbances, which in turn could contribute to obesity and increased inflammation [ 80 , 81 ]. In addition, PCOS is strongly is strongly associated with metabolic dysfunctions such as insulin resistance and obesity, making the impact of MVPA on regulating metabolic processes in PCOS more biologically plausible. In contrast, the etiology of endometriosis is complex and not fully understood, involving a range of factors including immune, genetic, and environmental influences. Consequently, the mechanisms connecting endometriosis to the timing of MVPA may be less clearly defined than those associated with PCOS. Therefore, further research is needed to explore the true role of physical activity patterns in the management or progression of endometriosis. The present findings may have several public health implications. The evidence fills the association between exercise patterns and female reproductive diseases, and informs a better understanding of optimal timing of MVPA to maximize health benefits of female reproduction, especially PCOS. Our findings can help guide and improve strategy development and practice in women’s health. Additionally, we can easily measure the timing of MVPA through wearable devices and now technology [ 34 ]. The objectivity and convenience of exercise patterns measured by accelerometers increases the reproducibility and feasibility of MVPA timing measurements in the future of public health. Although our study has some advantages including the large population-based sample size, the extensive covariate correction, objective measurement of MVPA in a free-living setting, and the consistent results in several sensitivity analyses, this study should be carefully considered for some potential limitations. Firstly, we lacked accurate information on exercise intensity, which may affect the relationship between exercise patterns and female reproductive diseases. In addition, the UK Biobank does not provide specific timing data on behavioral or environmental factors, making it impossible for us to verify any potential reverse causal relationships. Secondly, there is an important limitation that the UK Biobank is not representative of the general population due to the voluntary participation [ 31 ]. Further studies are needed to confirm our findings, particularly in populations that are more representative of the UK population. Thirdly, we routinely used 7-day monitoring periods in exercise monitor studies [ 99 ], but it remains unclear whether 7-day accelerometer measurements represent long-term behavior [ 100 ], especially PA timing. Fourthly, as mentioned earlier, the sample size for specific female reproductive system diseases, especially PCOS and stillbirth, spontaneous miscarriage, or termination, is relatively small, which may have a significant impact and error on the results. Therefore, in the future, it is necessary to conduct larger sample size studies on the corresponding diseases to further determine their relationships. Fifth, the dietary data of UK Biobank used in our sensitivity analyses were obtained by participants answering questions such as “How often do you eat oily fish? (e.g. sardines, salmon, mackerel, herring)” through touchscreen, but data on variables such as total energy intake or other macro/micronutrient intake are not available in UK Biobank. However, it is currently unclear whether the covariates adjusted in sensitivity analyses may represent total energy intake, especially the entire dietary factors. Sixth, our study focused only on individuals with regular daily routines, excluding those who frequently engage in nighttime activities. Therefore, it primarily analyzed diurnal patterns. Future research should include experimental interventions and analyses of potential relationships to enhance the generalizability of the findings. Furthermore, although this study accounted for various potential confounders, there remain some unavailable factors that could impact the relationship between activity patterns and female reproductive diseases. Among these, sleep quality and the use of hormone treatments are two key factors worth considering. Poor sleep or inadequate sleep quality has been widely linked to numerous physiological processes [ 101 ], including hormonal regulation [ 78 ], immune function, and metabolic health [ 102 , 103 ]. Research has demonstrated that sleep quality may have a significant impact on female hormone levels, menstrual cycles [ 104 ], and overall reproductive health [ 105 ]. Furthermore, hormonal treatments are commonly used to regulate women’s reproductive health and may directly influence hormonal levels and reproductive system function [ 106 ]. As such, failure to control for sleep quality and the use of hormone treatments may confound the potential effects of physical activity on reproductive system diseases, when investigating the timing of exercise on physiological functions. Therefore, future research should consider including these potential confounders and further explore their roles in the relationship between activity patterns and female reproductive health. A systematic evaluation of these factors will allow for a more comprehensive understanding of how exercise timing influences reproductive system diseases, providing stronger evidence for the development of personalized intervention strategies.

In conclusion, our study indicates that morning exercise and mixed-timing exercise were associated with decreased risks for female reproductive diseases and PCOS, compared with midday-afternoon exercise group. Meanwhile, we found that women with mixed-timing exercise was associated with a lower risk for HMB compared with the reference group.

The female reproductive system is driven by a complex combination of endocrine mechanisms, and the primary etiologies of these gynecological conditions are thought to be hormonal disorders and metabolic disorders [ 1 – 3 ]. In women, obesity and metabolic disorders are associated with an increased prevalence of gynecological conditions [ 4 ], including polycystic ovary syndrome (PCOS) [ 1 , 5 , 6 ], heavy menstrual bleeding (HMB) [ 3 , 7 ], endometriosis [ 8 ], pregnancy complications such as preeclampsia and eclampsia [ 9 ], infertility [ 10 ] and miscarriage [ 11 ]. Moreover, studies suggested that exercise was a primary measure to improve the treatment of female reproductive disorders [ 5 , 12 – 14 ], because physical activity has positive impacts on metabolism [ 15 ], oxidative stress and inflammation [ 16 , 17 ]. Reviews of exercise on women with PCOS showed that vigorous aerobic exercise improved insulin sensitivity, androgen levels, body composition, and cardiopulmonary health [ 18 , 19 ]. However, there are still numerous challenges in achieving therapeutic goals for PCOS-related metabolic disorders [ 20 ], suggesting that in addition to the intensity and type of exercise, timing of exercise that are not accounted for in therapeutic approaches may also have an impact on women’s reproductive health. In recent years, there has been increasing interest in the long-term effects of timing of the day when exercise is performed on health [ 21 , 22 ]. Animal studies consistently suggested that timing of exercise may have an impact on metabolic function [ 23 – 25 ]. However, results from human studies regarding the potential link between activity timing and physical health, such as glucose metabolism, are inconsistent [ 26 – 29 ]. Several randomised trials reported that afternoon or evening exercise is more effective than morning exercise at improving blood glucose levels [ 26 , 27 ], whereas other studies found that morning exercise is more likely to reduce hyperglycemia than afternoon exercise [ 28 ]. There are even studies suggesting no differences in glucose responses across different exercise timings [ 29 ]. Therefore, the optimal timing of exercise to maximize long-term health benefits, particularly for female reproductive health, remains unclear. Moreover, while exercise timing may influence female reproductive health through its potential effects on metabolic health [ 30 ], to our knowledge, the relationship between activity patterns and reproductive diseases in women is still poorly understood. In this study, we utilize data from the UK Biobank to assess the association between the time-of-day of moderate-to-vigorous intensity physical activity (MVPA) and female reproductive disorders. Meanwhile, extensive confounding including socioeconomic factors, lifestyle factors and biological factors were adjusted for accurate evaluation. This study may add insights into relationships between exercise patterns and women’s reproductive diseases for the first time, for clinical and public health reference.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1