Women's reproductive life patterns, intrinsic capacity, and the risk of all-cause and cause-specific dementia: a prospective cohort study.

This study included participants from the UK Biobank, a large-scale prospective cohort that enrolled over half a million participants across England, Scotland, and Wales between 2006 and 2010. Recruitment was conducted through postal invitations sent to individuals identified from the UK National Health Service patient registers who lived within approximately 25 miles of one of 22 assessment centers. At recruitment, participants completed touchscreen questionnaires and underwent physical measurements and biological sample collection. The baseline assessment collected detailed information on sociodemographic characteristics, lifestyle factors, sex-specific reproductive factors, psychosocial factors, medical history, medication usage. Participants’ health outcomes were ascertained through linkage to national health records, including hospital inpatient data, primary care system, and death register. Detailed descriptions of UK Biobank were available elsewhere [ 22 ]. The UK Biobank received the ethical approval from the Northwest Multicenter Research Ethics Committee and written informed consents were obtained from all participants. The present analyses were conducted using the UK Biobank Resource under Application Number 66,354. Among the 273,294 female participants in the UK Biobank, we included women who were post-menopause when recruited ( n  = 197,234), excluding those with no complete information to construct reproductive life sequence ( n  = 19,439), with implausible reproductive age sequence ( n  = 1,386), with implausible age at menarche or age at natural menopause ( n  = 4,418), with no information for any of IC components ( n  = 18,021), with dementia at baseline ( n  = 51), and those who were withdrawal from the study ( n  = 10). A total of 153,909 women were included in the final analysis (Fig.  1 ). Baseline characteristics of participants included or not included in the final analysis were compared in Table S1. Fig. 1 Study design and population selection A . The concept framework of the study design. B The flow chart of the population selection process Study design and population selection A . The concept framework of the study design. B The flow chart of the population selection process Female-specific reproductive factors were self-reported by participants, including age at menarche, age at first live birth, age at last live birth, number of live births, age at natural menopause, history of hysterectomy, age at hysterectomy, history of oophorectomy, and age at oophorectomy. Age at first and last live birth were both reported for multiparous women. For primiparous women, they only reported “age at childbirth”, and we used “age at childbirth” to define age at first live birth in these women. We further defined birth interval as the difference between age at last live birth and age at first live birth only among multiparous women. Oophorectomy menopause was defined as menopause due to bilateral oophorectomy, or hysterectomy and bilateral oophorectomy. Hysterectomy menopause was defined as menopause occurring due to hysterectomy with no concurrent bilateral oophorectomy. Reproductive life span was calculated as the difference between age at natural or surgical (due to oophorectomy or hysterectomy) menopause and age at menarche. Following the guidelines for sequence and cluster analysis [ 23 ], these individual reproductive factors were further used to construct the reproductive life sequence by converting them into different reproductive states across women’s lifetime: before menarche, menstrual cycle, birth span, after natural menopause, after hysterectomy menopause, and after oophorectomy menopause (Fig.  1 , A). Years between menarche and menopause were defined as menstrual cycle. During menstrual cycle, we further defined “birth span” as the years between first live birth and last live birth for multiparous women; the single year of childbirth for primiparous women; and no such state for nulliparous women. We used sequence analysis to construct the reproductive temporal sequence according to the reproductive states from age 8 to 60 years, which can maintain the order and consider the onset time and duration of reproductive factors [ 23 , 24 ]. The examples of reproductive life sequence data for nulliparous women, primiparous women, and multiparous women were shown in Figure S1. To identify the potential patterns of such reproductive temporal sequence, we used partitioning around medoids (PAM) clustering method based on pairwise sequence dissimilarities calculated using the generalized Hamming distance. We compared clustering solutions with different numbers of clusters ranging from 2 to 10 and evaluated model fit using multiple goodness-of-fit statistics (details in Table S2) [ 25 ]. The final 5-cluster solution was selected as the optimal model across statistical fit, cluster separation, cluster size, and substantive interpretability. The packages TraMineR [ 26 ] and WeightedCluster [ 27 ] were used for sequence analysis and cluster analysis. Baseline IC of each woman was assessed according to the concept proposed by World Health Organization (WHO) [ 28 ] and previous studies [ 29 – 31 ], which included five domains: psychological, sensory, vitality, locomotor, and cognitive capacities. Due to the lack of information on cognitive capacity in the UK Biobank baseline assessment and the measurement of dementia as the follow-up outcome, we considered four of the five domains (psychology, sensory, vitality, and locomotion) alongside previous studies measuring IC in UK Biobank [ 32 , 33 ] (Table S3). Seven items of IC in these four domains were measured: self-reported exhaustion and unhealthy sleep duration ( 9 h per day) in the psychology domain; eye problems and hearing impairment in the sensory domain; recent weight loss and low grip strength in the vitality domain; and slow walking pace in the locomotion domain. Participants with ≥ 4 of the seven impaired items were considered having IC impairment based on previous studies [ 21 , 33 ]. In subsequent analyses, IC was considered as a binary impairment variable, and as domain-specific impairment variables to examine its independent, joint, and modifying associations with reproductive life sequences and dementia. The primary outcome of this study was all-cause and cause-specific dementia. Diagnoses and dates of dementia were ascertained through the linkage of hospital inpatient records, primary care system, and death register updated to September 1, 2023. The International Classification of Diseases 10th Revision (ICD-10) codes F00, F01, F02, F03, G30, G31.0, and G31.8 were used to identify participants with all-cause dementia; the ICD-10 codes F00 and G30 were used to identify participants with Alzheimer’s disease (AD); and the ICD-10 code F01 was used to identify participants with vascular dementia (VD). In the analyses, all-cause dementia, AD, and VD were treated as separate outcomes. Individuals meeting the case definition of AD and VD were included as cases for both and AD and VD were not mutually exclusive. Baseline covariates included age (continuous variable), ethnicity (white/non-white), educational level (college or above/below high school), employment status (unemployed/employed), total household income before tax (less than £18000/£18000-£30999/£31000-£51999/£52000-£100000/>£100000), current smoking status (never/previous/current), current alcohol drinking status (never/previous/current), physical activity (low/moderate/high), body mass index (BMI, underweight/healthy weight/overweight/obese), ever usage of hormone-replacement therapy (HRT) (never/ever), baseline health conditions of cardiovascular disease (CVD), type 2 diabetes (T2D) and cancer, and polygenic risk score (PRS) for AD (low/medium/high). Physical activity was estimated by Metabolic Equivalent Task minutes per week for moderate and vigorous activity and was classifies into tertiles as low, moderate, and high physical activity. BMI was categorized into underweight (< 18.5 kg/m 2 ), healthy weight (18.5–24.9 kg/m 2 ), overweight (25–29.9 kg/m 2 ), and obesity (≥ 30 kg/m 2 ). Baseline health conditions, including CVD, T2D, and cancer, were ascertained through hospital inpatient records, primary care system, and self-reported disease history. The PRS for AD was generated using a Bayesian approach applied to meta-analysed summary statistics from external genome-wide association study data [ 34 ], and was provided by the UK Biobank in the data field 26,206. The PRS was further divided to tertiles as low, medium, and high risk of AD. Baseline cognitive performance was assessed based on reaction time and pairs matching. We additionally extracted participants’ diseases history of polycystic ovary syndrome (PCOS), infertility, diabetes mellitus in pregnancy, hypertensive disorders in pregnancy, endometriosis, uterine fibroids, and ovarian cysts through the linkage to national health registers. The ICD-10 code E28.2 was used to identify participants with PCOS; the ICD-10 code N97 was used to identify participants with infertility; the ICD-10 code O24 was used to identify participants with diabetes mellitus in pregnancy; the ICD-10 codes O10, O11, O13, O14, O15 were used to identify participants with hypertensive disorders in pregnancy; the ICD-10 code N80 was used to identify participants with endometriosis; the ICD-10 code D25 was used to identify participants with uterine fibroids; the ICD-10 code N83.0 was used to identify participants with ovarian cysts. For the covariates, responses of “do not know” or “prefer not to answer” were considered as missing values and were assigned to a separate group of “unknown”. Baseline characteristics of participants were described as means with standard deviations (SD) for continuous variables and as counts with percentages for categorical variables. Differences of variables across groups by reproductive life sequences were compared by analysis of variance for continuous variables and chi-squared test for categorical variables. We used Fine and Gray’s subdistribution hazard models to assess the associations of reproductive life patterns with all-cause and cause-specific dementia, with death treated as a competing event. The fully adjusted model included age at baseline, ethnicity, educational levels, employment status, household income levels, current smoking and alcohol drinking status, physical activity, BMI, ever usage of HRT, baseline CVD, T2D, and cancer, and PRS for AD. Days from baseline to the diagnosis of dementia, the ascertainment of death, or the end of follow-up, whichever came first, were calculated as the time scale. The cumulative incidence function (CIF) curves were used to compare the cumulative incidence of all-cause and cause-specific dementia across different patterns of reproductive life sequences. In the secondary analysis, we assessed the association of joint patterns of reproductive life sequences and IC impairment with the risk of dementia using Fine and Gray’s subdistribution hazard models adjusting for aforementioned covariates. Participants were categorized into ten groups by crossing five reproductive sequences and IC impairment, and those with standard reproductive sequence and no IC impairment were specified as the reference group. We conducted logistic regression models to assess the association of reproductive life sequences with IC impairment and specific IC domains. To examine the potential role of IC impairment in the association between reproductive life sequences and dementia, we conducted a subgroup analysis among sub-populations with specific IC impairment domains and calculated P for interaction by including interaction terms between reproductive life sequences and each IC impairment domain in the models. A series of additional analyses were conducted to test the robustness of the results. First, we used restricted cubic spline models to assess the association between number of IC impairment items and the risk of all-cause dementia within subpopulations with specific reproductive patterns. Second, subgroup analyses of the association between reproductive life sequences and dementia were conducted stratified by age, ethnicity, educational level, employment status, total household income, ever usage of HRT, PRS for AD, baseline CVD, T2D, and cancer. In addition, several sensitivity analyses were conducted to assess the association between reproductive life sequences and the incidence of dementia by: (1) including IC impairment as a covariate in the model, (2) excluding participants who have developed dementia during the first two years of follow-up, (3) repeating the analysis among participants aged ≥ 55 years at baseline, (4) excluding participants with missing covariates, (5) imputing covariates using multiple imputation for five times, (6) including the history of PCOS and infertility in the covariates, (7) including the history of diabetes mellitus in pregnancy and hypertensive disorders in pregnancy in the covariates, (8) including the history of endometriosis, uterine fibroids, and ovarian cysts in the covariates, (9) including cognitive performance of reaction time and pairs matching as covariates in the model. Hazard ratios (HRs), odds ratios (ORs), and 95% confidence intervals (95% CIs) were reported in this study. Statistical analyses in the observational study were conducted using SAS (version 9.4, SAS Institute Inc., NC, USA) and R (version 4.3.3 R, Foundation for Statistical Computing). Significance tests were evaluated at the 0.05 level using two-sided tests. A Benjamini-Hochberg false discovery rate correction was applied for multiple comparisons in the main and secondary analyses, and false discovery rate (FDR) adjusted q values are reported.

Among the 153,909 women included in the study, the mean age was 59.7 (SD = 5.7) years. Figure  2 shows the five reproductive life sequences identified among the participants, including the standard sequence (Pattern 1, n  = 71,398, 46.4%), the early childbearing and oophorectomy menopause sequence (Pattern 2, n  = 9,979, 6.5%), the short reproductive span with natural menopause sequence (Pattern 3, n  = 27,208, 17.7%), the early childbearing and hysterectomy menopause sequence (Pattern 4, n  = 17,532, 11.4%), and the high parity with long birth span sequence (Pattern 5, n  = 27,792, 18.1%). In the standard sequence, the mean (SD) age at menarche, number of live births, age at first live birth, age at last live birth, birth span, age at menopause, and reproductive life span was 12.9 (1.5) years, 1.6 (1.0), 26.6 (5.2) years, 29.3 (5.0) years, 3.3 (1.9) years, 52.5 (2.4) years, and 39.6 (2.8) years, respectively. Women with early childbearing and oophorectomy menopause (Pattern 2), and those with early childbearing and hysterectomy menopause (Pattern 4) had younger ages at first live birth (mean ages of 24.4 and 23.8 years, respectively), early ages of surgical menopause (mean ages of 44.6 and 40.9 years, respectively), and shorter reproductive life span (mean years of 30.9 and 26.7, respectively). Women in the short reproductive span with natural menopause pattern (Pattern 3) had early age at natural menopause (mean age: 44.6 years) and short reproductive life span (mean years: 31.6). Women in the high parity with long birth span pattern (Pattern 5) had higher number of live birth (mean number: 3.0) and long birth span (mean years: 8.6). Fig. 2 Patterns of reproductive life sequences among UK female population. A Identified five reproductive life patterns in UK female population. The x-axis represents the age from the beginning to the end of the reproductive life sequence. The y-axis represents the frequency of population with different reproductive states in each age group. B Distribution of age at menarche, number of live births, age at first live birth, age at last live birth, birth interval, age at menopause, and reproductive life span across different patterns of reproductive life sequences. Black vertical lines indicate the median age of each factor Patterns of reproductive life sequences among UK female population. A Identified five reproductive life patterns in UK female population. The x-axis represents the age from the beginning to the end of the reproductive life sequence. The y-axis represents the frequency of population with different reproductive states in each age group. B Distribution of age at menarche, number of live births, age at first live birth, age at last live birth, birth interval, age at menopause, and reproductive life span across different patterns of reproductive life sequences. Black vertical lines indicate the median age of each factor During a median follow-up of 14.5 years, 2,940 dementia cases (including 1,509 AD and 577 VD cases) were documented. The characteristics of participants by different patterns of reproductive life sequences were shown in Table  1 . Compared to women with the standard sequence, those with the other four patterns were more likely to be non-white, lower educated, unemployed, with lower household income, smokers, non-drinkers, less physically active, obese, ever using HRT, and have higher prevalence of baseline CVD, and T2D. Table 1 Baseline characteristics of participants by reproductive life sequences Total ( N  = 153909) Standard sequence ( N  = 71398) Early childbearing and oophorectomy menopause ( N  = 9979) Short reproductive span with natural menopause ( N  = 27208) Early childbearing and hysterectomy menopause ( N  = 17532) High parity with long birth span ( N  = 27792) P value Age at baseline, mean (SD) 59.7 (5.7) 60.3 (4.8) 57.9 (6.7) 58.6 (6.5) 58.8 (7.0) 60.6 (5.4) < 0.001 Ethnicity, n (%) < 0.001  White 148,182 (96.3) 69,374 (97.2) 9572 (95.9) 26,003 (95.6) 16,903 (96.4) 26,330 (94.7)  Non-white 5407 (3.5) 1866 (2.6) 387 (3.9) 1145 (4.2) 597 (3.4) 1412 (5.1)  Unknown 320 (0.2) 158 (0.2) 20 (0.2) 60 (0.2) 32 (0.2) 50 (0.2) Educational level, n (%) < 0.001  College or above 44,325 (28.8) 23,581 (33.0) 2365 (23.7) 8024 (29.5) 3604 (20.6) 6751 (24.3)  Below high school 91,408 (59.4) 39,226 (54.9) 6428 (64.4) 15,999 (58.8) 11,863 (67.7) 17,892 (64.4)  Unknown 18,176 (11.8) 8591 (12.0) 1186 (11.9) 3185 (11.7) 2065 (11.8) 3149 (11.3) Employment status, n (%) < 0.001  Unemployed 10,062 (6.5) 3808 (5.3) 851 (8.5) 1867 (6.9) 1422 (8.1) 2114 (7.6)  Employed 142,647 (92.7) 67,036 (93.9) 9050 (90.7) 25,133 (92.4) 15,988 (91.2) 25,440 (91.5)  Unknown 1200 (0.8) 554 (0.8) 78 (0.8) 208 (0.8) 122 (0.7) 238 (0.9) Total household income before tax, n (%) < 0.001  Less than £18,000 34,969 (22.7) 15,272 (21.4) 2315 (23.2) 6460 (23.7) 4203 (24.0) 6719 (24.2)  £18,000-£30,999 37,280 (24.2) 17,797 (24.9) 2328 (23.3) 6467 (23.8) 4219 (24.1) 6469 (23.3)  £31,000-£51,999 30,652 (19.9) 14,600 (20.4) 2027 (20.3) 5430 (20.0) 3377 (19.3) 5218 (18.8)  £52,000- £100,000 19,244 (12.5) 9358 (13.1) 1267 (12.7) 3421 (12.6) 1966 (11.2) 3232 (11.6)  >£100,000 4588 (3.0) 2302 (3.2) 260 (2.6) 885 (3.3) 382 (2.2) 759 (2.7)  Unknown 27,176 (17.7) 12,069 (16.9) 1782 (17.9) 4545 (16.7) 3385 (19.3) 5395 (19.4) Current smoking status, n (%) < 0.001  Never 88,868 (57.7) 42,485 (59.5) 5665 (56.8) 14,555 (53.5) 9940 (56.7) 16,223 (58.4)  Previous 52,070 (33.8) 24,201 (33.9) 3308 (33.1) 9587 (35.2) 5828 (33.2) 9146 (32.9)  Current 12,528 (8.1) 4547 (6.4) 970 (9.7) 2989 (11.0) 1708 (9.7) 2314 (8.3)  Unknown 443 (0.3) 165 (0.2) 36 (0.4) 77 (0.3) 56 (0.3) 109 (0.4) Current alcohol drinking status, n (%) < 0.001  Never 8527 (5.5) 3324 (4.7) 623 (6.2) 1593 (5.9) 1038 (5.9) 1949 (7.0)  Previous 5672 (3.7) 2248 (3.1) 477 (4.8) 1084 (4.0) 792 (4.5) 1071 (3.9)  Current 139,617 (90.7) 65,794 (92.2) 8874 (88.9) 24,510 (90.1) 15,689 (89.5) 24,750 (89.1)  Unknown 93 (0.1) 32 (0.0) 5 (0.1) 21 (0.1) 13 (0.1) 22 (0.1) Physical activity, n (%) < 0.001  Low 20,038 (13.0) 9171 (12.8) 1444 (14.5) 3675 (13.5) 2424 (13.8) 3324 (12.0)  Moderate 50,276 (32.7) 24,203 (33.9) 3022 (30.3) 8899 (32.7) 5342 (30.5) 8810 (31.7)  High 46,147 (30.0) 21,652 (30.3) 2924 (29.3) 8305 (30.5) 5022 (28.6) 8244 (29.7)  Unknown 37,448 (24.3) 16,372 (22.9) 2589 (25.9) 6329 (23.3) 4744 (27.1) 7414 (26.7) Body mass index, n (%) < 0.001  Underweight 4665 (3.0) 2325 (3.3) 235 (2.4) 1063 (3.9) 352 (2.0) 690 (2.5)  Healthy weight 53,690 (34.9) 26,548 (37.2) 2911 (29.2) 10,023 (36.8) 4939 (28.2) 9269 (33.4)  Overweight 59,122 (38.4) 27,138 (38.0) 3841 (38.5) 10,084 (37.1) 7003 (39.9) 11,056 (39.8)  Obese 36,432 (23.7) 15,387 (21.6) 2992 (30.0) 6038 (22.2) 5238 (29.9) 6777 (24.4) Age at menarche, mean (SD) 12.9 (1.5) 12.9 (1.5) 12.8 (1.6) 12.9 (1.6) 12.8 (1.6) 13.0 (1.5) < 0.001 Natural menopause, n (%) 116,670 (75.8) 65,831 (92.2) 0 (0.0) 25,832 (94.9) 0 (0.0) 25,007 (90.0) < 0.001 Age at natural menopause, mean (SD) 49.9 (4.6) 52.5 (2.4) 44.6 (4.3) 44.6 (3.1) 40.9 (3.7) 50.4 (3.5) < 0.001 Reproductive lifespan, mean (SD) 35.7 (5.9) 39.6 (2.8) 30.9 (5.4) 31.6 (3.4) 26.7 (5.2) 37.4 (3.9) < 0.001 Number of live births, n (%) < 0.001  0 24,921 (16.2) 14,278 (20.0) 1945 (19.5) 6614 (24.3) 2084 (11.9) 0 (0.0)  1–2 89,975 (58.5) 47,861 (67.0) 5961 (59.7) 16,962 (62.3) 10,480 (59.8) 8711 (31.3)  ≥ 3 36,585 (23.8) 9052 (12.7) 1953 (19.6) 3527 (13.0) 4654 (26.5) 17,399 (62.6)  Unknown 2428 (1.6) 207 (0.3) 120 (1.2) 105 (0.4) 314 (1.8) 1682 (6.1) Age at first live birth, n (%) < 0.001  < 21 19,341 (12.6) 7097 (9.9) 1644 (16.5) 2885 (10.6) 3542 (20.2) 4173 (15.0)  21–22 17,751 (11.5) 6710 (9.4) 1299 (13.0) 2676 (9.8) 2815 (16.1) 4251 (15.3)  23–24 21,011 (13.7) 7309 (10.2) 1368 (13.7) 2603 (9.6) 2917 (16.6) 6814 (24.5)  25–26 21,675 (14.1) 7970 (11.2) 1265 (12.7) 2713 (10.0) 2437 (13.9) 7290 (26.2)  27–29 25,662 (16.7) 12,458 (17.4) 1424 (14.3) 4187 (15.4) 2329 (13.3) 5264 (18.9)  > 29 23,548 (15.3) 15,576 (21.8) 1034 (10.4) 5530 (20.3) 1408 (8.0) 0 (0.0)  No child or unknown 24,921 (16.2) 14,278 (20.0) 1945 (19.5) 6614 (24.3) 2084 (11.9) 0 (0.0) Age at last live birth, n (%) < 0.001  < 26 20,421 (13.3) 10,801 (15.1) 1728 (17.3) 4062 (14.9) 3830 (21.8) 0 (0.0)  26–28 24,619 (16.0) 12,362 (17.3) 1834 (18.4) 4035 (14.8) 3693 (21.1) 2695 (9.7)  29–31 27,027 (17.6) 9046 (12.7) 1522 (15.3) 3184 (11.7) 3168 (18.1) 10,107 (36.4)  32–35 23,593 (15.3) 7884 (11.0) 1050 (10.5) 2698 (9.9) 1970 (11.2) 9991 (35.9)  > 35 14,019 (9.1) 5929 (8.3) 441 (4.4) 1891 (7.0) 759 (4.3) 4999 (18.0)  No child or only one child or unknown 44,230 (28.7) 25,376 (35.5) 3404 (34.1) 11,338 (41.7) 4112 (23.5) 0 (0.0)  Birth interval, mean (SD) 4.9 (3.8) 3.3 (1.9) 4.4 (3.2) 3.5 (2.1) 4.6 (3.3) 8.6 (4.5) < 0.001  History of hysterectomy, n (%) 36,601 (23.8) 5332 (7.5) 9768 (97.9) 1262 (4.6) 17,532 (100.0) 2707 (9.7) < 0.001  Age at hysterectomy, mean (SD) 44.3 (7.7) 54.7 (4.2) 43.8 (5.1) 52.0 (7.3) 39.6 (4.9) 52.6 (5.4) < 0.001  History of oophorectomy, n (%) 16,824 (10.9) 3542 (5.0) 9979 (100.0) 828 (3.0) 947 (5.4) 1528 (5.5) < 0.001  Age at oophorectomy, mean (SD) 47.7 (7.3) 55.3 (3.9) 43.7 (5.2) 52.9 (6.8) 47.0 (8.7) 53.5 (4.9) < 0.001 Ever usage of hormone replacement therapy, n (%) < 0.001  Never 78,474 (51.0) 413,434 (57.9) 1190 (11.9) 13,769 (50.6) 6813 (38.9) 15,359 (55.3)  Ever 75,193 (48.9) 29,962 (42.0) 8782 (88.0) 13,393 (49.2) 10,683 (60.9) 12,373 (44.5)  Unknown 242 (0.2) 93 (0.1) 7 (0.1) 46 (0.2) 36 (0.2) 60 (0.2)  Baseline CVD, n (%) 59,408 (38.6) 26,241 (36.8) 4361 (43.7) 9683 (35.6) 7951 (45.4) 11,172 (40.2) < 0.001  Baseline type 2 diabetes, n (%) 5892 (3.8) 2356 (3.3) 539 (5.4) 1048 (3.9) 833 (4.8) 1116 (4.0) < 0.001  Baseline cancer, n (%) 19,747 (12.8) 8608 (12.1) 1680 (16.8) 3691 (13.6) 2453 (14.0) 3315 (11.9) < 0.001 Polygenic risk score for Alzheimer’s disease, n (%) < 0.001  Low PRS 49,956 (32.5) 23,004 (32.2) 3238 (32.4) 8812 (32.4) 5791 (33.0) 9111 (32.8)  Medium PRS 49,956 (32.5) 23,405 (32.8) 3168 (31.7) 8836 (32.5) 5641 (32.2) 8906 (32.0)  High PRS 49,955 (32.5) 23,176 (32.5) 3286 (32.9) 8735 (32.1) 5681 (32.4) 9077 (32.7)  Unknown 4042 (2.6) 1813 (2.5) 287 (2.9) 825 (3.0) 419 (2.4) 698 (2.5) Baseline characteristics of participants by reproductive life sequences Women with short reproductive span with natural menopause, early childbearing and hysterectomy menopause, and high parity with long birth span had higher cumulative incidence of all-cause dementia during follow-up (Fig.  3 ). After adjusting for potential confounders, we observed that reproductive life sequences characterized by short reproductive span with natural menopause (adjusted HR: 1.30, 95% CI: 1.18–1.44), early childbearing and hysterectomy menopause (adjusted HR: 1.17, 95% CI: 1.04–1.32), and high parity with long birth span (adjusted HR: 1.13, 95% CI: 1.03–1.25) were associated with higher risks of all-cause dementia, compared to the standard sequence (Table  2 ). The associations remain significant after FDR correction. Similar associations between these three reproductive life sequences and dementia subtypes were identified, including AD, VD, and non-AD and non-VD dementia. However, the associations of the early childbearing and oophorectomy menopause sequence with all-cause dementia (adjusted HR: 0.97, 95% CI: 0.82–1.15) and dementia subtypes were not statistically significant. Fig. 3 Cumulative incidence of all-cause and cause-specific dementia according to reproductive life sequences Cumulative incidence of all-cause and cause-specific dementia according to reproductive life sequences Table 2 Association of reproductive life sequences with all-cause and cause-specific dementia All-cause dementia Alzheimer’s disease Vascular dementia Non-AD and non-VD dementia Events (IR) HR (95% CI) Events (IR) HR (95% CI) Events (IR) HR (95% CI) Events (IR) HR (95% CI) Model 1 Standard sequence 1206 (11.9) 1.00 (reference) 636 (6.3) 1.00 (reference) 223 (2.2) 1.00 (reference) 407 (4.0) 1.00 (reference) Early childbearing and oophorectomy menopause 151 (10.7) 0.89 (0.75–1.06) 74 (5.2) 0.81 (0.63–1.05) 23 (1.6) 0.78 (0.50–1.22) 58 (4.1) 0.92 (0.67–1.26) Short reproductive span with natural menopause 577 (15.1) 1.26 (1.14–1.39) * 300 (7.8) 1.22 (1.06–1.41) * 118 (3.1) 1.38 (1.09–1.76) * 182 (4.7) 1.13 (0.93–1.37) Early childbearing and hysterectomy menopause 375 (15.1) 1.26 (1.12–1.41) * 163 (6.6) 1.02 (0.85–1.22) 83 (3.3) 1.57 (1.21–2.06) * 147 (5.9) 1.31 (1.06–1.63) High parity with long birth span 631 (16.1) 1.35 (1.22–1.48) * 336 (8.6) 1.35 (1.17–1.54) * 130 (3.3) 1.57 (1.25–1.97) * 197 (5.0) 1.23 (1.02–1.48) Model 2 Standard sequence 1206 (11.9) 1.00 (reference) 636 (6.3) 1.00 (reference) 223 (2.2) 1.00 (reference) 407 (4.0) 1.00 (reference) Early childbearing and oophorectomy menopause 151 (10.7) 1.00 (0.84–1.19) 74 (5.2) 0.95 (0.73–1.22) 23 (1.6) 0.84 (0.54–1.33) 58 (4.1) 1.06 (0.77–1.46) Short reproductive span with natural menopause 577 (15.1) 1.30 (1.18–1.43) * 300 (7.8) 1.27 (1.10–1.46) * 118 (3.1) 1.41 (1.11–1.79) * 182 (4.7) 1.19 (0.98–1.44) Early childbearing and hysterectomy menopause 375 (15.1) 1.21 (1.07–1.36) * 163 (6.6) 0.99 (0.83–1.18) 83 (3.3) 1.42 (1.09–1.86) * 147 (5.9) 1.30 (1.04–1.63) High parity with long birth span 631 (16.1) 1.14 (1.04–1.26) * 336 (8.6) 1.14 (0.99–1.31) 130 (3.3) 1.28 (1.01–1.61) 197 (5.0) 1.07 (0.89–1.30) Model 3 Standard sequence 1206 (11.9) 1.00 (reference) 636 (6.3) 1.00 (reference) 223 (2.2) 1.00 (reference) 407 (4.0) 1.00 (reference) Early childbearing and oophorectomy menopause 151 (10.7) 0.97 (0.82–1.15) 74 (5.2) 0.92 (0.71–1.18) 23 (1.6) 0.80 (0.51–1.26) 58 (4.1) 1.04 (0.75–1.43) Short reproductive span with natural menopause 577 (15.1) 1.30 (1.18–1.44) * 300 (7.8) 1.27 (1.10–1.46) * 118 (3.1) 1.41 (1.11–1.79) * 182 (4.7) 1.19 (0.97–1.44) Early childbearing and hysterectomy menopause 375 (15.1) 1.17 (1.04–1.32) * 163 (6.6) 0.97 (0.81–1.16) 83 (3.3) 1.36 (1.04–1.77) 147 (5.9) 1.28 (1.02–1.60) High parity with long birth span 631 (16.1) 1.13 (1.03–1.25) * 336 (8.6) 1.13 (0.98–1.30) 130 (3.3) 1.27 (1.00-1.60) 197 (5.0) 1.06 (0.88–1.28) Model 1: unadjusted Model 2: adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, and ever usage of hormone-replacement therapy Model 3: additionally adjusted for disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease AD Alzheimer’s disease, VD Vascular dementia IR Incidence rate (per 10, 000 person-years) *indicates false discovery rate-adjusted q  < 0.05 within the corresponding model, based on Benjamini-Hochberg correction across 16 comparisons (4 non-reference reproductive sequence patterns × 4 dementia outcomes) Association of reproductive life sequences with all-cause and cause-specific dementia Model 1: unadjusted Model 2: adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, and ever usage of hormone-replacement therapy Model 3: additionally adjusted for disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease AD Alzheimer’s disease, VD Vascular dementia IR Incidence rate (per 10, 000 person-years) *indicates false discovery rate-adjusted q  < 0.05 within the corresponding model, based on Benjamini-Hochberg correction across 16 comparisons (4 non-reference reproductive sequence patterns × 4 dementia outcomes) When considering reproductive factors separately, we found that early age at first live birth, longer birth interval, early age at natural menopause, and shorter reproductive life span were associated with a higher risk of all-cause dementia, which were consistent to the findings from reproductive life sequences (Table S4). Among the included participants, 3.0% ( n  = 4,653) women had IC impairment at baseline. Women with IC impairment had higher risks of developing all-cause dementia (adjusted HR: 1.79, 95% CI: 1.54–2.07) and cause-specific dementia during follow-up (Table S5). These associations were consistent for specific IC impairment domains. The odds of IC impairment at baseline were higher among women with reproductive sequences characterized by early childbearing and oophorectomy or hysterectomy menopause, and short reproductive span with natural menopause (Table S6). In the joint association analysis (Table  3 ), we observed the highest hazard of all-cause dementia among participants had short reproductive span with natural menopause combined with IC impairment (adjusted HR: 2.40, 95% CI: 1.77–3.24), followed by high parity with long birth span combined with IC impairment (adjusted HR: 1.89, 95% CI: 1.39–2.57), and early childbearing and hysterectomy menopause combined with IC impairment (adjusted HR: 1.88, 95% CI: 1.34–2.64), compared to women with the standard sequence and no IC impairment. Table 3 Joint association of reproductive life patterns and intrinsic capacity with all-cause and cause-specific dementia Reproductive life sequences IC impairment All-cause dementia Alzheimer’s disease Vascular dementia Non-AD and non-VD dementia IR HR (95% CI) IR HR (95% CI) IR HR (95% CI) IR HR (95% CI) Standard sequence IC impairment (-) 11.5 1.00 (reference) 6.2 1.00 (reference) 2.1 1.00 (reference) 3.8 1.00 (reference) IC impairment (+) 34.0 1.95 (1.53–2.48) * 10.5 1.11 (0.71–1.72) 8.2 1.85 (1.04–3.29) 16.4 2.90 (1.95–4.30) * Early childbearing and oophorectomy menopause IC impairment (-) 10.3 0.98 (0.82–1.18) 5.1 0.92 (0.70–1.19) 1.5 0.76 (0.47–1.24) 4.0 1.13 (0.81–1.57) IC impairment (+) 19.8 1.50 (0.85–2.65) 8.2 1.09 (0.41–2.92) 4.9 2.11 (0.66–6.68) 6.6 0.96 (0.24–3.84) Short reproductive span with natural menopause IC impairment (-) 14.3 1.30 (1.17–1.44) * 7.5 1.25 (1.08–1.45) * 2.7 1.36 (1.06–1.76) * 4.5 1.23 (1.00-1.51) IC impairment (+) 43.0 2.40 (1.77–3.24) * 19.5 1.90 (1.18–3.07) * 14.2 3.28 (1.78–6.04) * 11.5 1.96 (1.03–3.72) Early childbearing and hysterectomy menopause IC impairment (-) 14.3 1.18 (1.04–1.33) * 6.2 0.94 (0.78–1.13) 3.2 1.36 (1.03–1.81) 5.6 1.34 (1.06–1.69) * IC impairment (+) 33.6 1.88 (1.34–2.64) * 14.9 1.54 (0.89–2.65) 7.4 2.18 (1.04–4.56) 13.9 1.90 (0.97–3.71) High parity with long birth span IC impairment (-) 15.5 1.14 (1.03–1.26) * 8.3 1.11 (0.97–1.28) 3.1 1.26 (0.99–1.60) 4.8 1.10 (0.91–1.34) IC impairment (+) 36.2 1.89 (1.39–2.57) * 16.3 1.71 (1.08–2.71) 10.6 2.54 (1.36–4.77) * 11.4 1.86 (1.01–3.42) All models were adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, ever usage of hormone-replacement therapy, disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease IC Intrinsic capacity, AD Alzheimer’s disease, VD Vascular dementia IR Incidence rate (per 10, 000 person-years) *indicates false discovery rate-adjusted q  < 0.05, based on Benjamini-Hochberg correction across 36 comparisons (9 non-reference joint groups × 4 dementia outcomes) Joint association of reproductive life patterns and intrinsic capacity with all-cause and cause-specific dementia All models were adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, ever usage of hormone-replacement therapy, disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease IC Intrinsic capacity, AD Alzheimer’s disease, VD Vascular dementia IR Incidence rate (per 10, 000 person-years) *indicates false discovery rate-adjusted q  < 0.05, based on Benjamini-Hochberg correction across 36 comparisons (9 non-reference joint groups × 4 dementia outcomes) According to the stratified analysis (Fig.  4 ), the associations between reproductive life patterns characterized by short reproductive span with natural menopause, early childbearing and hysterectomy menopause, and high parity with long birth span with all-cause and cause-specific dementia were more evident among those with IC impairment. Overall, IC impairment across locomotor and vitality domains appeared to show stronger joint associations with reproductive life sequences and dementia risk. For the interaction effects between reproductive life sequences and IC, impairment in the vitality domain appeared to show the most consistent modifying role. Similar results were observed for dementia subtypes (Figure S2-S4). In addition, the associations between reproductive sequences and dementia risk were stronger among women with more items of IC impairment (Figure S5). Fig. 4 Association of reproductive life patterns with all-cause dementia stratified by impairment of specific IC domains. All models were adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, ever usage of hormone-replacement therapy, disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease. *indicates false discovery rate-adjusted q  < 0.05 within each dementia outcome, based on Benjamini-Hochberg correction across 36 comparisons (4 intrinsic capacity domains × 9 non-reference joint groups) Association of reproductive life patterns with all-cause dementia stratified by impairment of specific IC domains. All models were adjusted for age at baseline, ethnicity, educational levels, employment status, income levels, smoking status, alcohol drinking status, physical activity, body mass index categories, ever usage of hormone-replacement therapy, disease history of cardiovascular disease, type 2 diabetes, and cancer, and polygenic risk score for Alzheimer’s disease. *indicates false discovery rate-adjusted q  < 0.05 within each dementia outcome, based on Benjamini-Hochberg correction across 36 comparisons (4 intrinsic capacity domains × 9 non-reference joint groups) Subgroup analyses showed consistent results between reproductive life sequences and dementia across different populations stratified by age at baseline, ethnicity, educational level, employment status, total household income, PRS for AD, and baseline CVD, T2D, and cancer (Figure S6). Notably, the HR for the association between early childbearing and oophorectomy menopause sequences with all-cause dementia was relatively higher among women never using HRT (HR: 1.37, 95% CI: 0.92–2.03), compared to those with ever usage (HR: 0.91, 95% CI: 0.75–1.10). The results of the associations between reproductive life sequences and all-cause and cause-specific dementia were robust according to the sensitivity analyses (Table S7).

In this large-scale prospective cohort study, we identified five patterns of women’s life-course reproductive factors from adolescence to menopause, which could present the typical reproductive sequences among the UK female population. Through the linkage of reproductive life sequences with dementia, we found that sequences characterized by short reproductive span with natural menopause, early childbearing and hysterectomy menopause, and high parity with long birth span were associated with higher risks of all-cause and cause-specific dementia, compared to the standard sequence. These independent associations of reproductive life patterns with dementia were strengthened in women combined with IC impairments. In contrast, the association between the early childbearing and oophorectomy menopause sequences and dementia was not observed. In general, the high-risk reproductive life sequences we identified for dementia were characterized by early age at childbearing, high parity, long birth span, early age at natural menopause, early age at hysterectomy, and short reproductive life span. The majority of these factors remain to show significant associations with dementia when they were assessed separately. Moreover, the levels of risk were higher for these identified high-risk sequences than single reproductive factors, suggesting the potentially cumulative effects of multiple reproductive factors on the development of dementia [ 35 ]. The associations between reproductive life sequences and dementia were broadly consistent with previous findings. Early age at menopause and short reproductive life span were widely explored as female-specific reproductive factors linking with women’s later life risk of cognitive decline and dementia [ 5 , 6 , 8 , 10 , 11 ]. Similarly, our study showed that the short reproductive span with natural menopause sequence was associated with the highest risk of dementia, including AD and VD, among the five reproductive sequences. Such association could mainly be explained by the short exposures of endogenous estrogen in these women, as these sex steroid hormones are neuroprotective and anti-aging [ 12 ]. It was also suggested that estrogen would protect the cardiovascular system and lower the risk of cardiovascular disease, which is also a risk factor for dementia [ 36 ]. For menopause due to surgery, our results revealed an increased risk of dementia among women with the early childbearing and hysterectomy menopause sequence. Although women with premenopausal hysterectomy conserve the ovaries, they tended to reach natural menopause at an earlier age [ 37 ]. A large-scale pooled study combined UK Biobank data with cohorts from Australia, Sweden, England, and the Netherlands found that the type of menopause was not associated with dementia after adjusting for age at menopause [ 5 ], suggesting that early menopause may be a more important predictor of dementia regardless of the menopause types. Corresponding associations were not observed for the early childbearing and oophorectomy menopause sequence, despite a similarly early mean age at menopause as in the short reproductive span with natural menopause sequence. One possible explanation is the high prevalence of HRT use (88%) in this sequence. According to the timing hypothesis, the influence of HRT on cognition may depend on the timing of treatment initiation relative to menopause, with use initiated before or around the average age of natural menopause may be more likely to exert neuroprotective rather than harmful effects [ 38 – 40 ]. In this context, our null finding may reflect a potential neuroprotective window, in which early hormonal intervention after bilateral oophorectomy may partly mitigate the adverse neurocognitive consequences of abrupt estrogen deprivation. This hypothesis is aligned with the subgroup analysis, which suggested that the association between the early childbearing and oophorectomy menopause sequence and dementia appeared relatively stronger among those never used HRT than ever used. Consistently, a previous study using data from the UK Biobank also revealed an association between ever usage of HRT and decreased odds of AD among women with bilateral oophorectomy [ 41 ]. Nevertheless, the UK Biobank does not have detailed information on the initiation times, durations, and medication types of HRT, this explanation remains inferential and should be interpreted cautiously. Further studies are warranted to examine whether timely HRT initiation after surgical menopause may modify long-term dementia risk. Reproductive sequences characterized by an early age at first childbirth have been associated with an increased risk of dementia in the current study. Similar results were also identified in a previous study based on a population from UK Biobank, which revealed that younger age at first birth < 22 years was associated with an increased risk of dementia, in particular for VD, when assessing this factor individually [ 9 ]. Such association was also reported for cognitive decline [ 42 , 43 ]. The socioeconomic disadvantages of women with early childbirth may explain these associations [ 44 , 45 ], as women with these reproductive sequences were more likely to be lower educated, unemployed, with lower household income in this study. Besides, women with these sequences had higher rates of exhibiting unhealthy lifestyles, including smoking and physical inactivity, which are also predictors of dementia and cognitive decline [ 46 , 47 ]. Our study, coupled with previous studies [ 9 , 42 , 48 – 50 ], revealed that the high parity with long birth span sequence was associated with a higher risk of all-cause dementia. Women’s parity is an indicator of cumulative estrogen exposure, with a higher parity referring to a dramatic fluctuation of hormones over the lifetime [ 48 , 49 ]. Such fluctuation due to the repetitive pregnancies may directly harm women’s brain [ 48 ]. In addition, higher parity has been associated with cardiometabolic disorders in women’s later life [ 51 – 53 ], which are risk factors for dementia. From the social perspective, women with the high parity with long birth span sequence were more likely to have lower socioeconomic status [ 48 ], e.g., we observed that women with such sequence tended to be lower educated, unemployed, and with lower household income. These socioeconomic disadvantages have also been linked to the risk of dementia [ 46 , 47 ]. Therefore, the reproductive sequence may be viewed not only as a female-specific biological exposure, but also as a maker of cumulative socioeconomic adversity. This perspective may better capture the interrelated biological and social mechanisms through which reproductive life patterns shape women’s later-life dementia risk. IC, reflecting an individual’s attributes that might influence the healthy ageing, has been associated with higher risks of dementia in the general population [ 21 ]. This association remains evident among females according to the current study. We further conducted joint analyses to examine potential interactions between women’s reproductive sequences and IC on the incidence of dementia. Our results supported significant joint associations of high-risk reproductive sequences and IC impairment with increased risks of all-cause and cause-specific dementia, which were stronger than the independent associations. The subgroup analyses also suggested that corresponding associations were stronger among women with baseline IC impairment. These results emphasize the importance of prioritizing women with high-risk reproductive life sequences in mid-to-late ages, with special focus on their IC (e.g., through regular screening for vision and hearing loss [ 5 ], continuous monitoring of psychological health [ 54 ], to maximise the health benefits. When considering specific IC domains, the locomotor domain showed the largest effect estimates, suggesting that impaired mobility may be a particularly important marker of dementia risk among women with high-risk reproductive patterns. This interpretation is consistent with previous studies showing that slow walking pace or gait impairment is associated with a higher risk of incident dementia [ 55 ]. Meanwhile, the vitality domain showed more consistent evidence of interaction across reproductive patterns, which may reflect the role of vitality as a marker of underlying physiological reserve [ 56 ]. By contrast, the psychology and sensory domains showed weaker or less consistent modification roles. These results suggest that different IC domains may capture different pathways linking reproductive life sequences to dementia risk in later life. The main strengths of this study include the prospective study design, the large sample size, the adjustment of multiple confounders, the consideration of multiple reproductive factors from a life-course perspective, and the examination of the joint association of reproductive life sequences and IC. The usage of the sequence analysis and cluster analysis allowed us to integrate the initiation times, durations, and sequences of multiple reproductive factors, and to identify typical patterns of reproductive sequences in UK female populations. Some limitations should also be noted in this study. First, the cases of cause-specific dementia were relatively small, which may result in the null association of reproductive life patterns with several cause-specific dementia. Second, the information on reproductive factors were self-reported by participants, which was vulnerable to recall biases. Besides, although we have included multiple reproductive factors in the construction of reproductive life sequences, there remain some important factors we have not considered due to the lack of information in UK Biobank, e.g., age at each live birth, age at each pregnancy, age at stillbirth, age at miscarriage, initiation time and duration of HRT, etc. Future studies are warranted to construct a more comprehensive reproductive life sequence by incorporating detailed information of reproductive factors. Third, although we have adjusted for a large set of covariates in this study, residual confounders were still a possibility. In sensitivity analyses, we further adjusted for baseline cognitive performance in sensitivity analyses using reaction time and pairs matching, however, other baseline cognitive tests in the UK Biobank had substantial missing data and were not additionally included. Therefore, residual influence of cognitive function could not be fully addressed. Forth, this is an observational study and causality cannot be inferred. Last, the generalizability of our findings may be limited. The included participants were restricted to post-menopausal women and those with missing or implausible information were not included. Compared to women excluded from the main analysis, those included participants differed in some sociodemographic and lifestyle factors and baseline health conditions. Therefore, selection bias may exist in this study. In addition, about 96% participants were White, which may limit the generalization of our results to other population. Although “unknown” responses for household income and physical activity differed modestly across reproductive sequence patterns, differential missingness across patterns cannot be entirely excluded. Further studies are needed in populations from different countries and regions with more complete reproductive and functional data. This study has critical public health implications. The individual associations of reproductive factors with the risk of dementia have been widely explored in previous studies based on populations from different sociodemographic contexts. Our study extended previous studies by integrating the initiation times, durations, and sequences of women’s life-course reproductive factors, identifying five typical patterns of reproductive sequences in a UK female population and linking them to the risk of dementia. Such results could facilitate future considerations of which women to be considered as the high-risk groups in clinical practice, which would inform timely prevention and intervention of dementia. This study also provided methodological clues for future research that the construction of reproductive life sequences would help estimate the cumulative effects of life-course reproductive factors on health outcomes. In addition, through the examination of joint associations of reproductive sequences and IC, our study suggests that the specific screening and monitoring of IC in middle and older ages should be prioritized in women with high-risk reproductive sequences.

Our study showed cumulative associations of women’s life-course reproductive factors with the risk of dementia in later life. Reproductive life patterns characterized by short reproductive span with natural menopause, early childbearing and hysterectomy menopause, and high parity with long birth span were associated with higher risks of all-cause and cause-specific dementia, compared to the standard sequence. The independent associations would be exaggerated when combining with IC impairments. These results highlight the importance of prioritizing women with high-risk reproductive sequences and IC impairments in clinical practice for the prevention and intervention of dementia.

Dementia is one of the leading causes of disability, dependency, and death worldwide [ 1 ], afflicting both patients and their families. The global burden of dementia continues to rise alongside the ageing population, with the number of people living with dementia been projected to increase from 57 million in 2019 to 153 million by 2050 2 . Women are disproportionately affected by dementia. The prevalence of dementia in females was estimated to be 1.69-fold higher than males [ 2 ], highlighting the need to identify female-specific risk factors for dementia that might lower the risk of onset differentially by sex [ 3 ]. Female-specific reproductive factors during the lifetime, such as age at menopause, reproductive life span (years between menarche to menopause), number of births, and age at childbearing [ 4 – 11 ], have been identified as critical factors affecting the risk of dementia. For example, a pooled analysis combining five longitudinal studies suggested that early menopause (< 40 years) was associated with a higher risk of dementia [ 5 ]. A systematic review and meta-analysis illustrated that women with a short reproductive life span (< 35 years) had a higher risk of developing dementia [ 10 ]. These associations could be explained by the shorter lifetime exposure to endogenous estrogen which appear to be neuroprotective [ 12 ]. For factors related to childbirth, a meta-analysis concluded that a higher parity (≥ 5) was associated with 32% increased risk of dementia [ 8 ]. There is also evidence supporting the association between early age at first childbirth (< 22 years) and higher risks of all-cause dementia and dementia subtypes [ 9 ]. Previous studies examining the separate associations of reproductive factors with dementia ignored that many of these factors prone to be inter-related [ 13 ]. So far, less is known about how life-course reproductive factors cumulatively inform women’s future risk of dementia when incorporating the sequences, initiation times, durations, and correlations of multiple reproductive factors. Women’s reproductive factors have also been linked to functional abilities in mid-to-late life. For instance, age at menarche, reproductive life span, menopause age, and parity, could potentially predict the physical frailty in middle and old ages [ 14 – 18 ]. Intrinsic capacity (IC), a concept comprising an individual’s mental and physical capacities, reflects the functional ability enabling a person to be and do what they value [ 19 ]. IC impairments have been associated with higher risks of neurodegenerative disorders and dementia [ 20 , 21 ]. Despite the separate association of IC with dementia, to date, we found no studies have examined the joint associations of women’s reproductive factors and IC on the risk of dementia. It is important to explore separate associations of reproductive life sequences and IC impairments, and further how they interact to influence women’s risk of dementia. Using data from the UK Biobank, this study aims to identify reproductive life sequence and its potential patterns in the UK female population, and to examine the independent and joint associations of female reproductive life sequences and IC with all-cause and cause-specific dementia.

Supplementary Material 1. Supplementary Material 1.