ABSTRACT
Objective
To investigate whether age at menopause is associated with the trajectory of multimorbidity progression.
Design
A retrospective cohort study.
Setting
The UK Biobank.
Population
121 017 postmenopausal women aged 40–69 years with complete baseline and menopausal data.
Methods
Cox proportional hazards models were used to explore the associations between menopausal age and both multimorbidity and mortality. Multi‐state models were further employed to evaluate the association of menopausal age with transitions from a health state to the first chronic disease (FCD), to multimorbidity, and to mortality from each state.
Main Outcome Measures
Multimorbidity (defined as ≥ 2 of 35 chronic conditions) and mortality.
Results
Over a median follow‐up period of 8.6 years, 86 821 women developed the FCD, 42237 multimorbidity, and 10 527 participants died. In the multi‐state models analysis, from health to FCD, compared with women who experienced menopause after age 50 years, the risks increased by 32%, 14%, and 3% among women with premature, early, and relatively early menopause, respectively (HR = 1.32, 95% CI: 1.28–1.36; HR = 1.14, 95% CI: 1.11–1.17; HR = 1.03, 95% CI: 1.01–1.05). Similarly, for the transition from FCD to multimorbidity, the risks increased by 22% (HR = 1.22, 95% CI: 1.17–1.27), 13% (HR = 1.13, 95% CI: 1.09–1.16), and 6% (HR = 1.06, 95% CI: 1.03–1.08), respectively.
Conclusions
Earlier menopause was associated with an increased risk of multimorbidity and mortality, mainly affecting two trajectories, from health to FCD and FCD to multimorbidity.
Keywords: age at menopause, menopausal age, multimorbidity progression, multi‐state model, postmenopausal women, UK biobank cohort
1. Background
Multimorbidity, defined as the co‐occurrence of two or more chronic diseases in an individual, poses a significant challenge to health systems worldwide [1]. Accumulating evidence demonstrates that patients with multimorbidity face a higher risk of premature mortality and longer hospital stays compared with those with a single condition [2, 3]. These patients have a two‐fold higher risk of mortality relative to individuals without multimorbidity [4]. Notably, the prevalence of multimorbidity is significantly higher in women than in men, particularly among postmenopausal women, where rates reach 60% [5]. Unfortunately, the factors that contribute to the elevated prevalence of multimorbidity among postmenopausal women remain poorly understood.
Menopause, a physiological marker of aging in women, has been reported to increase the risk of multiple diseases, including breast cancer and osteoporosis [6, 7, 8]. The majority of women experience menopause between the ages of 40 and 58, with the average age being 51 years [9, 10]. Evidence has demonstrated that age at menopause is associated with chronic diseases, with premature menopause specifically elevating the risk of developing multiple such conditions [11, 12, 13]. Several prior studies have examined the link between age at menopause and multimorbidity, demonstrating that there is an association between menopausal age and multimorbidity [14, 15]. However, it remains unclear whether age at menopause is associated with the dynamic disease trajectory, such as from health to a single disease, subsequent progression to multimorbidity, and eventual mortality.
To understand their correlation, we employed a multi‐state model using data from the UK Biobank to investigate whether age at menopause can influence the progression trajectory of multimorbidity, including transition from a healthy state (free of chronic diseases) to first chronic disease (FCD), subsequent progression to multimorbidity, and ultimately mortality. This approach can help us identify the relative risks at different transitions of multimorbidity and provide potential strategies for disease prevention and intervention.
2. Materials and Methods
2.1. Data Source and Study Cohort
The UK Biobank is a large cohort study including over 500 000 participants aged 40–73 years, enrolled between 2006 and 2010 across the United Kingdom. At the baseline study visit, data on demographic characteristics, surgical history, and menopausal history were collected via touchscreen questionnaires, verbal interviews, physical measurements, and biometric samples. All participants provided written informed consent, and the UK Biobank has received ethical approval from the National Research Ethics Service. Follow‐up data through June 2023 were derived from hospital inpatient records for disease diagnoses and from the national death register for mortality outcomes.
Of the total cohort of 273 165 women, we excluded those who were premenopausal (n = 64 043) or had unknown menopausal status (n = 445), those with missing age at menopause (n = 41 206), those with discordant timing of bilateral oophorectomy and menopause (n = 111), those with any of 35 chronic diseases at baseline (n = 45 993), and those lost to follow‐up (n = 350), leaving 121 017 participants for analysis (Figure 1).
FIGURE 1.
The flowchart of the UK Biobank participants included and excluded in the study.
2.2. Age at Menopause
Natural menopause was defined as the absence of menstruation over a period for 12 consecutive months and no history of hysterectomy or oophorectomy prior to this. Surgical menopause was defined as the removal of both ovaries [16]. Participants were grouped by age at menopause to assess the associations of earlier menopause timing with multimorbidity and mortality, rather than to reflect physiological age. Based on prior evidence, age at menopause was categorized into four groups: < 40 years (premature menopause), 40–44 years (early menopause), 45–49 years (relatively early menopause), and ≥ 50 years (reference group) [17, 18].
2.3. Outcomes
The outcomes of interest in this study were multimorbidity and mortality. Multimorbidity was identified as the presence of any two or more of the 35 chronic conditions during the follow‐up period (Methods S1). Information on 34 non‐cancer chronic diseases was obtained from self‐reported information and health‐related records, including primary care, hospital admission, and death registration data in the UK Biobank. Cancer information was obtained from the national central registry [19]. Mortality included both cause‐specific mortality (deaths attributed to one or more of the 35 chronic conditions) and all‐cause mortality. Diagnoses of chronic diseases and causes of death were confirmed using the ICD‐10 codes.
2.4. Covariates
Information on covariates was collected, including age, ethnic background, education level, body mass index (BMI), Townsend deprivation index, healthy diet score (HDS), hormone‐replacement therapy (HRT), smoking status, and alcohol consumption. These covariates were chosen based on potential associations with both age at menopause and multimorbidity outcomes (Figure S1). Townsend deprivation index is a measure of overall deprivation and reflects the socioeconomic status [20]. HDS was determined based on several factors, including daily vegetable intake of at least four tablespoons, daily fruit intake of at least three pieces, consuming fish at least twice a week, limiting unprocessed red meat consumption to no more than twice a week, and restricting processed meat intake to no more than twice a week. Each favourable diet factor received one point, resulting in a total diet score that can range from 0 to 5 [21].
2.5. Statistical Analysis
Baseline characteristics were presented as mean values with standard deviations (SDs) for continuous variables and as numbers and weighted percentages for categorical variables. Continuous variables were compared using t‐tests, and categorical variables were analysed using ANOVA or chi‐square tests. Heat maps were used to visualize chronic disease case counts and the most frequent dyad disease patterns.
In the survival analyses, we used restricted cubic spline plots to assess potential non‐linear associations between age at menopause and multimorbidity or mortality. Then we plotted Kaplan–Meier (KM) curves to evaluate the cumulative event rates and survival probability of different menopausal age groups, and used the log‐rank test to assess the differences between the curves. The proportional hazards assumption was tested using Schoenfeld residuals (p = 0.86). Cox proportional hazards regression models were employed to estimate the association between age at menopause and the risk of the FCD, multimorbidity, multimorbidity‐specific mortality, and all‐cause mortality. Results are presented as hazard ratios (HRs) with their corresponding 95% confidence intervals (CIs).
Subsequently, weighted multi‐state models (five possible states) and semi‐Markov models (Weibull distribution) were applied to estimate the associations between age at menopause and disease progression, including transitions from baseline health to intermediate events, and ultimately to mortality. Based on the disease's development and previous evidence, five transitions were specified in the multi‐state framework: (A) from baseline health to FCD; (B) from FCD to multimorbidity; (C) from baseline health to mortality due to non‐35 chronic conditions; (D) from FCD to any‐cause mortality; and (E) from multimorbidity to any‐cause mortality [22, 23]. These five transitions could occur at any time during follow‐up; however, once an individual transitioned to a given disease state, they were assumed to remain in that state until death or the end of follow‐up. The semi‐Markov model accounted for the time elapsed since the previous state. If participants transitioned between multiple states on the same day, the duration spent in the preceding state was calculated using fixed median intervals.
To control for potential confounding rather than to estimate a direct causal effect unmediated by intermediate variables, we applied three progressively adjusted models in Cox proportional hazards models and multi‐state analyses, with sequential adjustment for age, ethnicity, education levels, Townsend deprivation index, and lifestyle‐ and health‐related factors (including BMI, HDS, smoking status, alcohol consumption, and HRT) to examine the association between menopausal age and multimorbidity and mortality.
2.6. Sensitivity Analyses
We conducted a series of sensitivity analyses to test the stability of the results: (a) excluding participants who died or developed multimorbidity within 2 years from baseline; (b) excluding individuals with missing covariates data; (c) calculating the entering date of the prior state using different median intervals (0.5, 1, 3, and 5 days) for participants who entered different states on the same day.
All analyses were performed using R (version 4.2.3), and multi‐state models were implemented using the “mstate” package. The missing data in this study were assumed to missing at random; therefore, multiple imputations (five imputations) were performed using the “mice” package in R and combined according to the Rubin's rules to obtain the final results. A two‐tailed p‐value < 0.05 indicated statistical significance. Bonferroni correction was applied to control for the impact of multiple comparisons (p = 0.017).
3. Results
3.1. Baseline Characteristics of the Study Population
This study included 121 017 participants with an average age of 60.2 ± 5.57 years, 95% of whom were white (Table 1). Compared with women without multimorbidity, those with multimorbidity had a higher proportion of premature menopause (7.2% vs. 4.2%), early menopause (11.5% vs. 9.3%). Individuals with multimorbidity were more likely to have a history of hormone therapy use, to be older, obese, and smokers; less likely to have lower education levels, consume less alcohol, and have lower HDS.
TABLE 1.
Baseline characteristics of 121 017 participants by incident status of multimorbidity and mortality at the end of follow‐up.
| Characteristics | Total | Multimorbidity | Mortality | ||
|---|---|---|---|---|---|
| (N = 121 017) | No (N = 78 780) | Yes (N = 42 237) | No (N = 110 490) | Yes (N = 10 527) | |
| Age, Mean (SD), years | 60.2 (5.57) | 59.5 (5.52) | 61.6 (5.40) | 60.0 (5.56) | 62.6 (5.12) |
| Age at menopause, n (%), years | |||||
| < 40 | 6367 (5.3%) | 3339 (4.2%) | 3028 (7.2%) | 5627 (5.1%) | 740 (7.0%) |
| 40–44 | 12 157 (10.0%) | 7315 (9.3%) | 4842 (11.5%) | 10 977 (9.9%) | 1180 (11.2%) |
| 45–49 | 29 777 (24.6%) | 19 500 (24.8%) | 10 277 (24.3%) | 27 157 (24.6%) | 2620 (24.9%) |
| ≥ 50 | 72 716 (60.1%) | 48 626 (61.7%) | 24 090 (57.0%) | 66 729 (60.4%) | 5987 (56.9%) |
| Ethnic background, n (%) | |||||
| White | 115 873 (95.7%) | 75 552 (95.9%) | 40 321 (95.5%) | 105 693 (95.7%) | 10 180 (96.7%) |
| Non‐white | 5144 (4.3%) | 3228 (4.1%) | 1916 (4.5%) | 4797 (4.3%) | 347 (3.3%) |
| Education level, n (%) | |||||
| College or University degree | 33 839 (28.0%) | 24 326 (30.9%) | 9513 (22.5%) | 31 449 (28.5%) | 2390 (22.7%) |
| Any other qualification | 60 140 (49.7%) | 39 709 (50.4%) | 20 431 (48.4%) | 55 346 (50.1%) | 4794 (45.5%) |
| No qualification | 27 038 (22.3%) | 14 745 (18.7%) | 12 293 (29.1%) | 23 695 (21.4%) | 3343 (31.8%) |
| BMI categories, n (%) | |||||
| Underweight (< 18.5 kg/m2) | 895 (0.7%) | 539 (0.7%) | 356 (0.8%) | 764 (0.7%) | 131 (1.2%) |
| Normal weight (18.5–24.9 kg/m2) | 45 102 (37.3%) | 32 204 (40.9%) | 12 898 (30.5%) | 41 709 (37.7%) | 3393 (32.2%) |
| Overweight (25–29.9 kg/m2) | 45 975 (38.0%) | 30 220 (38.4%) | 15 755 (37.3%) | 42 169 (38.2%) | 3806 (36.2%) |
| Obese (≥ 30 kg/m2) | 29 045 (24.0%) | 15 817 (20.1%) | 13 228 (31.3%) | 25 848 (23.4%) | 3197 (30.4%) |
| Townsend deprivation index, Mean (SD) | −1.41 (3.00) | −1.60 (2.80) | −1.07 (3.17) | −1.46 (2.97) | −0.923 (3.28) |
| Healthy diet score (HDS), Mean (SD) | |||||
| 0 | 1867 (1.5%) | 1149 (1.5%) | 718 (1.7%) | 1655 (1.5%) | 212 (2.0%) |
| 1 | 10 861 (9.0%) | 6721 (8.5%) | 4140 (9.8%) | 9790 (8.9%) | 1071 (10.2%) |
| 2 | 25 033 (20.7%) | 16 049 (20.4%) | 8984 (21.3%) | 22 724 (20.6%) | 2309 (21.9%) |
| 3 | 30 714 (25.4%) | 19 951 (25.3%) | 10 763 (25.5%) | 28 004 (25.3%) | 2710 (25.7%) |
| 4 | 31 590 (26.1%) | 20 907 (26.5%) | 10 683 (25.3%) | 29 011 (26.3%) | 2579 (24.5%) |
| 5 | 20 952 (17.3%) | 14 003 (17.8%) | 6949 (16.5%) | 19 306 (17.5%) | 1646 (15.6%) |
| Smoking status, n (%) | |||||
| Previous | 41 167 (34.0%) | 25 682 (32.6%) | 15 485 (36.7%) | 37 282 (33.7%) | 3885 (36.9%) |
| Current | 10 927 (9.0%) | 5823 (7.4%) | 5104 (12.1%) | 9103 (8.2%) | 1824 (17.3%) |
| Never | 68 923 (57.0%) | 47 275 (60.0%) | 21 648 (51.2%) | 64 105 (58.1%) | 4818 (45.8%) |
| Alcohol consumption, n (%) | |||||
| Two or less times a week | 63 967 (52.8%) | 41 050 (52.1%) | 22 917 (54.3%) | 58 313 (52.8%) | 5654 (53.7%) |
| Three or more times a week | 44 877 (37.1%) | 31 230 (39.6%) | 13 647 (32.3%) | 41 552 (37.6%) | 3325 (31.6%) |
| Never | 12 173 (10.1%) | 6500 (8.3%) | 5673 (13.4%) | 10 625 (9.6%) | 1548 (14.7%) |
| Menopausal type, n (%) | |||||
| Natural | 111 550 (92.2%) | 73 220 (92.9%) | 38 330 (90.7%) | 101 823 (92.2%) | 9727 (92.4%) |
| Surgical | 9467 (7.8%) | 5560 (7.1%) | 3907 (9.3%) | 8667 (7.8%) | 800 (7.6%) |
| Hormone‐replacement therapy, n (%) | |||||
| Ever use | 59 384 (49.1%) | 35 751 (45.4%) | 23 633 (56.0%) | 53 912 (48.8%) | 5472 (52.0%) |
| Never use | 61 633 (50.9%) | 43 029 (54.6%) | 18 604 (44.0%) | 56 578 (51.2%) | 5055 (48.0%) |
Note: Multimorbidity is identified as the presence of any two or more of the 35 chronic conditions during the follow‐up period. The mortality refers to all‐cause mortality. All‐cause mortality refers to death from any cause.
Over a median follow‐up period of 8.6 years, 86 821 (71.7%) participants developed a chronic disease. Of these, 42 237 (48.6%) participants progressed to multimorbidity, while 3919 (4.5%) participants without multimorbidity died. Notably, among the 42 237 individuals with multimorbidity, 6427 (15.2%) deaths were recorded (Figure 2). Among participants with single diseases, cancer was the most prevalent, affecting 17 851 individuals, followed by hypertension with 11 515 cases. Among those with multimorbidity, the combination of cancer and hypertension was most common. Specifically, 1205 individuals developed hypertension first, followed by cancer, while 851 initially had cancer before developing hypertension (Figure 3).
FIGURE 2.
Numbers (percentages) of participants in transition patterns from health to first chronic disease (FCD), multimorbidity, and all‐cause mortality. Multimorbidity is defined as the coexistence of at least two chronic diseases.
FIGURE 3.
Dyad disease patterns of multimorbidity among 35 included chronic diseases (n = 42 237).
3.2. Survival Analyses
The restricted cubic spline analyses indicated a non‐linear association between age at menopause and the outcomes (Figures S2 and S3). Kaplan–Meier curves showed the cumulative event rates of FCD and multimorbidity, and overall survival probabilities across menopausal age groups. Compared with women who experienced menopause at ≥ 50 years, those who experienced earlier menopause had higher rates of adverse outcomes and lower overall survival, particularly those with premature menopause (log‐rank p < 0.001) (Figures S4 and S5).
Table 2 presents the numbers of outcome events and HRs (95% CIs) for postmenopausal women with a history of earlier menopause. After adjusting for all potential confounding variables (model 3), premature menopause (HR = 1.32, 95% CI: 1.29–1.36), early menopause (HR = 1.14, 95% CI: 1.12–1.17), and relatively early menopause (HR = 1.04, 95% CI: 1.02–1.05) were all associated with FCD occurrence. Similarly, the three menopausal age groups increased the risk of multimorbidity by 46% (95% CI: 1.41–1.52), 21% (95% CI: 1.17–1.25), and 8% (95% CI: 1.05–1.10), respectively. For mortality, women who experienced premature menopause had a 26% higher hazard for multimorbidity‐specific mortality (1.26, 95% CI: 1.15–1.39) and a 38% higher hazard for all‐cause mortality (1.38, 95% CI: 1.27–1.49), compared with those who experienced menopause at ≥ 50 years. The associations between age at menopause and both multimorbidity and mortality were stronger at younger menopausal ages (p for trend < 0.001).
TABLE 2.
Cox proportional hazards models to assess associations of age at menopause with first chronic disease, multimorbidity, and mortality.
| Age at menopause (years) | First chronic disease | Multimorbidity | Multimorbidity‐specific mortality | All‐cause mortality | ||||
|---|---|---|---|---|---|---|---|---|
| Case | HR (95% CI) | Case | HR (95% CI) | Case | HR (95% CI) | Case | HR (95% CI) | |
| Model 1 | ||||||||
| < 40 | 5254 | 1.55 (1.51, 1.60) | 3028 | 1.83 (1.76, 1.90) | 505 | 1.47 (1.34, 1.61) | 740 | 1.63 (1.51, 1.76) |
| 40–44 | 9252 | 1.25 (1.23, 1.28) | 4842 | 1.38 (1.34, 1.42) | 855 | 1.24 (1.15, 1.33) | 1180 | 1.29 (1.21, 1.37) |
| 45–49 | 21 047 | 1.08 (1.06, 1.10) | 10 277 | 1.15 (1.12, 1.17) | 1908 | 1.14 (1.08, 1.21) | 2620 | 1.19 (1.13, 1.24) |
| ≥ 50 | 51 269 | 1 (reference) | 24 090 | 1 (reference) | 4503 | 1 (reference) | 5987 | 1 (reference) |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
| Model 2 | ||||||||
| < 40 | 5254 | 1.48 (1.44, 1.53) | 3028 | 1.70 (1.63, 1.76) | 505 | 1.38 (1.26, 1.52) | 740 | 1.51 (1.40, 1.63) |
| 40–44 | 9252 | 1.22 (1.19, 1.25) | 4842 | 1.31 (1.27, 1.36) | 855 | 1.19 (1.11, 1.28) | 1180 | 1.23 (1.15, 1.31) |
| 45–49 | 21 047 | 1.06 (1.04, 1.08) | 10 277 | 1.12 (1.09, 1.14) | 1908 | 1.12 (1.06, 1.18) | 2620 | 1.15 (1.10, 1.21) |
| ≥ 50 | 51 269 | 1 (reference) | 24 090 | 1 (reference) | 4503 | 1 (reference) | 5987 | 1 (reference) |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
| Model 3 | ||||||||
| < 40 | 5254 | 1.32 (1.29, 1.36) | 3028 | 1.46 (1.41, 1.52) | 505 | 1.26 (1.15, 1.39) | 740 | 1.38 (1.27, 1.49) |
| 40–44 | 9252 | 1.14 (1.12, 1.17) | 4842 | 1.21 (1.17, 1.25) | 855 | 1.13 (1.05, 1.21) | 1180 | 1.16 (1.09, 1.24) |
| 45–49 | 21 047 | 1.04 (1.02, 1.05) | 10 277 | 1.08 (1.05, 1.10) | 1908 | 1.08 (1.03, 1.14) | 2620 | 1.12 (1.07, 1.17) |
| ≥ 50 | 51 269 | 1 (reference) | 24 090 | 1 (reference) | 4503 | 1 (reference) | 5987 | 1 (reference) |
| P for trend | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
Note: Multimorbidity is identified as the presence of any two or more of the 35 chronic conditions during the follow‐up period. Multimorbidity‐specific mortality refers to the cause of death due to one or more among 35 chronic diseases. All‐cause mortality refers to death from any cause. Model 1: analysis adjusted for age, ethnic background. Model 2: analysis adjusted for age, ethnic background, educational level, and Townsend deprivation index. Model 3: analysis adjusted for age, ethnic background, educational level, Townsend deprivation index, BMI, healthy die score (HDS), smoking status, alcohol consumption, and hormone‐replacement therapy (HRT).
Abbreviations: CI, confidence interval; HR, hazard ratio.
3.3. Multi‐State Model Analyses
The results from the multi‐state model showed that the age at menopause was associated with an increased risk of developing the FCD and multimorbidity (Figure 4). In Model 3, compared with women who experienced menopause after age 50, those with premature, early, and relatively early menopause had a 32% (HR = 1.32, 95% CI: 1.28–1.36), 14% (HR = 1.14, 95% CI: 1.11–1.17), and 3% (HR = 1.03, 95% CI: 1.01–1.05) higher risk of transitioning from a healthy state to FCD, respectively. From FCD to multimorbidity, the risk increased by 22% (HR = 1.22, 95% CI: 1.17–1.27), 13% (HR = 1.13, 95% CI: 1.09–1.16), and 6% (HR = 1.06, 95% CI: 1.03–1.08), respectively. These results showed that earlier menopause increased the risk of transitioning from health to FCD and from FCD to multimorbidity.
FIGURE 4.
Multi‐state model to assess associations of age at menopause with different transitions from health to first chronic disease (FCD), multimorbidity, and all‐cause mortality. Age at menopause was categorized into the following groups: < 40 years (premature menopause), 40–44 years (early menopause), 45–49 years (relatively early menopause), ≥ 50 years (reference). Model 1 was adjusted for age and ethnic background. Model 2 included additional adjustments for education levels and the Townsend deprivation index. Model 3 was further adjusted for BMI, HDS, smoking status, alcohol consumption, and HRT.
From multimorbidity to mortality, the HR was 1.09 (95% CI: 1.03–1.16) for the relatively early menopause group in model 3, while the HRs for the other two groups were not significant. The trend across all models showed decreasing HR values for the transitions from health to FCD and from FCD to multimorbidity with later menopausal age. Furthermore, age at menopause was not significantly linked to the transitions from health to mortality or from FCD to mortality.
3.4. Sensitivity Analyses
To assess the stability of the multi‐state model, we conducted a series of sensitivity analyses. These analyses confirmed the stability of our results. First, excluding events occurring within the first 2 years of follow‐up yielded consistent results (Table S1). Second, varying time intervals between the prior state and the subsequent state to calculate the entry date of the prior state for participants entering different states on the same day did not change the results (Table S2). Finally, excluding participants with incomplete covariate data also showed consistent outcomes (Table S3).
4. Discussion
In this study of 121 017 postmenopausal women, we observed a substantial burden of chronic disease, with 78.7% of participants developing at least one chronic condition and nearly half (48.6%) progressing to multimorbidity. Our findings indicated that earlier age at menopause was associated with an increased risk of multimorbidity and mortality. Compared to women experiencing menopause at or after age 50, those with premature menopause had a 1.46–fold higher risk of multimorbidity and a 1.38–fold higher mortality risk. The results of the multi state model further indicated that the earlier the menopause, the stronger the association during transitions from health to FCD and then to multimorbidity. Moreover, the strength of these associations followed a graded pattern, with the strongest effects observed in premature menopause.
To our knowledge, this is the first study to apply a multi‐state model to investigate whether the age at menopause is associated with the trajectory of multimorbidity progression. Multi‐state models are probabilistic frameworks used to analyse transitions between health states, assuming that transition rates are independent of both sojourn time (duration in the current state) and prior state sequences [24, 25]. Although the Cox proportional hazards models effectively quantify risk factor effects for terminal endpoints like mortality via hazard ratios, they cannot account for intermediate events along progression trajectories [26]. The multi‐state model bridges this gap through dynamic capture of transitions between distinct health states, thereby providing a more accurate description of multimorbidity and mortality dynamics [27]. Building on this methodological distinction, we first used Cox proportional hazards models to assess endpoint risks (multimorbidity and mortality), then applied multi‐state models to examine effects on intermediate transitions.
Previous studies have demonstrated an association between the age at menopause and the incidence of chronic diseases and mortality. A meta‐analysis of 310 000 women found that those who entered menopause before 45 had a higher risk of coronary artery disease than those after 45 [28]. Another cohort study in Taiwan suggested that relatively early menopause was associated with higher all‐cause and cardiovascular diseases‐specific mortality [29]. Studies from Canada and Australia on menopausal age and multimorbidity found that premature or early menopause is associated with an increased risk of developing multiple chronic conditions in women [30, 31]. Building upon these prior studies, the present study expanded the scope of chronic diseases to encompass 35 distinct conditions and further explored the impact of menopausal age on mortality risk. Similarly, our study revealed that age at menopause not only increased the risk of developing multimorbidity but also amplified risks of all‐cause mortality and multimorbidity‐specific mortality, following a dose–response gradient in which earlier menopausal age progressively increased both risks.
While the above findings underscored that age at menopause was associated with the occurrence of multimorbidity, its relationship with disease progression remained unclear. To address this, we employed a multi‐state model, finding that age at menopause was associated with transitions from health to FCD and from FCD to multimorbidity, but had little to no effect on direct transitions to mortality (i.e., from health to mortality, FCD to mortality, or multimorbidity to mortality). However, we also observed that the effect of age at menopause varied across transitions from health to FCD and from FCD to multimorbidity. Compared with women who experienced menopause at age ≥ 50 years, those with menopause before Age 40 had 32% and 22% higher risks for these two transitions, respectively, whereas women who reached menopause at ages 45–49 years showed only modest increases of 3% and 6%. This may be because menopause occurring at ages 45–49 years is closer to the typical age of natural menopause at 51 years. Emerging evidence indicates that postmenopausal women with multimorbidity face substantially higher mortality risk, which increases with each additional chronic disease [4, 32]. Therefore, during women's progression from health to multimorbidity, greater attention to those with earlier menopause and targeted interventions may help reduce subsequent mortality risk. Potential strategies include postmenopausal oestrogen supplementation [33, 34], vitamin D and K supplementation [35], and lifestyle modifications such as regular exercise and smoking cessation [36, 37]. Accordingly, prevention efforts may be prioritized for women who experienced menopause before age 45, while interventions for those aged 45–49 years may have limited clinical value. This study provides insights for preventive strategies to improve women's health and reduce the multimorbidity risk after menopause. In particular, targeted interventions for women with earlier menopause may lower their mortality risk during transitions to FCD and multimorbidity.
Several mechanisms may explain the association between earlier menopausal age and increased risks of multimorbidity and mortality. Menopause signifies the depletion of ovarian follicular reserves, accompanied by a substantial decline in endogenous oestrogen synthesis within the body [38, 39]. Menopause can change body fat distribution, increase visceral adipose tissue content, reduce insulin sensitivity, and elevate the risk of developing metabolic disorders [40, 41]. Concurrently, oestrogen loss promotes endothelial dysfunction and reduces anti‐inflammatory compound synthesis, accelerating atherosclerosis [42]. An earlier menopause extends the duration of exposure to these oestrogen‐deficient states, thereby amplifying disease risk and leading to variations in multimorbidity and mortality among postmenopausal women.
The strengths of our study were the use of a large sample design and the collection of extensive data, which allowed us to adjust for a variety of potential confounders. Additionally, while most previous studies focused on a single disease, this study specifically addresses multimorbidity and includes a broader set of chronic conditions. Furthermore, we used a multi‐state model, which describes in a more detailed way the transition of patients from health to intermediate events to the process of mortality, to explore the effect of age at menopause on the different transition stages.
Some limitations should be noted in the present study. Firstly, our study used self‐reported information about menopausal age, which may have introduced misclassification bias. However, misclassification may be random and may only lead to underestimation of the associations observed in this analysis [43]. Secondly, as a retrospective cohort study, this study is inherently subject to potential unmeasured confounding, as not all relevant variables may have been fully captured or adjusted for. Furthermore, owing to its observational nature, definitive causal relationships cannot be inferred. Thirdly, for participants transitioning to different states on the same date, the use of a fixed median proportion to calculate theoretical entry dates for prioritized states may have affected the accuracy of associations. However, sensitivity analyses using the median interval times for different states and excluding participants who transitioned on the same day yielded similar results, indicating the robustness of our findings. Fourthly, by excluding women with chronic diseases at baseline, the study population was healthier than the general population. Consequently, the observed associations may primarily reflect effects among relatively healthy individuals, potentially underestimating the true associations. Fifthly, although a Bonferroni correction was applied to address multiple comparisons, the possibility of Type I error cannot be entirely excluded.
5. Conclusions
Age at menopause was associated with the risk of progression from health to FCD and subsequently to multimorbidity, with a higher risk observed among women with earlier menopause. This highlights the importance of age at menopause as a key risk factor, particularly premature menopause.
Author Contributions
W.T. and X.L. conceived the idea. M.L. and X.R. designed the study and wrote the original manuscript. H.Z., T.L., Z.W. and X.H. analysed and interpreted data. X.Y., H.W. and L.Z. performed the validation analyses and critically revised important intellectual content. All authors have read and agree to the published version of the manuscript.
Funding
This work was supported by the Noncommunicable Chronic Diseases National Science and Technology Major Project (2023ZD0503500).
Ethics Statement
The UK Biobank received ethical approval from the NHS National Research Ethics Service (Ref21/NW/0382). All participants provided written informed consent to participate in the study.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1: List of the 35 chronic conditions included within the definition of multimorbidity.
Figure S1: Directed acyclic graph of the association between exposure, covariates, and outcomes.
Figure S2: Restricted cubic spline (RCS) analysis of the association between age at menopause and multimorbidity.
Figure S3: Restricted cubic spline (RCS) analysis of the association between age at menopause and all‐cause mortality.
Figure S4: Cumulative event rates of (A) multimorbidity and (B) all‐cause mortality associated with age at menopause.
Figure S5: Kaplan–Meier survival curves stratified by menopausal age groups over follow‐up time.
Table S1: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases excluding events in the first 2 years of follow‐up.
Table S2: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases for participants who enter different states on the same day, different time intervals (0.5, 1, 3, 5 days) were used to calculate the entry date of the previous state.
Table S3: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases excluding participants with incomplete covariates.
Acknowledgements
This research has been conducted using the UK Biobank Resource under Application Number 69815.
Contributor Information
Xiyun Ren, Email: renxiyunvip@163.com.
Xiangfeng Lu, Email: luxf@pumc.edu.cn.
Wenjing Tian, Email: tianwenjing@ems.hrbmu.edu.cn.
Data Availability Statement
All data are available in the UK Biobank data (https://www.ukbiobank.ac.uk/). UK Biobank data are publicly accessible for research use by application.
References
- 1. Valderas J. M., Starfield B., Sibbald B., Salisbury C., and Roland M., “Defining Comorbidity: Implications for Understanding Health and Health Services,” Annals of Family Medicine 7 (2009): 357–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Menotti A., Mulder I., Nissinen A., Giampaoli S., Feskens E. J., and Kromhout D., “Prevalence of Morbidity and Multimorbidity in Elderly Male Populations and Their Impact on 10‐Year All‐Cause Mortality: The FINE Study (Finland, Italy, Netherlands, Elderly),” Journal of Clinical Epidemiology 54 (2001): 680–686. [DOI] [PubMed] [Google Scholar]
- 3. Vogeli C., Shields A. E., Lee T. A., et al., “Multiple Chronic Conditions: Prevalence, Health Consequences, and Implications for Quality, Care Management, and Costs,” Journal of General Internal Medicine 22, no. Suppl 3 (2007): 391–395. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Singh K., Patel S. A., Biswas S., et al., “Multimorbidity in South Asian Adults: Prevalence, Risk Factors and Mortality,” Journal of Public Health (Oxford, England) 41 (2019): 80–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Yao S.‐S., Cao G.‐Y., Han L., et al., “Prevalence and Patterns of Multimorbidity in a Nationally Representative Sample of Older Chinese: Results From the China Health and Retirement Longitudinal Study,” Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 75 (2020): 1974–1980. [DOI] [PubMed] [Google Scholar]
- 6. Ilankoon I. M. P. S., Samarasinghe K., and Elgán C., “Menopause Is a Natural Stage of Aging: A Qualitative Study,” BMC Women's Health 21 (2021): 47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Collaborative Group on Hormonal Factors in Breast Cancer , “Menarche, Menopause, and Breast Cancer Risk: Individual Participant Meta‐Analysis, Including 118 964 Women With Breast Cancer From 117 Epidemiological Studies,” Lancet Oncology 13 (2012): 1141–1151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Brennand E. A., Scime N. V., Manion R., and Huang B., “Unilateral Oophorectomy and Age at Natural Menopause: A Longitudinal Community‐Based Cohort Study,” BJOG: An International Journal of Obstetrics and Gynaecology 132 (2025): 337–345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zhu D., Chung H.‐F., Pandeya N., et al., “Body Mass Index and Age at Natural Menopause: An International Pooled Analysis of 11 Prospective Studies,” European Journal of Epidemiology 33 (2018): 699–710. [DOI] [PubMed] [Google Scholar]
- 10. Jaspers L., Daan N. M. P., van Dijk G. M., et al., “Health in Middle‐Aged and Elderly Women: A Conceptual Framework for Healthy Menopause,” Maturitas 81 (2015): 93–98. [DOI] [PubMed] [Google Scholar]
- 11. Shin J., Han K., Jung J.‐H., et al., “Age at Menopause and Risk of Heart Failure and Atrial Fibrillation: A Nationwide Cohort Study,” European Heart Journal 43 (2022): 4148–4157. [DOI] [PubMed] [Google Scholar]
- 12. Welten S. J. G. C., Onland‐Moret N. C., Boer J. M. A., Verschuren W. M. M., and van der Schouw Y. T., “Age at Menopause and Risk of Ischemic and Hemorrhagic Stroke,” Stroke 52 (2021): 2583–2591. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Guo C., Li Q., Tian G., et al., “Association of Age at Menopause and Type 2 Diabetes: A Systematic Review and Dose‐Response Meta‐Analysis of Cohort Studies,” Primary Care Diabetes 13 (2019): 301–309. [DOI] [PubMed] [Google Scholar]
- 14. Jiao J., Feng X., Gong A., and Yao Y., “Association Between Reproductive Lifespan and Multimorbidity Among Chinese Postmenopausal Women,” Menopause 31 (2024): 945–951. [DOI] [PubMed] [Google Scholar]
- 15. Blümel J. E., Carrillo‐Larco R. M., Vallejo M. S., and Chedraui P., “Multimorbidity in a Cohort of Middle‐Aged Women: Risk Factors and Disease Clustering,” Maturitas 137 (2020): 45–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Zhu D., Chung H.‐F., Dobson A. J., et al., “Type of Menopause, Age of Menopause and Variations in the Risk of Incident Cardiovascular Disease: Pooled Analysis of Individual Data From 10 International Studies,” Human Reproduction 35 (2020): 1933–1943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Muka T., Asllanaj E., Avazverdi N., et al., “Age at Natural Menopause and Risk of Type 2 Diabetes: A Prospective Cohort Study,” Diabetologia 60 (2017): 1951–1960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Lee G. B., Nam G. E., Kim W., et al., “Association Between Premature Menopause and Cardiovascular Diseases and All‐Cause Mortality in Korean Women,” Journal of the American Heart Association 12 (2023): e030117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Chudasama Y. V., Khunti K. K., Zaccardi F., et al., “Physical Activity, Multimorbidity, and Life Expectancy: A UK Biobank Longitudinal Study,” BMC Medicine 17 (2019): 108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Tyrrell J., Jones S. E., Beaumont R., et al., “Height, Body Mass Index, and Socioeconomic Status: Mendelian Randomisation Study in UK Biobank,” BMJ 352 (2016): i582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Yang L., Feng H., Chen J., et al., “Association of Circadian Rest‐Activity Rhythms With Cardiovascular Disease and Mortality in Type 2 Diabetes,” Diabetes Research and Clinical Practice 197 (2023): 110262. [DOI] [PubMed] [Google Scholar]
- 22. Han Y., Hu Y., Yu C., et al., “Lifestyle, Cardiometabolic Disease, and Multimorbidity in a Prospective Chinese Study,” European Heart Journal 42 (2021): 3374–3384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Singh‐Manoux A., Fayosse A., Sabia S., et al., “Clinical, Socioeconomic, and Behavioural Factors at Age 50 Years and Risk of Cardiometabolic Multimorbidity and Mortality: A Cohort Study,” PLoS Medicine 15 (2018): e1002571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Keown‐Stoneman C. D., Horrocks J., Darlington G. A., Goodday S., Grof P., and Duffy A., “Multi‐State Models for Investigating Possible Stages Leading to Bipolar Disorder,” International Journal of Bipolar Disorders 3 (2015): 5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wu G., Cai M., Wang C., et al., “Ambient Air Pollution and Incidence, Progression to Multimorbidity and Death of Hypertension, Diabetes, and Chronic Kidney Disease: A National Prospective Cohort,” Science of the Total Environment 881 (2023): 163406. [DOI] [PubMed] [Google Scholar]
- 26. Tang H., Huang J., Zhang X., et al., “Association Between Triglyceride Glucose‐Body Mass Index and the Trajectory of Cardio‐Renal‐Metabolic Multimorbidity: Insights From Multi‐State Modelling,” Cardiovascular Diabetology 24 (2025): 133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Manevski D., Putter H., Pohar Perme M., Bonneville E. F., Schetelig J., and de Wreede L. C., “Integrating Relative Survival in Multi‐State Models‐A Non‐Parametric Approach,” Statistical Methods in Medical Research 31 (2022): 997–1012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Muka T., Oliver‐Williams C., Kunutsor S., et al., “Association of Age at Onset of Menopause and Time Since Onset of Menopause With Cardiovascular Outcomes, Intermediate Vascular Traits, and All‐Cause Mortality: A Systematic Review and Meta‐Analysis,” JAMA Cardiology 1 (2016): 767–776. [DOI] [PubMed] [Google Scholar]
- 29. Shen T.‐Y., Strong C., and Yu T., “Age at Menopause and Mortality in Taiwan: A Cohort Analysis,” Maturitas 136 (2020): 42–48. [DOI] [PubMed] [Google Scholar]
- 30. Xu X., Jones M., and Mishra G. D., “Age at Natural Menopause and Development of Chronic Conditions and Multimorbidity: Results From an Australian Prospective Cohort,” Human Reproduction 35 (2020): 203–211. [DOI] [PubMed] [Google Scholar]
- 31. Kirubarajan A., Sohel N., Mayhew A., Griffith L. E., Raina P., and Shea A. K., “The Association Between Primary Ovarian Insufficiency and Increased Multimorbidity in a Large Prospective Cohort (Canadian Longitudinal Study on Aging),” Fertility and Sterility 123 (2025): 289–299. [DOI] [PubMed] [Google Scholar]
- 32. Jani B. D., Hanlon P., Nicholl B. I., et al., “Relationship Between Multimorbidity, Demographic Factors and Mortality: Findings From the UK Biobank Cohort,” BMC Medicine 17 (2019): 74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Armeni E., Paschou S. A., Goulis D. G., and Lambrinoudaki I., “Hormone Therapy Regimens for Managing the Menopause and Premature Ovarian Insufficiency,” Best Practice & Research. Clinical Endocrinology & Metabolism 35 (2021): 101561. [DOI] [PubMed] [Google Scholar]
- 34. Davis S. R. and Baber R. J., “Treating Menopause–MHT and Beyond,” Nature Reviews. Endocrinology 18 (2022): 490–502. [DOI] [PubMed] [Google Scholar]
- 35. Rusu M. E., Bigman G., Ryan A. S., and Popa D.‐S., “Investigating the Effects and Mechanisms of Combined Vitamin D and K Supplementation in Postmenopausal Women: An Up‐To‐Date Comprehensive Review of Clinical Studies,” Nutrients 16 (2024): 2356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Daley A., Macarthur C., Stokes‐Lampard H., McManus R., Wilson S., and Mutrie N., “Exercise Participation, Body Mass Index, and Health‐Related Quality of Life in Women of Menopausal Age,” British Journal of General Practice 57 (2007): 130–135. [PMC free article] [PubMed] [Google Scholar]
- 37. Pant A., Gibson A. A., Marschner S., et al., “Age of Menopause, Healthy Lifestyle and Cardiovascular Disease in Women: A Prospective Cohort Study,” Heart 111 (2025): 262–268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Nelson H. D., “Menopause,” Lancet 371 (2008): 760–770. [DOI] [PubMed] [Google Scholar]
- 39. Davies R., Goyal A., Nash Z., et al., “Factors Associated With Non‐Vasomotor Menopause Symptoms Experienced by 7285 Women: A UK‐Wide National Survey,” BJOG: An International Journal of Obstetrics and Gynaecology (2025). 10.1111/1471-0528.18353. [DOI] [PubMed] [Google Scholar]
- 40. Xu B., Allard C., Alvarez‐Mercado A. I., et al., “Estrogens Promote Misfolded Proinsulin Degradation to Protect Insulin Production and Delay Diabetes,” Cell Reports 24 (2018): 181–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Hevener A. L., Zhou Z., Moore T. M., Drew B. G., and Ribas V., “The Impact of ERα Action on Muscle Metabolism and Insulin Sensitivity ‐ Strong Enough for a Man, Made for a Woman,” Molecular Metabolism 15 (2018): 20–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Kublickiene K., Fu X.‐D., Svedas E., Landgren B.‐M., Genazzani A. R., and Simoncini T., “Effects in Postmenopausal Women of Estradiol and Medroxyprogesterone Alone and Combined on Resistance Artery Function and Endothelial Morphology and Movement,” Journal of Clinical Endocrinology and Metabolism 93 (2008): 1874–1883. [DOI] [PubMed] [Google Scholar]
- 43. van der Plaat D. A., Pereira M., Pesce G., et al., “Age at Menopause and Lung Function: A Mendelian Randomisation Study,” European Respiratory Journal 54 (2019): 1802421. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: List of the 35 chronic conditions included within the definition of multimorbidity.
Figure S1: Directed acyclic graph of the association between exposure, covariates, and outcomes.
Figure S2: Restricted cubic spline (RCS) analysis of the association between age at menopause and multimorbidity.
Figure S3: Restricted cubic spline (RCS) analysis of the association between age at menopause and all‐cause mortality.
Figure S4: Cumulative event rates of (A) multimorbidity and (B) all‐cause mortality associated with age at menopause.
Figure S5: Kaplan–Meier survival curves stratified by menopausal age groups over follow‐up time.
Table S1: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases excluding events in the first 2 years of follow‐up.
Table S2: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases for participants who enter different states on the same day, different time intervals (0.5, 1, 3, 5 days) were used to calculate the entry date of the previous state.
Table S3: Sensitivity analysis of associations of age at menopause with transitions of 35 chronic diseases excluding participants with incomplete covariates.
Data Availability Statement
All data are available in the UK Biobank data (https://www.ukbiobank.ac.uk/). UK Biobank data are publicly accessible for research use by application.