Abstract
This study aimed to estimate the age, period, and cohort trends of multimorbidity among Chinese older adults from 2002 to 2022, and to explore how these three trends were affected by gender and whether or not living alone. Data were extracted from the China Longitudinal Healthy Living Survey (CLHLS) (2002–2022), and a total of 52,876 valid samples aged 65 to 105 were included. We measured 15 types of chronic diseases, and participants having two or more chronic diseases were considered to have multimorbidity. Hierarchical age–period–cohort cross-classified random-effects model (HAPC-CCREM) was applied to examine the age, period, and cohort dynamics of multimorbidity. The average age of participants is 85.91 ± 11.09 years. The temporal effect of age on multimorbidity is inverted U-shaped, with a higher probability in women aged 65–95 years and in men aged 95–105 years. The cohort trends of multimorbidity showed a rise, then a fall, then a rise again, and the period trends declined in fluctuations. Gender differences existed in age and cohort trends of multimorbidity. Moreover, the interaction of gender and living alone was significantly associated with the age trends of multimorbidity. Women living alone have a lower risk of multimorbidity than women living with others among all age groups. This study revealed the separate effects of age, period, and cohort on multimorbidity. The peaks of multimorbidity probability for different genders occur at different ages, and living alone had a protective effect on females, which both provide a scientific basis for the allocation of healthcare resources. The dynamic shifts of multimorbidity may help to forecast future multimorbidity trends and inform policies on population aging.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s11357-025-02001-0.
Keywords: Multimorbidity, Age–period–cohort, Chinese older adults, Gender difference, Living alone
Introduction
Population aging has emerged as a pivotal global trend. In 2020, over 18% of the Chinese population was 60 years or older, with estimations of 38.8% by 2050 [1]. Multimorbidity, defined as the coexistence of two or more chronic conditions [2], poses a substantial threat to elderly health. The coexistence of multiple chronic diseases not only impairs health, such as disability and reduced quality of life [3], but also places a huge burden on the healthcare system, leading to increased probabilities of hospitalization, prolonged lengths of stay, and readmission rates, along with elevated healthcare expenditures [4]. The global prevalence of multimorbidity was 37.2% for adults and 51% for older adults aged 60 and above [5]. There is considerable heterogeneity in the prevalence of multimorbidity, attributable to variations in participant age and the types and numbers of diseases considered in the assessments [6]. The prevalence of multimorbidity among older adults aged 60 or above was estimated at 40.6% in Malaysia [7], 53.1% in Brazil [8], and 46.2% in China [9]. Given the grave implications of multimorbidity and the pronounced heterogeneity in its prevalence, further investigation into the trends of multimorbidity among Chinese older adults is both necessary and timely.
Prior research has indicated that the prevalence of multimorbidity changes across time and age. Increasing temporal trends in multimorbidity have been documented in the USA (1998–2018) [10] and Canada (1978–2014) [11]. Kingston projected that the proportion of individuals with four or more chronic diseases is anticipated to reach 17% by 2035, escalating from 9.8% in 2018 [4]. Concerning age-specific trends, the prevalence of multimorbidity in England stands at 29.9% for the 60–69-year age group, 46.9% for the 70–79-year age group, and 68.2% for those aged 80 or above [12]. Similar age-related increases in multimorbidity have been observed in Chinese [9] and Japanese populations [13]. Nevertheless, in reality, age and period trends are interwoven and may be confounded by cohort compositional differences, potentially leading to spurious findings. To solve this problem, the age–period–cohort (APC) model can be used to disaggregate the temporal effect into distinct age, period, and cohort components [14]. Age effects reflect the biological process associated with age-driven changes in intrinsic bodily functions. Period effects are the results of economic, sociocultural, and technological transitions during certain periods. Cohort effects represent major historical events or social changes experienced by individuals born in the same cohort [15]. Although some research has examined the tendencies of multimorbidity, only a few studies explored the APC trends of multimorbidity [16].
Furthermore, substantial evidence supported that multimorbidity was associated with sociodemographic factors. The influence of gender difference on multimorbidity is consistent, with females exhibiting a higher prevalence than males across most age groups [17] and regions [18]. The status of living alone is also a significant factor in multimorbidity, often overlooked. Different living statuses influence the daily management of chronic diseases among older adults, which in turn affects the accumulation of multiple chronic conditions. Ingram [19] found that single-person households may contribute to a higher risk of multimorbidity. However, most of the previous research has not adopted the APC model to elucidate the differential effects of gender and living alone on age, period, and cohort trends in multimorbidity, and very few have explored the interactive effects of gender and living alone.
Little is known about the dynamics of multimorbidity prevalence across the life course, over time, and within birth cohorts, let alone when stratified by gender and whether or not living alone. Describing trends in multimorbidity is crucial for assessing the health status of older adults as well as for predicting the demand for long-term care services and the utilization of healthcare resources. In this study, we used the largest dataset of oldest-old cohorts in the world, the Chinese Longitudinal Healthy Longevity Survey (CLHLS), which is a national survey covering 80% of the Chinese elderly population. The objectives of this research were to (1) summarize the age, period, and cohort trends of multimorbidity of Chinese older adults from 2002 to 2022; (2) examine gender differences in the APC trends of multimorbidity; and (3) explore the effects of living alone and their interactions with genders on these multimorbidity trends.
Methods
Data source and study samples
Data of this study were retrieved from the China Longitudinal Healthy Living Survey (CLHLS), an ongoing national survey conducted by Peking University and Duke University together [20]. CLHLS adopted a multi-stage disproportionate and a targeted random-sample design to make samples more representative. There were nine face-to-face interviews performed by well-trained field workers of CLHLS in 1998, 2000, 2002, 2005, 2008–2009, 2011–2012, 2014, 2017–2018, and 2021–2022, respectively. Before each survey, voluntary agreements and informed consent had been obtained from all participants.
Since the first two waves of CLHLS (1998 and 2000) only included people over the age of 80, seven waves of CLHLS (2002, 2005, 2008–2009, 2011–2012, 2014, 2017–2018, and 2021–2022) were incorporated into our study. After the exclusion of samples that were younger than 65 or older than 105 (2664 observations) and had missing information on multimorbidity status (1936 observations), we obtained a total of 52,876 samples. The detailed process of sample selection and the number of excluded cases is shown in the Supplementary Fig. S1.
Measurement
Multimorbidity
Fifteen types of chronic diseases were measured through self-reported questionnaires in all seven waves, by asking “Are you suffering from …”. The 15 types of chronic diseases include hypertension, diabetes, heart disease, stroke or CVD, common respiratory disorders (bronchitis, emphysema, pneumonia, or asthma), tuberculosis, cataracts, glaucoma, cancer, prostate disease, gastric or duodenal ulcer, Parkinson’s disease, bedsore, arthritis, and dementia. Elderly people with two or more chronic diseases at the same time were considered to have multimorbidity.
Age, period, and cohort
As the age of participants ranged from 65 to 105 years, age was divided by 10, then minus 6.5 to facilitate model convergence when we add the quadratic term of age into model[21]. The period was regarded as the year in which the survey was conducted, including 2002, 2005, 2008–2009, 2011–2012, 2014, 2007–2018, and 2021–2022. The cohort was the year participants were born. We classified birth cohorts into 5-year groups and clustered the elderly born before 1905 or after 1946 separately to ensure sufficient numbers of participants in each group [22].
Sociodemographic variables
Sociodemographic characteristics included sex (male/female), whether or not living alone (yes/no), residence (urban/rural), nationality (Han/else), and marital status (married/widowed/else).
Statistical analysis
Data analyses were performed using SAS version 9.4 and R program 4.2.1. Descriptive analysis was used to describe the basic characteristics of participants in all waves using frequency (percentage). Then, we applied hierarchical age–period–cohort cross-classified random-effects models (HAPC-CCREM) to estimate the age, period, and cohort trends of multimorbidity for Chinese older adults. Traditional APC models have not truly resolved the “identification problem” caused by the perfect collinearity among age, period, and cohort[23]. HAPC-CCREM, proposed by Yang and Land [24], assumes that age is an individual-level attribute, while period and cohort can be treated as group-level variables, as people belonging to the same period or birth cohort experience similar social events and life circumstances that exert comparable influences at the group level [25]. Consequently, HAPC-CCREM specifies a fixed age effect at level 1 and random period and cohort effects at level 2 by adopting a mixed model framework, which breaks the linear dependency among the three dimensions and successfully solves the identification problem [21]. Moreover, HAPC-CCREM is well suited to repeated cross-sectional survey data and offers considerable flexibility in handling practical issues such as unbalanced samples and unequal cell sizes across periods and cohorts [26]. The mathematical expressions of the model have been explained in previous research [27].
In this study, we grouped the cohort every 5 years, transformed it into category variables, and added the quadratic term of age into the model. The prevalence of multimorbidity was regressed on age in linear and squared terms and other covariates as needed. The coefficients of period, cohort, gender, and whether or not living alone were allowed to have random effects. We conducted HAPC-CCREM for all participants and then did stratified analysis by gender. Model 1 tested the net APC effects on multimorbidity with a fixed effect for age and random effects for period and cohort. In model 2, we added random effects of gender to explore how gender disparities in multimorbidity varied by period and cohort. Models 3 and 4 included the interactions between age with gender and living alone to examine whether gender or living status difference exists in age trends of multimorbidity. Furthermore, we tested the association between multimorbidity and the interaction of age, gender, and living alone or not in model 5. In model 6, we included the fixed effects of other covariates based on model 5. Stratified HAPC-CCREM was conducted for females and males, respectively. We explored the period and cohort effects of whether or not living alone in models 7 and 10, the age effects of whether or not living alone in models 8 and 11, and conducted full models in models 9 and 12. Finally, based on the age of each subject at their last follow-up, we used the age corresponding to the peak of sex-specific age trends of multimorbidity as the cutoff to divide both males and females into two groups. HAPC-CCREM was then performed separately for these subgroups as a sensitivity analysis to examine the impact of survival bias on multimorbidity age trends. In our figures, the predicted probabilities of multimorbidity were calculated by computing probability = 1/(1 + exp(− β)) [28].
Results
Basic characteristics of samples
After excluding 1971 samples younger than 65 or older than 105 or without sufficient multimorbidity information, there were 52,876 samples in all survey waves, which contributed to 90,946 observations in total. The average age of participants is 85.91 ± 11.09 years. Most participants were females, had Han nationality, and of widowed marital status across all waves. The raw prevalence of multimorbidity increased over the year, ranging from 23.6 to 41.1% (Table 1).
Table 1.
Basic characteristics of participants from 2002 to 2022
| 2002 | 2005 | 2008–2009 | 2011–2012 | 2014 | 2017–2018 | 2021–2022 | |
|---|---|---|---|---|---|---|---|
| (N = 15,561) | (N = 15,051) | (N = 16,132) | (N = 9222) | (N = 6533) | (N = 14,805) | (N = 13,642) | |
| n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | |
| Sex | |||||||
| Male | 6694 (43.0) | 6505 (43.2) | 6914 (42.9) | 4232 (45.9) | 3043 (46.6) | 6525 (44.1) | 6139 (45.0) |
| Female | 8867 (57.0) | 8546 (56.8) | 9218 (57.1) | 4990 (54.1) | 3490 (53.4) | 8280 (55.9) | 7503 (55.0) |
| Age groups | |||||||
| 65–74 | 3250 (20.9) | 3264 (21.7) | 2893 (17.9) | 1907 (20.7) | 1121 (17.2) | 3176 (21.5) | 2352 (17.2) |
| 75–89 | 5725 (36.8) | 5447 (36.2) | 5642 (35.0) | 3805 (41.3) | 3249 (49.7) | 5819 (39.3) | 5526 (40.5) |
| 90–105 | 6586 (42.3) | 6340 (42.1) | 7597 (47.1) | 3510 (38.1) | 2163 (33.1) | 5810 (39.2) | 5764 (42.3) |
| Living alone | |||||||
| Yes | 2090 (13.4) | 2016 (13.4) | 2506 (15.5) | 1568 (17.0) | 1243 (19.0) | 2340 (15.8) | 2451 (18.0) |
| No | 13,471 (86.6) | 13,035 (86.6) | 13,626 (84.5) | 7654 (83.0) | 5290 (81.0) | 12,465 (84.2) | 11,191 (82.0) |
| Residence | |||||||
| Urban | 7196 (46.2) | 6763 (44.9) | 6424 (39.8) | 4381 (47.5) | 2947 (45.1) | 8290 (56.0) | 9754 (71.5) |
| Rural | 8365 (53.8) | 8288 (55.1) | 9708 (60.2) | 4841 (52.5) | 3586 (54.9) | 6515 (44.0) | 3888 (28.5) |
| Nationality | |||||||
| Han | 14,718 (94.6) | 14,167 (94.1) | 15,176 (94.1) | 7953 (86.2) | 5582 (85.4) | 12,118 (81.9) | 12,450 (91.3) |
| Else | 843 (5.4) | 884 (5.9) | 956 (5.9) | 1269 (13.8) | 951 (14.6) | 2687 (18.1) | 1192 (8.7) |
| Marital status | |||||||
| Married | 4664 (30.0) | 4684 (31.1) | 4888 (30.3) | 3411 (37.0) | 2560 (39.2) | 5872 (39.7) | 5139 (37.7) |
| Widowed | 10,327 (66.4) | 9896 (65.7) | 10,780 (66.8) | 5459 (59.2) | 3719 (56.9) | 8359 (56.5) | 8080 (59.2) |
| Else | 570 (3.7) | 471 (3.1) | 464 (2.9) | 352 (3.8) | 254 (3.9) | 574 (3.9) | 423 (3.1) |
| Multimorbidity | |||||||
| Yes | 3675 (23.6) | 3904 (25.9) | 4080 (25.3) | 2851 (30.9) | 2037 (31.2) | 5628 (38.0) | 5612 (41.1) |
| No | 11,886 (76.4) | 11,147 (74.1) | 12,052 (74.7) | 6371 (69.1) | 4496 (68.8) | 9177 (62.0) | 8030 (58.9) |
APC trends of multimorbidity
Shown as Table 2, in model 1, the effects of age, period, and cohort effect were all significant on multimorbidity, and age trends were curvilinear (age: OR = 0.115, P < 0.001; age2: OR = 1.001, P < 0.001). Shown as in Fig. 1, the predictive probability of multimorbidity rose and then fell with age, peaking at 80–85 years old. The period trend of multimorbidity presented an overall decreasing pattern. The cohort trend rose (born < 1915), then down (born 1915 ~ 1940), then up again (born > 1940).
Table 2.
HAPC-CCREM estimates of multimorbidity for the Chinese elderly
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
|---|---|---|---|---|---|---|
| Fixed effecta | ||||||
| Intercept |
0.010*** (0.330) |
0.010*** (0.338) |
0.011*** (0.333) |
0.009*** (0.332) |
0.010*** (0.335) |
0.016*** (0.335) |
| Age |
0.115*** (0.133) |
0.122*** (0.135) |
0.134*** (0.136) |
0.108*** (0.134) |
0.126*** (0.136) |
0.117*** (0.137) |
| Age2 |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
| Gender (male = 1) |
1.342*** (0.031) |
1.350*** (0.032) |
1.339*** (0.032) |
|||
| Living alone (yes = 1) |
1.012 (0.047) |
1.037 (0.047) |
0.992 (0.050) |
|||
| Age*gender |
0.891*** (0.041) |
0.883*** (0.014) |
0.896*** (0.014) |
|||
| Age*living alone |
1.059** (0.021) |
1.030 (0.022) |
1.035 (0.023) |
|||
| Age*gender*living alone | p |
1.058** (0.019) |
1.046* (0.019) |
|||
| Residence (urban = 1) |
0.648*** (0.015) |
|||||
| Ethics (Han = 1) |
0.659*** (0.027) |
|||||
| Marital status (reference = married) | ||||||
| Widowed |
1.071*** (0.021) |
|||||
| Else |
1.058 (0.043) |
|||||
| Random effectb | ||||||
| Variance components | ||||||
| Period effect | ||||||
| Intercept |
0.084** (0.050) |
0.092* (0.055) |
0.083* (0.049) |
0.085* (0.050) |
0.084* (0.049) |
0.069* (0.041) |
| Gender effect |
0.000 (/) |
|||||
| Living alone effect | ||||||
| Cohort effect | ||||||
| Intercept |
0.012** (0.007) |
0.018* (0.011) |
0.013* (0.007) |
0.012* (0.006) |
0.013* (0.007) |
0.013* (0.008) |
| Gender effect |
0.024* (0.012) |
|||||
| Living alone effect | ||||||
| Model fit | ||||||
| − 2 LOG Likelihood | 404,440 | 404,624 | 404,588 | 404,554 | 404,728 | 406,184 |
*P < 0.05; **P < 0.001; ***P < 0.0001 (two-tailed tests). The corresponding standard errors (SE) were presented within parentheses
aFixed effects are presented in the form of an odds ratio (OR)
bRandom effects are presented in the form of coefficients
Fig. 1.
Age (A), period (B), and cohort (C) effects on multimorbidity by gender. Note. To enhance the aesthetic of the figures’ layout, the years of the CLHLS waves were abbreviated as follows: 2008–2009 as 2008, 2011–2012 as 2011, 2017–2018 as 2018, and 2021–2022 as 2021
Model 2 and model 3 displayed that gender differences in multimorbidity varied significantly with age (OR = 0.891, P < 0.001) and cohort (coefficient = 0.240, P < 0.05). The probability of multimorbidity was higher for women than for men in the 65–95-year age group, and for men than for women in the 95–105-year age group (Fig. 1A). Regarding cohort trends (Fig. 1C), females born in 1915–1930 had a higher multimorbidity risk than males born in the same year. While the situation was the converse for people born after 1940. Both genders had similar decreasing trends of multimorbidity with period (Fig. 1B, Supplementary Table S1).
In Table 2, model 4 and model 5 indicated that the association between age and multimorbidity is significantly influenced by whether or not living alone (OR = 1.059, P < 0.01), and the interactive effect of gender and whether or not living alone had a significant impact on the age trends of multimorbidity (OR = 1.058, P < 0.01). Model 6 examined the effect of sociodemographic factors on multimorbidity. We found that urban residents (OR = 0.648, P < 0.001) and people with Han nationality (OR = 0.659, P < 0.001) have a lower probability of multimorbidity. Widowed people have a higher probability of multimorbidity compared with people who are married (OR = 1.071, P < 0.001).
In the sensitivity analyses, HAPC-CCREMs were conducted for four groups (men aged 65–85, men aged 85–105, women aged 65–80, and women aged 80–105). As shown in Supplementary Table S2, the effect of age on multimorbidity was statistically significant in every subgroup. The predictive probability of multimorbidity increased with age among men aged 65–85 and women aged 65–80, whereas it declined with age among men aged 85–105 and women aged 80–105. At the intersection point (men aged 85 and women aged 80), the multimorbidity probability in the oldest-old groups was lower than that in the non–oldest-old groups for both genders (Supplementary Fig. S2 and Table S3).
The interactive effect of gender and living alone
Table 3 presented the results of a stratified age–period–cohort analysis according to gender. In model 7 and model 10, we found that the effects of age and period effects were significant on multimorbidity for both genders. Only females had a significant cohort effect on multimorbidity. The period and cohort trends of multimorbidity had no significant associations with whether or not living alone. Illustrated as model 8 and model 11, the females’ age trend of multimorbidity was influenced significantly by whether or not they lived alone (OR = 1.062, P < 0.05). However, the association was not significant for males. Model 9 showed that widowed males have a higher probability of multimorbidity compared with males who are married (OR = 1.103, P < 0.001).
Table 3.
HAPC-CCREM estimates of multimorbidity stratified by gender
| Males | Females | |||||
|---|---|---|---|---|---|---|
| Model 7 | Model 8 | Model 9 | Model 10 | Model 11 | Model 12 | |
| Fixed effecta | ||||||
| Intercept |
0.023*** (0.492) |
0.022*** (0.493) |
0.039*** (0.496) |
0.007*** (0.425) |
0.007*** (0.425) |
0.011*** (0.428) |
| Age |
0.148*** (0.203) |
0.146*** (0.203) |
0.139*** (0.205) |
0.109*** (0.177) |
0.106*** (0.177) |
0.103*** (0.180) |
| Age2 |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
1.001*** (< 0.001) |
| Living alone (yes = 1) |
1.178* (0.077) |
1.061 (0.082) |
0.963 (0.060) |
0.961 (0.064) |
||
| Age*gender | ||||||
| Age*living alone |
1.033 (0.035) |
1.047 (0.036) |
1.062* (0.026) |
1.052 (0.027) |
||
| Residence (urban = 1) |
0.630*** (0.023) |
0.664*** (0.020) |
||||
| Ethics (Han = 1) |
0.619*** (0.040) |
0.697*** (0.036) |
||||
| Marital status (reference = married) | ||||||
| Widowed |
1.103*** (0.028) |
1.035 (0.030) |
||||
| Else |
1.070 (0.054) |
1.020 (0.073) |
||||
| Random effectb | ||||||
| Variance components | ||||||
| Period effect | ||||||
| Intercept |
0.084* (0.049) |
0.081* (0.048) |
0.068* (0.040) |
0.092* (0.055) |
0.092* (0.055) |
0.076* (0.046) |
| Living alone effect |
0.041 (0.038) |
0.009 (0.008) |
||||
| Cohort effect | ||||||
| Intercept |
0.008 (0.005) |
0.009 (0.005) |
0.008 (0.005) |
0.018* (0.010) |
0.017* (0.010) |
0.018* (0.011) |
| Living alone effect |
0.005 (0.014) |
0.000 (/) |
||||
| Model fit | ||||||
| − 2*LOG(Likelihood) | 177,689 | 177,691 | 178,497 | 226,995 | 227,032 | 227,708 |
The corresponding standard errors (SE) are presented within parentheses
*P < 0.05; **P < 0.001; ***P < 0.0001 (two-tailed tests)
aFixed effects are presented in the form of an odds ratio (OR)
bRandom effects are presented in the form of coefficients
Figure 2 and Supplementary Table S4 illustrate the interactive effect of gender and whether or not living alone on multimorbidity trends. We found that multimorbidity probability peaked at age 85 years for men and 80 years for women. Women living alone have a significantly lower risk of multimorbidity than women living with others. Living alone had a negative influence on males, although it was not significant (Fig. 2A, B). The probability of multimorbidity declined with the period for both sexes, regardless of whether they lived alone or not (Fig. 2C, D). Although the effect of whether or not living alone on cohort trends was not significant, it had an opposite impact on the cohort trends of multimorbidity probability by gender (Fig. 2E, F).
Fig. 2.
Age (A, B), period (C, D), and cohort (E, F) effects on multimorbidity by gender and whether or not living alone. Note. To enhance the aesthetic of the figures’ layout, the years of the CLHLS waves were abbreviated as follows: 2008–2009 as 2008, 2011–2012 as 2011, 2017–2018 as 2018, and 2021–2022 as 2021
Discussion
We explored the age–period–cohort trends of multimorbidity and their disparities by gender and living alone. Our findings indicated that the temporal effect of age on multimorbidity follows an inverted U-shaped distribution, with a higher probability in women aged 65–95 years and in men aged 95–105 years. In addition, the cohort trends in multimorbidity showed an initial increase followed by a decline and subsequent resurgence, and the period trends declined in fluctuations. Moreover, living alone had a significant effect on the age trend of multimorbidity, playing a protective role for females.
First, the age trends of multimorbidity were inverted U-shaped, reaching a zenith at 80–85 years old. Similar to previous findings [6], the probability of multimorbidity increased with age among 65–85-year-olds. The difference, however, is that the present study incorporated a substantial cohort of long-lived elderly individuals [29], thereby enabling a more nuanced understanding of multimorbidity among older adults. The probability of multimorbidity declined with age beyond 80 years old in women and 85 years old in men. The results of sensitivity analyses showed similar patterns, and at the intersection points (age 85 for men and 80 for women), the probability of multimorbidity was lower in the oldest-old group than in the non–oldest-old group. This phenomenon may be attributed to the role of longevity genes, suggesting that individuals capable of reaching advanced ages possess superior fitness to mitigate the risks of chronic disease co-occurrence [30]. At the same time, longer-lived people may adopt relatively healthier lifestyles, such as less smoking and drinking and regular exercise [31], to help them reduce the threat of multimorbidity [32].
In addition, gender disparities are evident in the age trends of multimorbidity. In the 65–95-year age group, women exhibited a higher likelihood of multimorbidity compared to men; however, this trend reverses beyond the age of 95 years. Females generally face a higher risk of multimorbidity across most age groups, aligning with previous studies [33]. They usually neglect their health in taking care of their family members and are more likely to be unemployed [34], widowed, and less engaged in physical activity. Conversely, after the age of 95 years, women have a lower probability of multimorbidity than men, potentially due to a higher prevalence of longevity genes that enable them to reach advanced ages with better health.
Moreover, we revealed significant interactions between age, gender, and living status on multimorbidity risk. Across all age groups, older women living alone demonstrated a significantly lower risk of multimorbidity compared to those cohabiting with others. As previously highlighted, older Chinese women often play the role of caregivers in the family, managing household chores and providing primary care for their grandchildren [35]. The burdensome family responsibilities may lead them to overlook their own health [36]. Besides, women who live with their children or spouses are likely to accommodate their families in their lifestyles. Even if they have the will to adopt a healthy lifestyle, they are unable to do so due to objective constraints [37], which may lead to the co-occurrence of multiple chronic diseases. Consequently, living alone may serve as a protective factor for older Chinese women, reducing their risk of multimorbidity. In contrast, we observed a higher likelihood of multimorbidity among older men living alone, although the difference was not statistically significant.
We also found significant cohort trends in multimorbidity, rising (born < 1915), then down (born 1915 ~ 1940), then up again (born > 1940). The variation in multimorbidity across cohorts is supposed to be associated with individual experiences and socio-economic circumstances in their early life [38]. The pre-1915 cohort experienced adverse events, including wars and famines, during their formative years, which indirectly contributed to an increased risk of multimorbidity [39]. Following the social stability and rapid economic development post–Chinese liberation in 1949, the probability of multimorbidity of the cohort of 1915–1940 gradually decreased [40]. With the progression of China’s reform and opening up policies, living standards improved, and the prevalence of chronic diseases, such as cardiovascular diseases, increased dramatically [41], which contributed to the rise of multimorbidity for the post-1940 cohort. Consistent with our findings, Bishop’s study on the multimorbidity burden among American older adults revealed that later-born US cohorts (1960–1965) faced a greater multimorbidity threat than those born in earlier periods [42]. Increased wealth inequality and the boost of obesity-related chronic diseases may contribute to the increased probability of multimorbidity in the later cohort [43].
Some limitations of this study warrant acknowledgment. First, the ascertainment of chronic diseases relied on self-reported questionnaires, which may introduce reporting biases affecting the precision of multimorbidity estimation. Second, the exclusion of samples lacking data on age or multimorbidity potentially compromised the representativeness of the samples. Third, only the 15 types of chronic conditions that were measured in every wave of the CLHLS were included in this study, while chronic conditions added after the 2008–2009 wave and those never covered by the CLHLS were disregarded, which may lead to an underestimation of the multimorbidity prevalence. Fourth, despite the incorporation of various factors in our age–period–cohort (APC) analyses, there may exist unmeasured confounding factors that were not taken into account. Fifth, although we performed sensitivity analyses by dividing participants into the oldest-old and the non–oldest-old groups based on their age at the last follow-up, and the results showed a certain degree of robustness, the influence of survival bias on the age trends cannot be fully ruled out since death might not be the only reason for loss of follow-up. Therefore, the multimorbidity age trends of this study should be interpreted with caution. Lastly, due to data constraints, the earliest and latest birth cohorts did not capture the complete age distribution, which may have introduced bias into the cohort trends of multimorbidity.
Notably, the present study has some significant strengths. Primarily, using the largest dataset of oldest-old cohorts in the world, we are the first to analyze age, period, and cohort trends of multimorbidity in Chinese older adults over 20 years. Additionally, we found that the probability of multimorbidity peaked at different ages for the elderly of different genders, with females peaking slightly earlier than males (aged 80 and 85 years, respectively). When implementing multimorbidity prevention programs at the primary care or community level, people at high risk of multimorbidity could be identified based on age, gender, and medical history, and subsequently monitored with targeted vigilance to mitigate the incidence of multimorbidity. Besides, older women living with others demonstrated a higher probability of multimorbidity compared to those living alone. Therefore, alongside living alone elderly, we should also pay attention to the chronic conditions of older people who live with their family members, particularly older women who bear the burden of household chores.
Conclusion
In conclusion, we found that the age trends of multimorbidity followed an inverted U-shaped pattern, the period trends declined in fluctuations, and the cohort trends experienced a rise followed by a fall and a subsequent resurgence. Gender differences significantly influence age and cohort trends of multimorbidity. Solitary living emerges as a protective factor for older females against multimorbidity. The dynamic shifts in multimorbidity help to forecast future multimorbidity trends, thereby guiding health resource allocation and informing policies on population aging.
Supplementary Information
Below is the link to the electronic supplementary material.
(DOCX 232 KB)
Acknowledgements
We would like to express our particular thanks to the working group of the Chinese Longitudinal Healthy Longevity Survey (CLHLS) who made this study possible.
Author contribution
XXS: conceptualization, funding acquisition, and resources; YYB: methodology and writing—original draft; WP: data curation and validation; YBW and HWM: writing—review and editing; ZH: data curation and software; YZ and XXZ: project administration and funding acquisition. All authors had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors approved the final version of the manuscript.
Funding
This work was funded by the National Natural Science Foundation of China (72474007).
Data availability
The CLHLS dataset analyzed during the current study is available data that are freely available in the ICPSR repository, https://www.icpsr.umich.edu/icpsrweb/NACDA/series/487.
Declarations
Ethics approval
The study was approved by the Biomedical Ethics Committee of Peking University, China (IRB00001052-13074).
Conflict of interest
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ni J. The aging population in China. Highlights in Business, Economics and Management. 2024;26:151–8. [Google Scholar]
- 2.Hernández B, Reilly RB, Kenny RA. Investigation of multimorbidity and prevalent disease combinations in older Irish adults using network analysis and association rules. Sci Rep. 2019;9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Zhang Y, et al. Associations among multimorbid conditions in hospitalized middle-aged and older adults in China: statistical analysis of medical records. JMIR Public Health Surveill. 2022;8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kingston A, et al. Projections of multi-morbidity in the older population in England to 2035: estimates from the population ageing and care simulation (PACSim) model. Age Ageing. 2018;47:374–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chowdhury SR, Das DC, Sunna TC, Beyene J, Hossain A. Global and regional prevalence of multimorbidity in the adult population in community settings: a systematic review and meta-analysis. E Clin Med. 2023. 10.1016/j.eclinm.2023.101860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ho IS-S, et al. Variation in the estimated prevalence of multimorbidity: systematic review and meta-analysis of 193 international studies. BMJ Open. 2022;12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ghazali SS, et al. Prevalence and factors associated with multimorbidity among older adults in Malaysia: a population-based cross-sectional study. BMJ Open. 2021;11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Melo LAd, Lima KCd. Prevalence and factors associated with multimorbidities in Brazilian older adults. Cienc Saude Coletiva. 2020;25:3869–77. [DOI] [PubMed] [Google Scholar]
- 9.Hu Y. et al. Prevalence and patterns of multimorbidity in China during 2002–2022: a systematic review and meta-analysis. Ageing Research Reviews, 2023;102165. [DOI] [PubMed]
- 10.Caraballo C. et al. Temporal trends in racial and ethnic disparities in multimorbidity prevalence in the United States, 1999–2018. The American Journal of Medicine 2022;135, 1083–1092. e1014. [DOI] [PubMed]
- 11.Chireh B, D’Arcy C. Contrasting trends in prevalence of chronic diseases and multimorbidity, Canada 1978–2014. SN Compr Clin Med. 2020;2:1563–72. [Google Scholar]
- 12.Valabhji J, et al. Prevalence of multiple long-term conditions (multimorbidity) in England: a whole population study of over 60 million people. J R Soc Med. 2024;117:104–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Saito Y, Igarashi A, Nakayama T, Fukuma S. Prevalence of multimorbidity and its associations with hospitalisation or death in Japan 2014–2019: a retrospective cohort study using nationwide medical claims data in the middle-aged generation. BMJ Open. 2023;13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Pan C, Cao N, Kelifa MO, Luo S. Age and cohort trends in disability among Chinese older adults. Front Public Health. 2023;11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Li T. Which generation is happier? An age-period-cohort analysis on the subjective well-being of Chinese residents. Population & Economics. 2018;1:90–102. [Google Scholar]
- 16.Lai FT, et al. Sex-specific intergenerational trends in morbidity burden and multimorbidity status in Hong Kong community: an age-period-cohort analysis of repeated population surveys. BMJ Open. 2019;9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wang X, et al. Multimorbidity among two million adults in China. Int J Environ Res Public Health. 2020;17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sharma P, Maurya P. Gender differences in the prevalence and pattern of disease combination of chronic multimorbidity among Indian elderly. Ageing Int. 2022;47:265–83. [Google Scholar]
- 19.Ingram E, et al. Household and area-level social determinants of multimorbidity: a systematic review. J Epidemiol Community Health. 2021;75:232–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zeng Y. Chinese longitudinal healthy longevity survey and some research findings. Geriatr Gerontol Int. 2004;4:S49–52. [Google Scholar]
- 21.Yang Y. & Land KC. Age-period-cohort analysis: new models, methods, and empirical applications. (Taylor & Francis, 2013).
- 22.Yu R, et al. Trends in activities of daily living disability in a large sample of community-dwelling Chinese older adults in Hong Kong: an age-period-cohort analysis. BMJ Open. 2016;6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Heckman J, Robb R, Mason W, Fienberg S. Cohort analysis in social research: beyond the identification problem. New York: Springer Verlag; 1985. [Google Scholar]
- 24.Yang Y, Land KC. A mixed models approach to the age-period-cohort analysis of repeated cross-section surveys, with an application to data on trends in verbal test scores. Sociol Methodol. 2006;36:75–97. [Google Scholar]
- 25.Yang Y, Land KC. Age–period–cohort analysis of repeated cross-section surveys: fixed or random effects? Sociol Methods Res. 2008;36:297–326. [Google Scholar]
- 26.Yang, Y. in Handbook of aging and the social sciences 17–30 (Elsevier, 2011).
- 27.Zhang S, et al. Marriage matching patterns associated with depressive symptoms among Chinese adults: a nationally representative age-period-cohort modeling study. J Affect Disord. 2024;351:341–8. [DOI] [PubMed] [Google Scholar]
- 28.Raudenbush SW. & Bryk AS. Hierarchical linear models: applications and data analysis methods. Vol. 1 (sage, 2002).
- 29.Zhang H, Wang Y, Wu D, Chen J. Evolutionary path of factors influencing life satisfaction among Chinese elderly: a perspective of data visualization. Data. 2018;3. [Google Scholar]
- 30.Longo VD, Anderson RM. Nutrition, longevity and disease: from molecular mechanisms to interventions. Cell. 2022;185:1455–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wang J, et al. Healthy lifestyle in late-life, longevity genes, and life expectancy among older adults: a 20-year, population-based, prospective cohort study. Lancet Healthy Longev. 2023;4:e535–43. [DOI] [PubMed] [Google Scholar]
- 32.Zhang, Y.-B. et al. Associations of healthy lifestyle and socioeconomic status with mortality and incident cardiovascular disease: two prospective cohort studies. Bmj 2021;373. [DOI] [PMC free article] [PubMed]
- 33.de Bezerra Souza DL, et al. Multimorbidity and its associated factors among adults aged 50 and over: a cross-sectional study in 17 European countries. PLoS One. 2021;16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ortiz-Ospina E, Tzvetkova S. & Roser M. Women's employment. Our world in data (2024).
- 35.Liu Y, Hughes MC, Roberto KA, Savla J. Physical and mental health of family caregivers of older parents and grandchildren in China. Aging and Health Research. 2022;2. [Google Scholar]
- 36.Adjei NK, Brand T. Investigating the associations between productive housework activities, sleep hours and self-reported health among elderly men and women in western industrialised countries. BMC Public Health. 2018;18:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Sialino L, et al. Perspectives of older women in the Netherlands: identifying motivators and barriers for healthy lifestyles and determinants of healthy aging. BMC Public Health. 2023;23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Tucker-Seeley RD, Li Y, Sorensen G, Subramanian S. Lifecourse socioeconomic circumstances and multimorbidity among older adults. BMC Public Health. 2011;11:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Liu H, Zhang M, Zhang X, Zhao X. Exposure to early-life adversity and long-term trajectories of multimorbidity among older adults in China: analysis of longitudinal data from the China health and retirement longitudinal study. BMJ Open. 2024;14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wang S, et al. Influence of socioeconomic events on cause-specific mortality in urban Shanghai, China, from 1974 to 2015: a population-based longitudinal study. CMAJ. 2018;190:E1153–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Wang L, Wang Z, Ma Q, Fang G, Yang J. The development and reform of public health in China from 1949 to 2019. Glob Health. 2019;15:1–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Bishop NJ, Haas SA, Quiñones AR. Cohort trends in the burden of multiple chronic conditions among aging US adults. The Journals of Gerontology: Series B. 2022;77:1867–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Zewde N, Crystal S. Impact of the 2008 recession on wealth-adjusted income and inequality for US cohorts. The Journals of Gerontology: Series B. 2022;77:780–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
(DOCX 232 KB)
Data Availability Statement
The CLHLS dataset analyzed during the current study is available data that are freely available in the ICPSR repository, https://www.icpsr.umich.edu/icpsrweb/NACDA/series/487.