Association between multimorbidity and young-onset dementia: a prospective study Authors: Durong Chen, Meiling Zhang, Yalu Wen, Hongjuan Han, Yao Qin, Rong Zhang, Hongyan Cao, Hongmei Yu Published: 2025 ABSTRACT AbstractBackgroundMultimorbidity contributes to an increased risk of late-onset dementia, but the multimorbidity effect on young-onset dementia (YOD, defined as dementia diagnosed before age 65) remains unclear. We aimed to explore the associations between multimorbidity (including its burden, pattern, and trajectory) and YOD.MethodsWe utilized the China Health and Retirement Longitudinal Study in China. Multimorbidity burden, patterns and trajectories were assessed based on 12 chronic diseases from 2011 to 2015. Patterns of multimorbidity were analysed by hierarchical clustering in 2015. Group-based trajectory modeling (GBTM) was utilized to identify the trajectories of the number of newly developed chronic conditions from 2011 to 2015. YOD and probable YOD were assessed by self-reported physician diagnosis of YOD and cognitive assessments from 2015 to 2020. Cox regression was used to estimate associations between multimorbidity and YOD.ResultsAmong 4,874 participants in 2015, 47.21% had multimorbidity; overall, 189 individuals developed YOD. The number of chronic conditions was associated with incident YOD (hazard ratio (HR) = 1.42, 95% confidence interval (95% CI):1.32–1.53). Three distinct multimorbidity patterns were found to be associated with YOD: the cardiometabolic cluster (HR = 3.57, 95% CI: 2.40–5.31), the gastric-arthritis cluster (HR = 2.43, 95% CI: 1.66–3.54), and the mixed cluster (HR = 4.05, 95% CI: 2.76–5.94). Multimorbidity trajectories were classified as “no new condition”, “slow growth”, and “rapid growth” by the GBTM. Compared to the “no new condition” group, both the “slow growth” (HR = 1.42, 95% CI: 1.04–1.94) and “rapid growth” group (HR = 2.67, 95% CI: 1.76–4.04) were associated with YOD.ConclusionsThe multimorbidity burden, patterns, and growth trajectories were associated with the risk of YOD. The study highlights the importance of addressing multimorbidity in reducing YOD risk and improving public health outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-025-25633-5.Keywords: Multimorbidity, Young-onset dementia, Healthy aging, Population-based study Background Multimorbidity refers to the presence of two or more chronic diseases and health conditions in an individual [1]. As the world faces an aging population, multimorbidity is becoming more prevalent, with its prevalence among older adults ranging from 4.8% to 93.1% [2]. This results in a lasting burden and significant impact on both individuals and society. Many studies have confirmed that several chronic diseases are linked to a higher dementia risk [3–6]. Although existing studies have investigated the relationships between multimorbidity and dementia risk, they have primarily focused on late-onset dementia (LOD) with inconsistent results [7, 8]. Notably, the association between multimorbidity and young-onset dementia (YOD) has not been clearly established. YOD is defined as dementia with an onset before the age of 65, with an age-standardized prevalence of 119 per 100,000 individuals [9, 10]. Unlike LOD, YOD poses a more profound socioeconomic burden, as it affects not only the health of patients but also their work capabilities, social lives, relationships, financial stability, and access to social security [11, 12]. This highlights the pressing need to mitigate the risk factors associated with YOD. Recent findings indicate a significant association between stroke, diabetes, cardiovascular disease, depression, and an increased risk of YOD [13, 14]. Nevertheless, these studies concentrate on one chronic disease at a time, and the connection between multimorbidity and YOD has yet to be sufficiently investigated. Previous studies demonstrated that specific multimorbidity patterns among older populations, resulting from shared risk factors and pathophysiological mechanisms [15], were associated heterogeneously with dementia [16–19]. There is reasonable suspicion that the different combinations of chronic diseases may also exert different effects on YOD. Furthermore, the temporal trajectory of multimorbidity accumulation may harbor more salient pathological insights. The speed of chronic condition accumulation reflects the pace of physiological system decline [20], and this accelerated pathological load could drive cognitive deterioration through multiple pathways. Therefore, exploring the relationship between multimorbidity and YOD is crucial. In this study, we aimed to assess the associations of multimorbidity and the risk for YOD from burden, patterns, and trajectories of multimorbidity using data from a large, nationally representative cohort of China. Such efforts are critical to provide a foundation for the development of targeted prevention strategies and integrated healthcare, and to reduce the burden of YOD for middle-aged population. Methods Study population For this population-based study, data were derived from the wave 1 (2011) to 5 (2020) of the China Health and Retirement Longitudinal Study (CHARLS), a nationwide population-based prospective cohort survey in China, focusing on the middle-aged and older adult community-dwelling adults aged 45 years and older, as well as their spouses when possible [21]. CHARLS used a multi-stage probability-proportional-to-size sampling and uniformly trained investigators to collect high-quality data, information including basic demographics, family information, health status, healthcare, employment, and household economy. The response rate for the CHARLS cohort was approximately 80.5%. In our study design, we collected information on self- or proxy-reported chronic conditions from wave 1 (2011) to 3 (2015) for analysis of multimorbidity and detected incident YOD and probable YOD from wave 3 (2015) to wave 5 (2020). Participants aged 45–60 in 2015, who had no history of self- or proxy-reported dementia or probable dementia, and who had available multimorbidity data from 2011 to 2015, were included. There was a total of 4,874 individuals enrolled in the current study. A summary of missing data distributions for covariates can be found in Supplementary Table S1. The study timeline and flowchart of the analytical sample are detailed in the flowchart (Fig. 1). Flowchart of study populations The research protocol of the CHARLS was granted by the Institutional Review Board of Peking University (approval number: IRB00001052-11015), and all participants have signed informed consent forms. Definitions of burden, pattern, and trajectory of multimorbidity In total, 12 chronic diseases, covering hypertension, dyslipidemia, diabetes, lung disease, hepatopathy, heart disease, stroke, kidney disease, gastric illness, psychological disorder, arthritis, and asthma, were defined based on self- or proxy-reported health history from 2011–2015. We quantified the number of chronic diseases in 2015. Multimorbidity refers to the co-concurrent in the same individual of two or more chronic conditions. We further categorized multimorbidity into five groups (≤ 1, 2, 3, 4, and ≥ 5 chronic conditions) to assess the burden of multimorbidity [22]. Hierarchical clustering was performed to identify multimorbidity patterns using data from the year 2015. Euclidean distance was used as the metric for clustering, and Ward’s linkage method was employed. The optimal number of disease clusters was determined based on the average silhouette width and visual inspection of the resulting dendrogram. Group-based trajectory modeling (GBTM) was utilized to identify the trajectories of the number of newly developed chronic conditions from 2011 to 2015 [8]. GBTM identifies latent subgroups with similar temporal patterns, without incorporating random effects. A newly developed condition was defined as any chronic condition not present in 2011 but first reported in 2013 or 2015. The optimal number of the classes was determined based on the minimum Bayesian Information Criterion (BIC), with the additional requirement that the size of each trajectory group exceeded 5% of the total sample. Definitions of YOD and probable YOD The outcome was all-cause YOD or probable YOD, defined as dementia diagnosed before the age of 65 years. Probable YOD was considered to address the potential under-diagnosis of YOD [22, 23]. Briefly, the diagnosis of YOD or a related memory disorder was diagnosed by medical doctors or specialists and reported by self- or proxy-respondents in response to the question, “Have you been diagnosed by a doctor with a memory-related disorder (such as Alzheimer’s disease, brain atrophy, or Parkinson’s disease)”. Probable YOD was assessed by cognitive assessments with total score of 25, including immediate and delayed word recall tests (maximum score of 20) and self-assessed memory abilities (maximum score of 5). Participants scoring 1.5 standard deviations (SD) below the mean cognitive score, adjusted for their levels of education, were judged to have probable YOD [22, 24]. Covariates Information on demographic characteristics and lifestyle factors in the CHARLS was collected using face-to-face questionnaires. Demographic characteristics included age, sex and education levels. Three levels were defined for education attainment: middle school and below, high school or equivalent education levels, college and above. Lifestyle factors included body mass index (BMI), smoking status (never or ever/current), and alcohol consumption (never or ever/current). BMI was divided into four groups according to the Chinese standard: underweight (BMI < 18.5 kg/m²), normal weight (BMI 18.5–23.9 kg/m²), overweight (BMI 24.0–27.9 kg/m²), and obese (BMI ≥ 28 kg/m²). All of these variables were measured at baseline and were adjusted for in the analyses. Statistical analyses The baseline characteristics of participants were described as mean (SD) for continuous variables and frequency (percentage) for categorical variables. The missing proportions for covariates were below 20%. The distribution of each chronic disease within the multimorbidity is presented in Table S2. Missing data were imputed using the mode. The restricted cubic spline (RCS), using four knots, was performed to investigate the potential non-linear relationship between the number of chronic conditions and dementia. Individual follow-up time was measured from the baseline (2015) to the year of incident YOD or probable YOD, death, or missing follow-up, or study endpoint (2020), whichever came first. Individuals without information on dementia status or complete cognitive assessment information were treated as censored. To assess the role of multimorbidity burden, patterns, and trajectories in YOD, Cox proportional hazard models were used to assess the hazard ratios (HRs) with 95% confidence intervals (CIs). To verify the proportional hazard assumption, the Schoenfeld residual method was applied. We constructed three models. Model 1 was unadjusted. Model 2 was adjusted for age, sex, and education levels. Model 3 extended Model 2 by further adjusting for BMI, smoking status, and drinking status. To examine the robustness of the results, we performed several subgroup analyses. We conducted subgroup analyses by sex (male vs. female), education level (middle school and below vs. high school/equivalent education level and above), BMI (non-overweight vs. overweight), alcohol consumption (never or ever vs. current), and smoking status (never or ever vs. current). We also conducted a set of sensitivity analyses by (1) treating mortality as the competing event in Fine & Gray’s model; (2) focusing on participants without 12 chronic diseases in 2011; (3) adding the number of chronic diseases in 2011 as a confounding variable in the models involving the trajectories of multimorbidity; (4) and using data without missing covariates to assess the robustness of the results. All statistical tests were conducted using R statistical software version 4.3.2 and Stata statistical software version 18. The threshold for statistical significance was set at a p value of less than 0.05. Results Baseline characteristics of study population At baseline, of 4,874 participants with a mean age of 54.22 (SD = 3.80) years, 55.66% were female (Table 1). A total of 189 participants progressed to YOD during a median follow-up of 5 years, including 177 with diagnosed dementia and 12 with probable dementia. In the total sample, the number of chronic conditions per individual was 1.71, and 2,301 (47.21%) participants were determined to have multimorbidity. Specifically, 1,059 (21.73%) participants had two chronic conditions, 629 (12.91%) had three, 327 (6.71%) had four, and 286 (5.87%) had ≥ 5 coexisting conditions. Participants with a greater number of chronic conditions were more likely to be female, to have a middle school and below, to have a normal weight, to be never-smokers or former smokers, and to be never or former alcohol consumers. Table 1Baseline characteristics of study participants in 2015Multimorbidity condition in 2015Total(N = 4874)≤ 1(N = 2573)2(N = 1059)3(N = 629)4(N = 327)≥ 5(N = 286)Age, year, mean ± SD53.97 ± 3.8254.43 ± 3.7654.48 ± 3.8154.73 ± 3.7654.61 ± 3.6054.22 ± 3.80Sex, n(%) Male1196 (46.48)459 (43.34)277 (44.04)132 (40.37)97 (33.92)2161 (44.34) Female1377 (53.52)600 (56.66)352 (55.96)195 (59.63)189 (66.08)2713 (55.66)Education level, n(%) Middle school and below2129 (82.74)916 (86.50)523 (83.15)267 (81.65)242 (84.62)4077 (83.65) High school or equivalent education level404 (15.70)135 (12.75)96 (15.26)55 (16.82)41 (14.34)731 (15.00) College and above40 (1.55)8 (0.76)10 (1.59)5 (1.53)3 (1.05)66 (1.35)BMI, n(%) < 18.572 (2.80)35 (3.31)13 (2.07)12 (3.67)7 (2.45)139 (2.85) 18.5–23.91497 (58.18)599 (56.56)319 (50.72)137 (41.90)119 (41.61)2671 (54.80) 24.0–27.9719 (27.94)293 (27.66)209 (33.23)106 (32.42)92 (32.17)1419 (29.11) ≥ 28285 (11.08)132 (12.46)88 (13.99)72 (22.02)68 (23.78)645 (13.23)Smoking, n(%) Never or ever1793 (69.69)775 (73.18)463 (73.61)255 (77.98)220 (76.92)3506 (71.93) Current780 (30.31)284 (26.82)166 (26.39)72 (22.02)66 (23.08)1368 (28.07)Alcohol consumption, n(%) Never or ever1581 (61.45)696 (65.72)405 (64.38)230 (70.34)214 (74.83)3126 (64.14) Current992 (38.55)363 (34.28)224 (35.61)97 (29.66)72 (25.17)1748 (35.86)Missing values for BMI, smoking, and alcohol consumption were imputed using the mode of each variable as described in the Methods section. Percentages in Table 1 reflect these adjustments Baseline characteristics of study participants in 2015 Missing values for BMI, smoking, and alcohol consumption were imputed using the mode of each variable as described in the Methods section. Percentages in Table 1 reflect these adjustments Association of multimorbidity burden with YOD The risk of YOD exhibited a linear increase with each additional chronic disease (Fig. 2A; Table 2). The relationships between chronic conditions and YOD were also observed in each stratified subgroup by sex, education levels, BMI, smoking status, and alcohol consumption (Fig. S1 and Table S3). Specifically, compared with individuals without multimorbidity, those with ≥ 2 chronic diseases had a 2.24-fold higher YOD risk in the fully adjusted model. Specifically, the HRs were 2.09 (95% CI: 1.36 to 3.20) for three chronic conditions, 3.21 (95% CI: 2.02 to 5.10) for four, and 5.58 (95% CI: 3.69 to 8.45) for five or more chronic conditions in the fully adjusted model (Table 2). Different measurements of multimorbidity A Multimorbidity burden. The solid line represents the hazard ratio from the Cox proportional hazards model, and the shaded area represents the 95% confidence interval (left y-axis). The gray bars represent the scaled probability density of number of chronic conditions in the study population, shown as probability density (right y-axis) B Multimorbidity patterns. Cardiometabolic Cluster (purple) includes heart disease, hypertension and dyslipidemia. Gastric-arthritis cluster (yellow) includes gastric illness and arthritis. Mixed cluster (red) includes diabetes, lung disease, hepatopathy, stroke, kidney disease, psychological disorders and asthma C Multimorbidity trajectories. Group-based trajectory modeling identified distinct patterns of chronic disease accumulation from 2011 to 2015 Table 2The associations of multimorbidity with risk of YODmultimorbidityEvents/ParticipantsModel 1Model 2Model 3HR (95% CI)HR (95% CI)HR (95% CI) Burden Each additional chronic condition1.43 (1.33–1.54)1.42 (1.33–1.53)1.42 (1.32–1.53) Status of multimorbidity 0–1 chronic condition62/25731.00 (Reference)1.00 (Reference)1.00 (Reference) ≥ 2 chronic condition127/23012.32 (1.71–3.14)2.27 (1.67–3.07)2.24 (1.65–3.04) Number of chronic conditions 0–162/25731.00 (Reference)1.00 (Reference)1.00 (Reference)232/10591.24 (0.81–1.90)1.23 (0.80–1.89)1.23 (0.85–1.89)332/6292.14 (1.40–3.28)2.08 (1.36–3.19)2.09 (1.36–3.20)426/3273.40 (2.15–5.38)3.25 (2.05–5.14)3.21 (2.02–5.10)>=537/2865.79 (3.85–8.70)5.65 (5.65–8.51)5.58 (3.69–8.45) Pattern Cardiometabolic clustera47/5493.67 (2.51–5.36)3.46 (2.37–5.09)3.57 (2.40–5.31) Gastric-arthritis clustera50/8352.52 (1.74–3.65)2.45 (1.68–3.56)2.43 (1.66–3.54) Mixed clustera48/4924.29 (2.94–6.26)4.19 (2.87–6.12)4.05 (2.76–5.94) Trajectory No new condition83/27531.00 (Reference)1.00 (Reference)1.00 (Reference) Slow growth75/17331.44 (1.05–1.97)1.43 (1.05–1.95)1.42 (1.04–1.94) Rapid growth31/3882.74 (1.82–4.14)2.68 (1.77–4.05)2.67 (1.76–4.04)Model 1: No variables for adjustment, Model 2: HRs were adjusted for age, sex, education levels, Model 3: HRs were adjusted for age, sex, education levels, BMI, smoking status, and alcohol consumption. Cardiometabolic cluster includes heart disease, hypertension and dyslipidemia, Gastric-arthritis cluster includes gastric illness and arthritis, Mixed cluster includes diabetes, lung disease, hepatopathy, stroke, kidney disease, psychological disorders and asthmaa Participants without multimorbidity were considered the reference group The associations of multimorbidity with risk of YOD Model 1: No variables for adjustment, Model 2: HRs were adjusted for age, sex, education levels, Model 3: HRs were adjusted for age, sex, education levels, BMI, smoking status, and alcohol consumption. Cardiometabolic cluster includes heart disease, hypertension and dyslipidemia, Gastric-arthritis cluster includes gastric illness and arthritis, Mixed cluster includes diabetes, lung disease, hepatopathy, stroke, kidney disease, psychological disorders and asthma a Participants without multimorbidity were considered the reference group Association between multimorbidity patterns and YOD Among participants with multimorbidity, three patterns of multimorbidity were generated in 2015 (Fig. 2B). The cardiometabolic cluster (23.86%) included heart disease, hypertension and dyslipidemia; the gastric-arthritis cluster (36.29%) included gastric illness and arthritis; the mixed cluster (21.38%) included diabetes, lung disease, hepatopathy, stroke, kidney disease, psychological disorders and asthma. Cox models further quantified the YOD risk: mixed cluster exhibited the highest risk (HR = 4.05, 95%CI: 2.76–5.94), cardiometabolic cluster showed a 3.57-fold increased risk (95%CI: 2.40–5.31), gastric-arthritis cluster had a 2.43-fold increased risk (95%CI: 1.66–3.54) after full adjustment of covariates (Table 2). Regarding analyses stratified by sex, education levels, BMI, smoking status, and alcohol consumption, the associations persisted (Table S4). Association of multimorbidity trajectories and YOD By GBTM analysis, we identified three multimorbidity trajectories using newly-developed chronic conditions: “no new condition” group (56.5%), “slow growth” group (35.6%), and “rapid growth” group (8.0%) (Fig. 2C and Table S5). In the fully adjusted model, compared to the “no new condition” group, individuals in the “slow growth” group (HR = 1.42, 95%CI: 1.04–1.94) and “rapid growth” group (HR = 2.67, 95%CI: 1.76–4.04) showed increased hazards of YOD (Table 2). Consistent results were also observed in subgroup analyses of sex, education levels, BMI, smoking status, and alcohol consumption in “rapid growth” group (Table S6). Sensitivity analyses The competing risk analysis showed that consistent results were observed for multimorbidity burden, patterns and trajectories (Table S7). When analyses were limited to participants who did not have any of the 12 chronic conditions in 2011, the estimates of individuals with four chronic conditions and equal to or greater than five chronic conditions remained significant (Table S8). Adding the number of chronic diseases in 2011 as a confounding variable in the models involving the trajectories of multimorbidity, the “rapid growth” multimorbidity trajectory maintained its association with YOD (Table S9). The analysis using data without missing covariates yielded consistent results (Table S10). Discussion In this analysis of 4,874 adults, a greater multimorbidity burden, diverse multimorbidity patterns, and the accumulation of multimorbidity trajectories were associated with a higher risk of YOD. These findings highlight the critical role of multimorbidity for prevention of YOD and support healthcare planning with evidence. As the number of chronic conditions grows, the overall health of the individual deteriorates [25, 26], which could cause the dysfunction of several physiological systems, ultimately harming brain health and raising the risk of developing dementia. In this study, we found that participants with multimorbidity showed a higher risk of YOD, which aligns with previous findings regarding LOD [8, 27], but the magnitude of this effect was more pronounced. A potential explanation is that LOD, which is characterized by more pronounced physiological brain atrophy, may thus have the impact of multimorbidity on the brain obscured by age-related physiological decline in this older population. This highlights that from a public health perspective, prioritizing the prevention and management of multimorbidity among middle-aged populations is critical for reducing the risk of YOD. Our study identified three types of multimorbidity patterns and found that all were risk factors for YOD. Dementia is a condition with its etiology not yet fully understood, and various chronic diseases play different roles in its onset. Hypertension, diabetes and high cholesterol levels are risk factors for heart disease, stroke, and dementia, while heart disease and stroke are associated with the development of dementia [4, 17, 28–30]. These diseases may contribute to the decline of brain function through shared common pathways, such as vascular injury, inadequate oxygen supply, and inflammatory response [31]. It is worth noting that in our study, patients with multimorbidity of heart disease, hypertension, and dyslipidemia were more likely to develop YOD. This is in alignment with the report from UK Biobank cohort [32]. In the current study, subjects with gastric illness and arthritis were found to experience high risks of dementia. Previous evidence suggested that digestive system diseases like liver disorders and gastric illness might contribute to the risk of dementia through brain insulin resistance, cerebrovascular dysfunction, gut dysbiosis, and neuroinflammation [33, 34]. Arthritis and asthma are thought to be associated with dementia through a common biological process associated with inflammation [35–37]. Mixed cluster may result from overlapping risk factors and increased treatment burden, which could subsequently contribute to neural injury [3, 6, 38, 39]. To the best of our knowledge, this study was the first to examine the trajectories of multimorbidity and their association with the risk of YOD. Our research identified three different trajectories of multimorbidity, and indicated that the different trajectories of multimorbidity accumulation were associated with an increased risk of YOD. Our findings also indicated that more rapid accumulation of chronic conditions was linked to a higher risk of dementia. And nearly half of the study population experienced chronic disease accumulation, marking them a high-risk group for dementia. These findings suggested that not only the presence of multimorbidity but also the speed of their accumulation significantly influences the onset of YOD. The results remained consistent after considering competing risk of death. Our findings were also in agreement with a study that indicated participants with developing multimorbidity trajectories had lower cognitive scores [40]. Mechanistically, the rapid accumulation of diseases may contribute to systemic metabolic disturbances, and heightened inflammation [41, 42], which could accelerate neurodegenerative processes. Compared with individuals exhibiting slow growth of multimorbidity, those with rapid growth of multimorbidity may exceed homeostatic capacity, resulting in a higher risk of dementia. In contrast to another population-based study showing an association between the rapid growth trajectory of multimorbidity and LOD, which had a median follow-up of eight years [8], our study found that even slow progression of multimorbidity trajectories was linked to YOD within a short period. This suggests that early detection and management of multimorbidity, regardless of its speed, could be crucial in reducing the risk of YOD. We also tested the interaction terms between multimorbidity and factors such as sex, BMI categories, education attainment, alcohol consumption, and smoking. However, no significant interactions were observed (p > 0.05). Despite this, we performed subgroup analyses to assess whether these factors influenced the incidence of YOD. Our results suggested that while sex, BMI categories, education, and lifestyle factors like smoking and alcohol consumption may not significantly interact with multimorbidity, they remain important independent risk factors for YOD. These findings highlight the need for further research to explore potential additive effects in larger, more diverse populations. Considering the impact of multimorbidity on YOD risk, effective management of chronic conditions may play a pivotal role in mitigating this risk. Previous studies have shown that effective management of multimorbidity—including adherence to medication, nutritional interventions, maintaining a healthy lifestyle, and timely access to treatment—can help manage and ameliorate chronic conditions, thereby slowing cognitive decline in individuals with YOD [43]. Moreover, most people with YOD are still in periods of high occupational and familial responsibility, facing substantial psychological stress and social obligations. Therefore, multimorbidity management interventions should not only address the medical conditions themselves but also incorporate family involvement and provide psychosocial support. Such an integrated approach can enhance the sustainability and effectiveness of interventions, ultimately minimizing dementia risk and improving quality of life. The current research provides comprehensive and valuable supporting evidence for the association between multimorbidity and YOD. Furthermore, subgroup analyses and four sensitivity analyses were performed to validate our main findings. However, there were some limitations in the interpretation of our study findings. First, the diagnosis of chronic diseases was based on self-reported or proxy-reported data, which may be subject to recall bias, potentially affecting the accuracy of the results [44, 45]. Therefore, it is crucial to establish objective dementia indicators to further investigate the relationships between multimorbidity and YOD. Second, the current definition of multimorbidity included 12 chronic conditions, but certain diseases that could potentially affect YOD have not been included. Moreover, future studies should also consider the role of intrinsic vulnerability that may interact with multimorbidity in shaping dementia risk. Third, despite adjusting for multiple confounding factors and conducting various sensitivity analyses, unmeasured confounders, such as whether chronic diseases were treated with medication, might lead to biased estimates. In addition, given the progressive decline in cognitive function prior to dementia diagnosis, underlying cognitive decline may trigger the onset of multiple chronic conditions, making reverse causality a possible factor. Instrumental variable could be used in the future to confirm causality. Finally, future research could investigate whether individuals who develop YOD possess inherent vulnerability, and whether multimorbidity has differential effects on YOD compared with LOD. Conclusions In conclusion, multimorbidity burden, patterns, and developmental trajectories were associated with the risk of YOD. These findings underscore that for middle-aged individuals, efforts to reduce the number and rate of development of multimorbidity may play a pivotal role in reducing YOD risk. Supplementary Information Supplementary Material 1. Acknowledgements We thank the staff of the China Health and Retirement Longitudinal Study for their contributions. The analyses use data or information from the Harmonized CHARLS dataset and Codebook, Version D as of June 2021 developed by the Gateway to Global Aging Data. The development of the Harmonized CHARLS was funded by the National Institute on Aging (R01 AG030153, RC2 AG036619, R03 AG043052). For more information, please refer to https://g2aging.org/. Abbreviations Young-onset dementia Late-onset dementia China Health and Retirement Longitudinal Study Group-based trajectory modeling Bayesian information criterion Standard deviations Hazard ratio 95% Confidence interval Authors’ contributions D.C. and H.Y. contributed to the conceptualization and design of the study. D.C. performed the statistical analyses and wrote the main manuscript text. M.Z. and H.H. prepared tables and Figures. Y.W., Y.Q., R.Z., and H.C. further revised the manuscript. All authors reviewed the manuscript. Funding National Natural Science Foundation of China, Grant/Award Number: 82574209, 82173632, 82504527, and 82473739. Data availability The CHARLS data set is freely available to all bona fide researchers. Researchers can gain access to the data (http://charls.pku.edu.cn/en). Declarations Ethical approval and consent to participate The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. The Biomedical Ethics Committee of Peking University approved CHARLS (IRB00001052-11015), and all participants gave written informed consent. Consent for publication All authors approved the final manuscript for submission and gave consent for publication. Competing interests 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 Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The CHARLS data set is freely available to all bona fide researchers. Researchers can gain access to the data (http://charls.pku.edu.cn/en).