Impact Statement
This study provided a more comprehensive understanding of the cognitive decline trajectories that occur across the adult lifespan, particularly in relation to the physical multimorbidity. The findings may have significant implications for the implementation of highlighting the extent to which different physical disease clusters increase the risk of poor cognitive functioning health across wealth status and the need for their early prevention, diagnosis and treatment through clinical and public health efforts.
Introduction
Cognitive decline has become a global issue with rapid population aging (Hsieh et al., Reference Hsieh, Chen, Chen, Chiou and Chen2023) and the prevalence of cognitive impairment will continue to increase (Du et al., Reference Du, Tao, Liu and Liu2024). It is estimated that more than 55 million people worldwide suffer from cognitive decline currently (Bai et al., Reference Bai, Chen, Cai, Zhang, Su, Cheung, Sha and Xiang2022). Regarding the previous studies, various chronic long-term diseases, including but not limited to diabetes (van Boxtel et al., Reference van Boxtel, Buntinx, Houx, Metsemakers, Knottnerus and Jolles1998), stroke (Dik et al., Reference Dik, Deeg, Bouter, Corder, Kok and Jonker2000), pulmonary diseases (Zhou et al., Reference Zhou, Yang, Zhang, Li, Zhang, Li, Ma, Hou, Lu and Wang2022) and peripheral atherosclerosis (Comijs et al., Reference Comijs, Kriegsman, Dik, Deeg, Jonker and Stalman2009), have been consistently shown to be associated with cognitive decline. It is worth noting that a single chronic disease was rare among middle-aged and older adult (Valletta et al., Reference Valletta, Vetrano, Calderón-Larrañaga, Kalpouzos, Canevelli, Marengoni, Laukka and Grande2024). Currently, physical multimorbidity, which refers to the presence of two or more chronic long-term conditions, is attracting more and more attention from researchers. It is estimated that 67.8% of the UK population aged 65 and over will experience physical multimorbidity by 2035 (Aarts et al., Reference Aarts, van den Akker, Tan, Verhey, Metsemakers and van Boxtel2011). The prevalence of physical multimorbidity increased up to 78% in subjects aged 80 and over (van den Akker et al., Reference van den Akker, Buntinx, Metsemakers, Roos and Knottnerus1998). The simultaneous presence of different long-term chronic diseases may be due to shared risk factors and common underlying pathophysiologic mechanisms, while different physical multimorbidity clusters with the same pathways may also have different effects on cognitive decline.
In recent years, there have been limited studies exploring the influence of different physical multimorbidity patterns on cognitive function decline, which generated inconsistent results. A large population-based cohort study reported that an increasing number of cardiometabolic multimorbidity (including diabetes, heart disease and stroke) dose-dependently accelerated cognitive decline and the onset of cognitive impairment by 2.3 years (Dove et al., Reference Dove, Marseglia, Shang, Grande, Vetrano, Laukka, Fratiglioni and Xu2022). Metabolic multimorbidity pattern was associated with non-amnestic cognitive impairment, whereas degenerative ocular multimorbidity pattern was only associated with Alzheimer’s disease (Ren et al., Reference Ren, Li, Tian, Liu, Dong, Hou, Liu, Han, Han, Wang, Vetrano, Ngandu, Marengoni, Kivipelto, Wang, Cong, Du and Qiu2023). Besides, a longitudinal observational study pointed out that vascular multimorbidity pattern has no significant impact on the progression of cognitive decline after 2 years (Eldholm et al., Reference Eldholm, Persson, Barca, Knapskog, Cavallin, Engedal, Selbaek, Skovlund and Saltvedt2018). However, existing studies had some limitations. Since no effective treatment exists for severe cognitive impairment, intervening in risk factors for cognitive decline beginning in middle-aged adults is imperative (Bloomberg et al., Reference Bloomberg, Brocklebank, Hamer and Steptoe2023), but few studies have focused on the impact of physical multimorbidity patterns on subsequent cognitive functioning in middle-aged populations, mostly in those older than 60 years. A combined analysis in both middle-aged and older adults may provide a more comprehensive understanding of the cognitive decline trajectories that occur across the adult lifespan, particularly in relation to physical multimorbidity. Meanwhile, some studies explored only one physical multimorbidity pattern, lacking studies of broader patterns.
In addition, specific physical multimorbidity patterns may differ in association with the risk of developing cognitive decline over time. An investigation from the Swedish national study on aging revealed that participants in the cardiovascular pattern exhibited an increased hazard of progression from cognitive impairment with no dementia (CIND) to dementia (Valletta et al., Reference Valletta, Vetrano, Xia, Rizzuto, Roso-Llorach, Calderón-Larrañaga, Marengoni, Laukka, Canevelli, Bruno, Fratiglioni and Grande2023). During the 7-year follow-up period, memory function in the osteoarthrosis group declined at the highest rate (Li et al., Reference Li, Hu, Han, Wang, Ma, Chu, He, Feng, Sun and Shen2024). In terms of the trajectory of physical multimorbidity status, the higher trajectory also posed an elevated risk of cognitive decline (Du et al., Reference Du, Tao, Liu and Liu2024). Even so, little research exists examining the rate of cognitive decline over time.
In the middle-aged and older population, lower wealth status may create conditions that are detrimental to health (e.g., severe pollution, occupational hazards), while limiting access to resources to protect and promote health(O’Neill et al., Reference O’Neill, Newsom, Trubits, Elman, Botoseneanu, Allore, Nagel, Dorr and Quiñones2023) (e.g., health care, nutritious food), increasing the burden of illnesses (e.g., cognitive impairment) (Depp and Jeste, Reference Depp and Jeste2006) and the risk of physical multimorbidity (Pathirana and Jackson, Reference Pathirana and Jackson2018). Wealth status and physical multimorbidity are interrelated, with lower levels of wealth strongly associated with physical multimorbidity (Ni et al., Reference Ni, Zhou, Kivimäki, Cai, Carrillo-Larco, Xu, Dai and Xu2023), although wealth status and physical multimorbidity might also influence cognitive function through distinct pathways (Grande et al., Reference Grande, Marengoni, Vetrano, Roso-Llorach, Rizzuto, Zucchelli, Qiu, Fratiglioni and Calderón-Larrañaga2021; Pan et al., Reference Pan, Gao, Zhu and Guo2023). Different combinations between wealth status and physical multimorbidity may be differentially correlated to cognitive decline. To the best of our knowledge, there is no evidence on the joint associations of wealth status and physical multimorbidity with cognitive function. The extent to which the association persists over time is therefore unclear.
In sum, the first objective of the current study was to identify physical multimorbidity patterns in middle-aged and older adults. The second is to estimate the association between physical multimorbidity and cognitive function and the rate of decline in the cognitive function over time in different physical multimorbidity patterns. Thirdly, we then aim to further explore the relationship between the combinations effects of wealth status and physical multimorbidity on cognitive function.
Methods
Study design and participants
The English Longitudinal Study of Aging (ELSA) is an ongoing study of the health, social and economic lives that began in 2002 and the population aged 50 years and older (Steptoe et al., Reference Steptoe, Breeze, Banks and Nazroo2013). The core sample in wave 1 was recruited from households that were included in the 1998, 1999, or 2001 Health Survey for England. At the first wave of ELSA, 11,391 core members were included from a total sample of 12,099 participants, with subsequent waves of data collection occurring every 2 years.
In the present study, participants were asked about physical long-term conditions in the 2002–2003 wave (wave 1). Cognitive function was assessed from wave 2 to wave 9 (Jan 1, 2004, to July 31, 2019). Respondents participating in wave 1 with at least three waves of cognitive testing were eligible for this study. The final sample consisted of 9,035 individuals in ELSA.
Measurements
Physical multimorbidity
In this study, physical long-term conditions at wave 1 were measured by self-report doctor diagnoses (Poole and Steptoe, Reference Poole and Steptoe2018; Ronaldson et al., Reference Ronaldson, de la Torre, Bendayan, Yadegarfar, Rhead, Douiri, Armstrong, Hatch, Hotopf and Dregan2023) and included hypertension, diabetes, stroke and coronary heart disease (CHD; which included myocardial infarction and/or angina), other cardiac illnesses (such as heart failure, arrhythmia and heart murmur), lung disease (such as chronic bronchitis and emphysema), asthma, arthritis, osteoporosis, cancer, Parkinson’ s disease, dementia, glaucoma and cataracts. A cut-off of two conditions for every participant in the study was used to define a comprehensive measure of physical multimorbidity (Johnston et al., Reference Johnston, Crilly, Black, Prescott and Mercer2019). To explore the potential dose–response association with cognitive function outcome, we formulated an ordinal variable representing the status of physical multimorbidity, categorizing participants into no or one condition (absence of multimorbidity/no multimorbidity); two conditions and three or more conditions (Ronaldson et al., Reference Ronaldson, Arias de la Torre, Prina, Armstrong, Das-Munshi, Hatch, Stewart, Hotopf and Dregan2021, Reference Ronaldson, de la Torre, Bendayan, Yadegarfar, Rhead, Douiri, Armstrong, Hatch, Hotopf and Dregan2023).
Cognitive function
Cognitive function is assessed in ELSA using a series of objective neurocognitive tests as described previously (Chen et al., Reference Chen, Lu, Wang, Zhang, Zhang and Li2022a, Reference Chen, Lu, Zhao, Wang, Zhang, Zhang and Li2022b; Zheng et al., Reference Zheng, Zhong, Song and Xie2018), including word recall (immediate and delayed) tests (Baars et al., Reference Baars, van Boxtel, Dijkstra, Visser, van den Akker, Verhey and Jolles2009), animal-naming test (Dregan et al., Reference Dregan, Stewart and Gulliford2013) and letter cancelation test (see supplementary material A). We generated a composite global cognitive z score in each wave (wave 2 to wave 9) to estimate the overall cognitive function (see supplementary material B). As a secondary outcome, low cognitive performance assessed by the objective neurocognitive tests had no gold standard on the cutoff scores. Therefore, low cognitive performance was defined using the lowest quartile of each test stratified by gender as the cutoff value, which was consistent with previous studies (Chen et al., Reference Chen, Lu, Wang, Zhang, Zhang and Li2022a, Reference Chen, Lu, Zhao, Wang, Zhang, Zhang and Li2022b; Li et al., Reference Li, Sun and Zhang2019).
Covariates
We selected covariates at baseline according to prior research (Agur et al., Reference Agur, McLean, Hunt, Guthrie and Mercer2016; Chen et al., Reference Chen, Lu, Wang, Zhang, Zhang and Li2022a, Reference Chen, Lu, Zhao, Wang, Zhang, Zhang and Li2022b; Ronaldson et al., Reference Ronaldson, Arias de la Torre, Prina, Armstrong, Das-Munshi, Hatch, Stewart, Hotopf and Dregan2021). Sociodemographic factors included age, gender and cohabitation status (not cohabiting and married/cohabiting). Education attainment level, lifestyle variable (including smoking status and alcohol intake), depressive symptoms and cognitive function at baseline were also included in this analysis (more details in supplementary material C). Wealth status was estimated by the quintiles of net non-pension wealth comprising financial, physical and housing wealth, while excluding pension wealth (Ronaldson et al., Reference Ronaldson, de la Torre, Bendayan, Yadegarfar, Rhead, Douiri, Armstrong, Hatch, Hotopf and Dregan2023). It was calculated net of debt and encompassed the valuation of any home and additional property (excluding mortgage), financial assets (including all forms of savings accessible in England), business assets and physical wealth like artwork and jewelry.
Statistical analysis
Analyses were done using Stata 16 and Mplus 8.3. We compared characteristics among different physical multimorbidity statuses and patterns using means with standard deviation (SD) or N with percentage (%) for descriptive analyses. Weights were derived by many of the variables that were used as covariates. Therefore, to avoid over-adjustment, we did not apply weights to the final analysis. One-way analysis of variance was used to compare the mean levels of cognitive functions among the physical multimorbidity status or patterns and Pearson chi-squared test or Fisher’s exact was conducted to compare the distribution of the categorical variables. We defined our complete case sample as those who had complete data on physical long-term conditions, baseline cognitive function estimation, covariates and the relevant outcome in at least two waves.
A latent class analysis was used to analyze physical multimorbidity patterns among the 14 physical conditions recorded for all participants who provided data at wave 1. The lowest Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Bootstrapped likelihood ratio test and Vuong-lo–Mendell–Rubin likelihood ratio test, as well as clinical interpretability, were used as criteria for finalizing the optimal number of classifications (Ronaldson et al., Reference Ronaldson, de la Torre, Bendayan, Yadegarfar, Rhead, Douiri, Armstrong, Hatch, Hotopf and Dregan2023). After determining the number of latent classes, each person in the sample was assigned to the class with the largest posterior probability (i.e., the class to which they most likely belonged). Multilevel models were carried out to explore the association between physical multimorbidity (wave 1) and subsequent cognitive function (waves 2 to 9), and we conducted multilevel logistic regression models for binary outcomes. Repeated cognitive function measurements were clustered within individuals, with random slopes according to time (time defined according to years of follow-up), and a random intercept for individuals (unstructured correlation matrix). Multilevel models are designed to optimize the utilization of longitudinal data, adjust for the intercorrelation between repeated measures within individuals, enhance statistical power and precision compared to traditional regression and incorporate weighted estimates to address missing data across waves (Lee et al., Reference Lee, Pearce, Ajnakina, Johnson, Lewis, Mann, Pitman, Solmi, Sommerlad, Steptoe, Tymoszuk and Lewis2021). Besides, to estimate the potential dose–response relationship, the linear trend test was conducted by treating the physical multimorbidity status as a continuous variable in the full-adjusted model for global cognitive function.
We built models in four stages. In the first model, we included physical multimorbidity (status or patterns), continuous linear time variables and cognitive function. We included the time quadratic term if we found evidence of a nonlinear influence of time on cognitive function. In the second model, we added age and sex. The third (fully-adjusted model) was additionally adjusted for couple status, net non-pension wealth, educational attainment, smoking status, alcohol intake, depressive symptoms and four cognitive test scores at baseline. To investigate whether the association between physical multimorbidity and cognitive function differed across timepoint, we calculated an interaction term between multimorbidity (status or patterns) and time as well, followed by calculating the average marginal effect through a marginal command.
As a secondary outcome, a binary variable for cognitive function was also included in the analyses of this study, using multilevel logistic regression models. Then, we defined low net non-pension wealth status using the lowest quintile as the cutoff value in joint association analysis for wealth status (Fry, Reference Fry, Firestone and Chakraborty2014). In the joint analysis, the combinations between physical multimorbidity status/patterns and wealth status were defined as categorical variables. Additionally, subgroup analyses were conducted to verify the associations between physical multimorbidity and subsequent cognitive function in different age and gender groups.
Results
Characteristics of the study participants by physical multimorbidity status and patterns
Table S1 provides baseline characteristics for the overall sample and participants with different physical multimorbidity statuses. Wave 1 of ELSA completed the data collection in a total of 11,391 participants. All participants with cognitive function measured at baseline and at least one subsequent follow-up comprised the overall sample (n = 9, 035). 3,762 (41.63%) participants had multimorbid physical long-term conditions. Participants with a higher number of coexistent conditions were more likely to be older, to be female, to have lower education attainment levels, to live with no cohabitation partners and to have lower net non-pension wealth. The likelihood of being a former or current smoker was highest in the individuals with two or more than three coexistent physical long-term conditions compared with those with no or one physical long-term condition. The proportion of participants who drank occasionally / weekly / monthly was found highest in all physical long-term condition groups. Likewise, depressive symptoms were more prevalent among those with more complex physical long-term conditions. Baseline cognitive function indicated that lower cognitive function was found among participants with more complex physical multimorbidity.
Sample characteristics for each physical multimorbidity pattern are shown in Table 1. Among the participants included, we identified four patterns of physical multimorbidity, namely, relatively healthy pattern (n = 6,400, 70.84%), complex/multisystem pattern (n = 428, 4.74%), respiratory/osteoporosis pattern (n = 1,005, 11.12%) and hypertension/diabetes pattern (n = 1,202, 13.30%) [Figure 1 and Table S3]. This process was based on the lowest AIC and BIC values (supplementary Table S2). The most prevalent physical long-term conditions in each pattern were used as pattern labels.
Sample characteristics by physical multimorbidity patterns at baseline
a P value was tested by one-way analysis of variance;
b P value was tested by Pearson chi-squared test or Fisher’s exact.
Probability of each disease in four physical multimorbidity patterns at baseline.
Participants in all physical multimorbidity patterns were older relative to the relatively healthy group, with the oldest in the complex/multisystem pattern (n = 232, 54.21%). Compared to all other patterns, there were more males in the hypertension/diabetes multimorbidity pattern (n = 590, 49.08%). The highest proportion of participants with no qualification/other education level was in the complex/multisystem pattern (n = 229, 53.50%). Participants in the respiratory/osteoporosis pattern were more likely to be no cohabiting compared to other patterns (n = 437, 43.48%), and those in the complex/multisystem pattern were more likely to have lower net non-pension wealth (n = 139, 32.48%). The complex/multisystem pattern had the highest proportion of former or current smokers (n = 308, 71.96%). The prevalence of depression at wave 1 was highest among individuals with complex/multisystem patterns. In the domain of executive function (P < 0.001) and letter cancelation (P = 0.005), mean cognitive function scores were lower in the complex/multisystem pattern.
Prospective associations between physical multimorbidity and cognitive function
Prospective associations between physical multimorbidity status/patterns at baseline and global or individual cognitive z-score are presented in Table 2, S3 and S4, and the changes in the global cognitive z-score at follow-up are shown in Figure 2.
Prospective associations between physical multimorbidity and composite global cognitive z score at follow-up
* Full-adjusted model: adjusted for age, gender, cohabitation status, quintiles of net non-pension wealth, education attainment level, smoking status, alcohol intake, four cognitive tests score at baseline and depressive symptom; β * time2 reported in full-adjusted plus interaction term model were the coefficients of the interaction terms between multimorbidity (status or patterns) and quadratic time (to test whether the association between multimorbidity and cognitive function differed across time-points).
Average composite global cognitive z score over time by physical multimorbidity status and patterns in fully adjusted models (reference: no multimorbidity status; relatively healthy pattern). Full-adjusted model: adjusted for age, gender, cohabitation status, quintiles of net non-pension wealth, education attainment level, smoking status, alcohol intake, four cognitive tests score at baseline and depressive symptoms.
Physical multimorbidity status
After multivariable adjustment, ≥three conditions group was significantly associated with an increased rate of decline in composite global cognitive z score, as compared to the no multimorbidity group (β ≥three conditions = −0.031 SD/year, 95%CI −0.063, −0.004; Table 2). The results of the linear trend test showed a clear dose–response association at follow-up (P for trend = 0.043) and the strength ranged from −0.014 for ≥three conditions group to −0.031 for two conditions group. For individual cognitive domains, ≥Three conditions group were related to an increased rate of decline in immediate recall, delayed recall and letter cancelation domains z score significantly, compared with the no multimorbidity group (Table S4).
Table 2 shows the results of multilevel analyses of physical multimorbidity status non-linear trajectory on cognitive function. The interaction effect between the status and the quadratic term of time indicated that the trajectories of cognitive decline differed among different multimorbidity statuses over time (P two conditions = 0.015, P ≥three conditions < 0.001). Figure 2 shows that ≥three conditions group showed a greater decrease in their cognitive function than the two conditions group, and the gaps widened as follow-up progressed.
Physical multimorbidity patterns
In the age- and sex-adjusted model, the complex/multisystem group (β = − 0.338 SD/year, 95%CI −0.424, −0.251) was most strongly related to the rate of decline in composite global cognitive function, followed by the hypertension/diabetes group (β = − 0.201 SD/year, 95%CI −0.255, −0.147). After adjusting all covariates, the hypertension/diabetes group was significantly related to an increased rate of decline in composite global cognitive z scores (−0.047 SD/year, 95% CI −0.081, −0.013). It is worth noting that the hypertension/diabetes group was related to the rate of decline in both immediate and delayed recall and letter cancelation domains z score (Table S4).
Similarly, there is evidence for non-linear trajectories of cognitive decline among physical multimorbidity patterns, and the magnitude of the association between complex/multisystem patterns and cognitive function increased the most as follow-up progressed (P complex/multisystem = 0.005). Throughout the follow-up, the complex/multisystem pattern had a lower composite global cognitive z score than the hypertension/diabetes group, followed by the respiratory/osteoporosis group (Figure 2).
Table S6 presents similar findings for the prospective associations between physical multimorbidity status/patterns and low cognitive performance.
Combination association analysis
Table S7 reveals the effects of the combinations of physical multimorbidity patterns/status and wealth status for composite global cognitive z score. Compared with the reference group, the combination of relatively healthy and low wealth status exhibited a significant association with cognitive function (β (95% CI): −0.073(−0.111, −0.035)). We observed non-linear trajectories of cognitive decline in the combinations of complex/multisystem pattern and high wealth status, respiratory/osteoporosis pattern and high wealth status and hypertension/diabetes pattern and high wealth status and cognitive function (P complex/multisystem = 0.004, P respiratory/osteoporosis < 0.001, P hypertension/diabetes = 0.009). In terms of individual cognitive domains, the combination of complex/multisystem pattern and low wealth status exhibited a stronger association with immediate recall, delayed recall and executive function domains z-score, while the combination of hypertension/diabetes pattern and low wealth status was more related to letter cancelation domain z-score (Table S5).
Figure 3 shows that, for the participants with low wealth status, the association with composite global cognitive z score appeared to be driven by the complex/multisystem pattern to a greater extent over time, as compared with respiratory / osteoporosis and hypertension/diabetes pattern. However, for the participants with high wealth status, the size of the association between complex / multisystem patterns and cognitive function increased most over time, followed by hypertension/diabetes pattern. In terms of physical multimorbidity status, more long-term conditions and low wealth status showed lower cognitive function scores over time.
Average composite global cognitive z score over time by the combinations between physical multimorbidity status/pattern and wealth status in fully-adjusted models (reference: no multimorbidity status / high wealth status; relatively healthy pattern / high wealth status). Full-adjusted model: adjusted for age, gender, cohabitation status, quintiles of net non-pension wealth, education attainment level, smoking status, alcohol intake, four cognitive tests score at baseline and depressive symptoms.
Stratified analyses
In the stratified analysis by age group (Table S8), the association between hypertension/diabetes patterns and cognitive function was significant (β (95% CI): −0.076(−0.133, −0.018)). Compared with the reference group, the non-linear influence of time on the association between two conditions group and respiratory/osteoporosis patterns and cognitive function were significant among those aged 50–59 years old and ≥ 70 years group, respectively. For participants with ≥three conditions and the hypertension/diabetes pattern, the association with cognitive function was significant and the coefficients were larger in females (Table S9). Moreover, we only found the non-linear of time on the association between physical multimorbidity status and cognitive function in females (Table S9).
Discussion
In this large and prospective cohort of older adults aged 50 years and over, we observed that physical multimorbidity was associated with a decline in cognitive function over a 17-year follow-up period, and the magnitude of the association increased as follow-up progressed. Physical multimorbidity status was related to cognitive function in a dose–response pattern, revealing that as the number of long-term conditions increased so did the likelihood of the risk of developing cognitive decline. We identified four patterns of physical multimorbidity in this study. Compared to the relatively healthy group, there was variation in the association between other patterns and cognitive function, which implies that certain patterns may be more likely to increase the likelihood of cognitive decline. In contrast, among those with high wealth status, participants in the complex / multisystems pattern had a faster rate of cognitive decline than those with the hypertension/diabetes pattern followed by the respiratory/osteoporosis pattern.
The observed associations between physical multimorbidity and future cognitive function decline in our study were consistent with the previous findings from animal models, clinical and epidemiological studies (Aarts et al., Reference Aarts, van den Akker, Tan, Verhey, Metsemakers and van Boxtel2011; Bunn et al., Reference Bunn, Burn, Goodman, Rait, Norton, Robinson, Schoeman and Brayne2014; Gerritsen et al., Reference Gerritsen, Bakker, Verhey, de Vugt, Melis and Koopmans2016; Vassilaki et al., Reference Vassilaki, Aakre, Cha, Kremers, St Sauver, Mielke, Knopman, Petersen and Roberts2015, Reference Vassilaki, Aakre, Mielke, Geda, Kremers, Alhurani, Machulda, Knopman, Petersen, Lowe, Jack and Roberts2016; Wang et al., Reference Wang, Wu, Tee and Lo2018). Our results also paralleled a previous study that reported a dose–response relationship between the number of chronic conditions and cognitive decline (Caracciolo et al., Reference Caracciolo, Gatz, Xu, Marengoni, Pedersen and Fratiglioni2013). Previous observational studies revealed that physical multimorbidity was associated with all stages in the progression of cognitive dysfunction, ranging from preclinical (Vassilaki et al., Reference Vassilaki, Aakre, Mielke, Geda, Kremers, Alhurani, Machulda, Knopman, Petersen, Lowe, Jack and Roberts2016), cognitive decline (Aarts et al., Reference Aarts, van den Akker, Tan, Verhey, Metsemakers and van Boxtel2011), mild cognitive impairment (Vassilaki et al., Reference Vassilaki, Aakre, Cha, Kremers, St Sauver, Mielke, Knopman, Petersen and Roberts2015) and diagnosed dementia (Wang et al., Reference Wang, Wu, Tee and Lo2018) to even early-onset Alzheimer’s disease (Gerritsen et al., Reference Gerritsen, Bakker, Verhey, de Vugt, Melis and Koopmans2016). The increase in the number of long-term conditions may precipitate the accumulation of more damage in organs and physiological systems with age, all of which challenges the body’s reserve capacity and resilience (Barnes, Reference Barnes2015; Yarnall et al., Reference Yarnall, Sayer, Clegg, Rockwood, Parker and Hindle2017). It has been hypothesized that physical multimorbidity, defined as chronic condition counts, may lead to cognitive decline through atherosclerosis, microvascular changes and inflammatory processes (Biessels et al., Reference Biessels, Staekenborg, Brunner, Brayne and Scheltens2006), which in turn increases the risk of mild cognitive impairment or dementia (Vassilaki et al., Reference Vassilaki, Aakre, Cha, Kremers, St Sauver, Mielke, Knopman, Petersen and Roberts2015).
Besides, certain clusters of physical multimorbidity may disrupt the process of cognitive decline in older adults. In the data of the UK Biobank, studies have reported that different multimorbidity patterns (e.g., cardiovascular/respiratory/metabolic/musculoskeletal/depressive disorders and oncologic/genitourinary/digestive disorders) may accelerate the process of developing Alzheimer’s disease (Hu et al., Reference Hu, Zhang, Aerqin, Ou, Wang, Cheng and Yu2022) by affecting plasma biomarkers of amyloid and neurodegeneration (Ren et al., Reference Ren, Li, Tian, Liu, Dong, Hou, Liu, Han, Han, Wang, Vetrano, Ngandu, Marengoni, Kivipelto, Wang, Cong, Du and Qiu2023). In this study, compared to the relatively healthy group, we found that cognitive function declined more rapidly and substantially over time in the complex/multisystem group, followed by the hypertension/diabetes group. The results suggested that there are possible mechanisms underlying to increase in the likelihood of cognitive decline. For example, hypertension and diabetes are two of the most common risk factors for cardiovascular disease that are often observed simultaneously (Cheung and Li, Reference Cheung and Li2012). Recent studies have shed light on the association between the brain and the cardiovascular system in which the brain and heart communicate with one another through the vasculature, known as the brain–heart axis (Riching et al., Reference Riching, Major, Londono and Bagchi2020). The regulators in the brain-heart axis, such as the angiotensin II converted by renin-angiotensin system, may lead to hypertension (Mehta and Griendling, Reference Mehta and Griendling2007) and insulin resistance and dyslipidemia thereby causing diabetes (Cheung and Li, Reference Cheung and Li2012). Moreover, the sedentarism, inflammation and frailty phenotype (Goodman et al., Reference Goodman, Banerjee, Rooks, McInerney, Sun, Getz, Kaur, Sun-Suslow, Junco and Levin2022) associated with hypertension and diabetes were likely factors in the development of cognitive impairment. However, evidence for the pathway of cognitive decline due to respiratory/osteoporosis multimorbidity pattern is limited. Osteoporosis is a common bone metabolism disease accompanied by respiratory disturbance in clinical practice (Ma et al., Reference Ma, Qiu and Zhou2022). Hypoxia (Yang et al., Reference Yang, Kimura-Ohba, Thompson, Salayandia, Cossé, Raz, Jalal and Rosenberg2018) and corticosteroid abuse (Savas et al., Reference Savas, Vinkers, Rosmalen, Hartman, Wester, van den Akker, Iyer, McEwen and van Rossum2020), the main causes of respiratory disease-induced osteoporosis, were also the risk factors for a decline in cognitive function. More researches are needed to explore the potential pathways.
Moreover, compared with the relatively healthy and high-wealth status group, our results showed that the poorer wealth status group had lower cognitive function and the magnitude of the association between the physical multimorbidity and subsequent cognitive function increased over time, which have not been reported elsewhere. Besides, the presence of ≥three conditions and high wealth status played a stronger driving role than the combination of two conditions and high wealth status. There was evidence of a positive correlation between chronic conditions and greater out-of-pocket health care expenditures (McRae et al., Reference McRae, Yen, Jeon, Herath and Essue2013; Sum et al., Reference Sum, Hone, Atun, Millett, Suhrcke, Mahal, Koh and Lee2018). Similarly, among those in lower wealth status, expenditures related to multimorbidity were higher (Bernardes et al., Reference Bernardes, Saulo, Fernandez, Lima-Costa and Andrade2020). A series of previous studies have shown that asset poverty or negative wealth shocks were associated with smaller total brain volume (Grasset et al., Reference Grasset, Glymour, Elfassy, Swift, Yaffe, Singh-Manoux, Zeki and Hazzouri2019), poorer microstructural integrity of the brain (Grasset et al., Reference Grasset, Glymour, Elfassy, Swift, Yaffe, Singh-Manoux, Zeki and Hazzouri2019), accelerated subsequent cognitive decline (Pan et al., Reference Pan, Gao, Zhu and Guo2023) and the higher risk of dementia (Nilaweera et al., Reference Nilaweera, Gurvich, Freak-Poli, Woods, Owen, Murray, Orchard, Britt, Wu, McNeil and Ryan2023). Furthermore, experimental evidence suggested that psychological stress associated with negative wealth shocks may increase activation of the hypothalamic–pituitary–adrenal axis, which may further lead to dysregulation of glucocorticoid levels and increase pathological cognitive impairment (de Souza-Talarico et al., Reference de Souza-Talarico, Marin, Sindi and Lupien2011) and other long-term conditions.
There are several strengths in our study. First, to the best of our knowledge, this study is the first large, longitudinal study in ELSA on the association between physical multimorbidity and subsequent cognitive function in middle-aged and older adults with multiple waves of follow-up. Repeated assessments of cognitive function every 2 years allowed us to investigate the ongoing association between baseline physical multimorbidity and cognitive function over the 17-year follow-up period. These suggested the need for early diagnosis and treatment of physical multimorbidity in the middle-aged and elderly population. Another strength of the study was that the physical multimorbidity pattern was defined using both a data-driven approach and quantitative approach, providing a more in-depth account of disease clustering with different health implications and facilitating the provision of targeted policy and measures. In addition, we highlighted the impact of different combinations of physical multimorbidity patterns/status and wealth status to further clarify the concern on the imperative to address physical multimorbidity that interacts with wealth status in complex ways to undermine cognitive function.
Several limitations need to be mentioned. A major limitation is the use of self-reported long-term conditions, which are of relatively lower accuracy and prone to bias. Another potential limitation is that the possibility of unmeasured confounders cannot be ruled out. The inclusion of confounders in this study was limited by the structure of the data, and future studies are expected to more comprehensive adjustments. Finally, participants in our study were only the middle-aged and older populations in the UK. This should be kept in mind when generalizing our findings to different study populations or cultural settings.
Conclusions
In the middle-aged and older populations, baseline physical multimorbidity, particularly complex and multisystem patterns, significantly accelerates the subsequent cognitive decline over time. Particular attention should be given to the impact of hypertension/diabetes multimorbidity patterns on cognitive function in high-wealth populations, and respiratory/osteoporosis multimorbidity patterns in low-wealth populations. The findings of our study may have significant implications for the implementation of highlighting the extent to which different physical disease clusters increase the likelihood of poor cognitive functioning health and the need for their early prevention, diagnosis and treatment through clinical and public health efforts. Future researches are expected to further elucidate the underlying pathophysiologic mechanisms.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2024.141.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/gmh.2024.141.
Data availability statement
The data are linked with the UK Data Archive and freely available through the UK data services and can be accessed here: https://discover.ukdataservice.ac.uk.
Acknowledgements
This study uses data from the English Longitudinal Study of Aging (ELSA). The authors appreciate the ELSA consortium for making the survey and making it available online freely, and all the participants for providing these data.
Author contribution
CC: Conceptualization, Methodology, Software, Data curation, Validation, Writing-Original draft preparation. NH: Methodology. SZ, MZ and JF: Writing- Reviewing and Editing. JG: Conceptualization, Methodology, Writing- Reviewing and Editing. All authors agreed on the final manuscript and the decision to submit it for publication.
Financial support
This study was supported by the Beijing Natural Science Foundation(L242145), National Key Research and Development Plan Project (2022YFC3600904), and the International Institute of Population Health, and Peking University Health Science Center (Number: JKGL202302). The founders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interest
The authors declare that they have no conflict of interest.
Ethics statement
The ELSA was approved by the London Multicentre Research Ethics Committee (MREC/01/2/91). Informed consent was obtained from all participants.
Open access
Comments
Dear Editor,
We are submitting the manuscript entitled “Effects of Physical Multimorbidity on Cognitive Decline Trajectories among Adults Aged 50 Years and Older with different wealth status: A 17-year Population-Based Cohort Study” for your consideration in Global Mental Health. This original article described has not been submitted elsewhere for publication, in whole or in part, and all authors have contributed to, read and approved the manuscript.The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this study.
Recently, the effect of the physical multimorbidity on cognitive function have attracted a lot of attention. However, relevant research focused on the impact of physical multimorbidity patterns on cognitive decline in different populations is limited and there are inconsistencies in findings. Existing studies have focused on the effect of only one type of physical multimorbidity pattern or of only the number of diseases on cognitive function without sufficient consideration of other demographic or socioeconomic characteristics. This study offered a pioneering and comprehensive examination of the effect of physical multimorbidity pattern / status on the trajectory of cognitive decline over 17 years in a large, nationally representative cohort. Additionally, it investigated the influence of wealth status, age groups and gender.
The key finding from this study are that baseline physical multimorbidity, particularly complex and multisystem patterns, significantly accelerates the subsequent cognitive decline over time in the middle-aged and older populations. Particular attention should be given to the impact of hypertension/diabetes multimorbidity pattern on cognitive function in high-wealth populations, and respiratory/osteoporosis multimorbidity patterns in low-wealth populations. The findings of our study may have significant implications for the implementation of integrated cognitive and physical healthcare interventions for populations, highlighting the extent to which different physical disease patterns increase the likelihood of poor cognitive function health and the need for their early prevention, diagnosis and treatment through clinical and public health efforts. In addition, the results are helpful to facilitate the development of targeted preventive interventions and treatments for people with physical multimorbidity.
We deeply appreciate your consideration of our manuscript, and we look forward to receiving comments from the reviewers.
Yours sincerely,
Jing Guo, Assistant Professor, Department of Health Policy and Management, School of Public Health, Peking University, Beijing, 100191, P. R. China. Tel: 86-18086471505, Email: jing624218@163.com (J Guo)